UNIVERSITY OF KWAZULU-NATAL

SCHOOL OF ENGINEERING

The study of process health and safety management deficiencies relative

to hazardous chemical exposure

James Xolani Nyawera

953013647

A thesis submitted in fulfilment of the degree of

Doctor of Philosophy in Engineering at the College of Agriculture,
Engineering and Science, University of KwaZulu-Natal.

November 2019
Supervisor: Professor Theodore C Haupt
Declaration by Supervisor

As the candidate’s Supervisor I agree/do not agree to the submission of this thesis.

Signed: _______________________ Date: ______________________

Professor Theodore C Haupt

ii
Declaration 1 - Plagiarism

I, James Xolani Nyawera, declare that

1. The research reported in this thesis, except where otherwise indicated, is my original
research.

2. This thesis has not been submitted for any degree or examination at any other university.

3. This thesis does not contain other persons’ data, pictures, graphs or other information,
unless specifically acknowledged as being sourced from other persons.

4. This thesis does not contain other persons' writing, unless specifically acknowledged as
being sourced from other researchers. Where other written sources have been quoted,
then:

a. Their words have been re-written but the general information attributed to them has
been referenced.
b. Where their exact words have been used, then their writing has been placed in italics
and inside quotation marks, and referenced.

5. This thesis does not contain text, graphics or tables copied and pasted from the Internet,
unless specifically acknowledged, and the source being detailed in the thesis and in the
References sections.

Signed: Date:
James Xolani Nyawera

iii
Declaration 2 - Publications

DETAILS OF CONTRIBUTION TO PUBLICATIONS that form part and/or include research

presented in this thesis (include publications in preparation, submitted, in press and published
and give details of the contributions of each author to the experimental work and writing of
each publication).

Publication 1
Mr J.X. Nyawera and Prof T.C. Haupt, Process Health and Safety Management Elements: An
initial review of literature - The 12th Built Environment Conference, Durban. August 6-7,
2018.
Contribution: J.X Nyawera - Research and Writing; Prof. TC. Haupt - Supervision.

Publication 2
Mr J.X. Nyawera and Prof T.C. Haupt, Process Health and Safety Management Deficiencies
- The 13th Built Environment Conference, Durban. September 2-3, 2019.
Contribution: J.X Nyawera - Research and Writing; Prof. TC. Haupt - Supervision.

Signed: ____________________ Date:

James Xolani Nyawera

iv
 Dedication

This thesis is dedicated to my late Son, Bayanda Nyawera “Mr President”. This
thesis is in Honour of you, with all my love.

v
 Acknowledgements

Firstly, I would like to express my deep sense of gratitude to my supervisor, Professor Theo
C. Haupt, for providing me his treasurable advices, recommendations and guidance to
complete this work within the planned time. His encouragement and support was very valuable
throughout the course of this research and during the publication of papers in conference
proceedings.

THANK YOU to all the participants in this research, may GOD bless you and your families.

I appreciate the YouTube videos on SPSS and AMOS by James Gaskin, THANK YOU.

I am grateful to my wife, Dr Bongiwe Pepu, for her support, understanding and allowing me
to work in extra hours during the study period. In addition, I am thankful to my Son Mfundo
Nyawera, my Daughter Sandiswa Nyawera, my Daughter Khwezilokusa Nyawera and my
Son Nkazimulo Nyawera for providing me invaluable pleasures and amusements. You are
AWESOME and I LOVE YOU.

Finally, yet importantly, I thank the Almighty GOD for granting me the power, courage and
wisdom to finish this study.

vi
 List of Abbreviation

ANOVA Analysis of Variance

AGFI Adjusted Goodness-of-Fit Index

AVE Average Variance Extracted

CB-SEM Covariant Based Structural Equation Modelling

CFA Confirmatory Factor Analysis

CFI Comparative Fit Index

COPD Chronic Obstructive Pulmonary Disease

CR Composite Reliability

DoL Department of Labour

EFA Exploratory Factor Analysis

FMEA Failure Mode and Effect Analysis

FRMS Fatigue Risk Management System

GFI Goodness-of-Fit Index

HAZID Hazard Identification

HAZOP Hazard and Operability

IFI Incremental Fit Index

ILO International Labour Organisation

JSA Job Safety Analysis

LC50 Lethal Concentration 50

LD50 Lethal Dose 50

vii
LOPA Layer of Protection Analysis

LPG Liquefied Petroleum Gas

NFI Normed Fit Index

OSHAct Occupational Health and Safety Act 85 of 1993

PCA Principal Component Analysis

PCFI Parsimony Adjusted Comparative Fit Index

PLS-SEM Partial Least Squares Structural Equation Modelling

PNFI Parsimony Adjusted Normed Fit Index

PVC Polyvinyl Chloride

RFI Relative Fit Index

RMSEA Random Measures of Sample Error Approximation

SEM Structural Equation Model

SIL Safety Integrity Level

SPSS Statistical Package for the Social Sciences

Std. Dev Standard Deviation

S.E Standard Error

STEL Short Term Exposure Limit

TLI Tucker Lewis Index

TLV Threshold Limit Value

USA United States of America

WHO World Health Organisation

VCM Vinyl Chloride Monomer

viii
 ABSTRACT

The main objectives of this study were to identify the main process health and safety
management deficiencies that require senior management’s attention. To identify the critical
drivers that could be used to improve health and safety to reach generative process health and
safety culture level five and to develop a model of effectively managing hazardous chemical
substance exposure in the petrochemical industry.

Ethical clearance to conduct the study was obtained from the University of KwaZulu - Natal
Humanities and Social Sciences Research Ethics Committee (HSS/1094/018D). The targeted
population was 800 employees in one major petrochemical enterprise in the KwaZulu-Natal
province of South Africa. The study was conducted by distributing 400 questionnaires
manually to the randomly selected potential participants of which 259 were returned duly
completed and used. The returned questionnaires were statistically analysed using descriptive
statistics in SPSS version 25.

The research was planned to first explore the concepts by qualitative research methods, such
as in depth literature review. The quantitative data collection and analysis is based on a
quantitative research method involving questionnaire survey and statistical data analysis
methods. The validation of the findings and related conclusions rely on the results from both
qualitative and quantitative research methods. The mixed method was considered the best
option for this study as it assisted to leverage the advantages of both quantitative and
qualitative research methods in achieving the research objectives.

The key process health and safety management deficiencies to be prioritized are, namely, poor
engineering design integrity, poor controls when working with suspended loads, poor controls
when working at heights, poor housekeeping, poor controls of source of ignition, verifying
energy isolation before starting to work on equipment, poor health and safety risk assessments,
handling of hazardous chemicals, human error and fatigue management. The key process
health and safety drivers to be prioritized for generative process health and safety culture are,
leadership commitment, chemical exposure management, health and safety risk assessment,
process hazard analysis and permit to work. The developed generative process health and
safety culture model was subjected to rigorous measurement analysis using structural equation
modelling, namely, principal component analysis, goodness-of-fit measure, assessment of
normality, discriminant validity, multicollinearity, model adequacy, reliability and validity.

ix
This study will assist senior management with a framework to reduce process health and safety
incidents in the petrochemical industry and improve health and safety towards generative
culture where organisations say, “Health and safety is the way we do business”.

Key words: Process Health and Safety Management Deficiencies.

x
 TABLE OF CONTENTS

Page
DECLARATION 1 – PLAGIARISM………………………………………………………………. .. iii
DECLARATION 2 – PUBLICATIONS……………………………………………………………… iv
DEDICATION………………………………………………………………………………………… v
ACKNOWLEDGEMENTS………………………………………………………………………....... vi
LIST OF ABBREVIATIONS.……………..………………………………………………………… vii
ABSTRACT………………………………..…………………………………………………………. ix
TABLE OF CONTENTS..……………………………………………………………………………. xi
LIST OF FIGURES………………..………………………………………………………………… xiii
LIST OF TABLES……………………………………………………………………………...…… xiv

CHAPTER ONE: INTRODUCTION ........................................................................................ 1

1.1 Introduction ........................................................................................................................... 1
1.2 Petrochemical Industry Hazards ............................................................................................. 2
1.3 Health and Safety Culture ...................................................................................................... 5
1.4 Management Systems ............................................................................................................. 6
1.5 Employers Accountability ...................................................................................................... 7
1.6 Problem Statement Discussion ............................................................................................... 8
1.7 Research Questions .............................................................................................................. 14
1.8 Research Objectives ............................................................................................................ 14
1.9 Methodological Approach .................................................................................................... 15
1.9.1 Literature Review .......................................................................................................... 15
1.9.2 Conceptual Model .......................................................................................................... 15
1.9.3 Questionnaire................................................................................................................. 16
1.9.4 Structural Equation Modelling ...................................................................................... 16
1.9.5 Reliability and Validity ................................................................................................. 16
1.9.6 Ethical Consideration .................................................................................................... 17
1.10 Study Limitations and Delimitations .................................................................................... 18
1.11 Overview of Research Study ................................................................................................ 18
1.12 Contribution of the Study Findings to Knowledge .............................................................. 21
1.13 Chapter Summary ................................................................................................................. 22
CHAPTER TWO: LITERATURE REVIEW......................................................................... 23
2.1 Introduction .......................................................................................................................... 23
2.2 Background to the Study ...................................................................................................... 23
2.2.1 Physical Hazards ........................................................................................................... 26
2.2.2 Ergonomic Hazards ....................................................................................................... 28
2.2.3 Chemical Hazards .......................................................................................................... 29
2.2.4 Psychosocial Hazards .................................................................................................... 33
2.3 Leadership Elements ............................................................................................................ 35

xi
 2.3.1 Leadership Commitment ............................................................................................... 35
2.3.2 Training and Competence .............................................................................................. 36
2.3.3 Contractor Management ................................................................................................ 37
2.3.4 Asset Integrity ............................................................................................................... 39
2.3.5 Effective Communication .............................................................................................. 40
2.4 Health and Safety Elements.................................................................................................. 41
2.4.1 Chemical Exposure Management .................................................................................. 42
2.4.2 Health and Safety Risk Assessment .............................................................................. 43
2.4.3 Incident Investigation .................................................................................................... 46
2.4.4 Emergency Preparedness and Readiness ....................................................................... 48
2.4.5 Audit Compliance .......................................................................................................... 49
2.4.6 Personal Protective Equipment ...................................................................................... 49
2.4.7 Fitness to Work.............................................................................................................. 51
2.4.8 Fatigue Management ..................................................................................................... 52
2.4.9 Housekeeping in Workplace .......................................................................................... 53
2.4.10 Noise Exposure Management ........................................................................................ 54
2.5 Technical Elements .............................................................................................................. 54
2.5.1 Management of Change ................................................................................................. 55
2.5.2 Process Hazard Analysis ............................................................................................... 57
2.5.3 Process Health and Safety Information ......................................................................... 58
2.5.4 Design Integrity ............................................................................................................. 59
2.5.5 Human Factor ................................................................................................................ 60
2.6 Operational Elements ........................................................................................................... 60
2.6.1 Pre-Startup and Shutdown Reviews .............................................................................. 62
2.6.2 Operating Procedure ...................................................................................................... 63
2.6.3 Control of Ignition Source ............................................................................................. 65
2.6.4 Control of Confined Space Entry .................................................................................. 65
2.6.5 Permit to Work .............................................................................................................. 67
2.7 Health and Safety Culture .................................................................................................... 68
2.8 Development of Conceptual Model ...................................................................................... 71
2.9 Conceptual Model ................................................................................................................ 72
2.10 Chapter Summary ................................................................................................................. 79
CHAPTER THREE: RESEARCH DESIGN AND METHODOLOGY ............................. 80
3.1 Introduction .......................................................................................................................... 80
3.2 Research Paradigm ............................................................................................................... 80
3.2.1 Ontology ........................................................................................................................ 81
3.2.2 Epistemology ................................................................................................................. 81
3.3 Research Philosophy ............................................................................................................ 82
3.3.1 Positivism ...................................................................................................................... 83
3.3.2 Interpretivism ................................................................................................................ 84
3.4 Research Approach ............................................................................................................... 85
3.4.1 Deductive Approach ...................................................................................................... 85
3.4.2 Inductive Approach ....................................................................................................... 86
3.5 Research Methodology ......................................................................................................... 86

xii
 3.5.1 Quantitative Research .................................................................................................... 87
3.5.2 Qualitative Research ...................................................................................................... 88
3.5.3 Mixed Approach ............................................................................................................ 89
3.6 Research Methods ................................................................................................................ 89
3.7 Research Strategies ............................................................................................................... 90
3.7.1 Case Study ..................................................................................................................... 90
3.7.2 Ethnographic Study ....................................................................................................... 91
3.7.3 Grounded Theory........................................................................................................... 92
3.7.4 Phenomenological Study ............................................................................................... 92
3.8 Research Instrument ............................................................................................................. 93
3.9 Questionnaire Design ........................................................................................................... 95
3.9.1 Scales and Measurement ............................................................................................... 97
3.9.2 Questinnaire Administration.......................................................................................... 99
3.10 Sampling ............................................................................................................................. 100
3.10.1 Probability Sampling Methods ............................................................................ 100
3.10.2 Non-Probability Sampling Methods .................................................................... 102
3.11 Data Collection ................................................................................................................... 104
3.12 Data Analysis...................................................................................................................... 106
3.13 Data Screening.................................................................................................................... 107
3.13.1 Sample Size ................................................................................................................ 107
3.13.2 Missing Data .............................................................................................................. 109
3.13.3 Univariate and Multivariate Normality ...................................................................... 110
3.13.4 Outliers ....................................................................................................................... 111
3.13.5 Item Parcelling ........................................................................................................... 112
3.14 Reliability and Validity ...................................................................................................... 114
3.15 Regression Analysis ........................................................................................................... 116
3.16 Principal Component Analysis ........................................................................................... 117
3.17 Structural Equation Modelling ........................................................................................... 119
3.17.1 Exploratory Factor Analysis ....................................................................................... 121
3.17.2 Confirmatory Factor Analysis .................................................................................... 121
3.18 Path Modelling Process ...................................................................................................... 122
3.19 Chapter Summary ............................................................................................................... 125
CHAPTER FOUR: PRESENTATION OF RESULTS ....................................................... 125
4.1 Introduction ........................................................................................................................ 126
4.2 Questionnaire Administration ............................................................................................ 126
4.3 Questionnaire Responses .................................................................................................... 126
4.4 Demographic Information of Participants .......................................................................... 127
4.5 Quantitative Data Analysis ................................................................................................. 128
4.5.1 Leadership Commitment ............................................................................................. 128
4.5.2 Chemical Exposure Management ................................................................................ 129
4.5.3 Health and Safety Risk Assessment ............................................................................ 129
4.5.4 Process Hazard Analysis ............................................................................................. 130
4.5.5 Permit to Work ............................................................................................................ 131
4.5.6 Training and Competence ............................................................................................ 131

xiii
 4.5.7 Process Health and Safety Information ....................................................................... 132
4.5.8 Control of Confined Space Entry ................................................................................ 132
4.5.9 Operating Procedure .................................................................................................... 132
4.5.10 Control of Ignition Source ........................................................................................... 133
4.6 Discussion........................................................................................................................... 133
4.7 Chapter Summary ............................................................................................................... 134
CHAPTER FIVE: MODEL DEVELOPMENT ................................................................... 134
5.1 Introduction ........................................................................................................................ 135
5.2 Model Development using SEM ........................................................................................ 135
5.2.1 Structural Equation Modelling .................................................................................... 136
5.2.2 SEM Analysis Selection .............................................................................................. 137
5.2.3 Reliability Checks ....................................................................................................... 138
5.2.4 Principal Component Analysis .................................................................................... 139
5.2.5 Consistency Test Results ............................................................................................. 141
5.2.6 Structural Equation Modelling Specification .............................................................. 143
5.2.7 Confirmatory Factor Analysis ..................................................................................... 143
5.2.8 Model Modification ..................................................................................................... 143
5.3 Model Fit ............................................................................................................................ 145
5.3.1 Goodness of Fit (GOF) ................................................................................................ 145
5.3.2 Leadership Commitment Goodness of Fit ................................................................... 148
5.3.3 Chemical Exposure Management Goodness of Fit ..................................................... 149
5.3.4 Health and Safety Risk Assessment Goodness of Fit .................................................. 150
5.3.5 Process Hazard Analysis Goodness of Fit ................................................................... 151
5.3.6 Permit to Work Goodness of Fit .................................................................................. 152
5.3.7 Generative Health and Safety Culture Model Goodness of Fit ................................... 153
5.4 Model Refinement .............................................................................................................. 154
5.5 Chapter Summary ............................................................................................................... 158
CHAPTER SIX: MODEL VALIDATION ............................................................................ 159
6.1 Introduction ........................................................................................................................ 159
6.2 Normality Assessment ........................................................................................................ 160
6.3 Multicollinearity ................................................................................................................. 161
6.4 Reliability and Validity ...................................................................................................... 165
6.4.1 Composite Reliability .................................................................................................. 165
6.4.2 Discriminant Validity .................................................................................................. 166
6.5 Model Parameter and Hypothesis Testing .......................................................................... 168
6.6 Model Acceptance .............................................................................................................. 170
6.7 Expert Validation................................................................................................................ 171
6.7.1 Leadership Commitment ............................................................................................. 171
6.7.2 Chemical Exposure Management ................................................................................ 172
6.7.3 Health and Safety Risk Assessment ............................................................................ 172
6.7.4 Process Hazard Analysis ............................................................................................. 173
6.7.5 Permit to Work ............................................................................................................ 173
6.8 Chapter Summary ............................................................................................................... 173

xiv
CHAPTER SEVEN: DISCUSSION ....................................................................................... 175
7.1 Introduction ........................................................................................................................ 175
7.2 Discussion of Findings ....................................................................................................... 175
7.2.1 Findings related to Leadership Commitment .............................................................. 177
7.2.2 Findings related to Chemical Exposure Management ................................................. 178
7.2.3 Findings related to Health and Safety Risk Assessment ............................................. 179
7.2.4 Findings related to Process Hazard Analysis............................................................... 180
7.2.5 Findings related to Permit to Work ............................................................................. 181
7.3 Research Questions ............................................................................................................ 182
7.3.1 Deficiencies relative to Hazardous Chemical Exposure.............................................. 182
7.3.2 Priority Deficiencies .................................................................................................... 183
7.3.3 Critical Drivers towards Generative Health and Safety Culture.................................. 184
7.3.4 Leadership Commitment Enablers .............................................................................. 184
7.3.5 Chemical Exposure Management Enablers ................................................................. 185
7.3.6 Health and Safety Risk Assessment Enablers ............................................................. 185
7.3.7 Process Hazard Analysis Enablers .............................................................................. 187
7.3.5 Permit to Work Enablers ............................................................................................. 187
7.4 Chapter Summary ............................................................................................................... 187
CHAPTER EIGHT: RECOMMENDATIONS AND CONCLUSION.............................. 189
8.1 Introduction ........................................................................................................................ 189
8.2 Key Research Findings and Conclusion ............................................................................. 189
8.3 Proposed Generative Process Health and Safety Culture Model ........................................ 192
8.4 Contribution of the Research .............................................................................................. 194
8.4.1 Theoretical Contribution.............................................................................................. 194
8.4.2 Methodological Contribution ...................................................................................... 194
8.4.3 Practical Contribution .................................................................................................. 194
8.5 Limitations of the Research ................................................................................................ 195
8.6 Suggestions for Future Research ........................................................................................ 196
8.7 Conclusion .......................................................................................................................... 196
REFERENCES .......................................................................................................................... 198
APPENDICES ................................................................................................................................ i
Appendix 1: Ethical Clearance Approval ........................................................................................ i
Appendix 2: Research Instrument................................................................................................... ii
Appendix 3: Publication 1 ............................................................................................................. vi
Appendix 4: Publication 2 ............................................................................................................ xv

xv
 LIST OF FIGURES

Page
Figure 1.1 Blowdown Drum and Isomeration Unit after Incident .................................................. 9
Figure 1.2 Bhopal Tragedy (Isocyanate Gas Leak) ...................................................................... 12
Figure 1.3 Phillips Petroleum Explosion ...................................................................................... 12
Figure 1.4 Amuay Oil Refinery Explosion ................................................................................... 13
Figure 2.1 Reason’s Swiss Cheese Model .................................................................................... 25
Figure 2.2 Impact of Chemical Disasters ..................................................................................... 43
Figure 2.3 Hierarchy of Hazard Controls ..................................................................................... 50
Figure 2.4 ARCO Channelview Compressor Explosion .............................................................. 55
Figure 2.5 Fertilizer Plant in Port Neal Explosion........................................................................ 61
Figure 2.6 Health and Safety Culture Ladder ............................................................................... 71
Figure 2.7 The Proposed Conceptual Model ................................................................................ 73
Figure 5.1 Initial Measurement Model ....................................................................................... 156
Figure 5.2 Refined Measurement Model .................................................................................... 157
Figure 6.1 Hypothesis Testing Model ........................................................................................ 168
Figure 9.1 Generative Process Health and Safety Culture Model (JX Nyawera Model) ........... 192

xvi
 LIST OF TABLES

Page
Table 4.1 Age and Years of Service .......................................................................................... 127
Table 4.2 Gender, Marital Status and Department .................................................................... 127
Table 4.3 Leadership Commitment Construct........................................................................... 128
Table 4.4 Chemical Exposure Management Construct ............................................................. 129
Table 4.5 Health and Safety Risk Assessment Construct.......................................................... 129
Table 4.6 Process Hazard Analysis Construct ........................................................................... 130
Table 4.7 Permit to Work Construct ......................................................................................... 131
Table 4.8 Training and Competence Construct ......................................................................... 131
Table 4.9 Process Health and Safety Information Construct .................................................... 132
Table 4.10 Control of Confined Space Entry Construct.............................................................. 132
Table 4.11 Operating Procedure Construct ................................................................................. 132
Table 4.12 Control of Ignition Source Construct ........................................................................ 133
Table 4.13 Main Deficiencies ..................................................................................................... 133
Table 5.1 Principal Component Analysis with all Loading ...................................................... 139
Table 5.2 Reliability Analysis before Principal Component Analysis for Constructs .............. 140
Table 5.3 Leadership Commitment ........................................................................................... 141
Table 5.4 Chemical Exposure Management .............................................................................. 141
Table 5.5 Health and Safety Risk Assessment .......................................................................... 142
Table 5.6 Process Hazard Analysis ........................................................................................... 142
Table 5.7 Permit to Work .......................................................................................................... 142
Table 5.8 Final Principal Component Analysis......................................................................... 144
Table 5.9 Reliability Analysis After Principal Component Analysis for Construct.................. 145
Table 5.10 Threshold Limits for Model Fit Indices .................................................................... 147
Table 5.11 Leadership Commitment Construct Goodness of Fit ................................................ 148
Table 5.12 Chemical Exposure Management Construct Goodness of Fit ................................... 149
Table 5.13 Health and Safety Risk Assessment Construct Goodness of Fit ............................... 150
Table 5.14 Process Hazard Analysis Construct Goodness of Fit ................................................ 151
Table 5.15 Permit to Work Construct Goodness of Fit ............................................................... 152
Table 5.16 Generative Process Health and Safety Culture Model .............................................. 153
Table 6.1 Assessment of Normality .......................................................................................... 160
Table 6.2 Correlation Matrix of Leadership Commitment Latent Variable .............................. 162
Table 6.3 Correlation Matrix of Chemical Exposure Management Latent Variable ................ 163
Table 6.4 Correlation Matrix of Health and Safety Risk Assessment Latent Variable ............. 163

xvii
Table 6.5 Correlation Matrix of Process Hazard Analysis Latent Variable .............................. 164
Table 6.6 Correlation Matrix of Permit to Work Construct Latent Variable ............................ 164
Table 6.7 Correlation Matrix of Latent Variables ..................................................................... 164
Table 6.8 Composite Reliability Index...................................................................................... 165
Table 6.9 Average Variance Extracted Value ........................................................................... 166
Table 6.10 Reliability and Validity ............................................................................................. 167
Table 6.11 Hypothesis Relationship Testing in the Research Model .......................................... 169
Table 6.12 Variances Estimated by the Model............................................................................ 170
Table 6.13 Parameter Estimated and Standard Errors ................................................................. 171
Table 7.1 Leadership Commitment Hypotheses Ranking ......................................................... 177
Table 7.2 Chemical Exposure Management Hypotheses Ranking .......................................... 178
Table 7.3 Health and Safety Risk Assessment Hypotheses Ranking ........................................ 179
Table 7.4 Process Hazard Analysis Hypothesis Ranking .......................................................... 180
Table 7.5 Permit to Work Hypotheses Ranking ........................................................................ 181
Table 7.6 Means and Standard Deviations of Deficiencies ...................................................... 183
Table 8.1 Deficiencies Ranking ................................................................................................ 190
Table 8.2 Generative Process Health and Safety Culture Toolkit (JX Nyawera Toolkit) ......... 191

xviii
 CHAPTER ONE
Introduction
_____________________________________________________________________
1.1 Introduction

The world’s oil and gas resources provide some of the richest, solidest, and valuable energy sources
available (Kim, 2016). The petrochemical industry as the name implies is based upon the production of
chemicals from petroleum such as natural gas and crude oil (Hughes & Ferret, 2007). According to
Eyayo (2014) the importance of occupational health and safety is often overlooked and people tend to
equate occupational illness with industrialisation and huge factories in urban areas. World Health
Organisation’s programme on workers’ health is concerned with the control of occupational health and
safety risks, the protection and promotion of the working populations and humanization of work.

The occupational health and safety plays a central role in industry as it protects all workers from health
and safety related issues in their working environment (Hughes & Ferret, 2007). The workers in the oil
and gas industry are exposed to a lot of hazards such as, physical hazards, chemical hazards, ergonomic
hazards, psychosocial hazards and radiological hazards (Kim, 2016). According to Vitharana et al.
(2015) health hazards are properties of a chemical that have the potential to cause adverse health effects
and exposure usually occurs through inhalation, skin contact or ingestion.

It is vital to protect workers from injuries on a social level, but there is also a positive economic impact
in reducing health and safety hazards (Hughes & Ferret, 2007). According to Eyayo (2014) globally,
there are 2.9 billion workers who are exposed to hazardous risks at their work place. Annually there are
two million deaths that are attributable to occupational diseases and injuries while 4% of gross domestic
product is lost due to occupational diseases and injuries. Health and safety is without doubt, the most
crucial investment and the question is not what it costs, but what it saves (Hughes & Ferret, 2007).

Chronic unease starts with openness, where bad news is welcomed and incidents are treated as an
opportunity to learn (Flin, 2015). Measures and strategies designed to prevent, control, reduce or
eliminate occupational hazards and risks have been developed and applied continuously over the years
to keep pace with technological and economic changes (Eyayo, 2014). Material substances, processes
or circumstances which pose threat to health and well-being of workers in any occupation are termed
as occupational hazards (Kulkarni, 2017).

A hazard is a potential source of harm or an adverse health effect on a person or persons. “Hazard” and
“Risk” are often used interchangeably (Vitharana et al., 2015). Chronic unease is a pre-occupation with

1
failure, it is about resetting tolerance to risk and understanding that small failures are signs that
something needs to be corrected (Flin, 2015). According to Elssayed et al. (2012) management actions
related to health and safety should be adequate and be prioritised to improve health and safety quality
efficiently. HAZOP study provide a safe system of work during inspection and maintenance operation
especially in boilers (Karthika, 2013).

Developing and implementing safe work practices provide for the control of hazards during operations
such as lockout/tagout; confined space entry; opening process equipment or piping; and control over
entrance into a facility by maintenance, contractor, laboratory, or other support personnel. These safe
work practices shall apply to employees as well as contractor employees (Hardy, 2013). According to
Stojkovic’ (2013) entry into hazardous confined spaces, especially the performance in such area poses
a major threat to the health and lives of people so in many technically developing countries operating
in such areas is subject to previous approval of the competent authority.

It is vital that potential ignition sources are identified, and there is feedback from operational experience
back into hazard assessment process to identify changes and deviations from original expectations
(Puttick, 2008). Avoidance of ignition sources can then appear to be an attractive option, but it has
limitations and can be a useful and reliable basis of safety in certain circumstances provided that it is
restricted to the inside of chemical plants, and certain well defined charging and discharging areas
(Puttick, 2008).

1.2 Petrochemical Industry Hazards

Petrochemical industries play a crucial role in various manufacturing sectors. However, potential
hazards associated with these industries have raised increased concern for societies (Sharma et al.,
2017). According to a six-year fatal occupational injuries census conducted by the US Bureau of Labour
Statistics, workers in the oil and gas industry from the Gulf countries could be up to seven times more
likely to be fatally injured than workers in other industry sectors. Petrochemical industry release large
quantities of toxic and deleterious substances as effluents into the atmosphere and generates solid waste
that is difficult both to treat and to dispose of (Sharma et al., 2017).

Typical acute health effects include headaches, nausea or vomiting and skin corrosion, while chronic
health effects include asthma, dermatitis, nerve damage or cancer. A person conducting business has
the primary duty to ensure, as far as is reasonably practicable, that the health and safety of workers and
other persons is not put at risk from work carried out (Vitharana et al., 2015). It is essential to develop

2
control and preventive measures which are to be taken at the planning stages in these industries (Sharma
et al., 2017). The employers should ensure the safe use, handling and storage of hazardous substances
(Vitharana et al., 2015).

According to Hughes et al. (2007) cited in Dabup (2012) the general duties of employers as stipulated
in the British health and safety work Act, include the following;

 Ensuring the health, safety and welfare of employees of all categories in the workplace
including, service providers and temporary employees.
 Ensuring the health, safety and welfare of all visitors to the workplace.
 Ensuring the health, safety and welfare of all persons permitted to use the organisation’s
equipment.
 Ensuring the health, safety and welfare of those affected directly or indirectly by the work
activity, such as host communities and the general public.
 Ensure the provision of a safe environment for workers, the general public and communities
where projects are executed.

Dangerous chemicals are often used and handled in workplaces. The risk of injury or ill health upon
exposure to the hazards of the chemicals at work depends on whether there are adequate safety measures
in place (International Labour Organisation, 2017). The emissions of harmful substances from the
petrochemical industries has reduced significantly in last few years because of using environmental
developments along with an increased awareness about the health and safety aspects of plant operations
(Sharma et al., 2017). According to Vitharana et al. (2015) adverse health effects can be acute (short
term) or chronic (long term). According to OSHAcademy (2017) unexpected releases of toxic, reactive,
or flammable liquids and gases in process involving highly hazardous chemicals have been reported for
many years.

According to Almanssoor (2008) toxicology data are available for most chemicals and the most
commonly used in the industry are LC50, LD50, TLV and STEL. Their definitions are, namely;

 LC50 median lethal concentration 50: calculated concentration of a chemical in air exposure,
which can cause the death of 50% of experimental animals in a specified period of time.
 LD50 median lethal dose 50; calculated dose of a chemical that is expected to cause the death
of 50% of experimental animals when administered by any route other than inhalation.
 TLV threshold limit value: concentration of a substance in the air to which workers can be
exposed without adverse effect.

3
  STEL short term exposure limit: is the maximum permissible concentration of a material,
generally expressed in ppm in air, for a defined short period of time (typically 5 or 15 minutes,
depending upon the country). This concentration is generally a time-weighted average over the
period of exposure.

According to the Safety Association for Canada upstream Oil and Gas Industry (2012) the following
questions should be used to identify potential harmful worker chemical exposure hazards, namely;

 Have all potentially hazardous products that may be used, handled, or stored at the work site
been identified?
 Are all material safety data sheets current, available, and readily accessible to all on site
workers?
 What effect will the work operations have on the chemicals used, handled, or stored?
 How may a worker be potentially exposed to chemical hazards?
 Are written work procedures for the operation available to minimise worker exposure to
chemicals?
 Have the workers involved in the operation been given the appropriate level of training in the
work procedures and controls in place?
 Is all equipment readily available and in good operating conditions?
 Have all workers been properly trained in the selection and correct care, use and maintenance
of the PPE required?
 Are all workers sufficiently trained in the hazards to which they may be exposed and how to
recognize them?

According to Eyayo (2014) health and safety professionals, working with process, chemical,
instrumentation, and metallurgical engineers assure that potential physical, mechanical and chemical.
Health hazards are recognised and provisions are made for safe operating practices and appropriate
protection measures. One of the major needs with regard to the construction industry is to enhance
professional’s interest in active safety management and implementation of awareness programs. These
must be developed and implemented among construction workers (International Labour Organisation,
2017).

4
1.3 Health and Safety Culture

According to Hardy (2013) the most important indicator of a positive health and safety culture is the
extent to which employees are actively involved in health and safety on a daily basis. An organisation
that has a good health and safety culture consist of a strong senior management commitment, leadership
and involvement in health and safety. Better communications between all organisational levels. Greater
hazard control, good induction and follow up on health and safety training as well as ongoing health
and safety schemes reinforcing the importance of health and safety, including near miss reporting
(Bond, 2007).

If there is very little involvement of senior management with health and safety and a solely depend upon
line management and safety representatives to be involved, the organisation fails to win people over to
the health and safety effort. Therefore the organisation will not have a very good health and safety
culture (Hardy, 2013). Health and Safety culture combines the whole approach of reporting health and
safety matters, monitoring equipment and procedures into a health and safety management system
which becomes a coherent structure that becomes acceptable and comprehensible to the employees and
hence to the public (Bond, 2007).

Occupational accidents can be reduced through effective preventative measures by investing on health
and safety equipment, training and educating the employees, process design and machinery and in order
to develop a good health and safety culture, attitude of the workers needs to be reoriented by applying
best practices, good housekeeping, change in work culture, and work practices (Beriha et al., 2012).

To achieve a total health and safety culture, where employees feel as strong a sense of responsibility
for the health and safety of their co-workers as they do for themselves, it is necessary to increase the
openness and frequency of health and safety communication (Holstvoogd et al., 2006). Potential
hazards are identified based on the knowledge of operations and past experience with similar work tasks
and this usually involves brainstorming – type sessions among team members having familiarity with
operational activities (Campbell, 2008).

According to McKenzie (2007) organisations with positive health and safety culture learn from previous
incidents and safety deficiencies and encourage reporting of health and safety concerns, issues, and
problems by all levels of staff and take visible and concrete actions to remedy the issues. Human
behaviour is a major contributor to occupational safety. Health and safety issues cannot be tackled
effectively without interference of employers with a particular pattern of behaviour as important criteria
needed to change employee’s behaviours (Zin, 2012).

5
1.4 Management Systems

The management systems control the interaction of people with each other and with processes. They
are the high-level procedure used to control major activities such as, conducting process hazard analysis,
management of change, writing operating procedures, training employees, evaluating fitness for duty,
and conducting incident investigations (Bridges and Tew, 2010).

Process health and safety management systems are a proactive management and engineering approach
to control risk of failures and chance of human error, however incidents still happen, people get injured,
assets get damaged and environment is polluted (Eyayo, 2014). If management systems are weak, then
layers of protection will fail and accidents will happen (Bridges and Tew, 2010).

Typical concerns that cause industry incidents are rule breaking, incorrect risk assessments, supervisors
who are technically competent but short on personal management skills, ineffective contract
management in health and safety (Holstvoogd et al., 2006). Regardless of the industry that uses the
highly hazardous chemicals, there is potential for an accidental release any time of hazardous chemicals
if they are not properly controlled and this, in turn, creates the possibility of disaster (OSHAcademy,
2017).

According to Bridges and Tew (2010) the process related activities where errors have the most influence
include designing of a process, engineering of a process, specifying the process components, receiving
and installing equipment, commissioning the process equipment, operating safeguards necessary to
control the risk at acceptable level and sustaining these safeguards for the life of the process,
troubleshooting and shutting down the process equipment and managing process changes.

According to Thomas and Babu (2014) the key safety management systems are, namely;

 Management commitment to safety – because the attitudes and actions of management can
significantly influence the entire staff, it is therefore critical that these leaders commit to the
success of a safety management system implementation.
 Proactive identification of hazards – early identification and reporting of hazards can save a
significant amount of time and resources down the road.
 Actions taken to manage risks – a system must be in place to determine logical approaches to
counteract known risks to safe operation.
 Evaluation of safety actions – an ongoing evaluation of the impacts of risk management actions
is necessary to determine if further remedial activities are required.

6
According to Okoye et al. (2016) health and safety knowledge and compliance alone are not enough to
cause behavioural changes required for safety performance but a certain aspects of health and safety
culture are required, there are:

 Enforceable regulatory framework

 Management commitment
 Workers involvement.

1.5 Employers Accountability

According to the Safety Association for Canada upstream Oil and Gas Industry (2012) employers are
responsible for ensuring the health and safety of employees at the work site under occupational health
and safety legislation and an employer may be a contractor, lease owner, licensee, or owner’s
representative. According to Okoye et al. (2016) a vital element in health and safety management
system is visible health and safety commitment from leadership and managers. Perceptions of the
commitment of leadership towards health and safety rather than just the intentions have a strong bearing
on the actual behaviours and performance of the people in the organisation (Holstvoogd et al., 2006).

According to Bond (2007) the health and safety policy encompasses leadership, management systems,
competence, responsibility and communication. Achieving health and safety excellence requires going
beyond the traditional health and safety focus of engineering and regulations (Holstvoogd et al., 2006).
Management commitment provides the motivation force and resources for health and safety activities
within the organisation and creating an environment of continuous improvement belongs to all levels
of management (Okoye et al., 2016).

When people fail and accidents happen, it is usually because they are working in a flawed system.
Blaming a person is a simple, expedient method of removing blame from the organisation and hence
hides the latent flaw in the system (Bond, 2007). Human behaviour is a contribution cause to most
incidents and injuries, health and safety excellence can only be achieved by addressing the human
dimensions of health and safety (Holstvoogd et al., 2006).

Most of accidental losses except for natural disasters begin with a human error (Flin, 2015). For
organisations to have the improved health and safety, the petrochemical industry has had to look at near
misses and investigate them to get the root cause (Bond, 2007).

7
To prevent injuries, hazard recognition methods are introduced to identify workplace hazards and
mitigate risk associated with these hazards through the use of procedural or physical controls
(Campbell, 2008). According to Albert, Hallowell and Kleiner (2014) occupational safety has gained
considerable attention following the OSHAct, which shifted substantial health and safety responsibility
to employers. According to the regulations, employers are to provide workers with a workplace free
from any recognised hazards (Campbell, 2008). There are specific requirements of employers,
depending on the hazards and the work to be done (Safety Association for Canada upstream Oil and
Gas Industry, 2012).

According to the Safety Association for Canada upstream Oil and Gas Industry (2012) employers are
responsible for the following,

 Identifying and assessing workplace hazards.

 Developing and implementing safe work and emergency procedures and controls
 Ensuring that workers and supervisors are adequately instructed and trained.
 Immediately investigating incidents, complete root cause analysis and close all corrective
action items.
 Ensuring controls e.g. personal protective equipment are used and, in some cases, provided.
 Ensuring competent supervision as specified by local legislation.

Health and safety issues cannot be tackled effectively without interference of employers with a
particular pattern of behaviours as important criteria needed to change employee’s behaviours (Zin,
2012).

1.6 Problem Statement Discussion

The human, social and economic costs of occupational accidents, injuries, diseases and major industrial
disasters have long been cause for concern at all levels from the individual workplace to the national
and international (Eyayo, 2014). Workers are usually exposed to risk either because of their lack of
knowledge about workplace hazards due to limited experience and knowledge or failure to behave
safely, which may be associated with the workers’ attitude toward health and safety or the
underestimation of perceived risk (Vitharana et al., 2015). Incidents continue to occur in various
industries that use highly hazardous chemicals which may be toxic, reactive, flammable and explosive
or may exhibit a combination of these properties (OSHAcademy, 2017).

8
Risk analysis remains an important tool in safety management for prioritizing actions and plans, in order
to commit top health and safety management systems, training courses need to be further emphasised
and improved to ensure better health and safety culture, performance and involvement of all workers in
top health and safety management of the company (Elssayed et al., 2012).

There are many process incidents that take place in petrochemical industry, BP Texas City incident is
one of them. According to Broadribb (2007) the incident was an explosion caused by heavier – than air
hydrocarbon vapours combusting after coming into contact with an ignition source, probably a running
vehicle engine. The hydrocarbons originated from liquid overflow from the blowdown stack following
the operation of the Raffinate Splitter overpressure protection system caused by overfilling and
overheating of the tower contents. The failure to institute liquid rundown from the tower, and the failure
to take effective emergency action, resulted in the loss of containment that preceded the explosion.
Supervisors assigned to the unit were not present to ensure conformance with established procedures,
which had become custom and practice on what was viewed as a routine operation.

Figure 1.1 Blowdown Drum and Isomerisation Unit after Incident

According to Broadribb (2007) March 23, 2005 explosion happened and huge fire, 15 deaths and 180
injuries. This incident took place during start-up, tower and blowdown drum overfilled. Liquid
hydrocarbon released, vapour cloud was formed and ignited. The three initiating conditions in this
incident were, namely;

 The overfilling of the distillation tower.

 The use of blow down drum and stack that is open to the atmosphere.
 The placement of the contractor work trailers adjacent to high hazard process units.

9
The root cause was reported to be the following,

 Human error, numerous underlying conditions influenced operator’s decision-making and

actions. Lack of shift communication policy or emphasis on communication.
 The start-up procedure required the tower level control valve be open and the tower be filled
within the range of the level transmitter. However, the board operator closed the tower level
control valve and filled the tower above the amount specified in the procedures. In a majority
of the start-up the tower was filled above the range of the level transmitter. Tower ran with
high level to protect equipment and none of these start-up was considered abnormal or
investigated to correct problems. Management did not revise out-dated procedures.
 Communication was ineffective between operations personnel, instructions for routing feed
led to the level control valve being closed; the condition of equipment was not communicated
from one shift to the next. The organisation had no policy for effective communication between
operations personnel during shift changes.
 Instrument malfunctioned, six pieces of instrument malfunctioned on the day of the incident,
a redundant high level alarm, a sight glass on the tower and miscalibrated level transmitter.
The level transmitter’s setting was incorrect, likely not altered since it was set 30 years ago.
 Operators were likely fatigued, operators worked 12 hours shift, 7 days a week, 29 days. Acute
sleep loss and cumulative sleep debt resulted. The organisation had no corporate or site specific
fatigue prevention policy or maximum shift work regulations. No fatigue prevention guidelines
widely used.
 Supervisor and operator staffing was insufficient, unit start-up are especially hazardous but no
supervisor assisted with start-up. 25% budget cut target in 1999 led to Isomerisation unit
staffing cuts – control room consolidation and increased workloads followed. Hazard review
recommended two board operators during all start-up, but only one was working March 23,
2005.
 Operator training was not effective, move to computer-based training without effective
verification methods of competency. Switch to computer-based training was a business
decision driven by cost. From 1998 to 2004 central training staff reduced from 28 to 8 and
budget cut in half, concurrent with instruction to cut costs by 25%. Audit and review from
2002 – 2005 identified on-going deficiencies in operator competency, yet managers adopted a
compliance strategy that relied on engineering controls to prevent accidents due to cost. No
effective training for abnormal situation management or simulation technology made available.

Worker fatigue can adversely impact personal health and safety as well as the efficiency and safety of
the operation (Lerman et al., 2012). The most effective method to minimizing worker fatigue is through

10
a comprehensive fatigue risk management system (FRMS). An FRMS requires the active participation
of all parts of the organisation, including management, support functions, and workers (Lerman et al.,
2012).

It is not enough to determine the level of health and safety knowledge and compliance among the
construction workers in the state and/or their impact on project performance, rather it is important to
establish a relationship between the level of health and safety knowledge and compliance; health and
safety knowledge and project performance; and the level of health and safety compliance and project
performance (Okoye et al., 2016).

According to Almanssoor (2008) determining the fire potential of a chemical substance is accomplished
through its flammability characteristics; no single factor, however, defines a substance’s flammability
and when a flammability comparison is to be made between different substances, the following factors
should be considered, namely;

 Flammability limits (or explosion limits)

 Flash Point
 Auto-ignition temperature
 Vapour pressure
 Burning velocity
 Ignition energy

According to Almanssoor (2008) the most important and widely used factors are the first three, namely;

 Flammability limits – flammability limits of a gas define the concentration range of a gas-air
mixture within which an ignition source can start a self-propagating reaction.
 Flash point – the flash point of liquid is the lowest temperature at which the liquid releases
vapour in a sufficient amount to form an ignitable mixture with air near its surface.
 Auto-ignition temperature – the auto-ignition temperature is the minimum temperature required
to cause or initiate self-sustained combustion independent of the source of heat. A substance
will ignite spontaneously when it reaches its auto-ignition temperature.

Safety measures are chosen to eliminate an element of the fire triangle whilst causing minimal
interference to plant operation (Puttick, 2008). According to Almanssoor (2008) to produce combustion,
three conditions must coexist: flammable substance, oxygen, and a source of ignition. Avoidance of
ignition sources can be useful and reliable basis of safety in certain circumstances provided that it is

11
restricted to the inside of chemical plants, and certain well defined charging and discharging areas
(Puttick, 2008).

Figure 1.2 Bhopal Tragedy (Isocyanate Gas Leak)

According to Peterson (2009) in December 2-3, 1984, a gas leak of some 40 tons of methyl isocyanate
gas mixed with unknown other gases from chemical plant at the Union Carbide India Limited pesticide
plant caused one of the highest casualty industrial accidents of the 20th century. Over 500,000 people
were exposed to methyl isocyanate gas and suffered respiratory problems and injuries of varying
severity and more than 2000 people died immediately. The methyl isocyanate release at Bhopal, India,
in 1984 that killed over 2000 people and injured tens of thousands is one of the reasons why we need
process health and safety management systems to be appreciated by the petrochemical industry. A fire
and explosion at a PEMEX LPG terminal in Mexico City, also in 1984, killed more than 600 people
and injured around 7000 people (Ness, 2015).

Figure 1.3 Phillips Petroleum Explosion

There was an accident October 1989 in Phillips Petroleum Company, Pasadena, incident resulting in 23
deaths and 132 injuries.
In a petrochemical industry control of ignition sources is vital (Puttick, 2008). According to Almanssoor
(2008) explosion is a sudden and violent release of energy, this energy could be physical energy,

12
chemical energy and nuclear energy and most chemical explosions involve a limited set of simple
reactions, all of which involve oxidation. Fire and explosion hazards prevention involves dealing with
the elements of the fire triangle and they fall into three categories, namely absence of flammable
atmospheres achieved by limited fuel quantities, control/avoidance of ignition sources and absence of
flammable atmospheres by limiting the quantities of oxidant present (Puttick, 2008). An explosion can
be spontaneous or initiated by light, heat, friction, impact, or a catalyst (Almanssoor, 2008).

Explosions are not confined to closed systems; explosions may occur in an open area such as a process
plant in which case the pressure wave will expand itself until the pressure gradient becomes insignificant
(Almanssoor, 2008).

Figure 1.4 Amuay Oil Refinery Explosion

On August 25, 2012 the Amuay Oil Refinery in north western Venezuela suffered an explosion. This
unprecedented incident in the national oil industry killed 48 people and injured 156.

These incidents call for highly increased focus to process health and safety management systems in the
petrochemical industry (Zin, 2012). Employer must develop and implement written operating
procedures that provide clear instructions for safely conducting operations and maintenance (Hardy,
2013).

According to Ness (2015) three contract workers were killed and a fourth worker was seriously injured
in an explosion and fire at the Patridge-Raleigh oilfield in Mississippi. The contractors, who were
employees of Stringer Oilfield Services, were tasked with installing a pipe between two oil production
tanks. Welding sparks ignited flammable vapour that was escaping from an open-ended pipe near the
welding activity. The root cause of this incident was hot work being conducted in the presence of
flammable atmosphere without using any safe work permitting procedure. A gas detector should have
been used to test for flammable vapour. The open pipe on tank was not capped or isolated. All of the

13
tanks were interconnected, and some of the tanks still contained flammable residue and crude oil. Key
lessons learned in this incident are, namely;

 Safe work practices, such as hot work permits, are necessary to ensure a safe work environment
when hazardous chemicals, in this case flammable vapours, are present. The contractor did not
require the use of safe work procedures, specifically hot work permits in this case.
 Contractors need to be managed in such a way as to ensure they know about and use safe work
practices. The owner of the wells and tanks relied on contractors to do most of its well
commissioning work, such as installing tanks, pumps, and piping. The owner should have
managed the contractors to make sure they used safe work practices.

1.7 Research Questions

1.7.1 What deficiencies exist in process health and safety management systems when dealing with
hazardous chemicals?
1.7.2 How should the major deficiencies be prioritised by top management to prevent process health
and safety incidents when handling hazardous chemicals?
1.7.3 What are the critical drivers to achieve generative process health and safety culture?

1.8 Research Objectives

The research objectives are,

 Review the process health and safety management systems within industries where hazardous
chemical substances are being used and identify the deficiencies;
 Identify health and safety hazards and describe the awareness of occupational health and safety
hazards to the employees in the petrochemical industry;
 Test the effectiveness of the existing process health and safety management systems in
petrochemical industry;
 Assess the existing process health and safety management systems deficiencies and prioritise
the major deficiencies that need urgent senior management’s attention;
 Identify critical drivers to achieve generative health and safety culture;
 Develop a conceptual model and validate it using structural equation modelling.

14
1.9 Methodological Approach

The objectives of the research were achieved by the following research approach,

 Literature review to identify the critical process health and safety management elements that
need to be prioritised by senior management.
 Development of a conceptual model from literature review.
 Questionnaire development for the survey.
 Questionnaire design, validity, types of validity is also discussed under research instrument
design. Systematic random sampling was used.
 Data collection strategies are discussed that includes cross sectional, longitudinal, survey
strategies, sample size and population. The data was collected by using a self - administered
questionnaire survey in this research.
 Structural Equation Modelling was used to validate the conceptual model and identify the best
fit generative process health and safety culture model.

1.9.1 Literature Review

An extensive literature review on health and safety in the petrochemical industry was conducted in order
to describe and understand the challenges in this industry. The health and safety management systems
were used to identify the process health and safety elements that influences process health and safety
incidents in the petrochemical industry. The literature review was also use to identify the process health
and safety management deficiencies to be prioritised by senior management in the petrochemical
industry.

1.9.2 Conceptual Model

The literature review was used to identify the top process health and safety management elements that
were used to develop a theorised conceptual model. The conceptual model included the process health
and safety management elements, namely, Leadership Commitment, Training and Competency,
Chemical Exposure Management, Health and Safety Risk Assessment, Process Hazard Analysis,
Process Health and Safety Information, Operating Procedure, Control of Ignition Source, Control of
Confined Space Entry and Permit to Work among others. The process health and safety management
elements were tested against a generative health and safety culture as latent variable. There were 35
observed variables in this research. The conceptual framework was tested and validated.

15
1.9.3 Questionnaire

The questionnaire was used as a research instrument and was shared with the petrochemical industry
employees to answer the questions. The research questions used a Likert scale. The neutral option was
used to accommodate participants that prefer not to comment or that were puzzled about the statement.
The five options were presented to the participants to choose the answer that expressed their feelings
most accurately about the statement. The respondents were requested to use any type of indication be it
a cross, or a tick or they could circle in the relevant answer.

The questionnaire used in this research had two parts. Part one was the particulars of respondents and
part two was the research questions. The study was conducted by distributing 400 questionnaires
manually to the randomly selected potential participants of which 259 were returned duly completed
and used. They were statistically analysed using SPSS version 25. The response rate was 64.75%.

1.9.4 Structural Equation Modelling

Structural equation modelling was used in this research. The observed variables were identified in this
research and tested against the latent variable. The observed variables were identified as rectangular
shape and the latent variable were displayed as an eclipse or circle in AMOS version 25. The regression
path was expressed by an arrow. Model specification was the first step in path models as it determines
every relationship and parameters. Model identification was the second step in this research and the
third step was model estimation. Model testing was the fourth step as it determines how well the data
fits the model. Model modification was carried out in order to improve the model so that a better fitting
model can be more interpretable.

1.9.5 Reliability and Validity

According Zohrabi (2013) one of the main requirements of any research process is the reliability of the
data and findings. Internal consistency gives an estimate of the equivalence of sets of items from the
same test and the coefficient of internal consistency provides an estimate of the reliability of
measurement and is based on the assumption that items measuring the same construct should correlate
(Kimberlin and Winterstein, 2008). Reliability refers to the extent to which our measurement process
provides consistent or repeatable results (Zohrabi, 2013). Validity is often defined as the extent to which
an instrument measures what it purports to measure and validity requires that an instrument is reliable
without being valid (Kimberlin and Winterstein, 2008).

16
According to Hair et al. (2010) cited in Zhao (2017) composite reliability is frequently used to quantity
how well all the measurement indicators reliably represent the corresponding latent construct.
Reliability indicates how closely the results of repeated measurements of the same concept agree: a
reliable measurement of quantity known not to have changed that is performed twice with the same
person will produce the same value both times and the same is true where two different people are
involved who do not differ on the measured characteristic (Zohrabi, 2013).

This research used, Cronbach’s alpha, correlation and corrected-item-correlation to measure reliability.
Validity was ensured in this research by a severe examination of literature review, adequacy of samples,
representatives of samples, adequacy of data processing and analysis, appropriate interpretation and
justifiable conclusions. Composite reliability, discriminant validity and convergent validity of the final
study model were evaluated.

Testing for discriminant validity can be done using one of the following methods, namely, O-sorting,
Chi-square difference test and the average variance extracted analysis (Zait and Bertea, 2011). This
research discriminant validity was used to assess the shared variance between the constructs by
calculating the average variance extracted.

1.9.6 Ethical Consideration

In this research an informed consent form for the participants was used to ensure that the respondents
are aware that their participation is voluntary by nature and the participant has the right to withdraw
partially or completely from the process. The questionnaire was explained to the small groups and they
were allowed to decide whether they were willing to participate to this research or not.

Confidentiality is vital when doing research and the respondents were assured that their response will
be held in confidence. Those that were willing to participate were then allowed to answer the
questionnaire and those that decided to not participate were allowed to leave the room without
answering the research questionnaire. The participants that were willing to proceed and participate were
requested to be truthful and honest when answering the questionnaire and were told everything about
the research without compromising the truth to avoid deception.

The researcher ensured that an informed consent form was used when questionnaire was distributed,
and confidentiality was highlighted to the research participants. The privacy of the research data was
ensured by using a secured and personal laptop that requires a password known by the researcher. The
final research data was then transferred to the desk computer in the supervisor’s office, where there is
no access given to anyone without supervisor’s permission. In order to ensure ethical standards, the

17
researcher obtained an ethical clearance certificate from the University of KwaZulu-Natal. Ethical
clearance to conduct the study was obtained from the University of KwaZulu - Natal Humanities and
Social Sciences Research Ethics Committee (HSS/1094/018D).

1.10 Study Limitations and Delimitations

The scope of data collection for this research was limited to South Africa KwaZulu-Natal Province. The
research was also limited to the petrochemical industry and the participants invited to participate in this
study were from petrochemical and chemical industry. Limitations of this research comprise the short
duration for the study period within which the questionnaire data were collected as planned.

Literature sources used in order to meet the secondary data objectives were limited to those readily
available in online search engines databases as well as publications available in libraries within the
boundaries of South Africa. This study was also limited to employees working in the petrochemical
industry and responsible for production, technical services, maintenance, health, safety and
environment, turnarounds or plant shutdowns and engineering projects. The research is based on survey
instrument which was designed by the authors and so have not been extensively validated. The
reliability and validity of the research instrument may be questioned as it was not sufficiently assessed
for face validity through a panel of experts. Predictive and constancy were not assessed.

International factors could influence health and safety culture, particularly for countries that do not have
trade unions, different culture, autocratic to mention a few. The examination did not include any
international stimuli to reduce possible difficulties and to achieve results that could provide starting
points for research in the future.

The study limitations need some attention concerning the magnitude to which the results can be
generalised beyond the scope of the empirical data due to the complex nature of health and safety
culture. This study was based entirely on quantitative research data. A mixed methods approach would
have provided a check on the validity of the findings by comparing results from a qualitative data
source. The proposed generative health and safety model was not validated by testing too establish its
effectiveness when used.

1.11 Overview of the Research Study

The thesis is organised in nine chapters, which are divided into three parts. Part One contains four
chapters:

18
The objectives addressed by Part One are, namely;

 Identify health and safety hazards and describe the awareness of occupational health and safety
hazards to the employees in the petrochemical industry;
 Test the effectiveness of the existing process health and safety management systems in
petrochemical industry;

Chapter One describes the problem statement and provides a topical introduction. This chapter also
discusses the significance of this research towards health and safety of petrochemical employees. This
chapter discusses objectives, research questions, limitations and knowledge contribution to academia.
The overview of all the other chapters in this research are also summarised in this chapter. The key
contribution of this study arises from the application and refinement of the framework for the empirical
analysis of the health and safety management systems in order to identify main deficiencies when
handling hazardous chemicals.

Chapter Two reviews the literature available in health and safety to develop the theoretical framework.
This chapter discusses the process health and safety elements and attempts to recommend the key ten
process health and safety management elements from literature to be prioritized for an effective process
health and safety management system. This chapter presents a concise literature review regarding the
importance of occupational health and safety in the petrochemical industry. This chapter reviews the
five levels in the health and safety culture ladder, and the concept of health and safety culture and the
development of health and safety culture. An extensive literature review will be conducted in order to
understand the process health and safety concerns and its effect on petrochemical and chemical
industries. This chapter also shares some of the challenges in the petrochemical industry with the focus
on the following hazards, namely, physical hazards, ergonomic hazards, chemical hazards, psychosocial
hazards and radiological hazards.

Chapter Three develops the theoretical framework. This chapter proposes conceptual process health and
safety elements model, based on the latent variable (Generative Health and Safety Culture) and observed
variables (Leadership Commitment, Chemical Exposure Management, Health and Safety Risk
Assessment, Process Hazard Analysis, Permit to Work, Training and Competency, Process Health and
Safety Information, Control of Confined Space Entry, Operating Procedure and Control of Ignition
Source) identified through literature review.

Chapter Four describes the methodological tools used and research design as well as data collection
techniques employed in the fieldwork. This chapter outlined the philosophical assumptions
underpinning this research. These correspond to the epistemological and ontological assumptions of

19
positivism research. This chapter discusses research design and methodology used in this study. The
topics discussed are, namely, research philosophy, research methodology and approach, research
instrument, data collection strategies, sampling techniques and data analysis. The research philosophy
expands on the concepts ontology, realism, relativism, epistemology, positivism, interpretivism,
methodology, quantitative and qualitative in the research paradigm.

This research also shares different research strategies, namely, Case study, Ethnographic study, Ground
theory study, Phenomenological study, Experimental study, Archival research and action research. In
this chapter data analysis include the following topics, Analysis techniques, Bias, Response Bias and
Types of Response Bias, Deliberate Falsification, Misrepresentation, Validity Reliability and
Challenges with data access.

Part Two, which contains Chapter Five, Six and Seven focuses on the analysis of results, model
development and model validation.

The objectives addressed by Part Two are, namely

 Review the process health and safety management systems within industries where hazardous
chemical substances are being used and identify the deficiencies;
 Assess the existing process health and safety management systems deficiencies and prioritise
the major deficiencies that need urgent senior management’s attention;
 Identify critical drivers to achieve generative health and safety culture;

Chapter Five focuses on the presentation of the results. This chapter discusses the analyses of the results
acquired from the survey. This includes descriptive statistics using frequencies, means, and standard
deviation.

Chapter Six addresses the model development and model development is considered an effective
research method as it assists investigators and scientists in relating more accurately to reality; it also
aids the researcher to describe, predict, test or understand complex systems or events (Shafique and
Mahmood, 2010). Fit indices found using exploratory factor analysis and confirmatory factor analysis.

Chapter Seven validates the model. This chapter discusses structural equation modelling and how it was
used to establish the interrelationships of the hypotheses. In quantitative research, the researcher
analyses the data in order to test one or more formulated hypotheses. The aim of which is to find out if
the relationships between the observed variables in one or more groups are statistically significant

20
(Gelo, Braakmann and Benetka, 2008). According to Kline (2011); Raykov (2006) cited in Harinarain,
(2013).The aim of structural equation modelling is to see how well the proposed model accounts for the
observed relationships among these variables These two chapters are responds to the last objective that
is to develop a model and validate it using structural equation modelling.

Part Three discusses the findings, concludes and recommends for future research for this research topic
in Chapter Eight and Nine.

1.12 Contribution of the Study Findings to Knowledge

The study offers an innovative analytical and methodological approach in process health and safety
culture assessment. The thesis is part of the larger discussion of increasing importance in health and
safety policymaking. This study aims at contributing to the literature in the field of health and safety by
incorporating management deficiencies that require senior management’s attention and the drivers
towards a generative health and safety culture. It offers an innovative methodology in assessing
petrochemical industry performance in health and safety.

Methodological contribution lies in the experience gained through the application of positivism
approach and techniques applied for data collection. The other methodological contribution relates to
the appropriateness of applying theoretical concepts and theories developed in other contexts. The
successful use of these theories in this study contributes towards providing examples of the positivism
research approach. The study contributes methodologically the five drivers and ten main deficiencies
that require senior management’s undivided attention. The research utilised reliability measures and
validity to ensure that the research instruments were reliable and valid.

One of the practical contributions of this research is the detailed insights provided by the literature
review done in this study. The literature review reveal that senior management need to acquire new
skills of improving health and safety culture in the petrochemical industry. The contribution of this
research is to understand, based on theoretical assumptions, how the health and safety can be
institutionalised in an organisation and how it contributes to happy employees. The due model can be
used as a practical tool.

21
1.13 Chapter Summary

This chapter introduced the research topic of process health and safety management systems
deficiencies relative to hazardous chemical exposure in petrochemical industry. This chapter also stated
the research questions and presented the objectives of the study. The research design and methodology
and research limitations were introduced in this chapter. The research structure was shared in this
chapter. This chapter also shared the envisioned contribution of the study findings to knowledge

This research also includes appendices that contain sample of the questionnaire, ethical clearance
approval for the research, and conference papers published through this research. In the next chapter,
literature review on process health and safety management systems is presented.

22
 CHAPTER TWO
Literature Review
___________________________________________________________

2.1 Introduction

This chapter is devoted to studying the literature on the key theories in the research topic. This chapter
also explores the magnitude to which the key research questions and objectives have been addressed in
the previous studies. The chapter includes deliberations on the concept of health and safety in the
petrochemical industry and key contributors to poor statistics that indicate high fatalities, lost time
injuries and major asset damage that has huge financial impact to the industry. Different hazards in
petrochemical industry are discussed in this chapter, namely, physical hazards, ergonomic hazards,
chemical hazards and psychosocial hazards. This research exposed a wide range of chemical hazards
of diverse nature, although this is not quite surprising given that a great deal of chemical substances is
associated with activities of the petrochemical industry. The chapter ends with a summary of literature
review and how the research intended to contribute to filling the knowledge gap.

2.2 Background to the Study

The petroleum industry began with the successful drilling of the first commercial oil well in 1859, and
the opening of the first refinery two years later to process the crude into kerosene (Kumar et al., 2017).
A petrochemical plant is a high risk industry with potential risks of fire, explosion and poisoning of
employees (Thomas and Babu, 2014). The evolution of petroleum refining from simple distillation to
today’s sophisticated processes has created a need for health and safety management procedures and
safe work practices (Kumar et al., 2017). The health and safety of workers is most important. Operations
and processes in petroleum industries are hazardous, due to properties of the petroleum products and
raw materials (Kulkarni, 2017). Industry becomes successful by not only meeting the production
requirements but also should have high employee satisfaction by providing the health and safety
requirements in the workplace (Purohit et al., 2018).

According Ezejiofor (2014) for many occupational toxicologists, industrial hygienists, and others with
stake in the field of occupational health and safety, the safety of the work place has always been a major
concern. Failures of process health and safety management systems are deadly and costly. Major
accidents have emphasized the need for process health and safety within the petrochemical industries
(Ness, 2015).

23
Health and safety issues should be based on job health and safety analysis or comprehensive hazard or
risk assessment, using established methodologies such as a hazard identification study (HAZID), hazard
and operability study (HAZOP), or a quantitative risk assessment (QRA) (International Labour
Organisation, 2017). The health and well-being of workforce of accompany which is their most valued
asset should not be ignored by the management (Eyayo, 2014).

Health and safety management systems are a systematic and continuous management system based on
proactive identification of hazards, and analyses of their risk (Thomas and Babu, 2014). In spite of the
immense benefits derived from work, work itself has become a source of several deaths, ill-health and
injuries, as clearly illustrated by data from relevant authorities including the World Health Organisation
(WHO) and International labour organisation (ILO). World Health Organisation report for year 2000
concluded that workplace hazards are responsible globally for 37% of back pain, 16% of hearing loss,
13% of chronic obstructive pulmonary disease (COPD), 11% of asthma, 10% of injuries, 10% of lung
cancer, and 2% of leukaemia. The occupational hazards are major issue in oil and gas extraction
industry. According to studies by an investigator, fatality rate in this industry is 2.5 times more than
construction industry and 7 times more than general industry (Kulkarni, 2017).

The occupational health and safety act 85 of 1993 is leading OHS legislation in South Africa. Its primary
aim is to provide for health and safety of employees and for health and safety of persons in connection
with the use of plant and machinery, the protection of persons other than persons at work against hazards
to health and safety arising out of or in connection with activities of persons at work (Thomas and Babu,
2014). Occupational health hazard which is different from occupational safety hazards is prevalently
on the rise as industrialization increases in the global world (Eyayo, 2014).

According Ezejiofor (2014) in spite of the difficulty in obtaining information concerning occupational
diseases and injuries in developing countries due to lack of comprehensive and harmonious data
collecting systems, ILO still estimates that 2 million workers die each year from work related injury
and illness. In 2002, in sub-Saharan Africa alone, ILO estimated more than 257,000 total work related
fatalities, including about 50,000 injuries.

According to Ness (2015) accidents almost always have more than one cause. For many years, safety
experts have used the Swiss cheese model to help managers and workers in the process industries
understand the events, failures, and decisions that can lead to a catastrophic incident or near miss.
According to the Swiss model each layer of protection is depicted as a slice of Swiss cheese, and the
holes in the cheese represent potential failures in the protection layers, namely;

24
  Human errors
 Management decisions
 Single-point equipment failures or malfunctions
 Knowledge deficiencies
 Management system inadequacies, such as a failure to perform hazard analyses, failure to
recognise and manage changes, or inadequate follow-up on previously experienced incident
warning signs.

Figure 2.1 Reason’s Swiss Cheese Model.

According to Kumar et al. (2017) health and safety professionals, working with process, chemical,
instrumentation, and metallurgical engineers, assure that potential physical, mechanical, chemical, and
health hazards are recognised and provisions are made for safe operating practices and appropriate
protective measures. These measures may include hard hats, safety glasses and goggles, safety shoes,
hearing protection, respiratory protection and protective clothing such as fire – resistant clothing where
required. These health hazards, usually, could be associated with most industrialised organisation and
the oil and gas refinery is not left out (Eyayo, 2014). A critical part of any safety and health program is
the identification, assessment, elimination and/or control of hazards in the workplace (Dunbar, 2014).

Health hazards which could result in the development of diseases and sickness is categorised into
physical health hazard, chemical health hazard, biological health hazard, mechanical/ergonomic health
hazard and psychosocial health hazard (Eyayo, 2014). As a general approach, health and safety
management planning should include the adoption of a systematic and structured approach for

25
prevention and control of physical, ergonomic, biological, chemical, psychosocial and radiological
health and safety hazards (International Labour Organisation, 2017).

According to Makin and Winder (2008) cited in Elssayed, Hassan and Hosny (2012) the term hazard
profile is used to indicate the particular blend of characteristics or exposures within a given work
environment that have the potential to cause harm or loss to those whom a duty of care is owed. It is
impossible to eliminate all hazards, so the goal is to eliminate and/or control the hazards with critical
and high potential and to reduce the rest of the hazards to the lowest reasonable risk level to protect
workers from harm (Dunbar, 2014).

2.2.1 Physical Hazards

According to Eyayo (2014) physical hazards are often said to be less important than chemical hazards
but this not so. A physical hazard is a factor within the environment that can harm the body without
necessarily touching it. Vibration and noise are examples of physical hazards (International Labour
Organisation, 2017). According to Ezejiofor (2014) hand arm vibration syndrome and noise induced
hearing loss are some of the health effects associated with occupational exposures to vibration and
noise. In areas nearby petrochemical industries, elevated sound levels induce noise pollution associated
with feelings of headache, annoyance, uneasiness, stress, impatience, displeasure, hypersensitivity,
extreme anxiety, anger, endangerment and violence (Sharma et al., 2017). The environmental factors
of heat stress are air temperature, water vapour pressure, radiant heat, and air velocity (Ezejiofor, 2014).

According to Eyayo (2014) the nature of physical agents is wide and should not be underrated but the
main ones capable of causing occupational disorders and injuries are, namely;

 Noise
 Illumination
 Vibration
 Radiation (ionizing and non-ionizing)
 Microclimatic conditions in the case of extreme heat and cold.

Employees may be at risk of heat stress when exposed to hot environment or extreme heat and this can
result in illnesses including heat stroke, heat exhaustion, heat syncope, heat cramps and heat rashes or
death (International Labour Organisation, 2017).

26
Heat stress is the aggregate of environmental and physical work factors that constitute the total heat
load imposed on the body (Ezejiofor, 2014). Heat also increased the risk of workplace injuries such as
those caused by sweaty palms, fogged-up safety glasses or dizziness, and may reduce brain function
responsible for reasoning ability, creating additional hazards (International Labour Organisation, 2017).

Different types of mechanical energy such as noise, vibration, radiation and temperature extremes can
physical injury hazards (Vitharana et al., 2015). According to Ezejiofor (2014) vibration and noise
militate against health generally, since exposure to both are linked to sundry physiological and
psychological health effects including annoyance, sleep disturbances, electroencephalographic changes
and cardiovascular disorders. Noise is inevitable in construction sites due to the nature of construction
activities (Vitharana et al., 2015). Bruises, sprains, fracture, concussions and lacerations are injuries
that result from some form of physical contact, usually the worker is moving and hits a stationary object
or the moving object hits a stationary worker (International Labour Organisation, 2017).

Vibration is oscillatory motion about a point and occupational/chronic exposure to hand transmitted
vibration results in various disorders sometimes collectively known as the hand arm vibration disease
or vibration syndrome (Ezejiofor, 2014). The syndrome includes vascular, neurological and
musculoskeletal disorders that may become manifest individually or collectively. Loss of hearing,
burns, hypothermia, heat stroke, skin rash and blistering are some of the injuries or illnesses caused by
exposure to unseen physical hazards, are slips, trips and falls, electricity, tools and machinery, cold
stress and fire (International Labour Organisation, 2017).

The wet bulb temperature index (WBGT) index is one of the most commonly used heat stress indices,
and the standard in South Africa in industrial settings and is mentioned in occupational health and safety
legislation (Joubert and Bates, 2008).

Physical work contributes to the total heat stress of the job by producing metabolic heat in the body,
heat being in proportion to the work intensity and not only the environment and physical factors, but
also age, gender, physical fitness, health status, clothing and acclimatization may be major factor
contributing to the changes which occur in human physiological response to heat stress (Ezejiofor,
2014). The World Health Organisation states that it is inadvisable to exceed a rectal temperature of 38
degree C during prolonged exposure to heavy work. The thermal work limit defined as the maximum
sustainable metabolic rate that euhydrated (adequately hydrated), clothed, acclimatised individuals can
maintain in a specific thermal environment, whilst maintaining a safe deep body core temperature
(<38.2 degrees C) and sweat rate (<1.2 kg/hr) (Joubert and Bates, 2008).

27
Human beings maintain thermal and other bio-physiological factors in the body within relatively narrow
limits and the balance between man and the environment is especially important when one is working
in extreme thermal conditions (Ezejiofor, 2014).

2.2.2 Ergonomic Hazards

Managers in the oil and gas business have frequently received worker complaints such as chronic
fatigues, back pains, headaches, upper body pains, sleep disorders, and stresses and these complaints
are indications of ergonomic deficiencies in the work system from the industry (Kim, 2016). An
ergonomic hazard is a physical factor within the environment that harms the musculoskeletal system
and include, among others, repetitive movement, manual handling, inappropriate workplace/job/task
design, uncomfortable workstation height and poor body position (International Labour Organisation,
2017). Ergonomic injuries include strains, sprains and other problems (Ezejiofor, 2014).

Workers in the oil and gas industry continuously experience the problems of noise, slippery surfaces,
and numerous manual material handling exposures of carrying, lifting, lowering, pushing and pulling
tasks (Kim, 2016). These injuries can be caused by performing the same motion over and over again,
using physical force when lifting heavy objects or being in an awkward position like twisting your body
to reach a light bulb (Ezejiofor, 2014).

There are electrical issues and fall protection challenges, as well as repetitive tasks such as valve
turning, which increases the force risks to the employees (Kim, 2016). According to Ezejiofor (2014)
it is recommended that one makes occasional breaks between sitting and standing and talking about
shift and call duty, or prolonged working hours, there are several dimensions of this, just as there are
several implications for both health and work. Workers from the oil and gas industry are largely
uncovered to ergonomic – related injuries and injury risks such as repeated bending, lifting heavy items,
pushing and pulling weight loads, reaching overhead, performing the same or similar tasks repetitively,
and working in awkward body postures (Kim, 2016).

According to International Labour Organisation (2017) workplaces and premises for workers shall
have, as far as possible, natural lighting and shall be supplied with adequate artificial or electric lighting
to ensure good vision to workers. According to Kim (2016) the following examples of ergonomic
solutions are suggested to successfully manage health and safety problems in the oil and gas industry,
namely;

 Use ergonomic concepts and information in the designing of jobs and even choosing equipment
and tools.

28
  Establish ergonomic policies, procedures and review programs throughout the corporation, no
matter what the size of industries.
 Train workers in ergonomics for the appropriate handling and use of the special tools required
during maintenance and turnaround activities.
 Ensure all workers on the location to understand the risks and dangers before starting any
maintenance activities. They should be carefully informed and trained in appropriate safety
procedures, including the use of safety equipment and breathing apparatus.
 Utilise ergonomics for the design and layout of control rooms to eliminate human errors and
increase comfort, fit, user performance, and functionality.
 Ensure all signs and symbols are placed in areas that everyone can see and read clearly. Use
larger fonts and consider both indirect as well direct glares.

According to World Health Organisation (2013) cited in Eyayo (2014) repetitive tasks and static
muscular load are also common among many industrial and service occupations and can lead to injuries
and musculoskeletal disorders and in many developed countries such disorders are main cause of both
short term and permanent work disability and lead to economic losses amounting to as much as 5% of
gross national product.

Many of the oil and gas corporations have instigated to comprise ergonomic interventions to their plants
and managerial areas (Kim, 2016). Royal Dutch Shell has developed several design engineering
procedures covering many aspects of ergonomics for their projects. Shell has adopted ergonomics as
part of its corporate health and safety management guidelines. Chevron Texaco is upgrading its safety
in design manual to integrate ergonomics and assesses subcontractors on their ability to perform in this
area.

2.2.3 Chemical Hazards

According to International Labour Organisation (2017) some of the duties of the employer are to
identify risk of physical or chemical reaction of hazardous chemical and ensuring the stability of
hazardous chemicals, ensure that exposure standards are not exceeded, provision of information,
training, instruction and supervision to workers, obtaining the safety data sheets from the manufacturer,
importer or supplier of the chemical. According to Almanssoor (2008) the issue of safely producing
hazardous chemicals is as important as the economics of producing and selling them. Personal factors
determine the effects of a chemical and these include generic factors, age, gender, health status,
hypersensitivity, personal habits, hygiene, pregnancy and lactation (Naafs, 2018).

29
According to the Safety Association for Canada upstream Oil and Gas Industry (2012) once chemical
hazards and workers risks are determined, the required control approach needs to be identified and
implemented as follows,

 Identifying how to protect workers from the health hazards of each chemical that is on site.
 Identifying what to do in the event of an uncontrolled release, leak, or spill.
 Knowing where to go for more information.
 Knowing how to dispose of the products safely.

According to Ezejiofor (2014) some of the chemical substances recorded as hazardous in the
petrochemical industry work environment include petroleum fumes, Benzene, toluene, xylene, gases
such hydrogen sulphide (H2S), Hydrogen Fluoride (HF), ammonia (NH3), and methane, polycyclic
aromatic hydrocarbons (PAH), asbestos, several salts and acids, to mention this few table 1.

Hazardous chemicals in the petrochemical industry are substances, mixtures and materials that can be
classified according to their health and physico-chemical risks and dangers and such hazards include
skin irritants, carcinogens or respiratory sensitizers that have an adverse effect on an employees’ health
as a result of direct contact with or exposure to the chemical, usually through inhalation, skin contact
or ingestion (International Labour Organisation, 2017).

In the European Union alone, approximately 16 million people are exposed to carcinogenic agents at
work and the most common cancers resulting from these exposures are cancers of the lung, bladder,
skin, mesothelium, liver, haematopoietic tissue, bone and soft connective tissue (World Health
Organisation, 2013 cited in Eyayo, 2014).

About 300 – 350 substances have been identified as occupational carcinogens and they include chemical
substances such as benzene, chromium, nitrosamines and asbestos (World Health Organisation, 2013
cited in Eyayo, 2014). The hazardous effect of chemicals comes through three ways, namely: fire,
explosion and toxicity and the first essential step towards greater plant health and safety is being aware
of the potentially dangerous properties of the substances, i.e. whether they are flammable, explosive or
toxic (Almanssoor, 2008).

Toxicology is the study of poisons and how they affect the body, and it is an inherited property of a
chemical that causes bodily injury or disease to a living organism as a result of physiochemical
interaction with living tissue (Naafs, 2018). Toxicity is defined as the ability of toxic (poisonous)

30
substances, when absorbed by living tissues (either ingestion or via the skin), to cause injury or destroy
life (Almanssoor, 2008).

All substances including chemicals are potentially poisons. However, all chemicals can be used safely
if exposure is kept below tolerable limits (Naafs, 2018). Injuries, caused by toxic effects of chemicals,
vary and occur both close to and distant from the point of release of these chemicals, especially when
the correct precautions to chemical releases are ignored and injuries include eye, skin, poisoning,
asphyxia and respiratory system injuries (Almanssoor, 2008).

There are various factors that influence the toxicity and the health effects of a chemical agent and these
include its physical state, dose or concentration, route of absorption, duration of exposure and presence
of other chemicals (Naafs, 2018).

According to Almanssoor (2008) examples of hazardous substances prevailing within the petrochemical
industry are, namely;

 Gases (flammable, toxic, compressed).

 Liquids (flammable, toxic, acidic, alkaline, cryogenic).
 Solids (flammable, volatile).
 Viscous materials.
 Oxidizing, reactive and corrosive substances.

Among certain occupational groups such as asbestos sprayers, occupational cancer may be the leading
factor in ill-health and mortality and due to the random character of effect, the only effective control
strategy is primary prevention that eliminates exposure completely, or that effectively isolates the
worker from carcinogenic exposure (World Health Organisation, 2013; Eyayo, 2014).

According to the Safety Association for Canada’s Upstream Oil and Gas Industry (2012) to address the
personal health and safety of workers, chemical hazard assessment need to consider the overall health
and safety of the process design as well as the manner in which project activities are managed and the
recommended methodology has seven steps, namely;

 Identify operations that involve chemical exposure.

o Processes from purchasing through to disposal.
o Type of operation.
o Equipment design and layout.

31
 Identify and confirm who is responsible for chemical control and use.
o Planning phase: The planner should assign a suitable representative to identify, assess,
control, and communicate the chemical hazards as well as the planner’s standards for
all work on the site.
o Implementation phase: Those persons responsible for organising workers and ensuring
that the chemical management process gets done. Implementation roles include but are
not limited to supplier, supervisor, and site supervisor.
o Execution phase: Those persons who physically manage the chemicals on the job site
and therefore are directly or indirectly exposed to the chemicals. Completing effective
chemical hazard assessments and implementing suitable control measures are the
responsibility of every stakeholder on the work site.

 Identify the potential chemical exposure hazards.

o Material safety data sheets are one of the most effective and efficient ways of
communicating and managing chemical hazards. They should be readily available for
all participants throughout the process.
o Risk phrases, there should be full listings of risk phrases.
o Industry knowledge and literature, additional information on hazardous substances,
types of exposure, and unique chemical situations in the oil and gas industry.

 Assess the chemical hazards.

o Properties of the chemical (e.g., flammable, health hazards)
o Where is the chemical being used?
o How much of the chemical is being used?
o How long is the worker exposed to the chemical?
o How are workers exposed?

 Evaluate and analyse the chemical hazards.

o Quantifying the severity of the hazard.
o Quantifying the likelihood of the hazard.
o Using a matrix or equation to quantify the risk.

 Decide on the control approach needed to remove or reduce chemical exposure risks.
o Selecting the appropriate control approach.
o Identifying and applying the appropriate guidance sheets.
o Eliminating/substituting.

32
 o Applying engineering controls.
o Applying administrative controls.
o Specifying personal protective equipment.

 Confirm that the controls work.

o Management of change (MOC).
o Monitoring effectiveness.
o Informing, instructing, and training.
o Keeping records.
o Operations maintenance.

2.2.4 Psychosocial Hazards

According to Ezejiofor (2014) psychosocial hazards stem from overloaded work pressure, prolonged
hours of work, shift duty and/or call duty schedules, and in some instances, monotomy/boredom
obtainable in some sectional operations marked with repetitive schedules. Risks to psychological health
due to work should be viewed in the same way as other health and safety risks and a commitment to
prevention of work related stress should be included in an organisation’s health and safety policies
(International Labour Organisation, 2017). Psychological stress and overload have been associated with
sleep disturbances, burn-out syndromes and depression (Eyayo, 2014).

Psychosocial hazards include, among others, stress, violence and substance abuse and there are
circumstances in which work can have adverse consequences for health and well-being (Ezejiofor,
2014). Epidemiological evidence exists of an elevated risk of cardiovascular disorders, particularly
coronary heart disease and hypertension in association with work stress (Eyayo, 2014). Risk to
psychological health at work may arise from organizational or personal factors, with the major factors
being poor design of work and jobs poor communication and interpersonal relationship, bullying,
occupational violence and fatigue (Ezejiofor, 2014).

There is high tendency for individuals under uncontrolled stress to violate laid-down regulations in the
performance of activities in their places of work and this situation, in turn, could lead to serious errors
that may have disastrous consequences for the workers and the industries (Nwachukwu and Nabofa,
2014).

Severe psychological conditions (psychotraumas) have been observed among workers involved in
serous catastrophes or major accidents during which human lives have been threatened or lost (Eyayo,
2014). High levels of stress are injurious to a worker’s health and productivity; workers must cope with

33
work in a petrochemical industry, which is naturally very stressful and it is therefore imperative that
functional and utilizable stress management systems be out in place so as to assist workers cope with
the highly stressful work conditions of the petrochemical (Nwachukwu and Nabofa, 2014).

According to the National Institute for Occupational Safety and Health in the USA causes of work
related stress are,

 Improper design of tasks, which implies heavy workload, infrequent rest breaks, long working
hours and shift work, hectic and routine tasks that have no inherent meaning, not utilizing
workers skills and there being little sense of control.
 Management style that is not transparent precludes participation of workers in decision making
and results in poor organisation of work and lack of family-friendly policies in the company.
 Career related anxieties that include, among other factors, job insecurity, lack of opportunity
for advancement or promotion, little recognition, as well as rapid changes for which workers
are unprepared.
 Strained interpersonal relations that are usually a sign of a poor social environment, lack of
support, communication and help from supervisors and co-workers.
 Conflicting and uncertain work roles, too much responsibility, too many hats to wear, whereby
individuals’ need for role clarity varies.
 Unpleasant or dangerous work environment such as overcrowding, excessive noise and air
pollution, or ergonomically inferior designed work places resulting in health problems.

Social conditions of work such as gender distribution and segregation of jobs and equality (or lack of)
in the workplace, and relationships between managers and employees, raise concerns about stress in the
workplace (Eyayo, 2014). Employees who make use of drugs or alcohol often do so in the misperception
that they help to reduce the stress of work, or for mood adjustment, performance enhancement, helping
to get over peer pressures, or socializing (International Labour Organisation, 2017)

Many service and public employees experience social pressure from customers, clients or public, which
can increase the psychological workload and measures for improving the social aspects of work mainly
involve promotion of open and positive contacts in the workplace, support of the individual’s role and
identity at work, and encouragement of teamwork (Eyayo, 2014).

Employees become used to their consumption as they belong to the habit – behaviours and if they are
not used, anxiety increases which further increased the stress level (International Labour Organisation,
2017). Stress management systems prevent a situation where workers are exposed to protracted

34
experience of stress, which may lead to frustrations and abnormal behaviour (Nwachukwu and Nabofa,
2014). Substance abuse generally leads to increased chances of accident, increase absenteeism, and
lower productivity and general performance of the company (International Labour Organisation, 2017).

According to Eyayo (2014) organisational psychosocial factors include but not limited to the following,

 Violence and aggression

 Lone working
 Shift and night work
 Long working hours
 Time zone changes

2.3 Leadership Elements

In this research there are five leadership process health and safety elements namely leadership and
commitment, training and competency, contractor management, asset integrity and effective
communication explored. According to Beriha et al.(2012) cited in Elssayed, Hassan and Hosny (2012)
occupational accidents can be reduced through effective preventative measures by investing on safety
equipment, training, and educating the employees, process design, and machinery.

2.3.1 Leadership Commitment

According to Hardy (2013) it is recognised that leadership is important in the creation of a culture that
supports and promotes a strong health and safety performance of an organisation. Employers and
employees with good safety behaviour are particularly playing a significant role in achievement of
safety compliance to occupational, safety and health improvement in industry (Fernandez, 2011;
Elssayed, Hassan and Hosny, 2012). According to Agumba and Haupt (2018), there are five health and
safety practices that will minimize health and safety incidents in small and medium enterprise projects
in the South African construction industry, namely,

 Upper management commitment and involvement in health and safety

 Employee involvement and empowerment in health and safety
 Project supervision
 Project health and safety planning

35
  Communication in health and safety, health and safety resources and training

It is imperative that leadership ensures that each employee is trained in an overview of the process and
in the operating procedures, emphasis on the specific safety and health hazards, emergency operations
including shutdown, and safe work practices applicable to the employee's job tasks (Hardy, 2013). One
of the fundamental points to note is that employers have a common law duty to ensure that a safe system
of work plan is in place, prior to the work being started on site (Spillane and Oyedele, 2013). Having
senior managers who take a proactive interest in establishing a health and safety culture has been
considered to be a key influence on organisational health and safety performance (Hardy, 2013).

The other element that requires leadership’s attention is contractor management, contractors working
in petrochemical industry face a great risk during maintenance work (Othman, Jabar, Murad &
Kamarudin, 2014). Safety behaviour depends directly on the communication that exists in the firm and
indirectly on management’s commitment (Fernandez, 2011; Elssayed, Hassan and Hosny, 2012). In
order to develop a good safety culture, attitude of the workers needs to be reoriented by applying best
practices, good housekeeping, change in work culture, and work practices (Beriha et al., 2012; Elssayed,
Hassan and Hosny, 2012).

2.3.2 Training and Competence

It is the responsibility of the employers to assure that the contractors who work in and around hazardous
chemicals have the appropriate skills and knowledge to perform those tasks without compromising
health and safety (Hardy, 2013). Process safety management is critical in the chemical process industry
and improving organisational knowledge and knowledge management capabilities is an important
means to prevent chemical accidents and improve organisations safety level (Chen, 2016).

Health and safety knowledge encompasses awareness of occupational health and safety risks, including
an evaluation of occupational health and safety programmes in an organisation (Okoye, Ezeokonkwo
and Ezeokoli, 2016). According to Hardy (2013), training provides employees with the knowledge and
tools to fully understand the risks in working with hazardous chemicals.

Knowledge is more than information, since it involves an awareness or understanding gained through
experience, familiarity or learning (Okoye, Ezeokonkwo and Ezeokoli, 2016). The employer must
obtain appropriate certifications from that contractor, and should screen the contractor to assure that
they can perform the tasks safety and the employer must manage the contractor to maintain control over
health and safety of the work (Hardy, 2013).

36
According Elssayed, Hassan and Hosny (2012) in order to commit top safety management, training
courses need to be further emphasised and improved to ensure better safety culture, performance and
involvement of all workers in top safety management of the company. The employer shall prepare a
record that contains the identity of the employee, the date of training, and the means used to verify that
the employee understood the training (Hardy, 2013).

The transient nature of the construction workforce makes it difficult to train workers effectively.
However, availability of trained or skilled workers directly contributes to the improvement of quality
of construction work, and indirectly contributes to improve the site safety (Vitharana, De Silva and De
Silva, 2015). All employees, including contractor employees and maintenance personnel, must be
trained to understand the material safety data sheets, as well as safe work practices and operating
procedures (Hardy, 2013). One of the leading methods to inform personnel of the various health and
safety woes present on each particular construction site is through the use of site specific inductions
(Spillane and Oyedele, 2013).

The employer, in consultation with the employees involved in operating the process, shall determine
the appropriate frequency of refresher training (Hardy, 2013). According to Vitharana, De Silva and De
Silva (2015) indicates that in the construction industry, workers have high mobility and they switch
from one organisation to another, frequently. A site induction provides the initial contact with new
individuals entering a construction site, the purpose of which is to provide the individuals with an
overview of the various site specific safety, health and welfare requirements and associated issues
(Spillane and Oyedele, 2013).

Training in emergency response, including conditions for evacuation and shelter in place and the use of
personal protective equipment is critical (Hardy, 2013).

2.3.3 Contractor Management

Contractors working in processing plants, apart from the complexity of construction working condition,
contractors face a great risk during maintenance tasks (Othman et al., 2014). Each contractor must be
trained in an overview of the process and in the operating procedures and the training shall include
emphasis on the specific safety and health hazards, emergency operations including shutdown, and safe
work practices applicable to the employee's job tasks (Hardy, 2013). Employees are exposed to a
number of inevitable hazards as most workers employed by contractors are unfamiliar with hazardous
materials and more work performed under high pressure (Othman et al., 2014).

37
According to Cooper (2006) cited in Spillane and Oyedele (2013) where there is lack of formal protocol
in the development and implementation of safe system of work plans, this can result in unsafe behaviour
while carrying out the hazardous task in question. Contractors could be exposed to a number of
inevitable hazards: large numbers of workers- mostly employed by the contractors who are unfamiliar
with the plant in a confined space; the presence of hazardous materials; large number of tasks performed
under high pressure, and in various weather conditions (Othman et al., 2014).

Where a safe system of work is void, there is an increased propensity for accidents or incidents to occur,
therefore, it is an essential requirement, both legally and productively (Cooper, 2006; Spillane and
Oyedele, 2013). Contractors and permanent employees are exposed to a number of inevitable hazards
as most workers employed by contractors are unfamiliar with hazardous materials and more work
performed under high pressure (Othman et al., 2014). It is vital that training is done and refresher
training shall be provided at least every three years, and more often if necessary, to each employee
involved in operating a process to assure that the employee understands and adheres to the current
operating procedures of the process (Hardy, 2013).

Contractors working in petrochemical industry face a great risk during maintenance work and the
number of workers involved in a processing plant maintenance shutdown can be anywhere between 700
and 3000 at peak time (Othman et al., 2014). According to Geldar (2010) cited in Elssayed, Hassan and
Hosny (2012) workplaces having joint health and safety committees with greater worker involvement,
both in numbers and executive capacity had lower injury rates. It is not surprising to find that the
majority of some contractors do not encompass safety costs in their tender and this seems to suggest
that these contractors find it difficult to develop and implement the most effective safety during the
construction phase of their projects, construction is a multi-stage process as it includes conceptual,
design construction, maintenance, replacement and decommissioning (Othman et al., 2014).

According to Loosemore and Lee (2002) cited in Spillane and Oyedele (2013) where misinterpretations
occur, there is a propensity for misguided instructions or signals, which may contribute or result in an
accident or incident on site. The main concern of a contractor is how to save money and reduce costs
and safety is usually considered a secondary priority in the company’s plans (Othman et al., 2014).

In most projects due to limited availability of resources and time constraints, contractors are forced to
select a subset of hazards recognition program elements for effective field implementation. However
little is known regarding the relative influence of available methods (Albert, Hallowell and Kleiner,
2014). The construction industry consists of various parties such as clients, designer, consultants,
general contractor, suppliers and subcontractors (Othman et al., 2014).

38
The clients who demand the lowest contract costs have influenced this scenario and the contractors
search for lower quality supplies and neglect safety issues (Othman et al., 2014). According to Albert,
Hallowell and Kleiner (2014) contractors often choose safety and hazard recognition program elements
based on their subjective intuition with little regard to relative effectiveness. An additional training for
the workers, which could be provided by contractors about equipment they use, before workers engage
in their duty, would also help to prevent accidents (Vitharana, De Silva and De Silva, 2015). As
contractors work under pressure to complete the task in a specified period, they may choose to take
short cuts by using inappropriate equipment for the job (Othman et al., 2014).

2.3.4 Asset Integrity

Asset integrity is a continuous process of knowledge and experience applied throughout the asset
lifecycle to manage risk from design, construction, installation, operation, maintenance and finally
abandonment phases of facility to maximize the benefit to the owner whilst safeguarding people, asset
and environment (Ramasamy and Yusof, 2015).

It is the duty of the leadership team to ensure assets are maintained effectively and are safe to be
operated at all times, mechanical equipment must be maintained to ensure that it will continue to operate
correctly and safely (Hardy, 2013). According to Hought, Fowler and Grindrod (2013) effective asset
integrity management programme is a prerequisite for continued safe operation of any chemical process
plant and the challenge is not only to ensure that containment systems remain intact through use of
appropriate inspection, testing, maintenance and repair strategies but that those strategies are
implemented by competent and motivated personnel, and that they remain suited to the equipment, age
and condition over time.

The objectives of asset integrity are compliance to all national regulatory bodies, company policies and
standards, stay fit for purpose during operations under all circumstances, ensure all assets operate in
safe manner and reliable within design parameters (Dutta and Madi (2014); Ramasamy and Yusof,
2015).

According to Hardy (2013) Process health and safety management standards require that an
organisation establish and implement written procedures for maintaining equipment such pressure
vessels, storage tanks, piping systems, pressure safety valves, emergency shutdown systems, and
controls (monitoring devices, sensors, alarms and interlocks). In the eyes of the regulators that
adherence to best practice in relation to process safety leadership, asset integrity management and
competence management the ageing plants and financial global competition is a problem and these
management systems are now hot topics and subjects for close examination by inspectors of high hazard
manufacturing sites (Hought, Fowler and Grindrod, 2013).

39
An effective quality assurance program must be implemented to assure conformance to standards and
codes, identify and record deficiencies and confirm that deficiencies have been corrected (Hardy, 2013).
According to Hought, Fowler and Grindrod (2013) it is becoming increasingly difficult to achieve
quality assurance programs for those companies that are having to operate plant well beyond planned
retirement dates, with fewer people, brought about by growing financial pressures in the face of fierce
global competition.

According to Hardy (2013) organisations must establish and implement written procedures to maintain
the on-going integrity of facility equipment and they should include the following,

 Test and inspections on equipment following recognized and generally accepted good
engineering practices, manufacturer’s recommendations and operating experience for the
conduct and frequency.
 Documentation of test and inspections – identifying date; name of the person performing test
and inspections; serial number or other identifier; description of the inspection or test; results.
 Equipment deficiencies – correct deficiencies in equipment that are outside acceptable limits
before further use or in a safe and timely manner when necessary means are taken to assure safe
operation.
 New Equipment – assure that equipment as it is fabricated is suitable for the process application
for which they will be used. Additionally, conduct appropriate checks and inspections to assure
that equipment is installed properly and consistent with design specifications and the
manufacturer's instructions.
 Material Control – assure that maintenance materials, spare parts and equipment are suitable
for the process application for which they will be used.

2.3.5 Effective Communication

According to Loosemore and Lee (2002) cited in Spillane and Oyedele (2013) communication is one
of the key components with regards to the management of personnel in the construction sector. Workers
on a work site will only be protected with adequate chemical hazard controls in place and the necessary
information about the chemicals they are working with (Safety Association for Canada upstream Oil
and Gas Industry, 2012).

According to Purohit et al. (2018) proper management of hazards sporadically identified in the
workplace can be done through effective process and individual or team who identified the hazard must
ensure proper communication of the hazard to the appropriate workplace authority (manager,

40
department head, or designated person). One of the key conclusions provided is that poor
communication within work teams contributed to incidents, therefore effective means of
communication, including verbally, visual and written must be adhered to at all times (Loosemore and
Lee, 2002; Spillane and Oyedele, 2013).

Effective communication from all levels in an organisation is critical (Safety Association for Canada
upstream Oil and Gas Industry, 2012). Petroleum refineries handle large quantities of hazardous
chemicals often at extreme conditions of temperature, concentration and pressure and any mis-operation
due to poor communication is prone to be a source of disaster that will cause heavy financial losses as
well as casualties (Thomas and Babu, 2014). According to the Safety Association for Canada upstream
Oil and Gas Industry (2012) it is vital that information feedback is available to the right people at the
right time means that an effective communication system needs to be in place.

According to Eyayo (2014) occupational health and safety is a means of protecting and maintaining the
physical, psychological and social health of workers and their families. The key to successful
development and implementation of a chemical management process is communication and the
systematic flow of information (Safety Association for Canada upstream Oil and Gas Industry, 2012).

2.4 Health and Safety Elements

Process health and safety elements investigated in this research include chemical exposure
management, health and safety risk assessment, incident investigation, emergency response and audit
compliance. According to Thomas and Babu (2014) petrochemical industry is a highly risk industry
that involves chemical reaction, hazardous material, flammable explosion and any other risk that can
occur to the permanent employees or the contractors. According to Kumar et al. (2017) occupational
health and safety should be an integral part of production processes on an organisation, any industrial
and production organisation in the oil and gas refinery should not be seen lacking in this area.

The primary objective of health and safety management system is to identify the major hazards
associated with work activities and to ensure that appropriate controls are in place before work
commences (Spillane and Oyedele, 2013).

Occupational health and safety can be viewed as the study of factors or conditions influencing the health
and well-being of workers not only in the place of work but also at home with the aim of promoting
health, safety and welfare of the workers and their family (Eyayo, 2014). It must be noted though that
health and safety incidents still happen in the work place and it is important to investigate those incidents

41
and appreciate root causes so that corrective actions can be assigned to employees that will close them
and the intention is to prevent the similar incidents from happening again (Kumar et al., 2017).

According to Abdelhamid and Everett (2000) occupational accidents occur due to one or more of the
following,

 Failing to identify an unsafe condition that existed before an activity was started or that
developed after an activity was started.
 Deciding to proceed with a work activity after the worker identifies an existing unsafe
condition.
 Deciding to act unsafe regardless of initial conditions of the work environment.

2.4.1 Chemical Exposure Management

Chemicals can be classified on the base of hazards and the Globally Harmonised System divides
hazardous chemicals in the workplace into different categories; physical hazards, health hazards and
environmental hazards (Naafs, 2018). According to Reddy and Yarrakula (2016) chemical units include
a wide range of hazards arising from the process itself, properties of the chemical and their handling,
such as fire, explosion and exposure to toxic substances. However, there is a still real potential of major
industrial accident with catastrophic impact. If avoidance of ignition sources is to be safely applied it is
vital to be fully conversant with the details of plant and operations and it can also be important to be
aware of material handling properties which are outside the scope of normal hazardous properties, but
can affect what occurs on plant (Puttick, 2008).

Process safety management is critical in the chemical process industry and improving organisation
knowledge and knowledge management capabilities is an important means to prevent chemical
accidents and improve organisations’ health and safety level (Chen, 2016). Not surprisingly, noise –
induced hearing represents the most frequent occupational disease (25.3 %) in the petrochemical
industry followed by the musculoskeletal diseases with (22.9%). Malignant tumors of the pleura and
peritoneum follow with a proportional rate of 19%, six times higher than that recorded for the total
industrial sectors (3.6%). Disease of the respiratory system are clearly proportionally more frequent
(16.5%) compared to data reported from the total industrial sector (6%) (Naafs, 2018).

The major consequences of chemical disasters include impact on livestock, flora/fauna, the environment
(air, soil, and water) and losses to industry (Reddy and Yarrakula, 2016). Chemical process hazards at
a chemical plant can give rise to accidents that affect both workers inside the plant and members of the
public who reside nearby (Chen, 2016). Biological monitoring of urine and blood provides an indication

42
of exposure to chemicals in the workplace and is key part of health surveillance and exposure
assessment (Lankulsen, Vichit - Vadakan and Taptagaporn, 2011).

According to Reddy and Yarrakula (2016) chemical disaster, though low in frequency but has the
potential to cause significant immediate or long term damage, frequency and severity of chemical
disasters has increased in last few years due to rapid development of chemical industries of a wider
range. Urinary phenol is used as a biomarker of benzene exposure in some companies (table 4).
However, urinary trans,trans-muconic acid (t,t-MA) or S-phenylmercapturic acid (S-PMA) is used as a
biomarker of benzene exposure for much more reliable and sensitive occupational investigations
(Lankulsen, Vichit - Vadakan and Taptagaporn, 2011). A small accident occurring at the local level
may be a prior warning signal for impending and the disaster (Reddy and Yarrakula, 2016).

Figure 2.2 Impact of Chemical Disasters

2.4.2 Health and Safety Risk Assessment

Risk assessment is the evaluation of hazards to determine their potential to cause an accident.
Identifying health and safety hazards in order to prevent and control them is very imperative to the
health and well-being of the workers (Eyayo, 2014). In reality, risk assessment is something people do
each and every day without giving it much thought, case in point, when driving, people assess the
condition and circumstances of the roadway and adjust speed accordingly, or when caught in a rainstorm
on the golf course, they assess the weather to determine if they play on or seek cover (Dunbar, 2014).

43
Hazard identification and risk assessment is carried for identification of undesirable events that can lead
to a hazard, the analysis of hazard of this undesirable event, that could occur and usually the estimation
of its extent, magnitude and likelihood of harmful effects (Purohit et al., 2018).

When most people hear the term risk assessment they immediately think of insurance and
indemnification (Eyayo, 2014). According to Kumar et al. (2017) health and safety risk assessment is
a management tool that allows the workplace comply with her occupational policy, helps the workers
do their jobs without damage to their health, enables the workplace meet her legal responsibilities,
enables the workplace show due diligence in the protection. Promotion of health and safety of the
workers, provides an auditable platform and involves the work force in protecting the health and safety
of the workers. It is relatively easy for investigation programs to focus primarily on making changes to
or correcting the physical environment, equipment, tools and machinery that may have contributed to
the safety incident (Wachter and Yorio, 2014).

According to the Safety Association for Canada upstream Oil and Gas Industry (2012) risk assessment
reports should reflect the detail of the assessment and provide sufficient information to show how the
decisions about risk and controls were made. The management of health and safety at work require
employers to make a suitable and sufficient assessment of the health and safety risks to employees and
non-employees, arising from their work activities (Kumar et al., 2017). According to Wachter and Yorio
(2014) these physical factors are typically easy to visually identify, understand and correct.

According to Dabup (2012) the risk assessment provides a systematic approach for the identification,
management and reduction of the risk to an acceptable level. Risk assessment is now a common
requirement of all health and safety legislation, the emphasis is now on preventing accidents and work
related ill health, rather than just reacting to incidents, and making improvements after the event (Kumar
et al., 2017). According to Albert, Hallowell and Kleiner (2014) a critical component in health and
safety risk management is to adequately identify hazards and mitigate its associated risk using safety
program elements.

According Dunbar (2014) there are six-steps in the process of risk assessment, namely;

 Identification of a hazard
 Identification of the associated risk
 Assessment of the risk, which includes
o The likelihood
o The severity
o Assigning a priority for correction

44
  Control of the risk, which includes
o Elimination
o Engineering a barrier
o Administration controls
o Personal protection equipment
 Documentation of the process
 Monitoring and review of the process.

According to the Safety Association for Canada upstream Oil and Gas Industry (2012) hazard
assessment records should include the following,

 The name of the assessor or assessment team personnel.

 Description of normal operations in the work area.
 Procedures used to assess exposure.
 Description of hazard and routes of entry to the body.
 Procedures used to assess existing control measures
 Conclusion from the assessment about whether the risk was significant or not.
 Action to be taken, including induction, training, emergency procedures, and health
surveillance.
 The circumstances in which reassessment will be required.
 Signature, date, and position of assessor or assessment team.
 Signature, date, and position of the relevant person accepting the assessment.

According to Dabup (2012) risk assessment is a critical step in risk management and if done correctly,
it determines the minimum level of preparedness in order to respond effectively and it involves applying
qualitative or quantitative techniques to potential risks and it reduces the uncertainties in measuring risk
and it usually involves frequency and severity. Hazard identification and risk analysis is a collective
term that encompasses all activities involves in identifying hazards. Evaluating risk at facilities,
throughout their life cycle, to make certain that risks to employees, the public or the environment are
consistently controlled within the organisations risk tolerance level (Purohit et al., 2018).

According to Dabup (2012) different outcomes from risk assessments are, namely;

 Risk Aversion: This involves a conscious commitment and decision on an organisation’s part
to avoid completely a particular risk by discontinuing the operation producing the risk and it
presupposes that risk has been identified and evaluated.

45
  Risk Acceptance: This involves creating decision tables or standards for deciding what risks are
acceptable for individuals, organisations or the society. What is acceptable may differ for each
group.
 Risk Retention: The risk is retained in the organisation where any consequent loss is financed
by the organisation.
 Risk Transfer: this refers to the legal assignment of the cost of certain potential losses from one
party to another. The most common way of affecting such transfer is by insurance. Hence an
insurance company undertakes to compensate the insured organisation against losses resulting
from the occurrence of an event specified in the insurance policy.
 Risk Reduction: A goal of zero injuries is a challenge, but may not be credible since risk can
only be reduced and not completely eradicated. Organisation can design a management system,
which will reduce or eliminate all aspects of accidental loss that lead to wastage of an
organisation’s assets. It relies on the decline of risk within the organisation by the
implementation of loss control programme whose basic aim it to protect the company’s assets
from wastage caused by accidental loss.

2.4.3 Incident Investigation

According to the OSHAct 85 of 1993 an incident investigation team shall be established and consist of
at least one person knowledgeable in the process involved, including contract employee if the incident
involved work of the contractor, and other persons with appropriate knowledge and experience to
thoroughly investigate and analyse the incident. According to Hardy (2013) organisations must have an
active, aggressive incident evaluation program to identify the underlying causes of these incidents and
break the sequence of events that can lead to an accident and such incidents include those that could be
called a near miss and incident or problem investigation should be factored back into the hazard
analysis.

According to Bond (2007) the sole objective of the investigation of an accident or incident under
regulations shall be the prevention of accidents and incidents. According to Oakley (2003) cited in
Wachter and Yorio (2014) accident investigation programs can differ in the actual content that is
focused on during the investigation. A failure to investigate the incidents and fix the root cause allows
the opportunity for the incidents to reoccur (Hardy, 2013).

According to Wachter and Yorio (2014) more than 80% of accidents are still attributed to unsafe acts,
it is critical that accident investigations effectively explore the reasons a worker’s behaviour or
performance led to an incident in an effort to correct those management system deficiencies and some
of these reasons could be lack of knowledge, lack of motivation or job distractions that caused unsafe

46
behaviours and human error to occur. ISO 18001 suggest that the process of investigating incidents
should provide the collective with overlapping knowledge sets on what defined an incident for the
organisation and the types of corrective actions that are applicable and it further suggests that the
process must be impartial and objective.

As a primary element to any investigation, the exact casual factors that led to the safety incident must
be determined in order to effectively make changes that mitigate future risk (Oakley, 2003; Wachter
and Yorio, 2014). It shall not be the purpose of such an investigation to apportion blame or liability and
organisations should focus on preventing accidents, not just reporting problems, and this requires root
cause analysis (Bond, 2007). According to Wachter and Yorio (2014) accident investigations can vary
in the length of time between the time the incident occurred and the time investigation is initiated.

According to Lerman et al. (2012) when evaluating an incident that may be due to fatigue, the two
major steps are first to evaluate if the individual was susceptible to fatigue and second to evaluate
whether the performance, behaviour, and details of the event would be consistent with inaction or
inattention.

According to ISO 18001 the overall impact of accident investigation characteristics on safety
performance statistics is relatively minor. A major reason to investigate quickly is to accurately record
all the factors that contributed to the incident (Wachter and Yorio, 2014). Logically, there can be a
resulting difference in investigation quality between investigations that are limited immediately
following an incident and those that are initiated days following the incident (Wachter and Yorio, 2014).

Human error is the starting point of an investigation, not the end point and to do something about error,
one must look at the system in which people work and focus must extend past “what occurred” to why
do it occur (Bridger, 2015). Determining that the individual was susceptible to fatigue should not
support fatigue as a cause or contributing factor to the incident unless the behaviours are also consistent
with fatigue (Lerman et al., 2012).

According to Vitharana, De Silva and De Silva (2015) failure to appoint a safety officer is often
identified as a cause of scarcity of site safety. However effects of safety attitude of workers and safety
training has not often been studied, although no willingness to follow safety norms by workers is
identified as a cause of poor safety practices.

47
2.4.4 Emergency Preparedness and Readiness

The first step for emergency preparedness and maintaining a safe workplace is defining and analysing
hazards (Purohit et al., 2018). According to Hardy (2013) organisations must plan for an emergency
and be prepared to respond and no matter how hard organisations try to build their systems safety,
designs will be flawed, people will make mistakes, components will fail, software will do the
unexpected, and environment conditions will exist that are beyond the company’s control.

According to Stojkovic’ (2013) practical exercises for members of the rescue group for casualty rescues
in confined spaces must be maintained regularly and within the deadlines. Organisations must plan for
an emergency and be prepared to respond, employers must develop an emergency action plan that
includes evacuation and shelter in place instructions and training in the use of personal protective
equipment (Hardy, 2013).

In case a worker in a confined space is threatened, the worker who observes should not enter unless the
emergency alert informed other workers (Stojkovic’, 2013). Employees must be trained to this plan for
it to be effective, and alarm systems should be implemented to warn employees that emergency
conditions exist (Hardy, 2013). All appropriate personal protective equipment, such as protective
breathing masks, protective zones with associated rope, lamp, etc. must be ready and near the entrance
to the confined space (Stojkovic’, 2013). Training must accompany the operating procedures, with an
emphasis on what employees should do in case of emergency (Hardy, 2013).

According to Stojkovic’ (2013) a worker-observer should act in accordance with predetermined rescue
mode, as follows,

 To immediately alert the surrounding workers, rescue group, party fire and health centre.
 To slip inside the tube inlet for clean air, thereby providing increased ventilation and closed the
endangered area.
 Worker, which should indicate a vigilant eye, cannot enter into hazardous areas without
adequate means of personal protection and if not provided workers who will watch from the
doorway.
 After the rescue worker enters, other workers outside buildings urgently need to prepare all the
aids to pull the killed and fingering first aid.
 Workers, observers must continuously through the inlet view monitor employee-rescuers, and
in the case of invisibility must be with him regarding using agreed signals.

48
  If the injured worker is unconscious, it is necessary to immediately proceed with artificial
respiration mouth to mouth, or using breathing apparatus. If there is a respirator at hand, in
these cases the doctor does not have to always be near. When CPR must be renewed and during
transport the injured health centre or hospital, and artificial respiration was discontinued as soon
as the victim for consciousness returned.

2.4.5 Audit Compliance

Audit compliance is a proactive attempt to identify gaps to comply and intervene when non-compliance
is ascertained (Hardy, 2013). Once the unsafe acts are discovered, then investigations are needed to
understand why these behaviours occurred that led to human performance errors that often involve
psychology and employee perception factors (Wachter and Yorio, 2014). Compliance audits provide a
means for assuring that the procedures and practices in process health and safety management systems
are being followed and are adequate (Hardy, 2013).

Compliance audits must be conducted at specific periods depending on which tier, 1, 2 or 3 (Hardy,
2013). Workplace hazards contribute significantly to the overall population’s morbidity, mortality,
financial and social costs, which are all principle reasons for government, private and public sector
support of occupational health and safety (Ford, Haskins and Wade, 2014). The audit must be conducted
by a trained individual or team, and the auditing effort should be planned to ensure success (Hardy,
2013).

2.4.6 Personal Protective Equipment

Personal protective equipment means any equipment which is intended to be worn or held by a person
at work and which protects him/her against one or more risks to his/her health or safety and any
additional accessory designed to meet that objective (Purohit et al., 2018). Using personal protective
equipment is one of the health and safety measures, and is often regarded as the last resort,
supplementary to control measures, in providing protection to the employees (International Labour
Organisation, 2017). According to Vitharana, De Silva and De Silva (2015) dislike to wear personal
protective equipment, which are categorised under safety equipment, is often identified as the cause of
poor safety practices. Personal protective equipment is usually chosen to provide protection appropriate
to each of type of hazard present (Purohit et al., 2018).

49
Low level of awareness toward using personal protective equipment is also frequently identified as a
possible cause of poor safety practices (Vitharana, De Silva and De Silva, 2015). Lack of awareness
about site safety and dislike to wear personal protection equipment are main causes of poor safety
practices in construction sites (International Labour Organisation, 2017). Workers must be trained to
use and maintain personal protective equipment properly (Purohit et al., 2018). Unavailability of
personal protective equipment also contributes to poor safety practices (Vitharana, De Silva and De
Silva, 2015).

Figure 2.3 Hierarchy of Hazard Controls

According to the Safety Association for Canada’s upstream oil and gas industry (2012) the hierarchy
of hazard controls is as follows:

 Eliminate or substitution: controls the chemical hazard by removing the chemical outright from
the work site. Processes are avoided or adjusted to eliminate the need for the chemical, or a
safer alternative is used in place of a more hazardous chemical. If elimination or substitution is
not possible, engineering controls are the next possible choice.
 Engineering controls: These are used to remove a hazard or place a barrier between the worker
and the hazard. Engineering controls should always be considered first, where elimination or
substitution is not possible. Administrative controls are often used together with engineering
controls. The basic types of engineering controls are process controls, enclosure and/or isolation
of emission source, and ventilation.
o Basic engineering controls are methods that keep the chemical “in” and the worker
“out” (or vice versa). Examples include basic ventilation or isolation processes that can
be done on the spot without assistance, such as opening a window or door.

50
 o Advanced engineering controls are methods that are built into the design of a plant,
equipment, or process to minimize the hazard. They include designs or modifications
to plants, equipment, ventilation systems, and processes that reduce the source of
exposure.
 Administrative controls: Involve the work process and worker, and include such measures as
company policies, safe work procedures, training, work rotation, and signage.
 Personal protective equipment: this is the last line of defence against the hazard, and is used
where the hazard cannot be eliminated or sufficiently reduced by engineering or administrative
controls. Personal protective equipment does not remove the hazard; it only inserts a barrier
between the work and the hazard. Personal protective equipment includes but is not limited to
specified protective clothing and respiratory protective equipment.

The employer and workers must understand the limitations of the personal protective equipment
(Purohit et al., 2018). Personal protective equipment include hard hats, safety glasses and goggles,
safety shoes, hearing protection, respiratory protection, and protective clothing such as fire resistant
clothing where required and in addition, procedures should be established to assure compliance with
applicable regulations and standards such as hazard communications, confined space entry and process
health and safety management (Kulkarni, 2017). Personal protective equipment and clothing is used
when other controls measures are not feasible and where additional protection is needed (Purohit et al.,
2018).

2.4.7 Fitness to Work

The increasing healthcare cost and workers’ compensation expenses relating to modifiable health
behaviour such as alcohol use, smoking and obesity, occupational health professionals are moving
toward occupational health programs that not only address workplace hazards, but also support health
lifestyle behaviours (Ford, Haskins and Wade, 2014). Problematic substance use is a social care issue
and it is not just a health or criminal justice issue (Galvani, 2015).

People start using for many reasons, among these are experiences of sexual and domestic violence and
abuse, feelings of low self-esteem and self-worth, as well as mental and physical distress (Galvani,
2015). People with problematic substance use will often be socially isolated and are likely to have a
number of co-existing social and health care problems (Galvani, 2015).

According Ford, Haskins and Wade (2014) given the increase in the rate of obesity and the fact that
employed adults spend a quarter of their lives at work, the contribution of obesity to morbidity and

51
mortality in working populations is epidemiologically, productively and economically significant.
Economically, obesity directly contributes to the rising cost of health insurance as well as workers’
compensation expenses (Ostbye, Dement and Krause, 2007).

A NIOSH report reviewed studies that examined the association between long working hours and
illnesses, injuries, health behaviours, and performance. In general, overtime was associated with poorer
perceived general health, unhealthy weight gain, increased injury rates, more reported illnesses,
increased mortality, and poorer neuropsychological test performance (Lerman et al., 2012).

2.4.8 Fatigue Management

The organisation must arrange schedules of work that provide sufficient opportunities for rest, training
to support fatigue management, and procedures for monitoring and managing fatigue within the
organisation (Lerman et al., 2012). The major cause of fatigue is not having obtained adequate rest or
recovery from previous activities and in simple terms, fatigue largely results from inadequate quantity
or quality of sleep (Canada Aviation Industry, 2007).

The potential benefits of a nap in improving alertness and performance during routine operations, with
a resulting increase in health and safety margin, may outweigh the potential negative effects of a short
period of sleep inertia (Lerman et al., 2012). The quality (how good) and the quantity (how long) of
sleep are important for recovery from fatigue and maintaining normal alertness and performance
(Canada Aviation Industry, 2007). Variation in work activity across a shift can help to relive fatigue,
especially where the worker has a range of tasks to complete, each with differing mental and physical
demands and rotating routine sedentary mental tasks with physical tasks can promote alertness or
conversely help to relieve physical fatigue (Lerman et al., 2012).

According to Canada Aviation Industry (2007) we live in a 24 hour society where many different work
patterns have developed beyond the traditional Monday to Friday 9 to 5 routine. Safety and productivity
in the workplace are intimately related to worker health (Lerman et al., 2012). An increasing proportion
of the workforce is engaged in shift work and non-traditional schedules and between 15-30 percent of
the workforce of industrialised countries is engaged in shift work (Canada Aviation Industry, 2007).

According to Lerman et al. (2012) rates of absenteeism in extended hours industries range from 6% to
12% compared with the national average of about 2% and an absent worker in a 24/7 operation must
either be replaced (often with someone paid overtime) or co-workers must pick up the slack (potential

52
for fatigue, safety and morale problems). Fatigue is an experience of physical or mental weariness that
results in reduced alertness (Canada Aviation Industry, 2007).

2.4.9 Housekeeping in Workplace

According to Ness (2015) effective housekeeping can eliminate some workplace hazards and assist in
getting a job done healthy and safe. The unsafe conditions are usually present until corrected and they
are unlike unsafe acts that require “catching” workers committing unsafe acts, which could be sporadic
over time and place (Wachter and Yorio, 2014).

According to Health and Safety Ontario (2011) good housekeeping at work benefits both employers
and employees alike and good housekeeping can achieve the following,

 Eliminate clutter which is a common cause of accidents, such as slip, trips and fall, fires and
explosions.
 Reduce the chances of harmful materials entering the body (e.g., dusts, vapours)
 Improve productivity (the right tools and materials for the job will be easy to find)
 Improve your company’s image (good housekeeping reflects a well-run business. An orderly
workplace will impress all who enter it – employees, visitors, customers, etc.)
 Help your company to keep its inventory to a minimum (good housekeeping makes it easier to
keep an accurate count of inventories).

One of the most common findings in workplaces is poor housekeeping i.e. untidiness, disorder, poor
storage of materials and stock (Ness, 2015).

According to Health and Safety Ontario (2011) there are signs of poor housekeeping and they are,

 Cluttered and poorly arranged work areas

 Untidy or dangerous storage of materials (e.g., materials stuffed in corners, overcrowded
shelves, etc.)
 Dusty, dirty floors and work surfaces
 Items that are in excess or no longer needed
 Blocked or cluttered aisles and exits
 Tools and equipment left in work areas instead of being returned to roper storage places
 Broken containers and damaged materials

53
  Overflowing waste bins and containers
 Spills and leaks

The smallest accidents can happen from the most innocuous activity and it is very important to keep the
work area tidy from debris and from the materials and substances that are part of the everyday work
process (Ness, 2015).

2.4.10 Noise Exposure Management

Exposure to noise for an extended period of time may produce subjective feelings of unpleasantness
and an increase in complaints of fatigue (Lerman et al., 2012). Occupational noise exposures causes
between 7 and 21 percent of the hearing loss among workers in general lowest in industrial countries,
where the incidence is going down and highest in the developing countries (Naafs, 2018).

Noise may ameliorate performance decline due to circadian cycle effects by increasing the general level
of arousal and this effect is dependent on variables such as the nature of the noise, the nature of the task,
the time of day, and on personal factors (Lerman et al., 2012). According to Naafs (2018) hearing loss
due to noise exposure in the workplace is a significant health problem with economic consequences.

Noise induced hearing loss is the occupational disease most frequently reported to the petrochemical
safety authority (Naafs, 2018). Studies of the effects of noise on performance have been primarily
conducted in laboratory settings and sound can be sedating (continuous droning/humming) or
stimulating (music of varying tempos, conversation) in its effect (Lerman et al., 2012).

It is difficult to distinguish between noise induced hearing loss and age-related hearing loss at an
individual level and most of the hearing loss is age-related but men lose hearing more than women do
(Naafs, 2018).

2.5 Technical Elements

Five technical elements are considered important in this research, namely, management of change,
process hazard analysis, process health and safety information, design integrity and human errors.

54
2.5.1 Management of Change

Organisations should not assume that these minor changes have no impact on health and safety and
many accidents have resulted from small changes that did not appear to have an effect on health and
safety prior to the incident (Hardy, 2013). According to the Safety Association for Canada upstream
Oil and Gas Industry (2012) employers must maintain own management of change process but must
also consider chemical management and it is important to note that temporary as well as permanent
changes must be considered. The changes include modifications to process equipment, procedures and
processing conditions and these changes must be thoroughly evaluated to assure that health and safety
in the process industry is maintained (Hardy, 2013).

The changes must be thoroughly evaluated to assure that health and safety in the process industry is
maintained and the process health and safety information is critical to assist in identifying the hazards
and risks associated with the process (Hardy, 2013). Accidents can take place even where process health
and safety management systems exist and the probability of such occurrence increase if documentation
is deficient (Tzou et al., 2004).

Information related to toxicity, permissible exposure limits, physical data, reactivity, corrosivity,
chemical stability, thermal stability, and hazards associated with inadvertent mixing should be
appreciated by all involved (Hardy, 2013). According to Tzou et al. (2004) managing health and safety
related information inadequately has been cited as a significant factor to industrial accidents.
Information on the chemicals used or produced, technology and equipment should be understood by
everyone working in the process plant (Hardy, 2013).

The other vital information is material safety data sheets, maximum upper/lower limits in material of
construction, relief systems, safety systems, and ventilation systems (Hardy, 2013).

Figure 2.4 ARCO Channelview Compressor Explosion

55
According to Ness (2015) an explosion in ARCO Channelview took place July, 1990. A waste water
tank exploded during the restart of a compressor. The nitrogen purge had been significantly reduced
during maintenance, and a temporary oxygen analyser failed to detect the build-up of a flammable
atmosphere in the tank. When the compressor was restarted, flammable vapours were sucked into the
compressor and ignited. The flashback of the flame into the headspace of the tank caused an explosion
that killed 17 people and damages were estimated to be $100 million. The waste water tank was not
considered part of the operating plant. Hence, the management and workers did not understand that a
chemical reaction was taking place in the tank, generating oxygen. The lack of understanding enabled
a series of poor decisions, such as discontinuing the nitrogen purge, poor design and location of the
temporary oxygen probe, no management of change review of the decisions, and no pre-startup safety
review. Key lessons in this incident are as follows,

 Ensure that proper management of change procedures are followed before any maintenance
work is performed.
 In this incident, the workers did not know that a chemical reaction that could produce oxygen
build up was taking place in the tank. Therefore, they did not comprehend the importance of
continuing an effective nitrogen purge.

According to the Safety Association for Canada upstream Oil and Gas Industry (2012) the following
changes related to chemical hazards are examples of changes that would require a technical
management of change, namely;

 Replacing an original chemical with one from a different chemical supplier or following
different material specifications.
 Changing equipment temporarily or permanently, which adds new exposure points e.g., adding
tanks, vessels, or block/bleed valves or replacing old equipment with a different type.
 Changing the operating temperature and pressure range to exceed design and planned operating
limits.
 Changing or creating products at the work site.
 Transforming or changing the formation of chemicals (may affect compatibility, flammability,
toxicity).
 Using materials that are changed in a formation or produced in a process.
 Field – mixing or blending chemicals.
 Changing operating procedures that impact chemical exposure.

56
2.5.2 Process Hazard Analysis

According to Hardy (2013) process hazard analysis is defined as a systematic approach for identifying,
evaluating and controlling the hazards of processes involving highly hazardous chemicals. The purpose
of hazard identification is to highlight the critical operations of tasks, that is, those tasks posing
significant risks to the health and safety of employees as well as highlighting those hazards pertaining
to certain equipment due to energy sources, working conditions or activities performed (Purohit et al.,
2018).

The process hazard analysis is vital to the health and safety efforts as it provides information to assist
management and operations team to improve health and safety by reducing risk (Hardy, 2013). Analyses
examine the system, subsystems, components, and interrelationships, the elements involved are
training, maintenance, operational and maintenance environments and system/component disposal
(Department of Transport [USA], 2000).

Team findings and recommendations must be documented with resolutions and actions communicated
to operations and maintenance along with a written schedule of when these actions are to be completed
(Department of Labour [USA], 2016). Commonly used study methodologies are hazard identification
(HAZID), hazard and operability (HAZOP), What-If analysis, safety integrity level (SIL), failure mode
and effects analysis (FMEA) and layer of protection analysis (LOPA).

According to Department of Transport [USA] (2000) steps in performing a hazard analyses are,

 Describe the system.

 Perform functional analysis if appropriate to the system under study.
 Develop a preliminary hazard list.
 Identify contributory hazards, initiators, or any other causes.
 Establish hazard control baseline by identifying existing controls when appropriate.
 Determine potential outcomes, effects, or harm.
 Perform a risk assessment of the severity of consequence and likelihood of occurrence.
 Rank hazards according to risk.
 Develop a set of recommendations and requirements to eliminate or control risks.
 Provide managers, designers, engineers, and other affected decision makers with the
information and data needed to permit effective trade-offs.
 Conduct hazard trading and risk resolution of medium and high risks. Verify that
recommendations and requirements identified have been implemented.

57
  Demonstrate compliance with given health and safety related technical specifications,
operational requirements and design criteria.

To prevent or minimize process health and safety incidents in a petrochemical industry process hazard
analysis is conducted by a team with expertise in engineering and process operations, including at least
one employee who has experience and knowledge on the system (Hardy, 2013). A non-critical accident
that has a realistic chance of occurring may not require further study and frequency may be characterised
qualitatively by terms such “frequent” or “rarely” (Department of Transport [USA], 2000). The process
hazard analyses require process health and safety information to be clear and understood by all team
members involved (Hardy, 2013).

2.5.3 Process Health and Safety Information

According to Chen (2016) knowledge is critical organisation asset which will create value for improving
organisation competitive advantages and safety level. The relationship between knowledge and
information is interactive; health and safety knowledge therefore, encompasses awareness of
occupational health and safety risks, including an evaluation of occupational health and safety
programmes in an organisation (Okoye, Ezeokonkwo and Ezeokoli, 2016). Among knowledge
dimension, there are two kinds of knowledge, explicit knowledge and tacit knowledge, explicit
knowledge consists of facts, rules, relationships, and policies that can be faithfully codified in paper or
electronic form and shared without need for discussion (Chen, 2016).

Tacit knowledge represents knowledge based on the experience of individuals and tacit knowledge is
knowledge housed in the human brain, such as expertise, understanding, or professional insight formed
as a result of experience (Chen, 2016). Tacit knowledge is highly personal, context-specific, and
therefore hard to formalise and communicate (Chen, 2016).

According to Tzou et al. (2004) managing health and safety related information inadequately has been
cited as a significant factor to industrial accidents. There is clear distinction between data, information
and knowledge and data has commonly been seen as simple facts that can be structured to become
information (Chen, 2016). Accidents can still take place where process safety management systems
exist and the probability of such occurrence increase if documentation is deficient (Tzou et al., 2004).

Human behaviour is a major contributor to occupational health and safety issue cannot be tackle
effectively without interference of employers with a particular pattern of behaviours as important
criteria needed to change employee’s behaviours (Zin, 2012). To enhance safety practices, once of the

58
major needs in the construction industry is to enhance professionals’ interest in active safety
management and implementation of awareness programs, which must be developed and implemented
among construction workers (Vitharana, De Silva and De Silva, 2015).

According to Chen (2016) knowledge is one of the most important sources to prevent accidents and
guarantee process safety in the chemical process industry and there are numerous definitions of
knowledge in the knowledge domain. Awareness on possible risk factors and knowledge on how to
reduce these risk factors among workers and contractors will enhance site safety (Vitharana, De Silva
and De Silva, 2015).

Knowledge is a fluid mix of framed experience, values expertise, contextual information and insight
that provides a suitable environment and a structure for evaluating and incorporating new information
and experiences (Chen, 2016). Sources of health and safety knowledge include incident investigation,
teamwork, collaborations, and survey of safety culture (Okoye, Ezeokonkwo and Ezeokoli, 2016).
Knowledge is more than information, since it involves an awareness or understanding gained through
experience, familiarity or learning (Chen, 2016).

2.5.4 Design Integrity

Design integrity is assurance and verification function that ensure a product, process, or system meets
its appropriate and intended requirements under stated operating conditions (Hardy, 2013). According
to Baby (2008) cited in Ramasamy and Yusof (2015) design integrity provides assurance that facilities
are design in accordance to governing standards and meet specified operating requirements without
compromising on health and safety, accessibility, operability and maintainability

According to Duguid (2008) the most frequent specific problems during design for safety inadequate
are, namely;

 Alarms to detect high temperatures at critical locations or loss of cooling, particularly where
runaway reactions are possible. Also lack of skin thermocouples to detect high temperatures in
fired heater tubes.
 High level alarms/shutdowns totally independent of the normal level control system to
minimize the risk of overflow from tanks or vessels.
 Spring closed valves to remind operators to remain in attendance while draining water bottoms
from equipment containing flammables or toxics to an open drain. Also installing a totally

59
 independent low interface level shutdown system where such drainage is under automatic
control.
 Over reliance on check valves to prevent reverse flow.
 Inadequate attention to design for safety when implementing modifications to plant or
operation.

Process design and procedures often change in a petrochemical industry and the associated risks may
increase (Hardy, 2013). Most human machine interface design principles and guidelines focus on
aspects of usability, whilst usability is important for reducing the incidence of human error, more is
required of a safe human machine interface (Hardy, 2013).

2.5.5 Human Factor

The human element of the system has one of the biggest potentials for either causing or preventing an
accident and safe job performance by operating, maintenance personnel and contractors has a
tremendous positive impact on health and safety (Karthika, 2013). Complex interaction between
humans and machines is limited by the fact that whereas humans have natural intelligence, which
enables us to interpret situations according to the context, this ability is absent in most machines and
very restricted in even the most advanced (Hardy, 2013).

In general, software does not allow machines to adapt to unforeseen conditions, so computers are limited
in their actions and cannot adapt to given situations (Hardy, 2013). Employers and employees with good
health and safety behaviour are particularly playing a significant role in achievement of safety
compliance to occupational health and safety improvement in industry (Zin, 2012). According to Bond
(2007) it is clear that in the vast majority of accidents or near misses employees were acting with the
best intentions and did not expect a serious event to occur from their actions. Human error is not the
sole cause of failure, but it is a symptom of a deeper trouble (Bridger, 2015). It has to be accepted that
to err is human but one can learn from these events and share with others the lessons learned (Bond,
2007).

2.6 Operational Elements

Operational elements include pre-start up and shutdown reviews, operating procedure, control of
ignition source, control of confined space entry and permit to work (Hardy, 2013). The evolution of

60
petroleum refining from simple distillation to today’s sophisticated processes has created a need for
health and safety management procedures and safe work practices (Kumar et al., 2017).

Often accidents occur in the transition between operational phases, rather than when the process system
is up and running in “steady state” mode, start up and shutting down new and existing process systems
can be hazardous because changes to design or operations may be made in real time to meet schedule
temperatures and pressures, potentially introducing new hazards (Hardy, 2013). Hazard identification
and risk assessment studies performed once a process is near start up, during operation or before
decommissioning are typical done in a plant environment (Purohit et al., 2018).

A pre-start up and shutdown review is a valuable tool to assure that operating procedures are in place,
hazards are understood, engineering drawings are updated, and emergency shutdown procedures have
been communicated accordingly (Hardy, 2013). Human exposure to ammonia may cause the short term
health effects like burn the skin and damage eyes, inhaling ammonia can irritate the nose and throat
causing coughing and wheezing and higher exposures may cause a build-up of fluid in the lungs
(pulmonary oedema) with severe shortness of breath (Ezejiofor, 2014).

Figure 2.5 Fertilizer Plant in Port Neal Explosion

According to Ness (2015) a massive explosion occurred in the ammonium nitrate portion of Terra
Industry’s fertilizer plant in Port Neal Dec 13, 1994. Ammonia may cause asthma like allergy and can
cause asthma attacks and permanent lung damage (Ezejiofor, 2014).

According to Ness (2015) the explosion occurred after the process had been shut down and ammonium
nitrate solution was left in several vessels. Multiple factors contributed to the explosion, including
strongly acidic conditions in the neutralizer, application of 200 – psig steam to the neutralizer vessel,
and lack of monitoring of the plant when the process was shut down with materials in the process

61
vessels. Four people were killed and 18 injured. Serious damage to other parts of the plant caused the
release of nitric acid into the ground and anhydrous ammonia into the air.

According to Ezejiofor (2014) ammonia acts as an alkali, and anhydrous ammonia reacts with moisture
in mucosal surface (eyes, skin and respiratory tract) to produce ammonium hydroxide, which may cause
injury and ammonia is a severe respiratory tract irritant with acute inhalation effects including dry
mount with sore throat, and eyes, tight chest, headache, ataxia and confusion.

According to Ness (2015) the investigation concluded that the conditions that led to the explosion
occurred due to lack of safe operating procedures. There were no procedures for putting the vessels into
a safe state at shutdown, or for monitoring the process vessels during shutdown. The investigating team
found that other producers either emptied the process vessels during a shutdown or maintained the pH
above 6.0. Also, other producers either did not allow steam sparges or, if steam sparges were used, they
were conducted under direct supervision of operators. The investigating team also noted that no hazard
analysis had been done on the ammonium nitrate plant, and that personnel interviewed indicated they
were not aware of many of hazards of ammonium nitrate. Key lessons in this incident were, namely;

 Operating procedures need to cover all phases of operation. The lack of procedures for
shutdown and monitoring the equipment during shutdown led operators to perform actions that
sensitized the ammonium nitrate solution and provided energy to initiate the decomposition
reaction.
 Because there had been no hazard identification study, personnel did not know about the
conditions that sensitise ammonium nitrate to decomposition. A hazard assessment of the
shutdown step would have revealed that the pH of the neutralizer could not be measured if there
was no solution flowing through the overflow line, and that the temperature of the neutralizer
could not be accurately measured without any circulation in the tank. A complete hazard
identification study would have covered backflow of ammonium nitrate into the nitric acid line,
and better design solution could have been identified.

2.6.1 Pre-Startup and Shutdown Reviews

According to Duguid (2008) during plant shutdown and start-up operators should be reminded of the
following,

 To be aware that over half of all incidents occur during shutdown, start-up, maintenance and
abnormal operations.

62
  To be aware of the risks of runaway reactions due to changes in operations such as charging an
incorrect reactant, changing the order or rate of charging, temporary shutdowns or loss of
cooling.
 To be aware that they should not depart from written operating procedures without advising
supervision. In emergency situations this may have to be after the event.
 To be aware of the need to only use alarms/shutdowns as a backup, rather than relying on them
completely.
 To be aware of the need to physically check that equipment is de-pressured and drained in the
presence of the maintenance crew directly before work on it starts. Also that the crew know
exactly what they may work on.

2.6.2 Operating Procedure

Through the completion of job hazard analysis, hazards are identified and sometimes cannot be
eliminated or engineered out of a particular task, safe work procedure are step by step instructions that
allow workers to conduct their work safely when hazards are present (Purohit et al., 2018). Operating
procedures describe the tasks that must be performed, data to be recorded, and operating conditions to
be maintained (Hardy, 2013).

According to Kumar et al. (2017) procedures should be established to assure compliance with
applicable regulations and standards such as hazard communication, confined space entry and process
safety management. The procedures also identify the health and safety precautions, operating
procedures must be clear, concise, accurate and consistent with process safety information derived from
the process hazard analysis (Hardy, 2013).

According to Ness (2015) an explosion and fire at the Formosa Plastics Corporation occurred April 23,
2004 and killed five workers and seriously injured two others. The vent destroyed most of the polyvinyl
chloride manufacturing facility and ignited polyvinyl chloride resins stored in an adjacent warehouse.
Concerns about the ensuing smoke from the fire forced a two-day community evacuation. On the day
of the incident, the reaction and the power washing had been completed in reactor D306 and the operator
went downstairs to drain the reactor. It is believed that, at the bottom of the stairway, he turned in the
wrong direction, toward an identical set of four reactors that were in the reaction phase of the process.
By mistake, the operator likely attempted to empty reactor D310 by opening the bottom and drain valve.
The bottom valve, however, was above 10 psi. Because this tank was currently processing a batch of
Polyvinyl Chloride (PVC) at high pressure, the valve did not open. In case of an emergency, operators
could follow an emergency transfer procedure that required them to open the bottom valve and the

63
transfer valve to connect the reactor to an empty reactor. However, during an emergency transfer, the
reactor pressure is greater than 10 psi, and the safety interlock would prevent the opening of the bottom
valve. Therefore, the company added a manual interlock bypass so that operators could open the valve
and reduce reactor pressure in an emergency. The bypass incorporated quick-connect fittings on air
hoses so that operators could disconnect the valve actuator from its controller and open the valve by
connecting an emergency air hose directly to the reactor. It is likely that the operator thought he was at
the correct reactor D306 and that its bottom valve was not functioning. When the bottom valve did not
open, he switched to the backup air supply and overrode the interlock. He did not contact the upstairs
reactor operator or shift foreman to check the status of the reactor before doing this. Once the bottom
valve was opened, Vinyl Chloride Monomer (VCM) poured out of the reactor and the building rapidly
filled with liquid and vapour. A deluge system in the building activated and a shift supervisor came to
the area to investigate. The VCM detectors in the building were reading above their maximum
measureable levels. The shift foreman and reactor operators took measures to slow the release, rather
than evacuate. The VCM vapours found an ignition source and several explosions occurred. The root
cause was that the operator overrode an interlock, which led to a release of hot, pressurised VCM.
Formosa Plastics did not have comprehensive written procedure, such as requiring shift supervisor
approval, for managing interlocks on the vessels. Employees were unprepared for a major accident at
the facility. The reactor groupings had similar layouts. The operators on the lower levels were not given
radios, which would have made communication with the reactor control operators on the upper level
easier. Formosa eliminated an operator group leader position and gave its responsibilities to the shift
supervisors, who were not always as available as the group leaders used to be. Key lessons learned in
this incident are,

 Operators and engineers must follow operating procedures and protocols intelligently, and,
when the process moves outside the operating envelope, stop work, get experienced advice as
needed, and shutdown as appropriate.
 Operator should have obtained supervisory approval to override the interlock.
 Operators were not given tools (radios for communication between floors) to make it easier for
them to follow their procedures. It is management’s responsibility to provide the tools and
controls necessary for operators to do their jobs safely.
 When Formosa plastics took over the plant, it made staffing changes, such as reductions in staff
and changes in responsibilities. It did not conduct a formal management of organisational
change review to analyse the impact of these changes.
 This explosion also illustrates the importance of emergency response planning. When the VCM
release occurred, gas detectors in the building and a deluge system were activated. Operators

64
 responded by trying to mitigate the release. The proper response to these activations would
have been to evacuate.

Operating procedures shall be readily accessible to employees and the operating procedures shall be
reviewed as often as necessary to assure that they reflect current operating practice (Hardy, 2013).
According to Kumar et al. (2017) work practices are procedures that limit worker exposure by reducing
exposure times or keeping workers away from contaminants. The procedures should be formally
reviewed and updated as necessary to assure that they are consistent with existing processes (Hardy,
2013). A well-meaning “no blame” approach is not altogether desirable as it is impractical to condone
reckless non-compliance with operational procedures (Bond, 2007).

According to Okoye, Ezeokonkwo and Ezeokoli (2016) health and safety have become an integral
component in the workplace as employers, labour unions and other engage in trainings and procedures
to ensure compliance with health and safety standards and also to keep a healthy workplace. The
employer shall certify annually that these operating procedures are current and accurate (Hardy, 2013).

2.6.3 Control of Ignition Source

According to Puttick (2008) fire and explosion hazard assessment flammable and potentially flammable
atmospheres must be identified and compared with the potential ignition sources present and with
knowledge of the possible flammable atmospheres, their sensitivity to ignition and the possible ignition
sources present and the incendivity of these sources a robust basis of safety may be selected. When heat
with combustible materials reaches a combustion temperature, self-ignition and burning with flames
occur (Stojkovic’, 2013). Avoidance of ignition sources reliability depends on having relatively
insensitive atmospheres and the main applicability will be counteracting electrostatic and some
mechanical ignition sources (Puttick, 2008). Avoidance of ignition sources depends on having relatively
insensitive atmospheres and the main applicability will be counteracting electrostatic and some
mechanical ignition sources (Puttick, 2008).

2.6.4 Control of Confined Space Entry

The other high risk operational activity in the petrochemical industry is the confined space entry and it
defined as an enclosed or partially enclosed area that is big enough for a worker to enter (Stojkovic’,
2013). The hazards may not be obvious and it is imperative that the assessments must be done by a
qualified person familiar with the confined space and the work to be done in that space (Karthika, 2013).
According to Kumar et al. (2017) workers are often exposed in confined spaces, exposure levels to
workplace hazards are often much higher than exposures to hazards in the general environment.

65
Confined space is not designed for someone to work in regularly, but workers may need to enter the
confined space for tasks such as inspection, cleaning, maintenance, and repair, a small opening or a
layout with obstructions can make entry and exit difficult and can complicate rescue procedures
(Stojkovic’, 2013).

In a workplace, hazard has been identified and the risk to health and safety assessed, an appropriate
prevention or control strategy done, but many times people still get killed inside tanks and other
confined spaces because of misunderstandings like entering without permission to do some job or
merely put their head inside to inspect the inside (Karthika, 2013). Confined spaces may contain
hazardous atmospheres, including insufficient oxygen, toxic air, and an explosive atmosphere
(Stojkovic’, 2013). According to Puttick (2008) preventative bases of safety (absence of flammable
atmosphere and avoidance of ignition sources) are the most economic and so there will always be driver
to choose them over protective bases of safety (venting, suppression and containment) and it is not
always possible to use absence of flammable atmosphere due to insufficient fuel.

It is necessary to first control working conditions by an occupational safety expert and if there is danger
for workers in confined space, it is necessary prior to their entry, to clean and ventilate this space
(Stojkovic’, 2013). According to Karthika (2013) even though accidents can never be eliminated
completely, employers can prevent many of the injuries and fatalities that occur each year. Confined
spaces may contain hazardous atmospheres, including insufficient oxygen, toxic air, and an explosive
atmosphere and these spaces may also have physical hazards that may result, for example, in workers
falling, being crushed or buried or drowning (Stojkovic’, 2013). Inert brings its own set of problems, as
well as expense, and possible difficulty of implementation (Puttick, 2008).

According to Stojkovic’ (2013 as a worker could enter into a hazardous confined space it is necessary
to execute a sequence of procedures, some of which should be allocated as follows,

 Testing of the air

 Cleaning and ventilation
 Separation, breaking links with other technological devices
 Personal protective equipment for workers
 Entry control
 Blocking of mobile devices
 Rescue plan

66
2.6.5 Permit to Work

A permit to work is a document which specifies the task to be performed, associated foreseeable hazards
and the safety measures (Reddy and Reddy, 2015). According to Navadiya (2017) design of permit to
work is very significant but most key thing is definition of roles and responsibilities of involved
employees in procedure part and preparing checklist which is to be covered in synchronize way. The
permit to work system has been widely used to ensure safety during maintenance and/or construction
activities in almost every major hazard industry worldwide (Reddy and Reddy, 2015). Language must
be simple such that it can be understood to each level of user at shop floor personnel (Navadiya, 2017).

According to Karthika (2013) major hazards arise while working and root causes regarding permit to
work are,

 Wrong type of work permit used, wrong information about work required on the work permit,
failure to recognise the hazards where work is carried out (e.g. flammable substances).
 Introduction of ignition source in controlled flameproof area (e.g. welding, non-spark proof
tools, non-intrinsically safe equipment used in intrinsically safe zones)
 Terms of work permit not adhered to (e.g. failure to isolate plant and/or drain lines of hazardous
substances) failure to hand-over plant is safe condition on completion of work/cancelling of
work permit.
 Unauthorised staff performing work permits functions, poor management of the work permit
system, and insufficient monitoring of the work permit system.

Effective implementation of a comprehensive permitting program certainly helps preventing several

undesirable incidents. However, deficiencies in implementing a permit to work system have been a
contributing factor in several catastrophic incidents (Reddy and Reddy, 2015). Defined roles and
responsibilities in procedure of permit to work helps actual work to be smooth and without miss
understanding that may further lead to accident (Navadiya, 2017).

According to Reddy and Reddy (2015) major steps involved in the permit system include identifying
tasks requiring permit(s), develop permit forms, define roles, train and maintain competency of
personnel, create work order, fill out permit forms, identify associated hazards and mitigation, execute
the task within the constrained listed on the permit and adhering to specified safety measures, closeout
permit, review and monitoring, main aspects generally considered for permit to work system are as
follow,

67
  Complexity of the operation including risk involved
 Human factors including personnel skills
 Types of permits required and content of each permit.

According to Karthika (2013) normally all maintenance, repair, construction work shall be carried out
with a proper work permit and jobs where work permit is required include but not limited to the
following,

 Major and minor maintenance work

 Inspection
 Construction
 Alteration
 Any hot work (including use of normal battery driven equipment in operating areas)
 Entry into confined space
 Excavation
 Vehicle entry into process areas
 Work at height
 Handling of materials using mechanised means in operating areas
 Erecting and dismantling of scaffolding
 Radiography

According to Navadiya (2017) basic challenges of permit to work are,

 Steps of permit to work should be followed as per pre-defined procedure in safety management
system. Deviation of one of the step may invite accident, further may damage reputation of the
organisation.
 Short cuts are most dangerous in the petrochemical industry as they invite accidents. Once pre-
defined procedure is established, it should be followed strictly.
 All attachments should be clearly elaborated. Lock out tag out procedure, and other related
procedures.

2.7 Health and Safety Culture

Major hazard industries especially oil and gas operators are putting considerable efforts to ensure safe
operations thereby protecting health and safety of workforce and the environment (Reddy and Reddy,

68
2015). Knowledge and compliance alone are not enough to cause behavioural changes required for
safety performance but a certain aspects of health and safety culture are required and these other
essential safety factors include: enforceable regulatory framework, management commitment and
workers involvement (Okoye, Ezeokonkwo and Ezeokoli, 2016).

Health and safety culture combines the whole approach of reporting health and safety matters,
monitoring equipment and procedures into a health and safety management system which becomes a
coherent structure that becomes acceptable and comprehensible to the employees and hence to the
public (Bond, 2007).

According to Albert, Hallowell and Kleiner (2014) hazard recognition program elements are,

 Pre-job safety meeting quality measurement tool – tool that evaluates the crew’s hazard
identification capability and communication to create hazard awareness.
 Senior leadership engagement in JSA process – a quantitative measure of the management’s
involvement in the JSA process through resource allocation and commitment.
 Augmented and interactive virtuality training environment – Computer based simulation tool
that trains workers to identify hazards using a representative virtual environment.
 Safety situational awareness training – a worker centric program in which various potential
hazards are detailed to the work crew prior to initiating work.
 JSA post kick off audit – evaluation of JSA after task completion to obtain feedback on
unidentified hazards.
 Hazard identification board – a waterproof board displayed at the work site to communicate
potential hazards as work progresses.
 Precursory visual cues – using visual aids such as tapes, signals, signs and LEDs to
communicate hazards to the workforce proactively.
 Physical area hazard simulation – an active exercise by the crew that simulates work to be done
as way of identifying associated hazards.
 Foreman one on one with employee – a one on one walkabout through the work facility, where
an experienced foreman points to hazards in the environment.
 Video/Photo monitoring and feedback – a continuous feedback process received through the
review of previous work captured as videos or photographs.
 Job safety/hazard analysis – a formal technique that focuses on specific work tasks as a way of
identifying hazards before work in initiated.
 Task demand assessment – an evaluative method in which task difficulty is assessed and better
and efficient work practices are proposed.

69
  What-if analyses – use of a systematic, but loosely structured form of brainstorming sessions
guided by what-if questions.
 Action plan critique – a feedback mechanism involving the critiquing of established plans to
improve implementation work plan.
 Recordkeeping and accident analyses – creation of a database that records lessons learned from
past injuries and experiences to avoid recurrence of accidents.
 Safety checklists – survey of work area or construction process to ensure conformance to certain
established criterion.
 Method statement review/work permitting – audit of written work plan elaborating on work
tasks and conditions before a written permit to work is issued.
 Walk-through safety and health audit – an observational method to identify active hazardous
conditions, unsafe behaviour through walkabout sessions.
 Worker to worker observation program – a peer to peer safety observation program to provide
feedback on worker performance with respect to health and safety.
 Proactive safety alert systems – incorporation of detection technology into equipment that
sounds an alarm, or is disabled when hazard is detected.
 Pre-use analysis and planning – a formal study conducted prior to any process modification, or
the use of new equipment or chemical in the job site.

According to Holstvoogd et al. (2006) there is a health and safety culture ladder with five levels, namely
pathological (level 1), reactive (level 2), calculative (level 3), proactive (level 4) and generative (level
5). Pathological culture level 1 is where nobody cares to understand why accidents happen and how
they can be prevented and at generative culture level 5, health and safety is seen as a profit centre and
it is how business is done. Reactive culture level 2 says safety is important, and people do a lot when
there has been an accident while the proactive culture is where employees work safely because they
intrinsically motivated to do the right things naturally. Level 3 is calculative where an organisation says
there are management systems in place to manage all hazards.

70
Figure 2.6 Health and Safety Culture Ladder
Adapted from Holstvoogd et al. (2006).

The health and safety culture ladder in Figure 2.1 shows the various levels of cultural maturity, and the
change process required to achieve a lasting change in personal and organisational culture (Holstvoogd
et al., 2006). Health and safety knowledge automatic ensure compliance, but health and safety
knowledge and compliance alone cannot substantially improve health and safety culture (Okoye,
Ezeokonkwo and Ezeokoli, 2016).

2.8 Development of Conceptual Model

Models differ from theories in that the role of a theory is explanation whereas the role of a model is
representation. Conceptual modelling is the abstraction of a simulation model from the part of the real
world it is representing (‘the real system’). The real system may, or may not, currently exist. Abstraction
implies the need for simplification of the real system and for assumptions about what is not known
about the real system (Robinson, 2013).

The path diagram was utilised in this study to express the model in a visual form that is easily
understood. In path diagrams the observed variables are enclosed by rectangles and the latent variables
are depicted in ellipses. Line with a single arrowhead represents a hypothesised relationship or direct
effect of one variable on another. The two-way curved arrow with an arrowhead at each end is used to
represent co-variation between two variables.

71
2.9 Conceptual Model

The conceptual model proposed in this research includes the following latent variables, namely,

 Generative Health and Safety Culture (Main Latent Variable)

 Leadership Commitment
 Chemical Exposure Management
 Health and Safety Risk Assessment
 Process Hazard Analysis
 Permit to Work
 Training and Competency
 Process Health and Safety Information
 Control of Confined Space Entry
 Operating Procedure
 Control of Ignition Source

72
In this study there are 35 observed variables as indicated in figure 2.7

Figure 2.7 The proposed Conceptual Model

73
Latent Variable 1: Leadership Commitment

Leadership commitment alone can make health and safety a key business goal and provide the necessary
resources to achieve health and safety objectives. If leadership commitment consistently treats health
and safety as a priority, so will the rest of the work group.

LC – Leadership Commitment

dLC – Leadership Commitment Deviation

LCH – Leadership Commitment questions from the research instrument

errh – errors for all the observed variables

H7: Senior Management prioritises Health and Safety in my organisation.

H8: Senior Management has an open door policy on Health and Safety issues.
H10: Senior Management communicates Health and Safety policy to all employees.
H11: Senior Management allocates enough time to address Health and Safety concerns.
H17: Senior management prioritizes mechanical/asset integrity of our processing plant.
H38: Poor housekeeping in my organization is the cause for many health and safety incidents.
H40: Audit compliance is an excellent practice to prevent most of health and safety incidents in the
petrochemical industry.

Latent Variable 2: Chemical Exposure Management

Controlling exposures to chemical hazards and toxic substances is the fundamental method of protecting
workers. It is the responsibility of every employer to ensure employee exposures to chemical hazards
are maintained below local permissible exposure limits as defined by local law through the application
of appropriate controls, such as engineering, administrative controls or personal protective equipment.

CEM – Chemical Exposure Management

dCEM – Chemical Exposure Management Deviation

CEMH – Chemical Exposure Management questions from the research instrument

errh – errors for all the observed variables

H6: My organization has excellent chemical exposure management systems.

H12: Most employees are aware of hazardous chemicals in their work environment.
H14: Most employees know how to handle hazardous chemicals in the work place.

74
H15: Contractor’s on boarding appreciates all hazardous chemicals in my organization
H16: Most contractors know how to handle hazardous chemicals in my organization.
H30: All employees are aware that when you handling hazardous chemicals you need to use prescribed
personal protective equipment.

Latent Variable 3: Health and Safety Risk Assessment

Health and safety risk assessment is the process of evaluation of the risks arising from a hazard, taking
into account the adequacy of any existing controls and deciding whether or not the risks is acceptable.
It is vital that employers know where the risks are in their organisations and control them to avoid
putting in risk employees, customers and the organisation itself.

HSRA – Health and Safety Risk Assessment

dHSRA – Health and Safety Risk Assessment Deviation

HSRAH – Health and Safety Risk Assessment questions from the research instrument

errh – errors for all the observed variables

H9: There are effective noise exposure management systems in my organization.

H32: Most of the health and safety incidents in the petrochemical industry are due to not verifying
energy isolation before you start working on equipment.
H33: My organization diligently manages fatigue in both permanent employees and contractors.
H34: My organization has all management systems in place to manage substance misuse.
H39: Poor health and safety risk assessments are responsible for most of health and safety incidents in
the petrochemical industry.

Latent Variable 4: Process Hazard Analysis

A process hazard analysis is an organised effort to identify and analyse the significance of hazardous
scenarios associated with a process or activity. Process hazard analysis are used to pinpoint weaknesses
in the design and operation of facilities that could lead to accidental chemical releases, fires, or
explosions and to provide organisations with information to aid in making decisions for improving
health and safety.

PHA – Process Hazard Analysis

dPHA – Process Hazard Analysis Deviation

PHAH – Process Hazard Analysis questions from the research instrument

75
errh – errors for all the observed variables

H20: In my organization all engineering changes undergo a comprehensive management of change.

H21: The organization does comprehensive process hazard analysis before engineering changes are
made.
H23: The organization does comprehensive process hazard analysis before engineering changes are
made.
H24: In my organization we have a comprehensive pre-activity start up review and pre-activity
shutdown review.

Latent Variable 5: Permit to Work

Permit to work is a management system used to ensure that work is done safely and efficiently. Permit
to work is an essential part of control of work, hazard identification and risk assessment. It is designed
to provide protection for employees who are working in hazardous situations.

PTW – Permit to Work

dPTW – Permit to Work Deviation

PTWH – Permit to Work questions from the research instrument

errh – errors for all the observed variables

H25: Most of the health and safety incidents in petrochemical industry are due to poor controls when
working at heights.
H28: All the work activities in my organization are done after a valid permit to work has been signed
by the authorities.
H29: In my organization before you start excavation or entering a trench you need to obtain
authorization.
H31: In my organization all safety critical equipment is disabled with permission from the authorities.

Latent Variable 6: Training and Competency

Education and training are essential to an effective health and safety management system. Workers who
understand the hazards and risks of the assigned tasks are far less likely to be injured or become ill from
occupational disease.

TC – Training and Competency

76
dTC – Training and Competency Deviation

TCH – Training and Competency questions from the research instrument

errh – errors for all the observed variables

H13: Employees undergo comprehensive training on health and safety in my organization.

H19: The organization closed all corrective action items effectively after the root cause analysis.
H35: Most of the health and safety incidents are due to human error in my organization.

Latent Variable 7: Process Health and Safety Information

Occupational Health and Safety Act requires compiling of technical information on the process and
equipment prior to conducting a process hazard analysis. Occupational Health and Safety Act has three
categories, namely, hazards of the chemicals and flammables in the process, information related to the
technology of the process and lastly information pertaining to the equipment in the process.

PHSI – Process Health and Safety Information

dPHSI – Process Health and Safety Information Deviation

PHSIH – Process Health and Safety Information questions from the research instrument

errh – errors for all the observed variables

H18: The organization closed all corrective action items effectively after the root cause analysis.
H22: Most of the health and safety incidents are due to human error in my organization.

Latent Variable 8: Control of Confined Space Entry

A confined space is large enough and so configured that an employee can bodily enter and perform
assigned work, it has limited or restricted means for entry or exit and it is not designed for continuous
occupancy by the employee. Fatalities in confined spaces often occur because the atmosphere is oxygen
deficient or toxic, confined spaces should be tested prior to entry and continually monitored.

CCSE – Control of Confined Space Entry

dCCSE – Control of Confined Space Entry Deviation

CCSEH – Control of Confined Space Entry questions from the research instrument

errh – errors for all the observed variables

77
H36: My organization has effective management systems to manage working in confined space.
H37: Most of the health and safety incidents are due to poor controls in place when working with
suspended loads.

Latent Variable 9: Operating Procedure

It is the duty of an employer to develop and implement written operating procedures that provide clear
instructions for safely conducting operations and maintenance. Operating procedures shall be readily
accessible to employees and reviewed as often as necessary to assure that they reflect current operating
practice.

OP – Operating Procedure

dOP – Operating Procedure Deviation

OPH – Operating Procedure question from the research instrument

errh – errors for all the observed variable

H26: In my organization all work activities have a detailed operating procedure or work instruction.

Latent Variable 10: Control of Ignition Source

It is vital that potential ignition sources are identified, and there is feedback from operational experience
back into hazard assessment process. What is required from an assessment perspective is to be able to
characterise a flammable atmosphere with respect to its ignition sensitivity, and to identify potential
ignition sources. Control of ignition sources is valuable and reliable basis of health and safety.

CIS – Control of Ignition Source

dCIS – Control of Ignition Source Deviation

CSIH – Control of Ignition Source question from the research instrument

errh – errors for all the observed variable

H27: Most of the health and safety incidents in petrochemical industry are due to poor controls of
source of ignition.

78
Main Latent Variable: Generative Health and Safety Culture

In a mature generative health and safety culture everybody has the aspiration to do all health and
safety critical tasks appropriately.

GHSC – Generative Health and Safety Culture

2.10 Chapter Summary

In this research process health and safety management elements from literature to be prioritized for an
effective process health and safety management system are, Leadership Commitment, Training and
Competency, Chemical Exposure Management, Health and Safety Risk Assessment, Process Hazard
Analysis, Process Health and Safety Information, Operating Procedure, Control of Ignition Source,
Control of Confined Space Entry and Permit to Work.

Health and safety standards and regulations reasonably cannot cover all the possible cases for different
types of works in a variety of hazards in a petrochemical industry. It is the duty of health and safety
experts to share as much as they can on this subject of health and safety in the petrochemical industry
that can expose employees and contractors to hazardous chemicals. This chapter also presented
conceptual model for this research. The proposed conceptual process health and safety elements model
is based on the main latent variable (Generative Health and Safety Culture) and latent variables
(Leadership Commitment, Chemical Exposure Management, Health and Safety Risk Assessment,
Process Hazard Analysis, Permit to Work, Training and Competency, Process Health and Safety
Information, Control of Confined Space Entry, Operating Procedure and Control of Ignition Source)
identified through literature review. The next chapter discusses the research design and methodology.

79
 CHAPTER THREE
Research Design and Methodology
_____________________________________________________________________

3.1 Introduction

This chapter discusses the research design and methodology to attain the research aim and objectives;
the research process, paradigm and methodology as well as the sampling methodology. The qualitative
and quantitative data analysis is also explained.

3.2 Research Paradigm

According to Leedy and Ormrod (2010) research is a systematic process of collecting, analysing and
interpreting information in order to increase understanding of a phenomenon of interest. According to
Denzil and Lincoln (2011) paradigms could be categorised by the core ontology and epistemological
position embraced in the research process. These definitions imply that research is a systematic process
that has objectives or purpose, aims to increase human knowledge and allows for the collection of
information and evaluation of it before making a final decision (Leedy and Ormrod, 2010).

Every paradigm is based upon its own ontological and epistemological assumptions (Scotland, 2012).
Researchers normally develop the research methodologies based on different ontological and
epistemological assumptions and the related paradigms (Denzil and Lincoln, 2011). Ontology concerns
the issues of the fundamental nature of reality and since all assumptions conjecture, the philosophical
underpinnings of each paradigm can never be empirically proven or disproven (Easterby-Smith and
Lowe, 2002).

The paradigm in turn commands the methodological approach to be embraced, it affords holistic insight
of how the researcher views knowledge, and articulates the methodological strategies to disclose it
(Denzin and Lincoln, 2011). Different paradigms inherently contain differing ontological and
epistemological views, therefore, they have differing assumptions of reality and knowledge which
underpin their particular research approach (Scotland, 2012).

80
3.2.1 Ontology

Knowledge is both socially constructed and influenced by power relations from within society and what
counts as knowledge is determined by the social and positional power of the advocates of that
knowledge (Scotland, 2012).

The two basic classifications within ontology are realism and nominalism; realism accepts that reality
exist out there freely of people and their insights and interpretations of it, while nominalism assumes
that people never openly experience the real world but through their own subjectivity and interpretations
(Neuman, 2014). The ontological position of the critical paradigm is historical realism (Scotland, 2012).
Internal realism assumes that the truth is out there but is challenging to spot directly, while relativism
assumes that there is much truth out there, dependent on the viewpoint of the viewers (Bryman, 2016).

According to Mouton (1996) cited in Bryman and Bell (2011) historical realism is the view that reality
has been shaped by social, political, cultural, economic, ethnic, and gender values, reality that was once
deemed plastic has become crystallised (Scotland, 2012). Realism argues that there are in fact
fundamental differences between the social and natural sciences but similarities also exist which justify
the use of similar research approaches.

3.2.2 Epistemology

Epistemology is the concept that is concerned with how knowledge is gained (Malhotra, 2017).
According to Scotland (2012) the belief about the nature of the world (Ontology) adopted by a
researcher will affect the belief about the nature of knowledge in that world (epistemology) which in
turn will influence the researchers belief as to how that knowledge can be revealed (methodology).
Epistemology and methodology are intimately related: the former involves the philosophy of how we
come to know the world and the latter involve the practice (Malhotra, 2017).

Epistemological assumptions are concerned with how knowledge can be created, acquired, and
communicated in other words, what it means to know (Cohen, Manion and Morrison, 2007). According
to Malhotra (2017) the term epistemology comes from the Greek word episteme, their term for
knowledge. In simple terms, epistemology is the philosophy of knowledge or of how we come to know.
Critical epistemology is one of subjectivism which is based on real world phenomena and linked with
societal ideology (Scotland, 2012).

81
The methodology is concerned with how we come to know but is much more practical in nature and is
focused on the specific ways - the methods - that we can use to try to understand our world better
(Malhotra, 2017).

There are two basic paradigm groups within epistemology, namely, positivism and constructivism;
positivism supports the view that truth or knowledge should be produced by way of a scientific approach
(Andrew, 2012). According to Mouton (1996) cited in Bryman and Bell (2011) positivism, also referred
to as ontological naturalism, suggests that social sciences are similar to natural sciences and should
therefore follow the logical and rigor of natural sciences in research.

While constructivism debates that reality is composed of the perspectives and views of people and
therefore reality is highly subjective and can only be inferred from viewpoints or opinions of those who
have experienced reality (Andrew, 2012).

3.3 Research Philosophy

According to Scotland (2012) what knowledge is, and ways of discovering it, are subjective. The
individual perception of reality affects how people gain knowledge of the world, and how people act
within it. Those who believe there is a single objective truth are usually referred to as Positivist
(Creswell, 2009). Each paradigm has its own ways of realizing its aims and philosophy is concerned
with views about how the world works and, as an academic subject, focuses primarily on reality,
knowledge and existence. The individual view of the world is closely linked to what is perceived as
reality (Scotland, 2012).

According to Saunders et al. (2012) research philosophies are concerned with the development of
knowledge (epistemology) and the nature of that knowledge (ontology). Regarding educational
research, the scientific paradigm seeks to generalise, the interpretive paradigm seeks to understand, and
the critical paradigm seeks to emancipate (Scotland, 2012). Research philosophies are epistemological
and ontological positions or “world views” or assumptions or theoretical frameworks about how
knowledge should be generated (Bryman and Bell, 2011).

If it is accepted that the understanding of reality affects the way people gain knowledge of reality, then
there is a need to accept that this will affect how people actually conduct research about reality
(Scotland, 2012). Epistemological philosophies determine the approach to questioning and discovery
in research and ontological philosophies are concerned with the nature of reality and the assumptions
made about how the world operates (Saunders et al., 2012).

82
The ontological position of positivism is one of realism (Scotland, 2012). Realism is therefore pro-
positivist in nature (Saunders et al., 2012). Realism is the view that objects have an existence
independent of the knower and therefore the positivist epistemology is one of objectivism (Scotland,
2012).

3.3.1 Positivism

Positivists generally assume that reality is objectively given and can be described by measurable
properties which are independent of the observer (researcher) and his or her instruments (Malhotra,
2017). According to Creswell (2009) in a scientific paradigm positivist methodology is directed at
explaining relationships and positivists attempt to identify causes which influence outcomes.

According to Creswell (2009) in a scientific paradigm positivist methodology is directed at explaining

relationships and positivists attempt to identify causes which influence outcomes. Positivist studies
generally attempt to test the theory, in an attempt to increase the predictive understanding of
phenomena. Induction strategy has its roots attached to the positivism approach where the observations
analysed help in driving out the result (Malhotra, 2017). Their aim is to formulate laws, thus yielding a
basis for prediction and generalisation and a deductive approach is undertaken (Creswell, 2009).

Positivism assumes an objective world hence it often searches for facts conceived in terms of specified
correlations and associations among variables (Malhotra, 2017). According to Creswell (2009)
positivist methodology is directed at explaining relationships and positivist attempt to identify causes
which influence outcomes, such people believe there are universal truths that are waiting to be
discovered. The positivist focus on experimental and quantitative methods used to test and verify
hypotheses have been superseded or complemented to some extent by an interest in using qualitative
methods to gather broader information outside of readily measured variables (Malhotra, 2017).

The foundation of positivism is that the reality and facts are out there; it is applicable to measure and
explore them by objective methods rather than being inferred by subjective sensation, intuition or
reflection (Blaikie, 2007). Based on the results of the studies, people may learn that theory doesn’t fit
the facts well and so they need to revise the theory to better predict reality (Malhotra, 2017). The key
approach of the scientific method is the experiment, the attempt to discern natural laws through direct
manipulation and observation (Malhotra, 2017).

Postpositivism is a recent evolution of positivism and postpositivism is consistent with positivism in

assuming that an objective world exists but it assumes the world might not be readily apprehended and
that variable relation or facts might be only probabilistic, not deterministic (Malhotra, 2017). Positivism

83
is typically associated to quantitative methods, while constructionism is often linked with qualitative
approaches (Cresswell and Clark, 2011). In a positivist view of the world, science is seen as the way to
get the truth, to understand the world well enough so that people might predict and control the world
(Malhotra, 2017).

Positivism philosophy is based upon the highly structured methodology to enable generalization of the
results with the help of statistical methods (Williams, 2011). According to Malhotra (2017) the world
and the universe are deterministic; they operated by laws of cause and effect that could discern if applied
the unique approach of the scientific method and science is largely a mechanistic or mechanical affair
and researchers use deductive reasoning to postulate theories that they can test

This research followed an epistemological positivist philosophy so that it can empirically test structural
relationships among latent variables of generative process health and safety culture. The positivist
believes in empiricism – the idea that observation and measurement is the core of scientific endeavour
(Malhotra, 2017).

3.3.2 Interpretivism

Interpretive researchers start out with the assumption that access to reality is only through social
constructions such as language, consciousness and shared meanings (Malhotra, 2017). The interpretive
epistemology is one of subjectivism which is based on real world phenomena and the world does not
exist independently of people’s knowledge of it (Grix, 2004). The philosophical base of interpretive
research is hermeneutics and phenomenology and interpretive studies generally attempt to understand
phenomena through the meanings that people assign to them and interpretive methods of research
(Malhotra, 2017).

Interpretive research does not predefine dependent and independent variables but focuses on the full
complexity of human sense-making as the situation emerges (Malhotra, 2017). According to Scotland
(2012) interpretive paradigm is sensitive to individual meanings that can become buried within broader
generalizations. According to Mouton (1996) cited in Bryman and Bell (2011) the interpretive paradigm
is also anti-positivist like the phenomenological paradigm.

Interpretive methodology is directed at understanding phenomenon from an individual’s perspective,

investigating interaction among individuals as well as the history and cultural contexts which people
inhabit (Creswell, 2009). Interpretive research rejects a foundation base to knowledge, bringing into
question its validity and cannot be judged using the same criteria as the scientific paradigm (Scotland,
2012). This research investigates how the objectives features of society emerge from depend on, and

84
are constituted by subjective meanings of individuals and intersubjective processes such as discourses
or discussions in groups (Malhotra, 2017).

If reality is subjective and differs from person to person, then research participants cannot be expected
to arrive at exactly the same interpretations as researchers (Rolfe, 2006). Constructionists have also
been particular concerned with the interplay of subjective, objective and intersubjective knowledge and
written texts (Malhotra, 2017). Constructivist philosophy takes the stance that reality, truth, is a
construction of an individual’s view of their world and that constructivist research accepts the truth
which is generated between the researcher and the participant (Williamson, 2006).

A key form of interpretive research is social constructionism which seeks to understand the social
construction dialectic involving objective, intersubjective and subjective knowledge (Malhotra, 2017).
Those who believe there is no reality other than what individuals create in their heads are known as
interpretivists or contructivists (Creswell, 2009). In a sense, interpretive constructivism seeks to show
how variations in human meanings and sense making generate and reflect differences in reified or
objective realities (Malhotra, 2017).

3.4 Research Approach

There are two research approaches, namely, deductive and inductive approaches (Saunders et al, 2003).
According to Mouton (1996) cited in Bryman and Bell (2011) deductive reasoning works from the
general truth or theory to logically arrive at a specific conclusion to test a hypothesis and it is used when
testing hypotheses from existing theories. The deductive approach is when a theory and hypothesis or
hypotheses are developed and a research strategy is designed to test the hypothesis (Saunders et al,
2003). It begins explicitly with a tentative hypothesis or set of hypotheses that form a theory which
could provide a possible answer or explanation for a particular problem, then proceeds to use
observations to rigorously test the hypotheses (Malhotra, 2017).

According to Judd et al. (1986) cited in Saunders et al. (2012) inductive reasoning works from a specific
observation to propose a generalisation or hypothesis or theory based on the observation. The inductive
approach is when data is collected and a theory is developed from the data analysis (Saunders et al,
2003).

3.4.1 Deductive Approach

Deductive approach is concerned with developing a hypothesis (or hypotheses) based on existing
theory, and the designing a research strategy to test the hypothesis (Creswell and Plano Clark, 2007).

85
According to Malhotra (2017) the deductive argument moves from premises, at least one of which is a
general or universal statement, to a conclusion that is a singular statement.

Deductive means reasoning from the particular to the general (Creswell and Plano Clark, 2007).
According to Malhotra (2017) the deductive argument moves from premises, at least one of which is a
general or universal statement, to a conclusion that is a singular statement. Deductive propositions form
a hierarchy from theoretical to observational; from abstract to concrete, the deductivist accepts that
observation is guided and presupposed by the theory (Malhotra, 2017).

3.4.2 Inductive Approach

Inductive approach starts with the observations and theories are proposed towards the end of the
research and when following an inductive approach, beginning with a topic, a researcher tends to
develop empirical generalisations and identify preliminary relationships (Creswell and Plano Clark,
2007).

According to Malhotra (2017) the inductive strategy assumes that all science starts with observations
which provide a secure basis from which knowledge can be derived and claims that reality impinges
directly on the senses, hence there is a correspondence between sensory experiences, albeit extended
by instrumentation, and the objects of those experiences.

There is no hypothesis at the initial stages of the research and the researcher is not sure about the type
and nature of the research findings until the study is finished (Creswell and Plano Clark, 2007). The
conclusion of an inductive argument makes claims that exceed what is contained in the premises and
so promises to extend knowledge by going beyond actual experience (Malhotra, 2017).

3.5 Research Methodology

Research methodology in principle is focused around the problems to be studied in a research study and
therefore varies according to the problems explored (Scotland, 2012). The literature review plays a
major role in justifying the research and identifying the purpose of the study in quantitative research
and can be used to identify the questions to be asked and to inform the hypotheses (Creswell and Plano
Clark, 2007). Identifying the research methodology that best suits a research in hand is vital as it will
serve establishing the credibility of the work (Scotland, 2012).

86
Research philosophy, approach, strategy, and techniques are inherent components of the research
methodology (Scotland, 2012). Literature reviews in quantitative research are more comprehensive and
more detailed than is the case in qualitative research (Creswell and Plano Clark, 2007). In quantitative
research, the researcher and what is being researched are viewed as independent of each other, whereas
in the qualitative research, they are interactive and inseparable (Teddlie and Tashakkori, 2009). The
literature review is typically brief and does not usually guide the research questions to the same extent
as literature reviews in quantitative research does (Creswell and Plano Clark, 2007). Quantitative
researchers believe that reality is single and tangible, whereas qualitative researchers view reality as
constructed and hence multiple (Teddlie and Tashakkori, 2009).

In qualitative research, the literature review is used to provide evidence for the purpose of the study and
to identify the underlying problem that will be addressed by the inquiry and this is done to ensure that
the literature does not limit the types of information the researcher will learn from the participants
(Creswell and Plano Clark, 2007). Differences in ontology and epistemology mean that different
research methods have been employed, with quantitative researchers using deductive approaches,
whereas, in contrast, qualitative researchers have tended to use inductive approaches (Teddlie and
Tashakkori, 2009).

3.5.1 Quantitative Research

A quantitative approach endorses the view that psychological and social phenomena have an objective
reality that is independent of the subjects being studied (Yilmaz, 2013). The quantitative research design
is used to examine relationships among variables using statistical analyses, it uses either experimental
or survey research strategies with questionnaires, structured interviews or structured observation
(Saunders et al., 2012). Quantitative research emphasises the measurement and analysis of causal
relationship between isolated variables within a framework which is value – free, logical, reductionistic,
and deterministic, based on a priori theories (Yilmaz, 2013).

Quantitative methods require the researcher to use a pre – constructed standardised instrument or pre-
determined response categories into which the participants varying perspectives and experiences are
expected to fit (Yilmaz, 2013). Accordingly to Trochim (2006) quantitative research often translates
into the use of statistical analysis to make the connection between what is known and what can be
learned through research. Quantitative methods generally demand randomly selected large
representative samples in order for researchers to generalise their findings from the sample (Yilmaz,
2013).

87
The major advantage of Quantitative methods is that it allows one to measure the responses of number
of participants to a limited set of questions, thereby facilitating comparison and statistical aggregation
of the data (Yilmaz, 2013). Collecting and analysing data using quantitative strategies requires an
understanding of the relationships among variables using either descriptive or inferential statistics
(Trochim, 2006). Quantitative research is informed by objectivist epistemology and thus seeks to
develop explanatory universal laws in social behaviours by statistically measuring what it assumes to
be a static reality (Yilmaz, 2013).

3.5.2 Qualitative Research

Unlike quantitative studies which are concerned with outcomes, generalisation, prediction, and cause-
effect relationship through deductive reasoning, qualitative studies are concerned with process, context,
interpretation, meaning or understanding through inductive reasoning (Yilmaz, 2013). According to
Hox and Boeije (2005) qualitative researchers examine how people learn about and make sense of
themselves and others and how they structure and give meaning to their daily lives. Therefore, methods
of data collection are used that are flexible and sensitive to the social context.

According to Yilmaz (2013) qualitative research is based on a constructivist epistemology and explores
what it assumes to be a socially constructed dynamic reality through a framework which is value-laden,
flexible, descriptive, holistic and context sensitive. According to Creswell (2005) qualitative research
is often said to employ inductive thinking or induction reasoning since it moves from specific
observations about individual occurrences to broader generalisations and theories. According to Yilmaz
(2013) qualitative research tries to understand how social experience is created and given meaning, in
qualitative research reality or knowledge is socially and psychologically constructed.

A popular method of data collection is the qualitative interview in which interviewees are given the
floor to talk about their experiences, views and so on. Instead of a rigidly standardised instrument,
interview guides are used with a range of topics or themes that can be adjusted during the study (Hox
and Boeije, 2005). The aim is to describe and understand the phenomenon studies by capturing and
communicating participants’ experiences in their own words via observation and interview (Yilmaz,
2013).

According to Hox and Boeije (2005) another widely used method is participant observation, which
generally refers to methods of generating data that involve researchers immersing themselves in a
research setting and systematically observing interactions, events and so on. The qualitative paradigm
views the relationship between the knower and the known as inextricably connected (Yilmaz, 2013).
Qualitative research can be defined as a study which is conducted in a natural setting (Creswell, 2005).

88
According to Yilmaz (2013) in making use of the inductive approach to research, the researcher begins
with specific observations and measures, and then moves to detecting themes and patterns in the data.
This allows the researcher to form an early tentative hypothesis that can be explored (Yilmaz, 2013).

The research, in effect, becomes the instrument for data collection it is up to the researcher to gather
the words of the participants and to analyse them by looking for common themes, by focusing on the
meaning of the participants, and describe a process using both expressive and persuasive language
(Creswell, 2005). Other well-known methods of qualitative data collection are the use of focus (guided-
discussion) groups, documents, photographs, film, and video (Hox and Boeije, 2005). The results of the
exploration may later lead to general conclusions or theories and qualitative findings are far longer,
more detailed and variable in content than quantitative ones (Yilmaz, 2013).

3.5.3 Mixed Approach

According Driscoll et al. (2007) mixed methods designs can provide pragmatic advantages when
exploring complex research questions. According to Youngshin et al. (2017) when combining
qualitative and quantitative data in a study there are four principles, namely,

 To determine the kind of data needed about the problem and the relevance of the problem to
the method chosen should be evident.
 The strengths and weaknesses of each method employed should complement each other.
 Methods should be selected on the basis of their relevance to the nature of the phenomena of
interest
 The methodological approach employed should be continually monitored and evaluated to
ensure that the first three principles are being followed.

The qualitative data provide a deep understanding of survey responses, and statistical analysis can
provide detailed assessment of patterns of responses (Driscoll et al., 2007). According to Youngshin et
al. (2017) in a measurement context, combining qualitative and quantitative data within a study can be
useful in developing a scale.

The analytic process of combining qualitative and survey data by quantitizing qualitative data can be
time consuming and expensive and thus may lead researchers working under tight budgetary or time
constraints to reduce sample sizes or limit the time spent interviewing (Driscoll et al., 2007).

89
According to Youngshin et al. (2017) triangulation refers to combining multiple methods when studying
the same phenomena, to minimise systematic bias in study findings and triangulation may help not only
in the development of an instrument, but also with respect to insight about the meaning of concepts.

3.6 Research Methods

The research was planned to first explore the concepts by qualitative research methods, such as in depth
literature review. The quantitative data collection and analysis is based on a quantitative research
method involving questionnaire survey and statistical data analysis methods. The validation of the
findings and related conclusions rely on the results from both qualitative and quantitative research
methods. The mixed method was considered the best option for this study as it assisted to leverage the
advantages of both quantitative and qualitative research methods in achieving the research objectives.

3.7 Research Strategies

According to Malhotra (2017) a research method is a strategy of enquiry which moves from the
underlying philosophical assumptions to research design and data collection and the choice of research
method influences the way in which the researcher collects data. The research strategy is an action plan
to achieve the research objectives and respond to the research questions and therefore links the research
philosophy with the methods for collection and analysis of the data (Saunders et al., 2012). Research
strategy provides the overall direction of the research including the process by which the research is
conducted (Malhotra, 2017).

Research strategies are influenced by the general direction which a research study may follow, namely,
exploratory or formulative, descriptive, and explanatory (Saunders et al., 2012; Sekaran and Bougie,
2010; Thakur, 1993). Specific research methods imply different skills, assumptions, and research
practices (Malhotra, 2017). Case study, ethnographic study, ground theory study, phenomenological
study, experimental study, archival research, and action research are examples for such research
strategies (Malhotra, 2017).

3.7.1 Case Study

The word case means an instance of and the central feature of case study research design is the
investigation of the one or more specific instances of something that comprise the cases in the study
(Yin, 2003). Case studies are commonly associated with qualitative research and qualitative data but
this need not be so and quantitative data can readily be incorporated into a case study where appropriate
(George and Bennet, 2005). Case study research can also facilitate a holistic perspective on causality

90
because it treats the case as a specific whole and thereby offers the possibility of investigating casual
complexity where there are many relevant factors but few observations (Yin, 2009).

The case study method in research demands a high degree of depth, breadth and rigour, with careful
attention to showing the way in which evidence supports the conclusions reached (George and Bennet,
2005). Case studies can also be used to research questions about process because the use of multiple
data sources supports the retrospective investigation of events (Yin, 2009). Research questions should
make it clear what aspects of the cases are of interest, it will not be feasible to investigate every aspect
of chosen cases (Yin, 2009).

In explanatory research, for instance, case studies offer the possibility of investigating casual
mechanisms and the specific contexts in which they are activated (George and Bennet, 2005). A case
can be something relatively concrete such as an organisation, a group or an individual, or something
more abstract such as an event, a management decision or a change programme (Yin, 2003). Case
studies involve the intensive examination of a single entity but can also take a comparative form when
they focus on more than a single entity (Murray, 2003; Thakur, 1993).

3.7.2 Ethnographic Study

The other research strategy investigated in this research is Ethnographic Study. According to Angrosino
(2007) ethnographic study is conducted on site or in a naturalistic setting in which real people live and
it is personalised since the researcher are both observer and participant in the lives of those people. The
ethnographic research comes from the discipline of social and cultural anthropology where and
ethnographer is required to spend a significant amount of time in the field (Malhotra, 2017).

According to Williamson (2006) ethnography follows constructivist philosophical principals because

ethnographic researchers gather data by studying people in their everyday contexts or by participating
in social interactions with them in order to understand their world. Ethnography also collects data in
multiple ways for triangulation over an extended period of time and the process is inductive, holistic
and requires a long term commitment (Angrosino, 2007). Ethnographers immerse themselves in the
lives of the people they study and seek to place the phenomena studied in their social and cultural
context (Malhotra, 2017). Ethnography is dialogic since conclusions and interpretations formed through
it can be given comments or feedback from those who are under study (Angrosino, 2007).

91
3.7.3 Grounded Theory

Ethnography and grounded theory studies begin with the same a posteriori principle that truth is found
through experience and grounded theory naturally fits with a constructivist philosophy because that can
also be used to understand people’s thoughts and behaviour (Charmaz, 2006). According to Malhotra
(2017) grounded theory is a research method that seeks to develop a theory that is grounded in data
systematically gathered and analysed. According to Charmaz (2006) grounded theory encourages
reflexivity when analysing data to check for events and reactions that may have happened due to the
presence of the researcher and therefore a grounded theory study is more appropriate for a study with
time constraints.

Grounded theory is an inductive theory discovery methodology that allows the researcher to develop a
theoretical account of the general features of a topic while simultaneously grounding the account in
empirical observations or data (Malhotra, 2017). According to Williamson (2006) selecting a sample
frame for grounded theory and for ethnographic studies does not involve random sampling for research
participants and either method is concerned with statistical representation; instead groups or individuals
are targeted who represent the important characteristics that researchers consider of interest to the study.

The major difference between grounded theory and other methods is its specific approach to theory
development and grounded theory suggests that there should be a continuous interplay between data
collection and analysis (Malhotra, 2017).

Grounded theory was initially devised as a set of explicit procedures for qualitative data analysis in
order to construct useful middle range theories from data (Charmaz, 2006). Both grounded theory and
ethnography seek to understand different people’s perceptions and other realities, seeing events and
actions through the eyes of the participants (Fetterman 2010; Charmaz 2006). Grounded theory
approaches are becoming increasingly common in the research literature because the method is
extremely useful in developing context-based, process-oriented descriptions and explanations of the
phenomenon (Malhotra, 2017).

3.7.4 Phenomenological Study

Phenomenology as a philosophical research tradition emerged within the early part of the 20 th century
and was built on the work of either philosophers who discussed human experience as a starting point
for philosophy (Trodres and Holloway, 2006). The researchers that believe there is no objective reality,
but that reality is constructed by individuals and therefore reality is subjective use phenomenology
research approach (Rolfe, 2006). The fundamental aim of phenomenological philosophy is to develop

92
a greater understanding of individual’s experiencer and by adopting this approach, the theory is that it
will allow human beings to be understood from inside their subjective experience (Todres and
Holloway, 2006).

Phenomenology argues that differences between social science and natural science are so fundamental
that the same methods for research would not suffice (Bryman and Bell 2011; Mouton, 1996). The main
emphasis of phenomenological research is to describe or interpret human experience as lived by the
experiencer in a way that can be used as a source of qualitative evidence (Todres and Holloway, 2006).

Based on the philosophy of phenomenology, research approaches in the social world should take
cognisance of the difference with the natural world occasioned by human behaviour by employing an
epistemology which acknowledges and capitalises on the differences and should be empathetic of the
research participants (Bryman and Bell, 2011; Saunders et al., 2012).

3.8 Research Instrument

There are various procedures of collecting data: tests, questionnaires, interviews, classroom
observations, diaries and journals (Zohrabi, 2013). The process of recruiting participants for
quantitative research is quite different from that of qualitative research and sample bias is a major failing
in research design and can lead to inconclusive, unreliable results (Zikmund, 2003). Many researchers
have focused on instrument development to measure health phenomena and as a result, appropriate
instruments can be easily found for use in research and practice (Youngshin, Youn-Jung and Doonam,
2015).

Quite often, quantitative designs use tests and closed ended questionnaires in order to gather, analyse
and interpret the data. However, the qualitative methods mostly make use of interviews, diaries,
journals, classroom observations and open-ended questionnaires to obtain, analyse and interpret the
data (Zohrabi, 2013). Use of existing instruments may provide the advantage of cost-effective and
knowledge accumulation, however, instruments should be used in the same way that they were
designed, to fit the situation in terms of place, time and population (Youngshin et al., 2015).

The researcher should focus on the format of the questionnaire with attention to layout, readability, time
demands on respondents, logic and clarity of content (Youngshin et al., 2015). The researcher can revise
the instrument as needed based on feedback provided and prepare a protocol for implementing the
questionnaire (Youngshin et al., 2015).

93
If a distortion of the measurement occurs because respondents’ answer are falsified or misrepresented,
either intentionally or inadvertently, the sample bias that occurs is a response bias (Zikmund, 2003). In
this research the questionnaire had mostly closed ended questions. This meant there was little or no
possibility of misinterpreting the response.

Nonresponses can affect the quality of the survey separately or jointly and increase the total number of
survey errors (Toepoel and Schonlau, 2017). Non-response error is always a potential problem with
mail surveys and individuals who are interested in the general subject, the survey are more likely to
respond than those with less interest or little experience (Zikmund, 2003). Nonresponse can stem from
the inability to contact potential respondents, from the unit’s refusal or lack of cooperation, or from
language or technical difficulties (Toepoel and Schonlau, 2017).

If some groups are misrepresented (in a way that is not corrected by applying sampling weights to adjust
for unequal probability of selection), and if these groups behave differently with respect to the
investigated question, then nonresponse is selective and results are biased (Toepoel and Schonlau,
2017). Despite the methodological challenges encountered by nonresponse follow-ups, compliance to
the measurement invariance and non-respondents representativity, they are good tools to comprehend
the nonresponse and nonresponse mechanism (Vandenplas et al., 2015).

In practice, indicators measuring the representativeness of the collected data help overcome possible
nonresponse biases and post survey adjustment techniques can then be implemented to help reduce
nonresponse biases (Toepoel and Schonlau, 2017). Identifying good participation indicators is a key
step in detecting and treating nonresponse and nonresponse bias (Vandenplas et al., 2015). Preventing
nonresponse also passes through devising the questionnaire acceptable to the respondents (Toepoel and
Schonlau, 2017).

There are many different types of bias and participation in questionnaire must be on a voluntary basis,
if only those people with strong views about the topic being researched volunteer then the results of the
study may not reflect the opinions of the wider population creating a bias (Zikmund, 2003). According
to Toepoel and Schonlau (2017) post survey adjustment techniques, including imputation and
weighting, are devised to reduce nonresponse biases.

Imputation methods rely on information available on individuals for other variables than those to impute
and missing values can be replaced by the mean of the variable to impute or by values forecast in a
regression by other explanatory variables (Toepoel and Schonlau, 2017). Missing values due to attrition
can be reduced by extrapolating from previous waves and deterministic imputation methods tend to
underestimate the variances (Toepoel and Schonlau, 2017). The introduction of a random component,

94
which increases the variances, has the merit to counterbalance this effect and different imputation
techniques are commonly used to contain the biases introduced by a specific technique (Toepoel and
Schonlau, 2017).

According to Leeuw, Hox and Huisman (2003) three main patterns can be discerned in item missing
data, namely;

 The data are missing systematically by design

 All the data are missing after a certain point in the questionnaire (partial nonresponse)
 Data are missing for some items for some respondents (Item nonresponse).

According to Toepoel and Schonlau (2017) there are techniques to deal with nonresponse biases,
namely;

 Post stratification consists of distributing the population into groups using auxiliary common
variables such as sex, age and education so that the auxiliary variables are distributed as in the
whole population. This is achieved by dividing the population percentage of a post stratification
cell by the sample percentage in that cell and using the ratio as weight.
 Linear regression involves neither joint nor marginal distributions but helps adjust sample
estimates to population parameters. The estimate from the sample is equated to the population
total output. The weights are chosen to fit the population totals and can be viewed as regression
coefficients.
 Adjustment by propensity score is devised to modify the mean values of the auxiliary variables
in the sample closer to those estimated from a higher-quality sample of reference. The common
procedure is to regress the indicator variable of the sample versus the sample of reference on
attitudinal or web-related variables. Inverse propensity scores can be used as weights. The
quality of the adjustment depends on the relevance of auxiliary variables to the question under
study and to their correlations with the response biases.

3.9 Questionnaire Design

Questionnaire design is more of an art than science and the importance during the process of the
questionnaire design is attention to the purpose of the questionnaire (Youngshin et al., 2015). Before
designing the survey questions, some features and format of the survey should also be touched upon as
the look of the questionnaire survey will affect the respondents’ reaction to it (Dillman, 2007). The flow
of items should be clear and easy to understand in order to gather precise information and when using

95
an existing questionnaire and performing cultural adaptation, psychometric properties and cultural
equivalence should be initially evaluated (Youngshin et al., 2015). A pilot test will help to evaluate
preliminary questions prior to administration to avoid later mistakes (Youngshin et al., 2015).

According Youngshin et al. (2015) most respondents have the tendency to respond to questionnaires
without considering how missing responses will be analysed, how they will contribute to answering
research questions and how researchers will account for questionnaires that are not returned.

Most researchers experience issues related to non-response when self-report questionnaires are used
and the literature has offered suggestions on how to avoid those problems and how to develop
questionnaires to measure psychological constructs more concisely (Youngshin et al., 2015).

The questionnaire used in this research had two parts. Part one was the particulars of respondents and
part two was the research questions. The research questions used a Likert scale. The neutral option was
used to accommodate participants that preferred not to comment or were puzzled about the statement.
The five options were presented to the participants to choose the answer that express their feelings most
accurately about the statement. The respondents were requested to use any type of indication be it a
cross, or a tick or they could circle in the relevant answer.

Validity of Likert scale is driven by the applicability of the topic concerned, in context of respondents
understanding and judged by creator of the response item and when the topic concerned is relevant to
the respondents’ context provision of more option, may add to the content & construct validity of the
scale (Joshi et al., 2015).

According to Vannette (2014) questions should be worded to be as follows,

 Be simple, direct, comprehensible

 Not use jargon
 Be specific and concrete
 Avoid ambiguous words
 Avoid negations
 Avoid leading questions
 Include filter questions
 Be sure questions read smoothly aloud
 Avoid emotionally charged words
 Avoid prestige names

96
  Allow for all possible responses

According to Youngshin et al. (2015) the process and steps for developing a scale vary depending on
what is being measured in a study and the eight steps in creating a questionnaire for a successful
epidemic study are, namely;

 Identify the leading hypotheses about the source of the problem

 Identify the information needed to test the hypotheses
 Identify the information needed for the logistics of the study and to examine confounding
factors.
 Write the questions to collect this information
 Organise the questions into questionnaire format
 Test the questionnaire
 Revise the questionnaire
 Train interviewers to administer the questionnaire

3.9.1 Scales and Measurement

According to Joshi et al. (2015) Likert scale was devised in order to measure ‘attitude’ in a scientifically
accepted and validated manner. An attitude can be defined as preferential ways of behaving/reacting in
a specific circumstance rooted in relatively enduring organisation of belief and ideas (around an object,
a subject or a concept) acquired through social interactions. According to Boone and Boone (2012) to
properly analyse Likert data, one must understand the measurement scale represented by each and
numbers assigned to Likert-type items express a “greater than” relationship, however, how much greater
is not implied and because of these conditions, Likert –type items fall into the ordinal measurement
scale.

This is clear from this discourse mentioned above that thinking (cognition), feeling (affective) and
action (psychomotor) all together in various combination/permutation constitute delivery of attitude in
a specified condition (Joshi et al., 2015). The issue is how to quantify these subjective preferential
thinking, feeling and action in a validated and reliable manner (Joshi et al., 2015).

The difficulty of measuring attitudes, character, and personality traits lies in the procedure for
transferring these qualities into a quantitative measure for data analysis purposes and the recent
popularity of qualitative research techniques has relieved some of the burden associated with the
dilemma, however, many social scientists still rely on quantitative measures of attitudes, character and

97
personality traits (Boone and Boone, 2012). According to Joshi et al. (2015) psychometrics techniques
are being developed, instituted and refined in order to meet the quantification of traits like ability,
perceptions, qualities and outlooks – the requirement of social sciences and educational researches.

According to Joshi et al. (2015) psychometrics operates through two ways, namely;

 The first is to formulate approaches (theoretical construct) for measurements, followed by

development of measuring instruments and their validation. Stanford Binet test (measures
human intelligence) and Minnesota Multiphasic Personality) are the example for the same. The
content in such instruments are rather ‘pre – fixed’.
 The other path is same up to formulation of theoretical construct for the measurement. This
conceptualization is followed by operational assembly of abstract ideas/experiences/issues
under investigation into some statements (items) largely guided by the aim of the study. This
permits the contents (items) in such scales/modes to be rather flexible and need based. Rasch
measurement model (use for estimation of ability), Likert scale (measures human attitude) are
the examples of such scales in Psychometrics used widely in the social science and educational
research.

Descriptive statistics recommended for ordinal measurement scale items include a mode or median for
central tendency and frequencies for variability, additional analysis procedures appropriate for ordinal
scale items include the chi-square measure of association (Boone and Boone, 2012).

According to Joshi et al. (2015) if one wishes to combine the items in order to generate a composite
score (Likert scale) of set of items for different participants, then the assigned scale will be an interval
scale. Likert scale data are analysed at the interval measurement scale and Likert scale items are created
by calculating a composite score (sum or mean) from four or more type Likert – type items, therefore,
the composite score for Likert scale should be analysed at the interval measurement scale (Boone and
Boone (2012). According to Joshi et al. (2015) the methods adopted for Likert scale analysis largely
depends on the item response variable assignment into ordinal or interval scale which in turn depends
on the construct of research instrument.

The measures for central tendency and dispersion for an interval scale are mean and standard deviation,
this data set can be statistically treated with Pearsons’ correlation coefficient (r), analysis of variance
(ANOVA) and regression analysis (Joshi et al., 2015). Descriptive statistics recommended for interval
scale items include the mean for central tendency and standard deviations for variability, additional data
analysis procedures appropriate for interval scale items would include the Pearson’s coefficient, t-test,
ANOVA, and regression procedures (Boone and Boone, 2012).

98
Construct of research instrument can be derived from objectives of study and objectives are the
operational form of theoretical construct of phenomenon under inquiry and designing of instruments
based upon objectives and frameworks of study decided further statistical treatment (Joshi et al., 2015).

3.9.2 Questionnaire Administration

In this research a questionnaire was used as a research instrument. Questionnaire are divided into two
different types, self-administered and interviewer administered (Saunders et al., 2003). According to
Youngshin et al. (2015) self-administration is the most popular method of administering questionnaires
in survey studies and self-administered questionnaires can be collected via post, email or electronically.
This research used a self-administered questionnaire. Self-administered questionnaires are easy to
implement, cost-effective, and protect confidentiality (Youngshin et al., 2015).

Self-administered surveys of all types typically cost less than personal interviews (Cooper and
Schindler, 2005). Moreover, they can be completed at the respondent’s convenience and administered
in standard manner (Youngshin et al., 2015).

Interview administered questionnaires allow participation by illiterate people and clarification of

ambiguity and the best method for administered questionnaires depends on who the respondents are
(Youngshin et al., 2015).

In this research, the questionnaire was distributed personally and the explanation was done face to face
with the respondents. The questionnaire was pilot tested first with 20 permanent and contractor
employees to find out completion time, any unclear questions, and other concerns from the
questionnaire. The workmates were invited to a 30 minutes engaging session during working hours after
receiving permission from the supervisors and managers. The questionnaire was handed over to the
randomly selected potential participants in groups of 15 – 20 per engaging session. The participants
included all different levels in the organization from senior management, junior management,
supervisors, engineers, inspectors, technicians, fitters, artisans, operators and others.

The challenges encountered in this research were to persuade people to spend 20 or less minutes of their
time answering the research questionnaire. The other problem was to get the managers and supervisors
to release the employees during working hours, especially those that are doing physical labour. There
was no incentive provided to employees that answered the questionnaire, may be if there was an
incentive, employees would have been attracted to participate. Employees that were systematic selected

99
in this research varied from top management of the organisations to the hard labour in the petrochemical
industry KwaZulu-Natal Province.

3.10 Sampling

According to Creswell (2015) cited in Guetterman (2015) sampling in quantitative research typically
follows random sampling procedures and researchers calculate the required sample size before
beginning the study and that size remains a constant target throughout the study. According to Alvi
(2016) the process through which a sample is extracted from a population is called sampling and in
investigation it is impossible to assess every single element of a population, so a group of people
(smaller in number than the population) is selected for the assessment. Qualitative sampling is not a
single planning decision and a reflexive researcher then makes adjustments and considers the
implications of sampling on interpretation (Emmel, 2013; Guetterman, 2015).

Sample population was selected using systematic sampling amongst permanent employees and
contractors permanently onsite. A systematic sample is designed to be an easy alternative to simple
random sampling and it is taken by deciding on what fraction of the population is to be sampled (Cooper
and Schindler, 2005). According to Alvi (2016) the inferences are drawn for the population and the
more the sample is representative of the population, the higher is the accuracy of the inferences and
better is the results (generalizable). A simple random sample is a sample taken in such a way that each
combination of individuals in the population has an equal chance of being selected and the simple
random sample is the simplest sampling plan to execute if one has a list of the population (Cooper and
Schindler, 2005).

A sample is representative when the characteristics of elements selected are similar to that of entire
target population. Sampling process may encounter the problem of systematic errors and sampling
biases (Alvi, 2016). Systematic errors can be defined as incorrect or false representation of the sample
and there are two main sampling categories, namely, probability sampling methods and non-probability
sampling methods (Alvi, 2016).

3.10.1 Probability Sampling Methods

According to Alvi (2016) probability sampling is also called random sampling or representative
sampling. In probability sampling every member of the population has a known (non-zero) probability
of being included in the sample. Some form of random selection is used. The probabilities can be
assigned to each unit of the population objectively. These techniques need population to be very
precisely defined. They cannot be used for the population that is too general a category found almost

100
everywhere in the world. If target population is defined as college students, it means person studying at
any college of the world is an element of population.

Advantages
 This sampling technique reduces the chance of systematic errors.
 The methods minimize the chance of sampling biases.
 A better representative sample is produced using probability sampling technique.
 Inferences drawn from sample are generalizable to the population.

Disadvantages
 The techniques need a lot of efforts.
 A lot of time is consumed.
 The techniques are expensive.

According to Alvi (2016) there are five methods used for probability sampling, namely;

 Simple Random Sampling

 Systematic Random Sampling
 Stratified Random Sampling
 Cluster Sampling
 Multistage Sampling

(a) Simple Random sampling: In this type of sampling each and every element of the population
has an equal chance of being selected in the sample. The population must contain a finite
number of elements that can be listed or mapped. Every element must be mutually exclusive
i.e. able to distinguish from one another and does not have any overlapping characteristics. The
population must be homogenous i.e. every element contains same kind of characteristics that
meets the described criteria of target population.

(b) Systematic Random Sampling: This type of sampling is also used for homogenous population.
Unlike simple random sampling, there is no equal probability of every element to be included.
The elements are selected at a regular interval; it may be in terms of time, space or order. This
regularity and uniformity in selection makes the sampling systematic.

(c) Stratified Random Sampling: this type of sampling method is used when population is
heterogeneous i.e. every element of population does not match all the characteristics of the

101
 predefined criteria. Sub groups are formed that are homogenous and they are called strata. The
common criterions used for stratification are gender, age, ethnicity, socioeconomic status.
However, the criterion varies greatly from investigation to investigation. The sample is selected
from each stratum randomly. There are two techniques that are used to allocate sample from
strata: proportional allocation technique and equal allocation technique. For proportional
allocation technique the sample size of stratum is made proportional to the number of elements
present in the stratum. For equal allocation technique same number of participants is drawn
from each stratum regardless of the number of elements in each stratum.

(d) Cluster Sampling: the group of elements in one geographical region is called a cluster and
sampling of cluster is called cluster sampling. This sampling technique is used when the
elements of population are spread over a wide geographical area. The population is dived into
sub-groups called clusters on the basis of their geographical allocation. The clusters ought to
be homogenous.

(e) Multistage Sampling: it is a sampling technique where two or more probability techniques are
combined. It is used when elements of population are spread over wide geographical region
and it is not possible to obtain a representative sample with only one aforementioned technique.

3.10.2 Non - Probability Sampling Methods

According Alvi (2016) every unit of population does not get an equal chance of participation in the
investigation and no random selection is made; the selection of the sample is made on the basis of
subjective judgement of the investigator.

According Alvi (2016) these techniques do not need the population to be very precisely defined and
they can be used for both types of population: the population that is too general category, and the
population that is a specific category (precisely defined).

Non – probability sampling is well suited for exploratory research intended to generate new ideas that
will be systematically tested later (Alvi, 2016) and the following are advantages and disadvantages;

Advantages
 Techniques need less effort.
 These techniques need less time to finish up.
 They are not much costly.

102
Disadvantages
 The sampling techniques are prone to encounter with systematic errors and sampling biases.
 The sample cannot be claimed to be a good representative of the population.
 Inferences drawn from sample are not generalizable to the population.

According to Alvi (2016) there are seven methods used for non-probability sampling, namely;

 Volunteer Sampling
 Convenient Sampling
 Purposive Sampling
 Quota Sampling
 Snowball Sampling
 Matched Sampling
 Genealogy Based Sampling

(a) Volunteer Sampling: the members of the sample self-selected themselves for being the part of
the study. It is not the investigator who approaches the participants rather participants
themselves reach the investigator.

(b) Convenient Sampling: it is also called accidental sampling or opportunity sampling. The
researcher includes those participants who are easy or convenient to approach. The technique
is useful where target population is defined in terms of very broad category. The target
population may be girls and boys, men and women, rich and poor.
(c) Purposive Sampling: it is not a mutually exclusive category of sampling technique rather many
other non-probability techniques is purposive in nature. In purposive sampling the sample is
approached having a prior purpose in mind. The criteria of the elements who are to include in
the study is predefined

(d) Quota Sampling: This type of sampling method is used when population is heterogeneous i.e.
every element of population does not match all the characteristics of the predefined criteria.
The elements differ from one another on a characteristic. The participants are selected non-
randomly from each group on the basis of some fixed quota.

(e) Snowball Sampling: The investigator selects a person who matches the criteria of the research.
The first participant is now asked to refer the investigator to another person who meets the same
criteria. The second participant approached is asked to refer the researcher to another one and
in this way a chain is made.

103
 (f) Matched Sampling: this technique is used in experimental researches. The main purpose of this
sampling is to take a control group to assess the effects of an intervention. The groups of
elements that resemble on a variety of variables are selected. Intervention is introduced on only
one group. The other group is used to compare with the first one to see what impacts the
intervention produced.

(g) Genealogy Based Sampling: This sampling technique has been mostly used for taking samples
from rural areas. Using this technique, instead of selecting household in an area, the members
of the entire families are selected (whether or not living in the same house). It gives a reasonable
cross section of the community by age and gender.

3.11 Data Collection

Most research begins with an investigation to learn what is already known and what remains to be
learned about a topic, including related and supporting literature, but one should also consider
previously collected data on the topic and data may already exist that can be utilised in addressing the
research questions (Creswell, 2009).

There are two types of data available, primary and secondary data. Secondary data is information
collected by others and primary data is information collected by the researcher (Ghauri and Gronhaung,
2002). The focus of this research was on the primary collected data. According to Hox and Boeije
(2005) primary data are data that are collected for the specific research problem at hand, using
procedures that fit the research problem best.

On every occasion that primary data are collected, new data are added to the existing store of social
knowledge (Hox and Boeije, 2005).

According to Hox and Boeije (2005) increasingly, this material created by other researchers is made
available for reuse by the general research community; it is then called secondary data and data may be
used for, namely;

 The description of contemporary and historical attributes

 Comparative research or replication of the original research
 Reanalysis (asking new questions of the data that were not original addressed)
 Research design and methodological advancement
 Teaching and learning

104
According to Hox and Boeije (2005) the other established primary data collection strategy is the
interview survey, a large and representative sample of an explicitly defined target population is
interviewed.

Characteristically, a large number of standardised questions are asked and the responses are coded in
standardised answer categories and a survey is carried out when researchers are interested in collecting
data on the observations, attitudes, feelings, and experiences, or opinions of a population (Hox and
Boeije, 2015). Information on subjective phenomena can be collected only by asking respondents about
these (Hox and Boeije, 2015).

In conducting research, the area of investigation and the research questions determine the method that
the researcher follows and the research method consists of how the researcher collects, analyses, and
interprets the data in the study (Creswell, 2009). Social scientists who intend to study a particular
theoretical problem or a specific policy issue have the choice to collect their own data or to search for
existing data relevant to the problem at hand (Hox and Boeije, 2005).

The most important advantage of collecting one’s own data is that the operationalization of the
theoretical constructs, the research design and data collection strategy can be tailored to the research
question, which ensures that the study is coherent and that the information collected indeed helps to
resolve the problem (Hox and Boeije, 2005).

According to Froelicher (2009) cited in Johnston (2014) the use of existing data sets can accelerate the
pace of research because some of the most time consuming steps of a typical research project, such as
measurement development and data collection are eliminated. The most important disadvantage of
collecting one’s own data is that it is costly and time-consuming, if relevant information on the research
topic is accessible, reusing it gains benefits (Hox and Boeije, 2005).

The disadvantage of using secondary data is that the secondary researcher did not participate in the data
collection process and does not know exactly how it was conducted (Johnston, 2014). The other
disadvantage of using secondary data is that the data were originally collected for a different purpose
and therefore may not be optimal for the research problem under consideration or, in the case of
qualitative data, may not be easy to interpret without explicit information on the informants and the
context; the advantage of using secondary data is a far lower cost and faster access to relevant
information (Hox and Boeije, 2005). The secondary researcher does not know how well it was done
and if data are affected by problems such as low response rate or respondent misunderstanding of
specific survey questions (Johnston, 2014).

105
3.12 Data Analysis

According to Ozgur et al. (2017) MegaStat can perform a multitude of statistical operations: descriptive
statistics, frequency distributions, probability, confidence intervals and sample size, hypothesis tests,
ANOVA, regression, time series/forecasting, chi-square, nine nonparametrics tests, quality control
process charts, and generate random numbers. SPSS and SAS, for example, have more advanced
options, especially in the area of multivariate statistics.

Statistical Packages for Social Science (SPSS) version (25) was used to analyse the data collected
through the surveys. The research applied the structural equation modelling (SEM) technique and
utilised AMOS version (25) tools to test the hypotheses among the variables in the model. Structural
equation modelling (SEM) is a statistical technique that allows the researcher to examine multiple
interrelated dependence relationships in a single model. Structural equation modelling is a popular
method in social science research, it has flexibility for interpreting the theory to be tested and the sample
data (Alshetewi et al., 2015). Descriptive statistics used in this research are median, standard deviation,
standard error of the mean, and correlations between variables.

SPSS, originally termed Statistical Package for Social Science, was released in 1968 as software
designed for social sciences and a series of companies subsequently acquired SPSS, ending with
International Business Machines (IBM), the current owner, during which time the product’s user base
was expanded (Ozgur et al., 2017). There, its former acronym was replaced with Statistical Product and
Services Solutions to reflect the greater diversity of its clients (Ozgur et al., 2017). Along with Minitab,
it is one of the leading statistical packages used in the social and behavioural sciences (Ozgur et al.,
2017).

Quantitative methods were used to analyse the data by means of statistical package SPSS. Frequency
plots were used to calculate the percentages. The mean and standard deviation were computed and the
main deficiencies were decided on when the mean was greater than 2.3. The mean that is greater than
2.3 was considered to be more towards disagree and strongly disagree. The person correlation was used
to assess relationships. A correlation coefficient (r) was used to quantify the strength of the relationship
between different process health and safety elements that affected generative health and safety culture
in the petrochemical industry. Pearson’s coefficient has a value +1 and -1 in statistics (Waters, 2001).
Correlation and experimentation are used to reduce complex interactions (Grix, 2004).

According to Waters (2001) a value of r = 1 shows that the two variables have a perfect linear
relationship with no noise at all, and as one increases so does the other. A lower positive value of r
shows that the linear relationship is getting weaker. A value of r = 0 shows that there is no correlation

106
at all between the two variables and lo linear relationship. A lower negative value of r = shows that the
linear relationship is getting stronger. A value of r = -1 shows that the two variables have a perfect linear
relationship and as one increases the other decreases. The above values of r were used in this research
as a guide to decide whether the relationship between the variables was strong or weak. The relationship
between the factors with correlation coefficients (r) less than 0.3 and greater than -0.3 are taken as
statistically not related. Correlation coefficient (r) less than 0.7 and greater than or equal to 0.3 or greater
than -0.7 and less or equal to -0.3, are taken as statistically having weaker positive and negative
correlations respectively. Correlation coefficients greater or equal to 0.7 and less or equal to 0.7 are
taken as statistically having strong positive and negative correlations respectively (Waters, 2001).

3.13 Data Screening

The data from the survey questionnaire was analysed using the following methods:
 Assessment of internal consistency using SPSS
 Confirmatory Factor Analysis using SPSS
 Structural Equation Modelling using AMOS

Data screening and preparation was carried out as follows, namely;

 Consideration for sample size

 Examination for missing data
 Evaluating univariate and multivariate normality
 Consideration of outliers, and
 Item parcelling

3.13.1 Sample Size

In most areas in life, it is very difficult to work with populations and hence researchers choose to work
with samples (Gogtay, 2010). The most important consideration in sampling is the planning of
appropriate technique to be used considering the situation on ground, and determination of sample size
adequate to ensure confidence on the inference made out of the results of the study within the limitations
under which the study (sampling) was conducted (Dahiru et al., 2006). According to Arthur et al. (2014)
it is clever to determine efficient sample size before the data collection as the studies generally estimate
the characteristics of large population and many determinants command the sample size including
object variability, sample type, estimation precision, the degree of confidence and the availability of
time and budget (Kuncel et al., 2005).

107
Sample size calculations begin with an understanding of the type of data and distribution dealt with and
very broadly, data are divided into quantitative (numerical) and categorical (qualitative) data (Goptay,
2010). In practice, this means that before carrying out any investigation one should have an idea of what
kind of change from the null hypothesis would be regarded as practically important and the smaller the
difference regarded as important to detect, the greater the sample size required (Cornish, 2006).

Sample size calculations enable researchers to draw strong robust conclusions from the limited amount
of information and also permit generalisation of results and it is however important to remember that
since it is very difficult to predict the outcome of any study, sample size calculations will always remain
approximate (Gogtay, 2010). According to Cornish (2006) one crucial aspect of study design is deciding
how big sample should be and if sample size is increased the precision of estimates increases, which
means that, for any given estimate/size of effect, the greater the sample size the more “statistically
significant” the result will be.

If an investigation is too small then it will not detect results that are in fact important and if a very large
sample is used, even tiny deviations from the null hypothesis will be significant, even if these are not,
in fact practically important (Cornish, 2006). It should be borne in mind that sample size calculation is
based on estimates and assumptions that can be inaccurate and is therefore subject to error (Patino and
Ferreira, 2016).

According to Arthur et al. (2014) the sample size can be computed by statistical formula with three
determinants, namely, the degree of confidence (historically 95 percent), the level of precision
(acceptable error), and the variability of the object (standard deviation).
The sample size was computed in the following equation:

n = {[z2 x σ (1 – σ)]/e2}/ {1 + [z2 x σ (1- σ)/e2N]} (3.1)

Where:

z: is Z-score which is equal to 1.96 associated with 95% confidence level

σ: is population standard deviation which usually uses 0.5 for safe decision
e: is margin of error with respect to a confidence level
N: is the population.

For the former, information on the mean responses in the two groups’ µ1 and µ2 are required as also
the common standard deviation for the two groups and for categorical data, ρ1 and ρ2 or information
on proportions of successes in the two groups is needed (Goptay, 2010).

108
It is also important to be realistic when choosing the estimates employed in calculating the sample size
(Patino and Ferreira, 2016). Highly optimistic choices about the effect size increase the risk of
calculating an insufficient number of participants for the sample, whereas highly pessimistic choices
can make the study unviable by resulting in a sample size that is too large to be practical (Patino and
Ferreira, 2016).

According to Cornish (2006) In most studies, particularly those involving humans, there is likely to be
a certain amount of data “lost” (or never gathered) from the original sample and this could be for a
variety of different reasons, non-response, subjects deliberately withdrawing from a study or getting
“lost” in some other way (e.g cannot be traced), subjects in clinical trial not following their allocated
treatment, or missing data (e.g on a questionnaire).

Allowance should be made for this when determining the sample size. The sample size should be
increased accordingly. The extent to which this is needed should be guided by previous experience or
a pilot study (Cornish, 2006).

Choosing the correct size of sample is not a matter of preference, it is a crucial element of the research
process without which the researcher may well be spending months trying to investigate a problem with
a tool which is either completely useless, or over expensive in terms of time and other resources (Fox,
Hunn and Matters, 2009). It is very important and mandatory that sample sizes are determined based
on the study design and the objectives of the study. Failure to calculate size of sample with reference to
particular study design may lead to incorrect results and conclusions (Dahiru, Aliyu and Kene, 2006).

3.13.2 Missing Data

Missing data is a problem because nearly all standard statistical methods presume complete information
for all the variables included in the analysis and a relatively few absent observations on some variables
can dramatically shrink the sample size (Soley-Bori, 2013). According to Tabachnick & Fidel (2013);
Hair et al. (2010) cited in Harinarain (2013) considering the relatively few missing values in the dataset
and the random pattern of missingness the researcher deemed it acceptable to ignore the missing data
As a result, the precision of confidence intervals is harmed, statistical power weakens and the parameter
estimates may be biased (Soley-Bori, 2013).

According to Tabachnick & Fidel (2013); Hair et al. (2010) cited in Harinarain (2013) if the data is
missing in a random pattern the choice off procedure to handle missing values is not important since
most procedures will yield comparable results.

109
Appropriately dealing with missing data can be challenging as it requires a careful examination of the
data to identify the type and pattern of missingness, and also a clear understanding of how the different
imputation methods work (Soley-Bori, 2013). Sooner or later all researchers carrying out empirical
research will have to decide how to treat missing data (Soley-Bori, 2013).

In survey, respondents may be unwilling to reveal some private information, a question may be
inapplicable or the study participant simply may have forgotten to answer it (Soley-Bori, 2013).

According Soley - Bori (2013) there are different assumptions about missing data mechanisms:

 Missing completely at random (MCAR): suppose variable Y has some missing values. Say
these values are MCAR if the probability of missing data on Y is unrelated to the value of Y
itself or to the values of any other variable in the data set. However, it does allow for the
possibility that on Y is related to the missingness on some other variable X.
 Missing at random (MAR) a weaker assumption than MCAR: The probability of missing data
on Y is unrelated to the value of Y after controlling for other variables in the analysis (say X).
 Not missing at random (NMAR): missing values do depend on unobserved values.

3.13.3 Univariate and Multivariate Normality

According Kline (2011); Schumacher & Lomax (2004) cited in Harinarain (2013) structural equation
modelling estimation methods assumes multivariate normality, which means that (1) all the univariate
distributions are normal (2) the joint distribution is bivariate normal and (3) bivariate scatterplots are
linear.

According to Kline (2011); Pallant (2010); Raykov & Marcoulides (2006); Schumacker & Lomax
(2004); Tabachnick & Fedel (2013) cited in Harinarain, (2013) skewness describes the symmetry of a
distribution, whereas kurtosis describes the shape of the distribution. Positive skew indicates that most
of the scores are below the mean, and negative skew indicates most of the scores are above the mean.
Positive kurtosis indicates heavier tails and a higher peak described as leptokurtic and negative kurtosis
indicates lighter tails and lower peak

According to Rousseeuw and Hubert (2011) multivariate M-estimators have a relatively low breakdown
value due to possible implosion of the estimated scatter matrix. More recently, robust estimators of
multivariate location and scatter include S-estimators, MM-estimators, and the orthogonalised
Gnanadesikan-Kettenring (OGK) estimator.

110
3.13.4 Outliers

An outlier is a data point that is significantly different from the remaining data. Outliers are also referred
to as abnormalities, discordants, deviants, or anomalies in the data mining and statistics literature
(Aggarwal, 2016).

In most applications, the data is created by one or more generating processes, which could either reflect
activity in the system or observations collected about entities and when the generating process behaves
unusually, it results in the creation of outliers (Aggarwal, 2016). Outlying observations may be error,
or they could have been recorded under exceptional circumstances, or belong to another population,
consequently they do not fit the model well and it is very important to be able to detect these outliers
(Rousseeuw and Hubert, 2011).

An outlier often contains useful information about abnormal characteristics of the systems and entities
that impact the data generation process and the recognition of such unusual characteristics provides
useful application-specific insights (Aggarwal, 2016).

There are many trade-offs associated with model choice, a highly complex model with too many
parameters will most likely over fit the data, and will also find a way to fit the outliers (Rousseeuw and
Hubert, 2011).

A simple model, which is constructed with a good intuitive understanding of the data (and possibly also
and understanding of what the analyst is looking for), is likely to lead to much better results and on the
other hand, an oversimplified model, which fits the data model, is perhaps the most crucial one in outlier
analysis (Aggarwal, 2016). When analysing data, outlying observations cause problems because they
may strongly influence the results. Robust statistics aims at detecting the outliers by searching for the
model fitted by the majority of the data (Rousseeuw and Hubert, 2011).

SPSS defines outliers if they extend more than 1.5 box – lengths from the edge of the box. Extreme
points are those that extend more than three box – lengths from the edge of the box and are indicated
with an asterisk (Harinarain, 2013).

In practice, one often tries to detect outliers using diagnostics starting from a classical fitting method,
classical methods can be affected by outliers so strongly that the resulting fitted model does not allow
detecting the deviating observations (Rousseeuw and Hubert, 2011). This is called the masking effect.
In addition, some good data points might even appear to be outlier, which is known as swamping and

111
to avoid these effects, the goal of robust statistics is to find a fit that is close to the fit that would have
been found without the outliers (Rousseeuw and Hubert, 2011).

According to Jung Kim (2017) practitioners often “toss out” such anomalous points, which may or may
not be a good idea and if it is clear that an outlier is the result of a mishap or a gross recording error,
then this may be acceptable and on the other hand, if no such basis may be identified, the outlier may,
in fact, be a genuine response.

There are other aspects to robust statistics apart from outlier detection (Rousseeuw and Hubert, 2011).
For instance, robust estimation can be used in automated settings such as computer vision, another
aspect is statistical inference such as the construction of robust hypothesis tests, ρ – values, confidence
intervals, and model selection (e.g., variable selection in regression) (Rousseeuw and Hubert, 2011).

According to Jung Kim (2017) there are strategies to deal with outliers and the suggestions are as
follows,

 Delete outliers and redo the analysis (new outliers may surface).
 Sometimes the purpose of the experiment is just to identify the outliers. In this case there is
no need to redo the analysis.
 Check the experimental circumstances surrounding the data collection for the outlying cases.
 Report the analysis both with and without the analysis and let the reader decide.

3.13.5 Item Parcelling

According to Sterba and MacCallum (2010) when the conservative unidimensionalty requirements for
the use of item parcels are met, there can be substantial variability in parameter estimates and fit indexes,
due to different allocations of items to parcels from a 12 item scale.

According to Brown (2006) cited in Harinarain (2013) the advantage of using parcels when specifying
models, is a more parsimonious model as fewer parameters need to be estimated and the distribution of
these parcels more closely resemble normal distribution than the original items. Item parcelling can
therefore be used as a remedial approach to address non-normality and the one disadvantage of
parcelling is that parcelling depends on the unidimensionality of the items being combined (Bandalos,
2002; Brown, 2006; Harinarain, 2013).

112
According to Marsh et al. (2013) it is recommended to,

 Avoid using parcels to camouflage method effects, cross loadings, and other sources of
misspecification at the item level. It is better to systematically model the misspecification at
the item level. Failure to do so might turn out to be substantively important and will typically
bias interpretations of substantively important parameter estimates.
 The use of item parcels is only justified when there is good support for the unidimensionality
of all the constructs at the item level for the particular models and sample being considered.
Tests for this requirement should be conducted for the complete model at the item level – not
simply the evaluation of each construct separately, which is likely to ignore many forms of
misspecification.
 If parcels are used, then applied researchers should use homogeneous rather than distributive
parcel strategies that confound sources of misfit with allocation items to parcels. They should
make explicit the parcelling strategy used, its justification, and particularly if sampling
variability is large. Justification for the use of item parcels should be accompanied by at least
a summary of the corresponding results based on item analyses and any substantively
important difference between the two.
 Applied researchers will continue to argue, sometimes with good justification, that there are
situations in which it is not reasonable to replicate fully the analyses based on parcels with
models based on items.

It is almost always preferable to evaluate latent variable models based on item-level data, particularly
when the sample size is appropriate (Marsh et al., 2013). A priori use of item parcels (an extreme
pragmatist perspective) is never justified without clear support for the a priori model at the item level
and unidimensionality in relation to the models and data under consideration (Marsh et al., 2013).

Expedient compromises between parsimony and accuracy in applied research when sample sizes are
modest (e.g., the use of parcel scores) are likely to be biased under typical conditions (Marsh et al.,
2013). They should be avoided unless there is clear evidence that the very restrictive unidimensinality
assumptions upon which they are based are met, or that the sizes of biases are trivially small and
substantively unimportant (Marsh et al., 2013).

113
3.14 Reliability and Validity

According to Zohrabi (2013) one of the main requirements of any research process is the reliability of
the data and findings. Reliability means consistency or the degree to which a research instrument
measures a given variable consistently every time it is used under the same condition with the same
subjects (Yilmaz, 2013). Reliability refers to the extent to which our measurement process provides
consistent or repeatable results and it indicates how closely the results of repeated measurements of the
same concept agree: a reliable measurement of quantify known not to have changed that is performed
twice with the same person will produce the same value both times (Zohrabi, 2013). The same is true
where two different people are involved who do not differ on the measured characteristic.

It is important to note that reliability applies to data not to measurement instruments and from different
perspectives or approaches, researchers can evaluate the extent to which their instruments provide
reliable data (Yilmaz, 2013).

Cronbach’s alpha test is the most generally used measurement to test the reliability or internal
consistency of the data collection instruments, while internal consistency normally tells how well the
items in the test measure the same construct (Zohrabi, 2013). The two essential elements to assess the
instrument are validity and reliability, an instrument measurement cannot be authenticated without
reliability (Yilmaz, 2013). The reliability and validity of the constructs must be scrutinized and assessed
before using them in the structural model (Schreiber et al., 2006).

The research used the questionnaire survey to gather the data, and the reliability of the questionnaire
was tested in order to produce validated and reliable results. Cronbach’s alpha test was used in the data
analysis to measure the internal consistency of the items in the questionnaires.

The concept of validity in quantitative study corresponds to the concept of credibility, trustworthiness,
and authenticity in qualitative study which means that the study findings are accurate or true not only
from the standpoint of the researcher but also from that of the participants and the readers of the study
(Cresswell & Miller, 2000; Yilmaz, 2013). The concept of reliability in quantitative study is
comparable, but not identical, with the concept of dependability and auditability in qualitative study,
which means that the process of the study is consistent over time and across different researchers and
different methods Yilmaz, 2013).

Validity refers to the extent to which our measurement process is measuring what we intend to be
measuring and there is content validity that confirms how well the sample of questions reflects the
domain of possible questions (Vannette, 2014). According to Lipsey (2009), a study with both strong

114
internal validity and strong external validity would require: a relatively large research sample randomly
drawn from the relevant population (external validity), randomly assignment to intervention and control
conditions with attrition (internal validity).

Criterion - related validity looks at what is the strength of the empirical relationship between question
and criterion, the construct validity is more on how closely does the measure behave like it based on
established measures or the theory of the underlying construct (Vannette, 2014). Construct validity is
concerned with the question on how much the construct measured by the research reveals the
hypothesised construct (Mona and Martin, 2016).

When considering the validity of a research study, there is a need to ask two basic questions. First, does
the study have sufficient controls to ensure that the conclusions drawn are truly warranted by the data?
And second, can what is observed be used in the research situation to make generalisations about the
world beyond that specific situation? The answers to these two questions address the issues of internal
validity and external validity, respectively (Leedy and Ormrod, 2005).

Validity is the extent to which an instrument measures what it is supposed to measure and performs as
it is designed to perform and it is rare, if nearly impossible, that an instrument be 100% valid, so validity
is generally measured in degrees (Vannette, 2014).

Internal validity is concerned with how much the variations in the dependent variables is caused by
independent variables (Grajo et al., 2016). To attain high internal validity, consideration should centre
on no significant related data being overlooked and suitable methods should be nominated (Schick and
Vaughn, 2002).

As a process, validation involves collecting and analysing data to assess the accuracy of an instrument
(Vannette, 2014). Survey design should begin with a data analysis plan that can provide structure and
help to avoid many problems, every question must have a purpose, and every question should produce
the best possible data for the purpose of the study (Vannette, 2014).

The measurement indicators or variables have different correlations or loading values as the indicators
measure the latent construct in different degrees. Measurement indicators having low loadings should
be eliminated since they offer small explanatory power to the model (Aibinu and Al-Lawati, 2010). A
common acceptable threshold value for a good indicator is having a loading higher than 0.5. The value
of Cronbach’s alpha ranges from 0 to 1 as the value become closer to 1 the higher reliability of the
items. The reliability test should be conducted before the construct validity analysis is commenced.
Constructs are considered reliable when Cronbach’s alpha is 0.7 or higher (Alshetewi et al., 2015). The

115
lack of researcher’s faithfulness can cause the significance of the information gathered to be partial,
raising doubts regarding its reliability and validity (Vannette, 2014).

3.15 Regression Analysis

According to Campbell and Campbell (2008) regression analysis is a statistical method to determine
the linear relationship between two or more variables and regression is primarily used for prediction
and casual inference. In its simplest (bivariate) form, regression shows the relationship between one
independent variable (X) and a dependent variable (Y), as in the formula below:

Y = β0 + β1X + u (3.2)

The magnitude and direction of that relation are given by the slope parameter (β1), and the status of the
dependent variable when the independent variable is absent is given by the intercept parameter (β0), an
error term (u) captures the amount of variation not predicted by the slope and intercept terms, the
regression coefficient (R2) shows how well the values fit the data (Campbell and Campbell, 2008).

According to Rousseeuw and Hubert (2011) the multiple linear regression model assumes that in
additional to the ρ independent x – variables also a response variable y is measured, which can be
explained by a linear combination of the x – variables.

According to Campbell and Campbell (2008) a regression analysis is usually centred on describing and
evaluating the relationship between a given variable Y (dependent variable) and one or more other
variables X1,X2,…..Xn (independent variable). Regression models include more than one independent
variables are called multi regression models. One of the vital assumptions for regression analysis is that
there is a linear relationship between the dependent variable and independent variable.

According to Jung Kim (2017) regression is one of the most widely used statistical technique, estimates
relationships among variables and models provide a very flexible framework for describing and testing
hypotheses about relationship between explanatory variables and a response variable.

According to Jung Kim (2017) typically, a regression analysis is used for the following purposes,
namely;

 Modelling the relationship between variables.

 Prediction of the target variable (forecasting).

116
  Testing of hypotheses.

The basis of regression analysis is the linear model. The model can be characterised as follows. Where
there are n sets of observations {X1i, X2i,….., Xpi, Yi}, i = 1, …., n, which represent a random sample
from a larger population. It is assumed that these observations satisfy a linear relationship

In this research the relationships between the process health and safety elements was tested against
generative health and safety culture. According to Jung Kim (2017) the regression analysis can deliver
a simple understanding of the relations. Algorithms for exploratory factor analysis include principal
component analysis, and an algorithm for structural equation modelling using the analysis of a moment
structures (AMOS).

3.16 Principal Component Analysis

According to Rousseeuw and Hubert (2011) cited in Zhao (2017) principal component analysis is a
very popular dimension – reduction method and it tries to explain the covariance structure of the data
by a small number of components. According to Zhang et al. (2013) cited in Zhao (2017) the principal
component regression methods combine linear regression with principal component analysis and it
transfers a set of highly correlated independent variables into a set of uncorrelated independent principal
components.

These components are linear amalgamations of the original variables and often allow for an
interpretation and a better understanding of the different sources of variation and PCA is often the first
step of the data analysis, followed by other multivariate techniques (Rousseeuw and Hubert, 2011; Zhao
2017).

The principal component analysis not only litigates the multicollinearity problem but also reveals which
independent variables should be predictors (Zhang et al., 2013; Zhao, 2017).

According to Zhang et al. (2013) cited in Zhao (2017) the spatial pattern’ new variables remove the
properties caused by multicollinearity after principal component analysis, as the ideal predictors to use
in a regression analysis and the new variables are mutually orthogonal and uncorrelated which are linear
combinations of the original variables.

In the classical approach, the first principal component corresponds to the direction in which the
projected observations have the largest variance. The second component is then orthogonal to the first

117
and again maximises the variance of the data points projected on it (Rousseeuw and Hubert, 2011; Zhao,
2017).

Continuing in this way produces all the principal components, which correspond to the eigenvectors of
the empirical covariance matrix, but unfortunately, both the classical variance matrix (which is being
decomposed) is very sensitive to anomalous observations. Consequently, the first components from
classical PCA are often attracted toward outlying points and may not capture the variation of the regular
observations (Rousseeuw and Hubert, 2011; Zhao, 2017).

According to Zhang et al. (2013) cited in Zhao (2017) the number of principal components is often
reduced due to only several eigenvectors being used where their corresponding eigenvalues are greater
or equal to 1 and this is why the principal component analysis can reduce the number of independent
variables.

According to Zhang et al. (2013) cited Zhao (2017) principal components are linear combinations of
the original variables, but they are orthogonal, the percentage of total variance in the original data is
represented by each principal component is the reasonable number of principal components that are
selected depends on the cumulative variance being over 85 – 90%.

λiai = Vai (3.3)

Where:

λ is the eigenvalue of V
ai is the corresponding eigenvector

The principal components can be computed in

PCt (i) = ai Xt, i = 1,2,……,p (3.4)

The percentage of total variance in the original data represented by each principal component is
calculated as follows:

Li = λi/(Ʃi = 1 λi) x 100% (3.5)

118
Where:

λi represents the ration of the component i to the total components

p is the total number of components
Li represents the variance of component i

According to Tabachnick and Fidell (2007) cited in Yong & Pearce (2013) principal component
analysis is used to extract maximum variance from the data set with each component thus reducing a
large number of variables into smaller number of components. Principal components analysis is data
reduction technique and the issues of whether it is truly a factor analysis technique has been raised
(Costello & Osborne, 2005; Yong & Pearce, 2013).

Researchers may use Principal Component Analysis as the first step to reduce the data, then follow-up
with a true factor analysis technique, the factor loadings are fairly similar and will need to perform
rotation regardless of the extraction technique (Tabachnick & Fidell, 2007; Yong & Pearce 2013). It is
best to pick the extraction technique based on research question and the ease of interpretation (Yong &
Pearce, 2013).

3.17 Structural Equation Modelling

Structural Equation Modelling is an attempt to model casual relations between variables by including
all variables that are known to have some involvement in the process of interest and the first step in
structural equation modelling is to specify a model (Raykov, 2006; Zhao, 2017). Structural equation
modelling can be engaged to capture complex relationships between one or more dependent and
independent variables that can be sources from qualitative or quantitative data (Hox and Kleiboer, 2007;
Zhao, 2017).

A model is simply a statement or set of statements about the relations between variables and structural
equation modelling combines multiple regression analysis and factor analysis together to analyse the
relationship between measured variables and latent variables (Raykov, 2006; Zhao, 2017). Structural
equation modelling while resembling the regression analysis mostly, is a very powerful statistical
technique that models interactions, it can cope with non-linear situations and lets correlation among
independent variables (Hox and Kleiboer, 2007; Zhao, 2017).

Missing data is a serious issue in SEM and must be discussed in any article. Also, given new
technologies, more options can handle missing data, such as maximum likelihood estimation (Schreiber

119
et al., 2006). Although the problem of missing values is not unique to structural modelling, estimating
a successful model necessitates the appropriate handling of missing data from a methodological, as well
as conceptual, perspective (Schumaker & Lomax, 1996).

Reliance on pairwise deletion can result in a nonpositive covariance matrix, and other methods,
including replacement with the mean, may result in heteroscedastic error (Schumaker & Lomax, 1996).

Structural Equation modelling estimates the degree to which a hypothesized model fits the data and in
a confirmatory factor analysis, goodness of fit indexes is estimated for each latent variable as a distinct
structural model (Schreiber et al., 2006).

The structural equation modelling normally encompasses the latent variables, the measures variable and
regression paths, the latent variable can be identified as an eclipse or circle, while the measured or
observed variable is showed as a rectangular shape while the regression path is expressed by an arrow
(Child, 2006; Young and Pearce, 2013).

According to Schreiber et al. (2006) although it is wise and appropriate for one to measure items found
in other studies to form a certain construct, it is not appropriate to assume that a certain group of items
found to form a valid and reliable construct in another study will form an equally valid and reliable
construct when measured in a different set of data.

Constructs tested on a national data set are valid in a new study in the rare instance when the new study
uses the identical observations analysis in the same data with the same theoretical under pinning.
Divergent choices addressing the problem of missing data will normally change construct validity
results such that a new confirmatory analysis is appropriate (Schreiber et al., 2006).

According to Byrne (2012) there three types of structural equation modelling analysis, namely;

 Strictly confirmatory: The researcher postulates a single model based on theory, collects the
appropriate data, and then tests the fit of the hypothesised model to the sample data. The results
of this test, the researcher either rejects or fails to reject the model. No further modifications to
the model are made.
 Alternative models: The researcher proposes several alternative or competing models, all of
which are grounded in theory. Following analysis of a single set of empirical data, the
researcher selects one model as most appropriate in representing the sample data.

120
  Model generating: The researcher postulate and reject a theoretically derived model on the
basis of its poor fit to the sample data, proceeds in an exploratory rather than confirmatory
fashion to modify and re-estimate the model. The primary focus, in this instance, is to locate
the source of misfit in the model and to determine a model that better describes the sample data.

The two main factor analysis techniques are Exploratory Factor Analysis (EFA) and Confirmatory
Factor Analysis (CFA) and it attempts to confirm hypotheses and used path analysis diagrams to
represent variables and factors, whereas EFA tries to uncover complex patterns by exploring the dataset
and testing predictions (Child, 2006; Young and Pearce, 2013).

3.17.1 Exploratory Factor Analysis

According to Yong and Pearce (2013) factor analysis is used to identify latent constructs or factors and
it is commonly used to reduce variables into a smaller set to save time and facilitate easier
interpretations. Exploratory factor analysis is used when a researcher wants to discover the number of
factors influencing variables and to analyse which variables go together (DeCoster, 1998; Yong and
Pearce, 2013).

A basic hypothesis of EFA is that there are m common latent factors to be discovered in the dataset,
and the goal is to find the smallest number of common factors that will account for the correlations
(McDonald, 1985; Yong and Pearce, 2013). There are many extraction techniques such Principal Axis
Factor and Maximum Likelihood and factor analysis is mathematically complex and the criteria used
to determine the number and significance of factors are vast (Yong and Pearce, 2013).

There are two types of rotation techniques – orthogonal rotation and oblique rotation and orthogonal
rotation (e.g., Varimax and Quartimax) involves uncorrelated factors whereas oblique rotation (e.g.,
Direct Oblimin and Promax) involves corrected factors (Yong and Pearce, 2013).

According to Yong and Pearce (2013) the interpretation of factor analysis is based on rotated factor
loadings, rotated eigenvalues, and scree test and in reality, researchers often use more than one
extraction and rotation technique based on pragmatic reasoning rather than theoretical reasoning.

3.17.2 Confirmatory Factor Analysis

According to Schreiber et al. (2006) confirmatory factor analysis and structural equation modelling are
statistical techniques that one can use to reduce the number of observed variables into a smaller number
of latent variables by examining the covariation among the observed variables. Confirmatory factor

121
analysis allows for the testing and refining of research constructs in relation to a theoretical framework
(Yale et al., 2015). Confirmatory factor analysis is carried out to test whether the measure of a latent
construct corresponds with the study of the nature of the individual factor and at this stage, the indicator
elimination and model re-specification are performed for each latent construct (Schreiber et al., 2006).

Confirmatory factor analysis and structural equation modelling can each be an iterative process by
which modifications are indicated in the initial results, and parameter constraints altered to improve the
fit of the model, if such changes are warranted theoretically (Schreiber et al., 2006). CFA is a
theoretically driven process as opposed to EFA which is data driven (Byrne, 2006). If parameter is freed
on the basis of a high modification index value, the researcher is called on to theoretically defend the
change indicated so that the final model does not deviate from the initial theoretical model (Schreiber
et al., 2006).

Confirmatory factor analysis was adopted to test the strength and appropriateness of the relationship
between latent constructs and corresponding measurement. CFA is also known as a measurement model
as it can be used to assess the fit between the collected data and the conceptual relationship between
process health and safety elements (Leadership Commitment, Chemical Exposure Management, Health
and Safety Risk Assessment, Process Hazard Analysis, Permit to Work, Training and Competency,
Process Health and Safety Information, Control of Confined Space Entry, Operating Procedure and
Control of Ignition Source) identified through literature review and the latent variable (Generative
Health and Safety Culture).

3.18 Path Modelling Process

According to Schumacker & Lomax (2004) cited in Harinarain (2013) Path models and structural
models can be conducted through five basic steps, namely;

 Model Specification
 Model Identification
 Model Estimation
 Model Testing
 Model Modification

122
 (a) Step 1: Research Model Specification

According to Schumacker and Lomax (2010) involves using all of the available relevant theory,
research and information to develop a theoretical model and prior to any data collection or analysis, the
researcher specifies a particular model that should be confirmed using variance –covariance data.

Model specification involves determining every relationship and parameter in the model that is of
interest to the researcher and is usually the first step in path models, structural equation modelling and
regression models as it determines every relationship and parameters (Schumacker and Lomax, 2004;
Harinarain 2013).

This is usually the hardest part of modelling as it involves developing a theoretical model and in order
to avoid misspecification the theoretical model needs to be consistent with the true model which requires
the specification of all relevant variables (Schumacker and Lomax, 2004; Harinarain 2013). Model
specification was the first step performed in this research. Path modelling was used since this study has
observed dependent variables.

(b) Step 2: Research Model Identification

During model identification process the researcher has to determine if there is sufficient information to
obtain a unique solution for the parameters to be estimated by the model (Byrne, 2012; Harinarain,
2013). Model identification was the second step done in this research. According to Brown (2006);
Byrne (2012); Hair et al. (2010); Kaplan (2000); Kline (2011); Schumacker & Lomax (2004) cited in
Harinarain (2013) there are three types of model identification, namely;

 Under –identified models: where one or more parameters cannot be estimated or uniquely
determined in the matrix due to lack of information and has negative degrees of freedom. This
type of model can be unstable and should be looked at with scepticism as it is difficult to
determine unique values for the model coefficients.
 Just-identified: (also referred to as a saturated model) models where all of the parameters are
uniquely determined because there is just enough information in the matrix. This model
contains zero degrees of freedom. The model perfectly reproduces the data resulting in a perfect
fit.
 Over-identified models: these are the most preferred type of identification, when there is more
than one way of estimating a parameter/s because there is more than enough information in the

123
 matrix and the model has positive degrees of freedom. (Brown, 2006; Byne, 2012; Hair et al.,
2010).

(c) Step 3: Research Model Estimation

According to Kline (2011) cited in Harinarain (2013) during the estimation process the model fit is
evaluated, the parameter estimates are interpreted and equivalent or near-equivalent models are
considered. Model estimation was the third step done in this research. The aim of estimation is to
generate numerical values for free parameters within the model that produces the implied matrix such
that the parameter values yield a matrix as close as possible to the sample covariance matrix (Kline,
2011; Schumacker & Lomax, 2004; Harinarain, 2013).

(d) Step 4: Research Model Testing

Model testing is the fourth step that determines how well the data fits the model. Various authors have
cautioned against the use of goodness-of-fit indices as rules of thumb or golden standards (Ho, 2006;
Hu and Bentler, 1999; Schmitt, 2011; Harinarain, 2013). This is because all research is unique and
sample sizes differ. These authors believe that consideration should be given to the adequacy of model
parameters and complexity and therefore the model fit statistics should be used as guidelines as they
only provide information on a models lack of fit and do not reflect the extent to which the model is
plausible. Even if a model fits well, data can never confirm a model, they can only fail to disconfirm a
model (Kline, 2011; Harinarain, 2013).

(e) Step 5: Research Model Modification

According to Schumaker and Lomax (2010) if the fit of the implied theoretical model is not as strong
as one would like (which is typically the case with an initial model), then the next step is to modify the
model and subsequently evaluate the new modified model. Model modification is carried out in order
to improve the model so that a better fitting model and/or more parsimonious model which were
substantively more interpretable can be obtained (Kline, 2011; Harinarian, 2013).

According to Kline (2011) cited in Harinarain (2013) model modification should be guided by
theoretical considerations and not just determined by statistical results, because adjusting a model after
initial testing increases the chance of making a type 1 error.

124
3.19 Chapter Summary

This chapter outlined the steps followed in the research as a research design and methodology. There
were different research designs and methodologies that were examined in this chapter to simplify this
research. The research approaches and methods that were best suited to the research topic were selected.
The type of data used was mainly primary data for the quantitative research. A questionnaire was used
as a research instrument to collect primary data and the literature review was used for the secondary
data.

125
 CHAPTER FOUR
Presentation of Results
_____________________________________________________________________
4.1 Introduction

This chapter analyses the data that has been obtained from the respondents to establish relationships
that are available in the data and cross-examining that with the literature review. The data analysis aims
to identify the main factors that need senior management’s attention to improve the behaviour and mind-
set of employees to generative health and safety culture.

4.2 Questionnaire Administration

The questionnaire was piloted with about 30 employees within a petro-chemical organization to find
out completion time and any concerns from the questionnaire. The completion time established be
between 15 and 25 minutes depending on each respondents speed. The questionnaire was handed over
to the randomly selected potential participants in groups of 15 to 30 per engaging session. The
participants included all different levels in the organisations, senior management, junior management,
supervisors, engineers, inspectors, technicians, fitters, artisans, operators and administration employees.
The questionnaire was explained to the group and employees were allowed to decide whether they are
willing to participate to the research or not. Those that were willing to participate in the research were
then allowed to answer the questionnaire and those that opted out were allowed to leave. The potential
participants received the questionnaire and 5 to 10 minutes was spent to explain the reason for the
survey and the potential respondents were given a week to return the answered questionnaire.

4.3 Questionnaire Responses

According to Fincham (2008), a survey response rate helps to ensure that survey results are
representative of the target population. Response rates are calculated by dividing the number of usable
responses returned by the total number eligible in the sample chosen. The study was conducted by
distributing 400 questionnaires manually to potential participants within a single petrochemical
organization and therefore considered a convenience sample. The targeted population was 800
employees. The employees to whom questionnaires were handed during health and safety talks and
production meetings were then requested to remain behind for an explanation. The duly completed
questionnaires were 259, and were analysed using SPSS version 25. The response rate was computed
to be 64.75%.

126
4.4 Demographic Information of Participants

Table 4.1 Age and Years of Service

Minimum Maximum Median

Age 22 66 38
Years of Service 1 46 11

From Table 4.1 it is evident that the median age of participants was 38 with a minimum age of 22 years
and a maximum age of 66 years. Further, the median number of years of service was 11 years with a
minimum of 1 year and a maximum of 46 years. The participants were therefore considered matured
with considerable years of experience in the petrochemical industry. This aspect increases the reliability
of the responses received from the participants in terms of their accuracy and completeness.

Table 4.2 Gender, Marital Status and Department

Gender (%)
Male 80,6
Female 19,4
Total 100,0
Marital Status (%)
Single 37,2
Married 61,6
Divorced 1,2
Total 100,0
Department (%)
Health, Safety and Environment 8,2
Operations 50,6
Maintenance 24,1
Technical 12,5
Others 4,7
Total 100,0

Table 4.2 indicates that 80.6% of the participants were males, 61.6% were married and most of the
respondents were from the Operations Department (50.6%), followed by the Maintenance Department
(24.1%).The results show that the petrochemical industry in the case of the sample organization is still
male dominated. Operations and maintenance generally have more employees that are exposed to health
and safety risks in petrochemical industry.

127
4.5 Quantitative Data Analysis

A quantitative research methodology and the deductive research approach in the form of questionnaire
instrument was used to collect the data. The observed variables were measured using a Likert scale
where 1=Strongly Agree, 2=Agree, 3=Neutral, 4=Disagree and 5=Strongly Disagree

4.5.1 Leadership Commitment

Table 4.3 Leadership Commitment Construct

Str Str
Leadership Agree Agree Neutral Disagree Disagree Std
Commitment % % % % % Mean Dev Rank
LCH10 64.5 32.4 1.9 1.2 0.0 1.40 0.591 1
LCH7 61.2 32.2 4.3 1.9 0.4 1.48 0.707 2
LCH40 57.1 30.1 9.3 1.5 1.9 1.61 0.866 3
LCH11 51.0 37.5 9.3 2.3 0.0 1.63 0.748 4
LCH8 50.8 37.6 9.3 1.9 0.4 1.64 0.763 5
LCH17 37.6 36.0 18.6 6.6 1.2 1.98 0.966 6
LCH38 8.5 25.2 31.0 26.4 8.9 3.02 1.103 7

Table 4.3 shows the responses to statements about leadership commitment. It is evident that LCH10
(Senior Management communicates Health and Safety policy to all employees) with a mean of 1.40
ranked highest out of the seven statements presented to the participants. Further, LCH 7 (Senior
Management prioritises health and safety in my organisation) with a mean of 1.48 ranked second
highest.

LCH40 (Audit compliance is an excellent practice to prevent most of health and safety incidents in the
petrochemical industry), LCH11 (Senior Management allocates enough time to address Health and
Safety concerns), LCH8 (Senior Management has an open door policy on health and safety issues),
LCH17 (Senior Management prioritises mechanical/asset integrity of our process plant), and LCH38
(Poor housekeeping in my organisation is the cause for many health and safety incidents) ranked 3 rd to
7th within the leadership commitment latent variable.

128
4.5.2 Chemical Exposure Management

Table 4.4 Chemical Exposure Management Construct

Chemical Str Str
Exposure Agree Agree Neutral Disagree Disagree Std
Management % % % % % Mean Dev Rank
CEMH30 71.0 24.7 2.7 1.5 0.0 1.35 0.612 1
CEMH6 47.9 44.4 6.2 1.2 0.4 1.62 0.692 2
CEMH14 42.6 42.2 12.8 2.3 0.0 1.75 0.766 3
CEMH12 42.1 44.4 9.7 3.9 0.0 1.75 0.783 4
CEMH15 22.0 40.4 32.2 5.1 0.4 2.22 0.858 5
CEMH16 15.5 28.3 40.3 13.2 2.7 2.59 0.991 6

Table 4.4 shows the responses to statements about chemical exposure management. It is evident that
CEMH30 (All employees are aware that when you handling hazardous chemicals you need to use
prescribed personal protective equipment.) with a mean of 1.35 ranked highest out of the six statements
presented to the participants. Further, CEMH6 (My organisation has excellent chemical exposure
management systems.) with a mean of 1.62 ranked second highest.

CEMH14 (Most permanent employees know how to handle hazardous chemicals in the work place.),
CEMH12 (Most employees are aware of hazardous chemicals in their work environment.), CEMH15
(Contractor’s on boarding appreciates all hazardous chemicals in my organisation.), and CEMH16
(Most contractors know how to handle hazardous chemicals in my organisation.) ranked 3rd to 6th within
the chemical exposure management latent variable.

4.5.3 Health and Safety Risk Assessment

Table 4.5 Health and Safety Risk Assessment Construct

Health and Str Str
Safety Risk Agree Agree Neutral Disagree Disagree Std
Assessment % % % % % Mean Dev Rank
HSRAH9 43.6 45.2 8.5 2.7 0.0 1.70 0.737 1
HSRAH34 45.7 42.2 8.5 3.1 0.4 1.70 0.784 2
HSRAH33 22.9 32.6 29.5 10.5 4.7 2.41 1.092 3
HSRAH39 14.0 32.2 22.1 24.8 7.0 2.79 1.169 4
HSRAH32 15.1 25.5 27.8 24.7 6.9 2.83 1.166 5

Table 4.5 shows the responses to statements about health and safety risk assessment. It is evident that
HSRAH9 (There are effective noise exposure management systems in my organisation.) with a mean
of 1.70 and standard deviation of 0.737 ranked highest out of the five statements presented to the

129
participants. Further, HSRAH34 (My organisation has all management systems in place to manage
substance misuse.) with a mean of 1.70 and standard deviation of 0.784 ranked second highest.

HSRAH33 (My organisation diligently manages fatigue in both permanent employees and contractors.)
HSRAH39 (Poor health and safety risk assessments are responsible for most of health and safety
incidents in the petrochemical industry.) and HSRAH32 (Most of health and safety incidents in the
petrochemical industry are due to not verifying energy isolation before you start working on
equipment.) ranked 3rd to 5th within the health and safety risk assessment latent variable.

4.5.4 Process Hazard Analysis

Table 4.6 Process Hazard Analysis Construct

Process Str Str
Hazard Agree Agree Neutral Disagree Disagree Std
Analysis % % % % % Mean Dev Rank
PHAH24 58.8 35.0 4.7 1.2 0.4 1.49 0.680 1
PHAH21 43.2 45.6 9.3 1.5 0.4 1.70 0.732 2
PHAH20 44.4 42.9 9.3 3.1 0.4 1.72 0.787 3
PHAH23 8.2 12.5 38.5 32.3 8.6 3.21 1.038 4

Table 4.6 shows the responses to statements about process hazard analysis. It is evident that PHAH24
(In my organisation we have a comprehensive pre-activity start up review and pre-activity shutdown
review.) with a mean of 1.49 ranked highest out of the four statements presented to the participants.
Further, PHAH21 (The organisation does comprehensive process hazard analysis before engineering
changes are made.) with a mean of 1.70 ranked second highest.

PHAH20 (In my organisation all engineering changes undergo a comprehensive management of

change.) and PHAH23 (Most of the health and safety incidents are due to poor engineering design
integrity.) ranked 3rd to 4th within the process hazard analysis latent variable.

130
4.5.5 Permit to Work

Table 4.7 Permit to Work Construct

Str Str
Permit to Agree Agree Neutral Disagree Disagree Std
Work % % % % % Mean Dev Rank
PTWH29 82.6 15.8 1.2 0.0 0.4 1.20 0.478 1
PTWH28 72.8 23.0 2.7 0.8 0.8 1.34 0.648 2
PTWH31 68.1 22.2 6.6 2.7 0.4 1.45 0.770 3
PTWH25 6.2 18.7 36.6 30.7 7.8 3.15 1.018 4

Table 4.7 shows the responses to statements about permit to work. It is evident that PTWH29 (In my
organisation before you start excavation or entering a trench you need to obtain authorisation.)
with a mean of 1.20 ranked highest out of the four statements presented to the participants. Further,
PTWH28 (All the work activities in my organisation are done after a valid permit to work has
been approved by the authorities.) with a mean of 1.34 ranked second highest.

PTWH25 (Most of the health and safety incidents in petrochemical industry are due to poor
controls when working at heights.) and PTWH31 (In my organisation all safety critical
equipment is disabled with permission from the authorities.) ranked 3rd to 4th within the permit to
work latent variable.

4.5.6 Training and Competence

Table 4.8 Training and Competence Construct

Str Str
Training and Agree Agree Neutral Disagree Disagree Std
Competence % % % % % Mean Dev Rank
TCH13 40.9 45.2 9.7 3.9 0.4 1.78 0.805 1
TCH19 38.2 42.5 15.4 2.7 1.2 1.86 0.856 2
TCH35 15.5 39.5 32.2 9.7 3.1 2.45 0.970 3

Table 4.8 shows the responses to statements about training and competence. It is evident that TCH13
(Employees undergo comprehensive training on health and safety in my organisation.) with a mean of
1.78 ranked highest out of the three statements presented to the participants. Further, TCH19 (The
organisation closes all corrective action items effectively after the root cause analysis for all incidents
happening onsite.) with a mean of 1.86 ranked second highest. TCH35 (Most of the health and safety
incidents are due to human error in my organisation.) ranked 3rd within the training and competence
latent variable.

131
4.5.7 Process Health and Safety Information

Table 4.9 Process Health and Safety Information Construct

Process
Health and Str Str
Safety Agree Agree Neutral Disagree Disagree Std
Information % % % % % Mean Dev Rank
PHSIH22 51.7 38.2 8.1 1.5 0.4 1.61 0.736 1
PHSIH18 55.2 32.8 6.6 3.5 1.9 1.64 0.897 2

Table 4.9 shows the responses to statements about process health and safety information. It is evident
that PHSIH22 (The organisation has all process health and safety information available to all
employees.) with a mean of 1.61 ranked highest out of the two statements presented to the participants.
Further, PHSIH18 (The organisation communicates effectively all lessons learned after the
occupational health and safety incidents.) with a mean of 1.64 ranked second highest.

4.5.8 Control of Confined Space Entry

Table 4.10 Control of Confined Space Entry Construct

Control of Str Str
Confined Space Agree Agree Neutral Disagree Disagree Std
Entry % % % % % Mean Dev Rank
CCSEH36 63.8 29.2 4.3 2.3 0.4 1.46 0.723 1
CCSEH37 8.5 20.2 29.8 30.2 11.2 3.16 1.129 2

Table 4.10 shows the responses to statements about control of confined space entry. It is evident that
CCSEH36 (My organisation has effective management systems to manage working in confined
space.) with a mean of 1.46 ranked highest out of the seven statements presented to the participants.
Further, CCSEH37 (Most of the health and safety incidents are due to poor controls in place
when working with suspended loads.) with a mean of 3.16 ranked second highest.

4.5.9 Operating Procedure

Table 4.11 Operating Procedure Construct

Str Str
Operating Agree Agree Neutral Disagree Disagree Std
Procedure % % % % % Mean Dev Rank
OPH26 55.2 36.3 5.8 2.3 0.4 1.56 0.741 1
OPH26 (In my organisation all work activities have a detailed operating procedure or work instruction.)

132
4.5.10 Control of Ignition Source

Table 4.12 Control of Ignition Source Construct

Control of Str
Ignition Str Agree Neutral Disagree Disagree Std
Source Agree% % % % % Mean Dev Rank
CISH27 10.5 23.6 31.4 25.2 9.3 2.99 1.133 1

CISH27 (Most of the health and safety incidents in petrochemical industry are due to poor controls of
source of ignition.)

4.6 Discussion

The frequency of responses were used to determine the major deficiencies existing in process health
and safety management systems when dealing with hazardous chemicals. The mean and standard
deviation were computed and the main deficiencies were decided on when the mean was greater than
2.3. The mean that is greater than 2.3 was considered to be more towards disagree and strongly disagree.

Table 4.13 Main Deficiencies

Str Str
Observed Agree Agree Neutral Disagree Disagree Std
Variables % % % % % Mean Dev Rank
PHAH23 8.2 12.5 38.5 32.3 8.6 3.21 1.038 1
CCSEH37 8.5 20.2 29.8 30.2 11.2 3.16 1.129 2
PTWH25 6.2 18.7 36.6 30.7 7.8 3.15 1.018 3
LCH38 8.5 25.2 31.0 26.4 8.9 3.02 1.103 4
CISH27 10.5 23.6 31.4 25.2 9.3 2.99 1.133 5
HSRAH32 15.1 25.5 27.8 24.7 6.9 2.83 1.166 6
HSRAH39 14.0 32.2 22.1 24.8 7.0 2.79 1.169 7
CEMH16 15.5 28.3 40.3 13.2 2.7 2.59 0.991 8
TCH35 15.5 39.5 32.2 9.7 3.1 2.45 0.970 9
HSRAH33 22.9 32.6 29.5 10.5 4.7 2.41 1.092 10

It is evident in Table 4.13 that most of the health and safety incidents are due to poor engineering design
integrity with a mean of 3.21 and standard deviation of 1.038 ranked highest out of the ten statements
identified as process health and safety deficiencies. Further, it is revealed that most of the health and
safety incidents are due to poor controls in place when working with suspended loads with a mean of
3.16 and standard deviation of 1.129 ranked second highest.

Poor controls when working at heights, Poor housekeeping, Poor controls of source of ignition,
Verifying energy isolation before start working on equipment, Poor health and safety risk assessments,

133
Effective handling of hazardous chemicals, Human error and Fatigue management for both permanent
employees and contractors were ranked 3rd to 10th within the key process health and safety management
deficiencies.

Many construction sites involve working at height. This is a major cause of accidents and caused almost
half of all construction site deaths over the last five years. Hazards from working at heights include lack
of guardrails, inadequate edge protection, unsecured ladders and loose tools. According to Department
of Labor (USA) in out of 4 779 workers fatalities in private industry in calendar year 2018, 1008
(21.1%) were in construction industry. The leading cause of private sector worker deaths in the
construction industry were falls when working at heights (33.5%), followed by struck by object
(11.1%), electrocution (8.5%) and caught in/between objects (5.5%). The main deficiencies revealed in
this research confirm some of the recent incident statistics.

4.7 Chapter Summary

This chapter presented the results of this research and outlined the major deficiencies that require senior
management’s attention to prevent process health and safety incidents when dealing with hazardous
chemicals. Descriptive statistics in the form of median, minimum, maximum, percentage and
cumulative percentage were calculated using SPSS (version 25). The ten observed variables were
assessed to be main deficiencies that require senior management’s attention reduce process health and
safety incidents. This chapter is responding to the following research question, namely, what are the
major deficiencies that should be priorities by top management to prevent process health and safety
incidents when handling hazardous chemicals? The next chapter discusses the model development.

134
 CHAPTER FIVE
Model Development
_____________________________________________________________________
5.1 Introduction

There are several statistical analysis tools considered in this research study with an intention to select
an appropriate analysis approach. Structural Equation Modelling (SEM) has been chosen over other
multivariable analysis tools due to its ability to consider the measurement errors inherent in subjective
operational measurement and to develop and explore the entire set of hypothesised relationships in the
model.

According to Molwus et al. (2017), the development of a SEM research model includes some basic
steps, namely;

 Define and identify the model components that include the latent constructs and measurement
indicators sourced from literature theory and empirical studies.
 Set up the hypothetical relationship depending on the aim and objectives of the study.
 Develop the initial model by using the data collected from the questionnaire survey.
 Verification of the model by evaluating the model estimates and goodness-of-fit GOF
measures.
 Validation of the model based on theoretical and empirical justification.

5.2 Model Development using SEM

The study reviewed the process health and safety elements from the literature and then grouped the
observed variables. The questionnaire survey was designed and disseminated to petrochemical industry
employees (permanent and contractors) to collect the data and then test data to fit the hypothesis

In SEM, a variable can be an independent variable or exogenous variable and a dependent variable or
endogenous variable in a chain of casual hypotheses (Groanland and Stalpers, 2012; Zhao, 2017). The
adoption of SEM for research and studies in the construction management –related field is increasing
in the recent years (Shanmugapriya and Subramanian, 2016; Zhao, 2017). In SEM, the explained
variance of the endogenous latent variables are estimated by assessing model relationships in an
iterative sequence of maximum likelihood regression (Hair et al., 2010; Zhao, 2017). SEM consists of
a measurement model that identifies the relationship between a latent variable, measurement attributes,
indicators and a structural model that identifies the relationship between latent variables (Molenaar and

135
Washington, 2000; Zhao, 2017). AMOS 25 software was used to analyse the research model and test
the reliability and validity of the research model. The required data were entered into SPSS version 25
software that links with AMOS version 25 software.

5.2.1 Structural Equation Modelling

Structural Equation Modelling is an effort to model casual relations between observed variables by
including all observed variables that are known to have connection in the process of curiosity. The first
step in structural equation modelling is to stipulate a model. A model is basically a statement or set of
statements about the relations between observed variables. According to Maydeu-Olivares and Garcia-
Forero (2010) Goodness of Fit has been more extensively developed in SEM than in other areas and
new developments in GOF assessment for multivariate discrete data are strongly related to SEM
procedures, and expect further developments on GOF assessment procedures for multivariate discrete
data along the lines of SEM developments.

SEM combines multiple regression analysis and factor analysis together to analyse the relationship
between measured variables and latent constructs or factors (Raykov, 2006). Basically, it provides a
quantitative method to test a hypothesised model (Byrne, 2016). SEM can be employed to capture
complex relationships between one or more dependent variables that can be sourced from qualitative or
quantitative data (Hox and Kleiboer, 2007). The fit of the correlation matrix of the hypothesised model
to that from data gathering is analysed and assessed by SEM analysis, and then creates a set goodness
of fit index that indicates how well the hypothesised model fits the data (Raykov, 2006).

Missing data is a serious issue in SEM and must be discussed in any article. Also, given new
technologies, more options can handle missing data, such as maximum likelihood estimation (Schreiber
et al., 2006). Although the problem of missing values is not unique to structural modelling, estimating
a successful model necessitates the appropriate handling of missing data from a methodological, as well
as conceptual, perspective. Reliance on pairwise deletion can result in a nonpositive covariance matrix,
and other methods, including replacement with the mean, may result in heteroscedastic error
(Schumaker & Lomax, 1996).

SEM estimates the degree to which a hypothesized model fits the data. In a confirmatory factor analysis,
goodness of fit indexes are estimated for each latent variable as a distinct structural model. Although it
is wise and appropriate for one to measure items found in other studies to form a certain construct, it is
not appropriate to assume that a certain group of items found to form a valid and reliable construct in
another study will form an equally valid and reliable construct when measured in a different set of data.
Similarly, constructs tested on a national data set are valid in a new study in the rare instance when the

136
new study uses the identical observations analysis in the same data with the same theoretical under
pinning. Divergent choices addressing the problem of missing data will normally change construct
validity results such that a new confirmatory analysis is appropriate (Schreiber et al., 2006).

5.2.2 SEM Analysis Selection

According to Hair et al. (2011) cited in Zhao (2017) there are two approaches to SEM modelling. The
first is the covariance – based SEM (CB-SEM) method; the other is the partial least squares SEM (PLS-
SEM). CB-SEM aims to reproduce the theoretical covariance matrix that matches the sample covariance
matrix; the objective of the PLS-SEM approach is to maximise the explained variance of dependent
latent constructs and both are suitable to test the hypothetical causal relationships between latent
constructs (Zhao, 2017).

Although a growing number of studies use PLS-SEM, this approach is not as rigorous as the CB-SEM
approach (Hair, 2017; Zhao, 2017). Therefore, it is less reliable when examining the relationships
between latent constructs (Zhao, 2017). Few researchers view the PLS-SEM approach as a robust
analytical approach for dealing with SEM challenges for reasons such as adequacy of sample size,
normality and homoscedasticity (Wong, 2016; Zhao, 2017). According to Rigdon (2016) PLS-SEM
does not provide the calculation for goodness-of-fit (GOF) measures which provide a reliable tool for
examining the goodness-of-fit of the proposed model to the empirical dataset. Without this goodness-
of-fit assessment for the model, there will be no basis for concluding that the model is valid (Barrett,
2007). PLS-SEM use in such a situation is not model-specific and might result in an unreliable estimate
of the sample size requirement. The ratio of measurement indicators to the latent constructs is a more
reliable approach to computing the minimum sample size required for CB-SEM application (Westland,
2010; Zhao, 2017).

n ≥ 50r2 – 450r + 1100 (5.1)

Where
n is the number of samples
r is the ratio of indicators to latent constructs

Using Equation 5.1, the minimum number of samples for this research was calculated as 138 with 35
being the number of observed variables and ten latent constructs as shown in Table 5.1. The value of r
in Equation 5.1 is 35/10 = 3.5; then Equation 5.1 evaluates n = [(50 * 3.52) – (450 * 3.5) +1100] = 138;
this is less than the 259 samples used in this research (i.e. 259 > 138). Therefore, qualifying the use of

137
covariant based structural equation modelling (CB-SEM) in place of partial least squares structural
equation modelling (Westland, 2010; Zhao, 2017).

According to Maydeu-Olivares and Garcia-Forero (2010) GOF is a very active area of research in
structural equation modelling and in classical SEM applications, multivariate models for continuous
data are estimated from some summary statistics. PLS-SEM does not offer the computing for goodness
of fit (GOF) measures which provide a reliable tool for analysing the goodness-of-fit of the proposed
model to the empirical dataset (Rigdon, 2016, Zhao, 2017). It is essential to assess how well the model
fits the data as a key decider on the appropriateness of the research structural equation modelling model
developed from the dataset (Barrett, 2007; Zhao, 2017). Without this GOF assessment for the model,
there will be no basis for concluding that the model is valid (Barrett, 2007; Zhao, 2017).

CB-SEM technique draws from its stringent requirements in meeting the assumption about the observed
variables to be multivariate and normally distributed. Accordingly to Byrne (2010) cited in Zhao (2017)
this is imperative to SEM estimation, particularly for Generalised Least Squares (GLS) and Maximum
Likelihood (ML) estimating subroutines of the CB-SEM analysis. Violation of this assumption might
lead to inaccurate calculations for the Chi-square and t-test. According to Hair et al. (2011) cited in
Zhao (2017) if the assumptions of the CB-SEM are satisfied with respect to the minimum sample size
and data distribution, the CB-SEM is a better option; otherwise, PLS-SEM is a good approximation of
CB-SEM. Satisfaction of the above strict requirements of the CB-SEM by the empirical data attributes
warranty the use of CB-SEM in this research in place of a PLS-SEM approach.

5.2.3 Reliability Checks

Mean ratings of observed variables, principal component factor analysis and Cronbach’s alpha test were
conducted in this research. The mean ratings of the observed variables were obtained to check the
acceptance of them by the participants; principal component factor analysis was carried out to check
the commonality within the data set; and Cronbach’s alpha test was performed to test the reliability of
the data set. SEM with AMOS version 25 software was employed to test the hypothesised relationship
between the observed variables, latent variables and generative health and safety culture.

In a research study, the reliability and validity of the data are vital. The Cronbach’s Alpha is a high-
quality test widely used for reliability testing and an important test for assessing a research instrument
(Zhao, 2017). The appropriateness of the categories was confirmed by reliability test. The quality of the
observed variables attributes and the indicators for corresponding latent constructs in the model should
be examined and evaluated (Zhao, 2017). Cronbach’s Alpha test is extensively –used method to test the
internal consistency of observed variables, based on the correlations between indicators (Zhao, 2017).

138
5.2.4 Principal Component Analysis

Table 5.1 Principal Component Analysis with all loading

Rotated Component Matrixa
Observed Latent Constructs
Variables PTW PHA LC HSRA CEM TC PHSI CCSE OP CIS
PTWH29 0.779
PTWH28 0.687
PTWH31 0.661
CCSEH36 0.649 0.304
PHAH24 0.630
PHSIH22 0.587 0.395
PHAH20 0.564 0.447
PHAH21 0.504 0.305 0.425
OPH26 0.586
PHSIH18 0.486 0.375
HSRAH34 0.420 0.309 0.404
LCH8 0.726
LCH10 0.725
LCH11 0.698 0.364
LCH7 0.305 0.660
TCH13 0.547 0.401
HSRAH9 0.373 0.524 0.317
LCH17 0.519 0.432
LCH40 0.333 ←←←←←←← 0.587
TCH19 0.392 0.478
CCSEH37 0.833
HSRAH32 0.788
PTWH25 0.785
CSIH27 0.739
PHAH23 0.690
LCH38 0.598 0.564
HSRAH39 0.573 0.536
HSRAH33 ←←←← 0.637
TCH35 ←←←←←←←←←←←←←←← 0.769
CEMH30 0.735 →→→→→→→→→→
CEMH6 0.413 0.604 0.310
CEMH16 0.763
CEMH15 0.731
CEMH14 0.695
CEMH12 0.654
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.a
a. Rotation converged in 6 iterations.

139
It is evident in Table 5.1 that five latent constructs were eliminated after principal component
analysis, namely;

 Training and Competency (TC)

 Process Health and Safety Information (PHSI)
 Control of Confined Space Entry (CCSE)
 Operating Procedure (OP)
 Control of Ignition Source (CIS)

The above five latent constructs were eliminated due to no observed variable allocated to it or because
the loading was less than 0.5. There are observed variables that were relocated from original latent
constructs to other latent constructs as shown Table 5.2 for reliability analysis after principal component
analysis.

According Zohrabi (2013) one of the main requirements of any research process is the reliability of the
data and findings. The reliability and validity of the data are vital for a research study. The Cronbach’s
Alpha is a high-quality test extensively used for reliability testing and a vital test for assessing a
questionnaire instrument. The quality of the measurement attributes and indicator for corresponding
constructs in the model should be analysed and assessed. Cronbach’s Alpha test is an extensively used
method to test the internal consistency of measurement indicators, based on the correlations between
indicators.

Table 5.2 Reliability Analysis before Principal Component Analysis for Constructs

Latent Constructs No of Observed Variables Cronbach's Alpha

10 (LCH7, LCH8, LCH9, LCH10, LCH11,
Leadership Commitment (LC) LCH13, LCH17, LCH19, LCH34, LCH40) 0.873
Chemical Exposure 6 (CEMH6, CEMH12, CEMH14, CEMH15,
Management (CEM) CEMH16, CEMH30) 0.797
9 (HSRAH23, HSRAH25, HSRAH27, HSRAH32,
Health and Safety Risk HSRAH33, HSRAH35, HSRAH37, HSRAH38,
Assessment (HSRA) HSRAH39) 0.829
Process Hazard Analysis 5 (PHAH20, PHAH21, PHAH22, PHAH24,
(PHA) PHAH26) 0.810
5 (PTWH18, PTWH28, PTWH29, PTWH31,
Permit to Work (PTW) PTWH36) 0.774

It is evident in Table 5.2 that the all latent constructs have Cronbach’s Alpha above 0.7, which confirms
reliability of the data.

140
5.2.5 Consistency Test Results

The consistency tests are imperative to the study as they ensure the reliability and validity of the data
set. The mean ratings of all the measurement indicators by the participants are used to determine the
acceptance of them by the participants.

Table 5.3 Leadership Commitment

(Str Disagree
Leadership (Str Agree and
Commitment and Agree)% Neutral% Disagree)% Mean Std Dev
LCH7 93.4 4.3 2.3 1.48 0.707
LCH8 88.4 9.3 2.3 1.64 0.763
LCH9 88.8 8.5 2.7 1.70 0.737
LCH10 96.9 1.9 1.2 1.40 0.591
LCH11 88.4 9.3 2.3 1.63 0.748
LCH13 86.1 9.7 4.2 1.78 0.805
LCH17 73.6 18.6 7.8 1.98 0.966
LCH19 80.7 15.4 3.9 1.86 0.856
LCH34 88.0 8.5 3.5 1.70 0.784
LCH40 87.3 9.3 3.4 1.61 0.866

From Table 5.3 it is evident that the mean ratings for leadership commitment construct range between
1.40 (LCH10) to 1.98 (LCH17) and the standard deviation is between 0.591 (LCH 10) to 0.966 (LCH17)
for the ten observed variables.

Table 5.4 Chemical Exposure Management

(Str Agree (Str Disagree
Chemical Exposure and Neutral and
Management Agree)% % Disagree)% Mean Std Dev
CEMH6 92.2 6.2 1.6 1.62 0.692
CEMH12 86.5 9.6 3.9 1.75 0.783
CEMH14 84.9 12.8 2.3 1.75 0.766
CEMH15 62.4 32.1 5.5 2.22 0.858
CEMH16 43.8 40.3 15.9 2.59 0.991
CEMH30 95.8 2.7 1.5 1.35 0.612

From Table 5.4 it is evident that the mean ratings for chemical exposure management construct range
between 1.35 (CEMH30) to 2.59 (CEMH16) and the standard deviation is between 0.612 (CEMH30)
to 0.991 (CEMH16) for the six observed variables.

141
 Table 5.5 Health and Safety Risk Assessment
(Str Disagree
Health and Safety (Str Agree and
Risk Assessment and Agree)% Neutral% Disagree)% Mean Std Dev
HSRAH23 20.6 38.5 40.9 3.21 1.038
HSRAH25 24.9 36.6 38.5 3.15 1.018
HSRAH27 34.1 31.4 34.5 2.99 1.133
HSRAH32 40.5 27.8 31.7 2.83 1.166
HSRAH33 55.4 29.5 15.1 2.41 1.092
HSRAH35 55 32.2 12.8 2.45 0.970
HSRAH37 28.7 29.8 41.5 3.16 1.129
HSRAH38 33.7 31 35.3 3.02 1.103
HSRAH39 46.1 22.1 31.8 2.79 1.169

From Table 5.5 it is evident that the mean ratings for health and safety risk assessment construct range
between 2.41 (HSRAH33) to 3.21 (HSRAH23) and the standard deviation is between 0.970
(HSRAH35) to 1.169 (HSRAH39) for the nine observed variables.

Table 5.6 Process Hazard Analysis

Process Hazard (Str Agree (Str Disagree
Analysis and Agree)% Neutral% and Disagree)% Mean Std Dev
PHAH20 87.2 3.5 9.3 1.72 0.787
PHAH21 88.8 9.3 1.9 1.70 0.732
PHAH22 90.0 8.1 1.9 1.61 0.736
PHAH24 93.8 4.6 1.6 1.49 0.680
PHAH26 91.5 5.8 2.7 1.56 0.741

From Table 5.6 it is evident that the mean ratings for process hazard analysis construct range between
1.49 (PHAH24) to 1.72 (PHAH20) and the standard deviation is between 0.680 (PHAH24) to 0.787
(PHAH20) for the five observed variables.

Table 5.7 Permit to Work

(Str Agree and (Str Disagree
Permit to Work Agree)% Neutral% and Disagree)% Mean Std Dev
PTWH18 88.0 6.6 5.4 1.64 0.897
PTWH28 95.7 2.7 1.6 1.34 0.648
PTWH29 98.4 1.2 0.4 1.20 0.478
PTWH31 90.3 6.6 3.1 1.45 0.770
PTWH36 93.0 4.3 2.7 1.46 0.723

From Table 5.7 it is evident that the mean ratings for permit to work construct range between 1.20
(PTWH29) to 1.64 (PTWH18) and the standard deviation is between 0.478 (PTWH29) to 0.897
(PTWH18) for the five observed variables. Having inveterate the acceptance, commonality, and
reliability of all the measurement indicators, the confirmatory factor analysis would be conducted to
test the measurement model.

142
5.2.6 Structural Equation Modelling Specification

According to Gainey and Klass (2003); Molenaar et al. (2000); Zhao (2017) a SEM specification should
first be developed based on the theoretical framework. The initial SEM specification followed the
research model, the main key factors influencing generative health and safety culture were found to be
leadership commitment, chemical exposure management, health and safety risk assessment, process
hazard analysis and permit to work. The model denotes a theoretical framework where the five
constructs were selected from the 10 constructs in the original conceptual framework. These five
constructs were selected because of loadings higher than 0.5 after principal component analysis and
reliability test computed above 0.7 for all observed variables in each latent variable.

5.2.7 Confirmatory Factor Analysis

Confirmatory Factor Analysis (CFA) and SEM can each be an iterative process by which modifications
are indicated in the initial results, and parameter constraints altered to improve the fit of the model, if
such changes are warranted theoretically (Schreiber et al., 2006). If a parameter is freed on the basis of
a high modification index value, the researcher is called on to theoretically defend the change indicated
so that the final model does not deviate from the initial theoretical model (Schreiber et al., 2006).

In this research CFA was adopted to test the strength and appropriateness of the relationship between
latent constructs and corresponding measurement indicators. CFA is also acknowledged as a
measurement model as it can be used to evaluate the fit between the collected data and the conceptual
relationship between measurement and latent variables. According to Schreiber et al. (2006) CFA and
SEM are statistical techniques that one can use to reduce the number of observed variables into a smaller
number of latent variables by examining the covariation among the observed variables.

5.2.8 Model Modification

CFA is done to test whether the measure of latent construct corresponds with the study of the nature of
the individual factor and at this stage, the indicator elimination and model re-specification are performed
for each latent construct (Zhao, 2017). The reliability and validity of the constructs must be examined
and evaluated before using them in the following structural model (Shanmugapriya and Subramanian,
2016; Zhao, 2017). The measurement model includes all latent constructs and specifies the
measurement indicators and attributes for corresponding constructs. The measurement indicators or
variables have different degrees. Measurement indicators having low loadings should be eliminated
since the offer small explanatory power to the model (Aibinu and Al-Lawati, 2010; Zhao, 2017). A

143
common acceptable threshold value for a good indicator is having a loading higher than 0.5 (Rahman
et al., 2013; Zhao, 2017).
Table 5.8 Final Principal Component Analysis
Rotated Component Matrixa
Component
Observed Variables
PTW PHA LC HSRA CEM
Permit to Work (PTWH29) 0,779
Permit to Work (PTWH28) 0,687
Permit to Work (PTWH31) 0,661
Permit to Work (PTWH36) 0,649
Process Hazard Analysis (PHAH24) 0,630
Process Hazard Analysis (PHAH22) 0,587
Process Hazard Analysis (PHAH20) 0,564
Process Hazard Analysis (PHAH21) 0,504
Process Hazard Analysis (PHAH26) 0,586
Leadership Commitment (LCH8) 0,726
Leadership Commitment (LC10) 0,725
Leadership Commitment (LCH11) 0,698
Leadership Commitment (LCH7) 0,660
Leadership Commitment (LCH13) 0,547
Leadership Commitment (LCH9) 0,524
Leadership Commitment (LCH17) 0,519
Leadership Commitment (LCH6) 0,604
Leadership Commitment (LCH40) 0,587
Health and Safety Risk Assessment 0,833
(HSRAH37)
Health and Safety Risk Assessment 0,788
(HSRAH32)
Health and Safety Risk Assessment 0,785
(HSRAH25)
Health and Safety Risk Assessment 0,739
(HSRAH27)
Health and Safety Risk Assessment 0,690
(HSRAH23)
Health and Safety Risk Assessment 0,598
(HSRAH38)
Health and Safety Risk Assessment 0,573
(HSRAH39)
Health and Safety Risk Assessment 0,637
(HSRAH33)
Health and Safety Risk Assessment 0,769
(HSRAH35)
Chemical Exposure Management (CEMH30) 0,735
Chemical Exposure Management (CEMH16) 0,763
Chemical Exposure Management (CEMH15) 0,731
Chemical Exposure Management (CEMH14) 0,695
Chemical Exposure Management (CEMH12) 0,654
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.a
a. Rotation converged in 6 iterations.

144
Convergent validity confirms that a set of indicators measure one and the same latent construct and not
another construct. Squared multiple correlations (SMC) is used as a criterion of convergent validity and
the cut-off value of 0.5 is considered as an acceptable indicator (Azen and Sass, 2008; Zhao, 2017).

Table 5.9 Reliability Analysis after Principal Component Analysis for Constructs

Latent Construct No of Observed Variables Cronbach's Alpha

9 (LCH6, LCH7, LCH8, LCH9, LCH10,
Leadership Commitment (LC) 0.867
LCH11, LCH13, LCH17, LCH40)
Chemical Exposure Management 5 (CEMH12, CEMH14, CEMH15, CEMH16,
0.781
(CEM) CEMH30)
9 (HSRAH23, HSRAH25, HSRAH27,
Health and Safety Risk
HSRAH32, HSRAH33, HSRAH35, 0.829
Assessment (HSRA)
HSRAH37, HSRAH38, HSRAH39)
5 (PHAH20, PHAH21, PHAH22, PHAH24,
Process Hazard Analysis (PHA) 0.810
PHAH26)
Permit to Work (PTW) 4 (PTWH28, PTWH29, PTWH31, PTWH36) 0.749

It is evident from Table 5.9 that the five latent constructs that are drivers to generative process health
and safety culture are Leadership Commitment (0.867), Chemical Exposure Management (0.781),
Health and Safety Risk Assessment (0.829), Process Hazard Analysis (0.810) and Permit to Work
(0.749), and which all have Cronbach’s Alpha above 0.7, which confirms reliability of the data.

5.3 Model Fit

A range of established fit indices should be introduced to decide upon the goodness of fit (GOF)
between the research model and empirical data. Broadly, fit indices can be classified into three
categories: overall model fit, goodness-of-fit, and badness-of-fit (Green, 2016, Zhao, 2017). The overall
model fit is measured by a chi-square statistic that is used to examine whether the statistical significance
exists between the observed and estimated variance-covariance matrix (Bagozzi & Yi, 2012; Zhao,
2017). However, chi-square statistics are sensitive and artificially inflated by sample size (Iacobucci,
2010; Zhao, 2017).

5.3.1 Goodness-of-Fit (GOF)

The concept of goodness of fit is used for identification of statistical model for different analysis, in
research area this concept is used in almost all the fields directly or indirectly (Jha et al., 2011). The
goodness of Fit (GOF) of a statistical model pronounces how well it fits into a set of observations. GOF
indices summarize the inconsistency between the observed values and the values expected under a
statistical model (Jha et al., 2011). GOF statistics are GOF indices with known sampling distributions,

145
usually obtained using an asymptotic methods, that are used in statistical hypothesis testing. There are
numerous methods used for goodness of fit test and most important among them are, namely,
Kolmogorov - Smirnov, Anderson - Darling and Chi – Squared (Jha et al., 2011).

The overall model fit is measured by a chi-square statistic that is used to examine whether the statistical
significance exists between the observed and estimated variance-covariance matrix (Bagozzi and Yi,
2012; Zhao, 2017). However, chi-square statistics are insufficient to determine the merit of a model
which is where goodness-of-fit and badness-of-fit indices should be introduced. Goodness-of-fit indices
include comparative and absolute fit indices (Green, 2016; Zhao, 2017).

Comparative indices comprises Comparative Fit Index (CFI) whilst absolute fit indices consist of
Tucker-Lewis Index (TLI), adjusted Goodness of Fit Index (AGFI), Incremental Fit Index (IFI) and
Normed Fit Index (NFI) (Bagozzi, 2010; Zhao, 2017). Badness-of-fit indices are indicated by the root-
mean-square error of approximation (RMSEA) with a 90% confidence interval (90% CI) and the
standardised root-mean-square residual (RMSR) (Hu and Bentler, 1999; Zhao, 2017).

Parsimonious fit indices were developed to penalise for model complexity because complex, nearly
saturated models are dependent on the sample during estimation. Non-parsimonious or complex models
are models which contain few paths. Complex models create a less rigorous theoretical model that
produces better fit indices. Parsimonious indices include the Parsimony Goodness-of-fit (PGFI), the
Parsimony Adjusted Normed Fit Index (PNFI) and the Parsimony Adjusted Comparative Fit Index
(PCFI). There are no widely accepted minimum thresholds of acceptance. Often values of 0.50 are
obtainable even when other indices exceed the 0.90 threshold, it is recommended to report them with
other indices.

146
 Table 5.10 Threshold Limits for Model Fit Indices
Acceptable
Model Fit Index Threshold Interpretation References
Absolute Fit Indices
Relative Normed Chi-Square Value <2 Good Fit Tabachnick and Fidell (2013)
Value < 0.05 Good Fit Brown (2006); Hoe (2008);
Hooper et al. (2008); Hsu et al.
Root Mean Square Error of
Value is 0.06 - (2012), Hu and Bentler (1999);
Approximation Acceptable Fit
0.08 Schreiber et al. (2006);
Schumacker and Lomax (2004).
Incremental Fit Indices
Value ≥ 0.95 Good Fit Brown (2006); Hooper et al.
(2008); Hsu et al. (2012), Hu and
Bentler Comparative Fit Index
Value is 0.90 - Bentler (1999); Schreiber et al.
(CFI) Acceptable Fit
0.95 (2006); Schumacker and Lomax
(2004).
Value ≥ 0.95 Good Fit Brown (2006); Hooper et al.
(2008); Hsu et al. (2012), Hu and
Incremental Fit Index (IFI) Value is 0.90 - Bentler (1999); Schreiber et al.
Acceptable Fit (2006); Schumacker and Lomax
0.95
(2004).
Value ≥ 0.95 Good Fit Brown (2006); Hooper et al.
(2008); Hsu et al. (2012), Hu and
Normed Fit Index (NFI) Value is 0.90 - Bentler (1999); Schreiber et al.
Acceptable Fit (2006); Schumacker and Lomax
0.95
(2004).
Value ≥ 0.95 Good Fit Brown (2006); Hooper et al.
(2008); Hsu et al. (2012), Hu and
Tucker - Lewis Index (TLI) Value is 0.90 - Bentler (1999); Schreiber et al.
Acceptable Fit (2006); Schumacker and Lomax
0.95
(2004).
Parsimonious Fit Indices
Parsimony Adjusted Normed Fit Value > 0.90 Good Fit
Hooper et al. (2008)
Index (PNFI) Value > 0.50 Acceptable Fit
Parsimony Adjusted Comparative Value > 0.90 Good Fit
Hooper et al. (2008)
Fit Index (PCFI) Value > 0.50 Acceptable Fit

In this research there was no threshold limit for Chi-square values as this fit statistic varies according
to the design complexity of the model. The results of the model fit and its interpretation will be presented
for each latent construct in order to assess model fit for the dependent variables.

147
5.3.2 Leadership Commitment Goodness-of-Fit

Table 5.11 Leadership Commitment Construct Goodness-of-Fit

Leadership Commitment Construct
First Final
Model Fit Index Threshold SEM Acceptability SEM Acceptability
Absolute Fit Indices
Marginal
CMIN/df <2 2.394 Not Accepted 2.151
Accepted
Root Mean Square Error Value < 0.05
0.073 Accepted 0.067 Accepted
of Approximation Value is 0.06 - 0.08
Incremental Fit Indices
Bentler Comparative Fit Value ≥ 0.95
0.956 Accepted 0.971 Accepted
Index (CFI) Value is 0.90 - 0.95
Incremental Fit Index Value ≥ 0.95
0.957 Accepted 0.972 Accepted
(IFI) Value is 0.90 - 0.95
Value ≥ 0.95
Normed Fit Index (NFI) 0.928 Accepted 0.949 Accepted
Value is 0.90 - 0.95
Tucker - Lewis Index Value ≥ 0.95
0.926 Accepted 0.948 Accepted
(TLI) Value is 0.90 - 0.95
Parsimonious Fit Indices
Parsimony Adjusted
Normed Fit Index Value > 0.50 0.557 Accepted 0.527 Accepted
(PNFI)
Parsimony Adjusted
Comparative Fit Index Value > 0.50 0.573 Accepted 0.540 Accepted
(PCFI)

It is evident in Table 5.11 that RMSEA = 0.067, CMIN/df = 2.151 indicates that the theoretical model
of leadership commitment fitted the empirically data satisfactory. The CFI (0.971), IFI (0.972), NFI
(0.949) and TLI (0.948) was indicative of good fit and therefore suggested acceptable fit. When
considering the construct validity, leadership commitment observed variables were strong and
statistically significant. Parsimony was assessed using PNFI and PCFI. The indices exceeded the
threshold of 0.50 suggested by Hooper et al. (2008) at PNFI (0.527) and PCFI (0.540). However, it may
be argued that the general threshold index of 0.9 which is widely accepted for all other indices might
be more appropriate. The model presented is not so parsimonious, but still acceptable. The researcher
decided to eliminate LCH40 to improve CMIN/df from 2.394 to 2.151 which was then marginally
accepted.

148
5.3.3 Chemical Exposure Management Goodness-of-Fit

Table 5.12 Chemical Exposure Management Construct Goodness-of-Fit

Chemical Exposure Management Construct
First Final
Model Fit Index Threshold SEM Acceptability SEM Acceptability
Absolute Fit Indices
CMIN/df <2 4.957 Not Accepted 7.244 Not Accepted
Root Mean Square Error Value < 0.05
0.124 Not Accepted 0.156 Not Accepted
of Approximation Value is 0.06 - 0.08
Incremental Fit Indices
Bentler Comparative Fit Value ≥ 0.95
0.942 Accepted 0.960 Accepted
Index (CFI) Value is 0.90 - 0.95
Incremental Fit Index Value ≥ 0.95
0.944 Accepted 0.961 Accepted
(IFI) Value is 0.90 - 0.95
Value ≥ 0.95
Normed Fit Index (NFI) 0.931 Accepted 0.955 Accepted
Value is 0.90 - 0.95
Tucker - Lewis Index Value ≥ 0.95
0.827 Not Accepted 0.798 Not Accepted
(TLI) Value is 0.90 - 0.95
Parsimonious Fit Indices
Parsimony Adjusted
Value > 0.50 0.310 Not Accepted 0.191 Not Accepted
Normed Fit Index (PNFI)
Parsimony Adjusted
Comparative Fit Index Value > 0.50 0.314 Not Accepted 0.192 Not Accepted
(PCFI)

It is evident in Table 5.12 that RMSEA = 0.156 and CMIN/df = 7.244 was indicative of poor model fit
for the theoretical model of chemical exposure management. The CFI (0.960), IFI (0.961), NFI (0.955)
was indicative of good fit but TLI (0.798) suggested not acceptable fit. When considering the construct
validity, chemical exposure management observed variables were not strong and statistically not
significant. Parsimony was assessed using PNFI (0.191) and PCFI (0.192) and thus the model presented
is not so parsimonious. The researcher decided to eliminate CEMH30 to improve incremental fit indices
CFI, IFI and NFI to acceptable threshold. Due to lack of construct validity, any interpretations based on
the chemical exposure management latent variable needs to be inferred carefully.

149
5.3.4 Health and Safety Risk Assessment Goodness-of-Fit

Table 5.13 Health and Safety Risk Assessment Construct Goodness-of-Fit

Health and Safety Risk Assessment Construct

First Final
Model Fit Index Threshold SEM Acceptability SEM Acceptability
Absolute Fit Indices
CMIN/df <2 2.729 Not Accepted 3.666 Not Accepted
Root Mean Square Error of Value < 0.05 Marginal
0.082 0.102 Not Accepted
Approximation Value is 0.06 - 0.08 Accepted
Incremental Fit Indices
Bentler Comparative Fit Value ≥ 0.95
0.933 Accepted 0.944 Accepted
Index (CFI) Value is 0.90 - 0.95
Value ≥ 0.95
Incremental Fit Index (IFI) 0.935 Accepted 0.945 Accepted
Value is 0.90 - 0.95
Value ≥ 0.95
Normed Fit Index (NFI) 0.901 Accepted 0.926 Accepted
Value is 0.90 - 0.95
Value ≥ 0.95 Marginal Marginal
Tucker - Lewis Index (TLI) 0.889 0.888
Value is 0.90 - 0.95 Accepted Accepted
Parsimonious Fit Indices
Parsimony Adjusted Marginal
Value > 0.50 0.541 Accepted 0.463
Normed Fit Index (PNFI) Accepted
Parsimony Adjusted
Marginal
Comparative Fit Index Value > 0.50 0.560 Accepted 0.472
Accepted
(PCFI)

In Table 5.13, RMSEA = 0.082 and CMIN/df = 2.729 was indicative of marginally acceptable
theoretical model fit for health and safety risk assessment construct. The CFI (0.933), IFI (0.935), NFI
(0.901) was indicative of good fit and TLI (0.889) suggested marginal acceptable fit. When considering
the construct validity, health and safety risk assessment construct observed variables were strong and
statistically significant. Parsimony was assessed using PNFI (0.541) and PCFI (0.560) and thus the
model presented is not so parsimonious, but still acceptable. The researcher did not accept the first SEM
model and attempted to improve the model for this construct by eliminating HSRAH33 and HSRAH35.
RMSEA = 0.102 and CMIN/df = 3.666 and was not accepted. However the CFI (0.944), IFI (0.945),
NFI (0.926) improved from the first SEM was indicative of good fit but TLI (0.888) remained marginal
acceptable fit. The elimination of HSRAH33 and HSRAH35 improved the overall model. Parsimony
assessment was marginally acceptable PNFI (0.463) and PCFI (0.472).

150
5.3.5 Process Hazard Analysis Goodness-of-Fit

Table 5.14 Process Hazard Analysis Construct Goodness-of-Fit

Process Hazard Analysis Construct
Model Fit Index Threshold Final SEM Acceptability
Absolute Fit Indices
CMIN/df <2 10.263 Not Accepted
Root Mean Square Error of Value < 0.05
0.189 Not Accepted
Approximation Value is 0.06 - 0.08
Incremental Fit Indices
Value ≥ 0.95 Marginal
Bentler Comparative Fit Index (CFI) 0.895
Value is 0.90 - 0.95 Accepted
Value ≥ 0.95 Marginal
Incremental Fit Index (IFI) 0.897
Value is 0.90 - 0.95 Accepted
Value ≥ 0.95 Marginal
Normed Fit Index (NFI) 0.887
Value is 0.90 - 0.95 Accepted
Value ≥ 0.95
Tucker - Lewis Index (TLI) 0.684 Not Accepted
Value is 0.90 - 0.95
Parsimonious Fit Indices
Parsimony Adjusted Normed Fit Index
Value > 0.50 0.296 Not Accepted
(PNFI)
Parsimony Adjusted Comparative Fit
Value > 0.50 0.298 Not Accepted
Index (PCFI)

It is evident in Table 5.14 that RMSEA = 0.189 and CMIN/df = 10.263 indicates that the theoretical
model of the process hazard analysis construct did not fit the empirical data satisfactorily. The CFI
(0.895), IFI (0.897), NFI (0.887) were indicative of marginal accepted fit and TLI (0.684) suggested
poor model fit. Parsimony was assessed using PNFI (0.296) and PCFI (0.298) and therefore the model
presented is not so parsimonious. The researcher decided to eliminate three observed variables, namely,
PHAH22, PHAH24 and PHAH26 to improve the final model. Process hazard analysis construct had
only 2 observed variable in the final research model, namely, PHAH20 and PHAH21.

151
5.3.6 Permit to Work Goodness-of-Fit

Table 5.15 Permit to Work Construct Goodness-of-Fit

Permit to Work Construct
Model Fit Index Threshold Interpretation Final SEM Acceptability
Absolute Fit Indices
CMIN/df <2 Good Fit 1.651 Accepted
Root Mean Square Error of Value < 0.05 Good Fit
0.050 Accepted
Approximation Value is 0.06 - 0.08 Acceptable Fit
Incremental Fit Indices
Bentler Comparative Fit Index Value ≥ 0.95 Good Fit
0.995 Accepted
(CFI) Value is 0.90 - 0.95 Acceptable Fit
Value ≥ 0.95 Good Fit
Incremental Fit Index (IFI) 0.995 Accepted
Value is 0.90 - 0.95 Acceptable Fit
Value ≥ 0.95 Good Fit
Normed Fit Index (NFI) 0.988 Accepted
Value is 0.90 - 0.95 Acceptable Fit
Value ≥ 0.95 Good Fit
Tucker - Lewis Index (TLI) 0.975 Accepted
Value is 0.90 - 0.95 Acceptable Fit
Parsimonious Fit Indices
Parsimony Adjusted Normed
Value > 0.50 Acceptable Fit 0.198 Not Accepted
Fit Index (PNFI)
Parsimony Adjusted
Value > 0.50 Acceptable Fit 0.199 Not Accepted
Comparative Fit Index (PCFI)

It is evident in Table 5.15 that RMSEA = 0.05, CMIN/df = 1.651 indicates that the theoretical model of
permit to work construct fitted the empirically data satisfactory. The CFI (0.995), IFI (0.995), NFI
(0.988) and TLI (0.975) was indicative of good fit and therefore suggested acceptable fit. When
considering the construct validity, permit to work observed variables were strong and statistically
significant. Parsimony was assessed using PNFI and PCFI. The indices did not exceed the threshold of
0.50 suggested by Hooper et al. (2008) at PNFI (0.198) and PCFI (0.199). The model presented is not
so parsimonious.

152
5.3.7 Generative Process Health and Safety Culture Model Goodness-of-Fit

Table 5.16 Generative Process Health and Safety Culture Model

Generative Health and Safety Culture Final Model

Acceptable First Final
Model Fit Index Interpretation Acceptability Acceptability
Threshold SEM SEM
Absolute Fit Indices
CMIN/df <2 Good Fit 2.341 Not Accepted 1.758 Accepted
Root Mean Square Value < 0.05 Good Fit
Error of Value is 0.06 0.072 Accepted 0.054 Accepted
Approximation Acceptable Fit
- 0.08
Incremental Fit Indices
Bentler Value ≥ 0.95 Good Fit
Comparative Fit Value is 0.90 0.824 Not Accepted 0.925 Accepted
Index (CFI) Acceptable Fit
- 0.95
Value ≥ 0.95 Good Fit
Incremental Fit
Value is 0.90 0.827 Not Accepted 0.927 Accepted
Index (IFI) Acceptable Fit
- 0.95
Value ≥ 0.95 Good Fit
Normed Fit Index Marginal
Value is 0.90 0.733 Not Accepted 0.846
(NFI) Acceptable Fit Accepted
- 0.95
Value ≥ 0.95 Good Fit
Tucker - Lewis
Value is 0.90 0.795 Not Accepted 0.908 Accepted
Index (TLI) Acceptable Fit
- 0.95
Parsimonious Fit Indices
Parsimony Adjusted
Normed Fit Index Value > 0.90 Good Fit 0.630 Accepted 0.689 Accepted
(PNFI)
Parsimony Adjusted
Comparative Fit Value > 0.90 Good Fit 0.708 Accepted 0.755 Accepted
Index (PCFI)

It is evident in Table 5.16 that the model fit indices for the refined model met the acceptable threshold
limits. The absolute fit was assessed using the relative normed Chi-square and the RMSEA. The
CMIN/df and RMSEA met the recommended acceptable limits with 1.758 and 0.054 respectively. The
relative normed Chi-square is recommended to be less than 2.00 (Tabachnick and Fidell, 2013) and
RMSEA is recommended to be less than 0.05 (Hu and Bentler, 1999), however it is still acceptable
when it is less than 0.08. The RMSEA is used to measure the square root of the residual that is the
difference between the collected data and model prediction (Anderson and Gerbing, 1984). It ranges
between 0 and 1 with the value smaller than limit value of 0.08 perceived as an acceptable fit Kline,
2005).

Incremental indices assessed were the CFI, IFI, NFI and the TLI. The CFI compares the fit of the
hypothesised model to the collected data with the fit of baseline model to the data (Iacobucci, 2010).
The IFI is the ratio of the difference of Chi-square between the hypothesised model and the baseline

153
model and the difference of the degree of the freedom of the two models (Bentler, 1990). The TLI
compares the discrepancy and degrees of freedom of the baseline model with that of the hypothesised
model (Bentler and Bonett, 1980).

The CFI (0.925), IFI (0.927) and TLI (0.908) all met the minimum threshold suggested by Hooper et
al. (2008) and Hu and Bentler (1999). However, the NFI (0.846) fell below the 0.90 threshold. The
three of four incremental fit indices assessed fell above the acceptable threshold to provide support for
acceptable model fit and therefore the model has acceptable incremental fit. Parsimony was assessed
using PNFI (0.689) and PCFI (0.755). The indices exceeded the threshold limit of 0.50 recommended
by Hooper et al. (2008). It may be argued that the general acceptable index limit of 0.90 which is widely
accepted for all other indices might be more appropriate. Due to complexity of the model assessed, it
was expected that these indices would be lower than the widely accepted limits of 0.90 and thus the
model presented is not parsimonious.

5.4 Model Refinement

The model was refined by eliminating awkward constructs and observed variables as recommended by
Hooper et al. (2008). In this research there are five latent constructs that were eliminated after principal
component analysis, namely, Training and Competency (TC), Process Health and Safety Information
(PHSI), Control of Confined Space Entry (CCSE), Operating Procedure (OP) and Control of Ignition
Source (CIS). The above five latent constructs were eliminated due to no observed variable allocated to
it or because the loading was less than 0.50.

There were three observed variables that were eliminated since their loadings were less than 0.5 and
they are namely, TCH18 - The organisation communicates effectively all lessons learned after the
occupational health and safety incidents. PTW19 - The organisation closes all corrective action items
effectively after the root cause analysis for all incidents happening onsite). TCH34 - My organisation
has all management systems in place to manage substance misuse. Other Observed variables were
allocated to different latent variables after principal component analysis.

Leadership commitment is the latent construct that had nine observed variables and only LCH40 was
eliminated to improve CMIN/df from 2.394 to 2.151 which was then marginally accepted. Under
chemical exposure management there was original five observed variables and only CEMH30 was
eliminated to improve incremental fit indices CFI, IFI and NFI to acceptable threshold. Health and
safety risk assessment is the latent construct that had nine observed variables and two HSRAH33 and

154
HSRAH35 were eliminated to improve incremental fit indices CFI, IFI and NFI for the final research
model.

Process Hazard Analysis is the latent construct had five observed variables, namely, PHAH20 – In my
organization all engineering changes undergo a comprehensive management of change. PHAH21 – The
organization does comprehensive process hazard analysis before engineering changes are made.
PHAH22 – The organization has all process health and safety information available to all employees.
PHAH24 – In my organization we have a comprehensive pre-activity start up review and pre-activity
shutdown review. PHAH26 – In my organization all work activities have a detailed operating procedure
or work instruction. The final model eliminated PHAH22, PHAH24 and PHAH26 and remained with
only PHAH20 and PHAH21.

155
Figure 5.1 Initial Measurement Model

156
Figure 5.2 Refined Measurement Model

157
5.5 Chapter Summary

This chapter discussed model development. The descriptive statistics, reliability analysis and
confirmatory analysis were used. Descriptive statistics used was the cumulative percentage, mean and
standard deviation using SPSS (version 25). Cronbach’s alpha was used for reliability analysis and
principal component analysis was used to provide the goodness of fit. In this research, there were three
observed variables that were eliminated since the loading was less than 0.5 and they were five latent
constructs that were identified as drivers towards generative process health and safety culture. This
chapter discussed model fit to empirical data set for all five latent constructs and final refined
measurement model. This chapter is responding to the following research question, namely, what are
the critical drivers to achieve generative health and safety culture? The next chapter seven discusses the
model validation.

158
 CHAPTER SIX
Model Validation
_____________________________________________________________________
6.1 Introduction

In this chapter, the final research model was evaluated further and validated. Internal reliability and
validity tests were carried out through model fitness tests and hypotheses testing involving a statistical
test of significance. The internal reliability and validity test results were used in this research. The
normal distribution and multicollinearity of the data were checked, and the validity of the model was
assessed via tests of the research hypotheses.

Whatever modelling paradigm or solution method is being used, the performance measures extracted
from a model will only have some bearing on the real system represented if the model is a good
illustration of the system. What constitutes a good model is subjective, but from a performance
modelling point of view the criteria judging the goodness of models is based on how accurately
measures extracted from the model correspond to the measures which would be obtained from the
represented system.

Structural equation modelling was chosen for the model development and internal model validation
because the technique offers the most robust reliability and validity checks on the developed model
(Hair et al., 2010; Zhao, 2017). Structural equation modelling (SEM) is multivariate data analysis
approach used to assess complex relationships among constructs. It graphically models hypothesised
relationships among constructs with structural equations (Byrne, 2006; Zhao, 2017). SEM institutes
how well the theoretical model is supported by empirical data using goodness-of-fit indices (Hu and
Bentler, 1990). The assessment of model fitness against empirical data and the estimation of the
regression parameters is the primary goal of SEM (Byrne, 2006; Hu and Bentler, 1999). In this research,
the assessment of model fitness is achieved using numerous model fit indices. The fit indices are
clustered into three distinct groups namely, absolute fit indices, incremental fit indices and
parsimonious fit indices.

The model fitness test subroutine of the analysis of moment of structures (AMOS) provides an advanced
test process for this purpose (Hafeez et al., 2006; Zhao, 2017). The model fitness test studied the degrees
of fit between the final model and the empirical dataset. The following subsections discuss the AMOS-
based battery of model fitness tests; these include goodness of fit (GOF), normality, multi-collinearity
and model adequacy.

159
6.2 Normality Assessment

One critical important assumption in an SEM approach, especially in AMOS, is that the data are of
multivariate normality (Arbuckle, 2015). As skewness severely affects the means and kurtosis tends to
impact variances and covariance, it is necessary to appraise this criterion for the final model (DeCarlo,
1997; Zhao, 2017). Histograms can provide insights on skewness, behaviour in the tails, presence of
multi-modal behaviour, and data outliers. Histograms can be compared to the fundamental shapes
associated with standard analytic distributions. SEM is a method that is based on the analysis of
covariance structures, hence, the multivariate kurtosis is exceptionally important in SEM analysis.
Prerequisite to the test of multivariate normality is the need for assessment of univariate normality
(DeCarlo, 1997; Zhao, 2017).

If an observed variable fails a normality test, it is vital to look at the histogram and the normal
probability plot to see if an outlier or small subset of outliers has caused the non-normality. If there are
no outliers, it is recommended to try a transformation (such as, the log or square root) to make the data
normal (DeCarlo, 1997; Zhao, 2017). If a transformation is not a feasible alternative, nonparametric
methods that do not require normality. The Kurtosis values and corresponding critical values of the
measurement indicators in the model can be computed by AMOS version 25 in the section of assessment
of normality. According to Hoyle (1995; Zhao, 2017), a kurtosis value greater than seven or equal to
seven is an indication of a violation of normality. Using this rule of thumb as a guide, an assessment of
the generating kurtosis value in Table 6.1 revealed two observed variables Permit to Work (PTWH28
and PTWH29) are greater than seven and that reveals violation of normality.

Table 6.1 Assessment of Normality

Assessment of Normality
Std Std
Observed Variables Skewness Error Kurtosis Error
Leadership Commitment (LCH6) 1.170 0.152 2.295 0.303
Leadership Commitment (LCH7) 1.735 0.152 3.793 0.302
Leadership Commitment (LCH8) 1.200 0.152 1.547 0.302
Leadership Commitment (LCH9) 0.945 0.151 0.809 0.302
Leadership Commitment (LCH10) 1.540 0.151 3.016 0.302
Leadership Commitment (LCH11) 1.061 0.151 0.722 0.302
Leadership Commitment (LCH13) 1.058 0.151 1.201 0.302
Leadership Commitment (LCH17) 0.803 0.152 0.058 0.302
Leadership Commitment (LCH40) 1.712 0.151 3.324 0.302
Chemical Exposure Management (CEMH12) 0.953 0.151 0.668 0.302
Chemical Exposure Management (CEMH14) 0.778 0.152 0.098 0.302
Chemical Exposure Management (CEMH15) 0.209 0.153 -0.470 0.304
Chemical Exposure Management (CEMH16) 0.090 0.152 -0.399 0.302
Chemical Exposure Management (CEMH30) 1.979 0.151 4.430 0.302

160
 Health and Safety Risk Assessment (HSRAH23) -0.401 0.152 -0.195 0.303
Health and Safety Risk Assessment (HSRAH25) -0.219 0.152 -0.418 0.303
Health and Safety Risk Assessment (HSRAH27) -0.033 0.152 -0.761 0.302
Health and Safety Risk Assessment (HSRAH32) 0.024 0.151 -0.912 0.302
Health and Safety Risk Assessment (HSRAH33) 0.473 0.152 -0.365 0.302
Health and Safety Risk Assessment (HSRAH35) 0.454 0.152 -0.026 0.302
Health and Safety Risk Assessment (HSRAH37) -0.194 0.152 -0.727 0.302
Health and Safety Risk Assessment (HSRAH38) -0.021 0.152 -0.744 0.302
Health and Safety Risk Assessment (HSRAH39) 0.157 0.152 -0.969 0.302
Process Hazard Analysis (PHAH20) 1.112 0.151 1.368 0.302
Process Hazard Analysis (PHAH21) 1.007 0.151 1.476 0.302
Process Hazard Analysis (PHAH22) 1.244 0.151 1.877 0.302
Process Hazard Analysis (PHAH24) 1.564 0.152 3.449 0.303
Process Hazard Analysis (PHAH26) 1.478 0.151 2.653 0.302
Permit to Work (PTWH28) 2.571 0.152 8.903 0.303
Permit to Work (PTWH29) 3.297 0.151 16.586 0.302
Permit to Work (PTWH31) 1.874 0.152 3.394 0.303
Permit to Work (PTWH36) 1.847 0.152 3.984 0.303

6.3 Multicollinearity

Multicollinearity is defined as two or more predicators being highly correlated with each other
(Lauridsen and Mur, 2006; Zhao, 2017). Multicollinearity arises from two different sources – one is the
high correlation among underlying constructs and the other case is where two or more measurement
variables are highly correlated as they both essentially represent the same latent construct (Temme et
al., 2006; Zhao, 2017). Multicollinearity influences the parameter estimates and standard errors so that
they are far the real estimates and large standard errors. Moreover, it also affects significant values of
hypotheses testing and then it is likely to poses difficulties for theory testing (Hwang, 2009; Zhao,
2017).

The multicollinearity difficult is well understood in traditional analysis methods for non-latent
variables. However, the detection and consequence of the multicollinearity in SEM are not sufficiently
addressed (Kelava et al., 2008; Zhao, 2017). This problem cannot vanish by using more progressive
analysis techniques like SEM. Moreover, it can render the aftermath uninterpretable and generate
erroneous conclusions (Kelava et al., 2008; Zhao, 2017). Specifically, this problem imposes aggregate
influences on non-linear latent variables rather than manifest variables (Kelava et al., 2008; Zhao,
2017). In defending the final research model, the study should provide adequate checking on the
multicollinearity in order to avoid unsuitable understanding and spurious conclusions.

161
According to Zhao (2017) a vital evaluation of the output of the AMOS analysis explores that no
enormously large correlations (r >1) exist between the variables, with all latent constructs and
measurement indicators. Moreover, the signs of the standard errors also signal no multicollinearity
problem in the variables as they are extremely small.

Correlations indicate both the strength and the direction of the relationship between a pair of variables
and ranges between -1 to +1 (Bryman and Cramer, 2005). The direction of the relationships is indicated
by the positive and negative signs while the strength of the relationship is indicated by the magnitude
of the value (Bryman and Cramer, 2005). Correlation does not necessarily indicate causation (Leedy
and Ormrod, 2010). According to Pallant (2010) correlation coefficients should be greater than 0.30.
Correlations should be checked to provide a more rigorous assessment of the convergent validity. If the
correlations exceed the value of 0.30, convergent is achieved (Robinson et al., 1991). According to
Kline (2011) correlations values less than 0.90 is indicative of discriminant validity.

Table 6.2 Correlation Matrix of Leadership Commitment Latent Variable

Leadership Commitment Correlation Matrix

Leadership
LCH6 LCH7 LCH8 LCH9 LC10 LCH11 LCH13 LCH17
Commitment
LCH6 1.000
LCH7 0.510 1.000
LCH8 0.538 0.586 1.000
LCH9 0.564 0.439 0.513 1.000
LC10 0.434 0.452 0.502 0.400 1.000
LCH11 0.412 0.545 0.579 0.421 0.565 1.000
LCH13 0.485 0.448 0.466 0.457 0.472 0.512 1.000
LCH17 0.401 0.515 0.489 0.424 0.407 0.507 0.429 1.000

The correlation values between eight observed variables for leadership commitment were greater than
0.30. The results displayed in Table 6.2 suggest the attainment of convergent validity. The correlation
values ranged from 0.400 to 0.586 for the observed variables LCH6, LCH7, LCH8, LCH9, LCH10,
LCH11, LCH13 and LCH17 (above the cut off), therefore this scale meets the requirement for
discriminant validity as the correlations were less than 0.9. The Cronbach’s alpha value was above the
minimum value of 0.7 with value of 0.881 and the convergent validity characterised by high correlation
values were found to be satisfactory. This latent construct (leadership commitment) satisfied the internal
reliability criteria and construct validity.

162
 Table 6.3 Correlation Matrix of Chemical Exposure Management Latent Variable

Chemical Exposure Management Correlation Matrix

Chemical Exposure
CEMH12 CEMH14 CEMH15 CEMH16
Management
CEMH12 1.000
CEMH14 0.485 1.000
CEMH15 0.443 0.469 1.000
CEMH16 0.454 0.469 0.666 1.000

The correlation values between four observed variables for chemical exposure management were
greater than 0.30 all the observed variables. The results displayed in Table 6.3 suggest the attainment
of convergent validity. The correlation values ranged from 0.443 to 0.666 for the observed variables
CEMH12, CEMH14, CEMH15 and CEMH16 (above the cut off), therefore this scale meets the
requirement for discriminant validity as the correlations were less than 0.9. The Cronbach’s alpha value
was above the minimum value of 0.7 with the value of 0.798 and the convergent validity characterised
by high correlation values were found to be satisfactory. This latent construct (Chemical Exposure
Management) satisfied the internal reliability criteria and construct validity.

Table 6.4 Correlation Matrix of the Health and Safety Risk Assessment Latent Variable
Health and Safety Risk Assessment Correlation Matrix
Health and
Safety Risk HSRAH23 HSRAH25 HSRAH27 HSRAH32 HSRAH37 HSRAH38 HSRAH39
Assessment
HSRAH23 1.000
HSRAH25 0.463 1.000
HSRAH27 0.398 0.511 1.000
HSRAH32 0.427 0.520 0.523 1.000
HSRAH37 0.498 0.602 0.598 0.609 1.000
HSRAH38 0.448 0.445 0.398 0.440 0.529 1.000
HSRAH39 0.340 0.378 0.409 0.415 0.497 0.606 1.000

The correlation values between seven observed variables for health and safety risk assessment were
greater than 0.30 for all the observed variables. The results displayed in Table 6.4 suggest the attainment
of convergent validity. The correlation values ranged from 0.340 to 0.606 for the observed variables
HSRAH23, HSRAH25, HSRAH27, HSRAH32, HSRAH37, HSRAH38 and HSRAH39 (above the cut
off), therefore this scale meets the requirement for discriminant validity as the correlations were less
than 0.9. The Cronbach’s alpha value was above the minimum value of 0.7 with the value of 0.865 and
the convergent validity characterised by high correlation values were found to be satisfactory. This
latent construct (Health and Safety Risk Assessment) satisfied the internal reliability criteria and
construct validity.

163
 Table 6.5 Correlation Matrix of the Process Hazard Analysis Latent Variable
Process Hazard Analysis Correlation Matrix
Process Hazard Analysis PHAH20 PHAH21
PHAH20 1.000
PHAH21 0.751 1.000

The correlation values in Table 6.5 indicates that both observed variables of the process hazard analysis
were related to each other. The correlation value was 0.751. Discriminant validity was achieved as the
correlations were less than 0.9 and the Cronbach’s alpha value was above the minimum value of 0.7
with the value of 0.858. The convergent validity characterised by high correlation values were found to
be satisfactory. This construct therefore satisfied the internal reliability criteria and the construct
validity.

Table 6.6 Correlation Matrix of the Permit to Work Latent Variable

Permit to Work Correlation Matrix
Permit to Work PTWH28 PTWH29 PTWH31 PTWH36
PTWH28 1.000
PTWH29 0.597 1.000
PTWH31 0.417 0.473 1.000
PTWH36 0.399 0.438 0.402 1.000

The correlation values in Table 6.6 indicates that all four observed variables of the permit to work were
related to each other. The correlation values ranged between 0.399 to 0.597. Discriminant validity was
achieved as the correlations were less than 0.9 and the Cronbach’s alpha value was above the minimum
value of 0.7 with the value of 0.769. The convergent validity characterised by high correlation values
were found to be satisfactory. This construct therefore satisfied the internal reliability criteria and the
construct validity.

Table 6.7 Correlation Matrix of Latent Variables

Correlation Matrix of Latent Variables
Latent Variables LC CEM HSRA PHA PTW
LC 1
CEM 0.604 1
HSRA -0.099 0.040 1
PHA 0.707 0.463 -0.002 1
PTW 0.658 0.241 0.004 0.668 1

The correlation values in Table 6.7 indicates that all five latent variables of the generative health and
safety culture were related to each other. The correlation values ranged between -0.002 to 0.707.

164
Discriminant validity was attained as the correlations were less than 0.9. The convergent validity
characterised by high correlation values were found to be acceptable. These latent constructs therefore
fulfilled the construct validity. The results of the correlations of the latent variables are revealed in Table
6.7. From the results, the final research model is free from multicollinearity problem.

6.4 Reliability and Validity

The validity of the research model should be evaluated satisfactorily from the results of SEM. Given
the above validation, the reliability and validity were further assessed. Composite reliability and
discriminant validity of the final research model were further evaluated.

6.4.1 Composite Reliability

𝐶𝑅 =∑ 𝜑i2 / (∑ 𝜑i2 + ∑𝛿 i2) (6.1)

Where
𝜑𝑖 is the regression factor loading for corresponding measurement indicator
𝛿𝑖 is the measurement error of the corresponding measurement indicator
𝛿 = (1-𝜑)

Table 6.8 Composite Reliability Index

Composite Reliability Index
Influencing Factors CR
Leadership Commitment (LC) 0.840
Chemical Exposure Management (CEM) 0.835
Health and Safety Risk Assessment (HSRA) 0.814
Process Hazard Analysis (PHA) 0.977
Permit to Work (PTW) 0.816

Composite reliability is usually used to measure how well all the measurement indicators consistently
represent the corresponding latent construct (Hair et al., 2010). Moreover, the evidence of composite
reliability is established if the value of CR is more than 0.7 (Linn, 1989). The composite reliability can
be calculated by equation 6.1. The results shown in Table 6.8 suggest that the CR values of all the
constructs exceed the rule of thumb value of 0.70, demonstrating the achievement of composite
reliability on the model of acceptability and appropriateness.

165
6.4.2 Discriminant Validity

According to Zait and Bertea (2011) it is imperative to make the distinction between internal validity
and construct validity. Internal validity refers to assuring a methodology that enables the research to
rule out alternative explanations for the dependent variables, while construct validity is more concerned
with the choice of the instrument and its capability to capture the latent variable (Zait and Bertea, 2011).
Construct validity has three components, namely, convergent, discriminant and nomological validity
(Zait and Bertea, 2011).

Discriminant validity assumes that items should correlate with other items from other constructs that
are hypothetically supposed not to correlate (Zait and Bertea, 2011). Testing for discriminant validity
can be done using one of the following methods, namely, O-sorting, Chi-square difference test and the
average variance extracted analysis (Zait and Bertea, 2011). It is recommended that Q-sorting procedure
be used in the phase of exploratory research when developing a scale for measuring latent variables,
while the AVE analysis and Chi-square difference test must be used in the confirmatory stage.

This research discriminant validity was used to examine the shared variance between the constructs by
computing the average variance extracted. Discriminant validity is achieved when the AVE is greater
than the cut-off criterion 0.5. The equation 6.2 was used to calculate AVE.

𝐴𝑉𝐸 =∑𝜑i2/ 𝑛 (6.2)

Where
𝜑𝑖 is the regression factor loading for corresponding measurement indicator
n is the number of measurement indicators of the corresponding construct

Table 6.9 Average Variance Extracted Value

Average Variance Extracted Value
Influencing Factors AVE
Leadership Commitment (LC) 0.650
Chemical Exposure Management (CEM) 0.499
Health and Safety Risk Assessment (HSRA) 0.467
Process Hazard Analysis (PHA) 0.751
Permit to Work (PTW) 0.466

It is evident from Table 6.9 that the constructs in the final research model are unique and show good
discriminant validity, as their values are mostly greater than the acceptable value of 0.5. The results
recommend that they possess acceptable discriminant validity. The resulting scores for AVE ranged

166
0.466 and 0.751 and can be seen in Table 6.9. Based on the extensively acceptable minimum threshold
of 0.50 by Fornel and Larcker (1981), three constructs fell below the thresholds. The affected constructs
are chemical exposure management (CEM), health and safety risk assessment (HSRA) and permit to
work (PTW) which had values of 0.499, 0.467 and 0.466 respectively. While 0.50 is the widely accepted
threshold for AVE, 0.40 is also consider marginally acceptable especially when other measures of
validity are adequately met (Chin, 1998). Based on this threshold of 0.40, all latent constructs met the
acceptable minimum threshold.

Table 6.10 Reliability and Validity

Observed Factor Cronbach’s
Latent Constructs CR AVE
Variable Loading Alpha
LCH6 0.682
LCH7 0.719
LCH8 0.742
Leadership Commitment LCH9 0.669
0.840 0.650 0.881
(LC) LCH10 0.647
LCH11 0.721
LCH13 0.680
LCH17 0.719
CEMH12 0.602
Chemical Exposure CEMH14 0.615
0.835 0.499 0.798
Management (CEM) CEMH15 0.799
CEMH16 0.785
HSRAH23 0.608
HSRAH25 0.684
HSRAH27 0.670
Health and Safety Risk
HSRAH32 0.693 0.814 0.467 0.865
Assessment (HSRA)
HSRAH37 0.828
HSRAH38 0.659
HSRAH39 0.620
Process Hazard Analysis PHAH20 0.854
0.997 0.751 0.858
(PHA) PHAH21 0.879
PTWH28 0.764
PTWH29 0.694
Permit to Work (PTW) 0.816 0.466 0.769
PTWH31 0.615
PTWH36 0.648

It is evident in Table 6.10 that factor loading CFA ranged from 0.602 to 0.879 confirming that all factor
loading was above the threshold limit of 0.50 recommended by Anderson and Gerbing (1988).
Composite reliability index ranged from 0.814 to 0.997 for the five latent constructs signifying the
attainment of composite reliability on the model of adequacy and appropriateness. Average variance
extracted value ranged from 0.466 to 0.751, AVE measures the level of variance captured by a construct
versus the level due to measurement error. Internal reliability is achieved when Cronbach’s alpha value
is above 0.7 and the range of 0.769 to 0.881 was realised from the five latent constructs.

167
6.5 Model Parameter and Hypothesis Testing

Structural equation modelling has been used to test the hypotheses in this study. The path diagram
describes the hypothesised relationships among the latent constructs. The goodness-of-fit indices
indicate the research model is reliable and can be used to test the hypothesised relationship set in this
study. The analysis results shown in figure 6.1 are based on the questionnaire data collected from the
major petrochemical industry in South Africa, KwaZulu-Natal Province.

Leadership
Commitment (LC)
8 Observed Variables
Chemical
Health and Safety
Exposure
Risk Assessment
Management
(HSRA)
(CEM)
7 Observed Variables
4 Observed Variables

Process Hazard Generative Permit to Work

Analysis (PHA) Health and (PTW)
Safety
2 Observed Variables 4 Observed Variables
Culture

Figure 6.1 Hypothesis Testing Model

In this research factor loading was used to consider model parameters. The direction of association
between the observed variable and the latent construct as well as the size of the statement can be
determined by using unstandardized and standardised factor loadings. Unstandardised factor loadings
indicate the relationships between the manifest variables and latent variables (Diamantopoulos and
Siguaw, 2000). Unstandardized factor loadings should always be used in conjunction with standardised
factor loadings. According to Schumacker and Lomax (2004) the standardised factor loadings are used
in determining the importance one variable to other variables and is easier to interpret. In this research
only standardised factor loadings greater than 0.50 were considered suitable. Variables greater than 0.50
indicate reasonably good convergent and construct validity of the model (Hair et al., 2010).

The standardised regression weights indicated in the path diagram were used to accept or reject the
hypotheses. According to Hair et al. (2005); Hung and Lu (2008); Lattin et al. (2009) cited in Zhao
(2017) the hypotheses corresponding to a standardised regression weight less than 0.1 were rejected.
The five hypotheses were supported at the five percent level of significance.

168
Figure 6.1 presents the final structural research model for generative health and safety culture with
standardised regression weights on structural paths of the proposed hypothesised relationships. The
proposed hypothesised relationships were established by using standard regression weights, standard
errors, critical ratios and their level of statistical significance to examine whether the hypothesised
relationships are supported by the collected data.

Table 6.11 Hypothesised Relationships Testing in the Research Model

Latent Constructs Estimate S.E. C.R. P - Value Conclusion

Leadership Commitment (LC) 0.257 0.040 6.415 *** Supported

Chemical Exposure
0.221 0.045 4.897 *** Supported
Management (CEM)
Health and Safety Risk
0.397 0.078 5.065 *** Supported
Assessment (HSRA)
Process Hazard Analysis
0.412 0.050 8.159 *** Supported
(PHA)
Permit to Work (PTW) 0.201 0.035 5.776 *** Supported
Note: *** Sig (p) value is infinitesimally small (close to 0) hence cannot be reported.

It is evident in Table 6.11 that the standardised regression weights were all above the threshold limit of
0.1 with the highest value of PHA (0.412) process hazard analysis. The standardised regression weights
ranged between 0.201 and 0.412. The standard errors do not present with any extremely large or small
values and all the hypothesised relationships are supported at the statistical significance level p < 0.05
except for health and safety risk assessment (HSRA) with 0.078 standard error. The hypothesised
relationship between process hazard analysis and generative health and safety culture with a CR value
of 8.159 and regression coefficient 0.412 strongly support this hypothesis. The hypotheses leadership
commitment (LC) had CR value of 6.415, permit to work (PTW) had CR value of 5.776, health and
safety risk assessment (HSRA) had CR value of 5.065 and chemical exposure management (CEM) had
CR value of 4.897 and were all supported by the analysis results.

The analysis results indicates that the participants consider the five latent constructs as vital for
generative health and safety culture. As reported in the preceding chapter six, ten observed variables
were eliminated during the model refinement/modification for the final model to achieve the best fit,
namely, TCH18, PTW19, TCH34, LCH40, CEMH30, HSRAH33, HSRAH35, PHAH22, PHAH24 and
PHAH26.

169
6.6 Model Acceptance

The model evaluation should mainly centre on the adequacy of the parameter estimates and the whole
research model (Byrne, 2010; Zhao, 2017). Looking at the adequacy of the parameters estimated by the
model, the feasibility of the parameters and the appropriateness of standard errors were checked. To
test the feasibility of the parameters, the parameters should have correct sign and size to represent the
underlying theory (Motawa and Oladokun, 2015; Zhao, 2017). Any deviations from these recommend
that the model is unable to sufficiently explain and capture the subject under examination. An example
of these is the negative variances.

Table 6.12 Variances Estimated by the Model

Variances Estimated by the Model
Standard
Latent Constructs Error Estimate Critical Ratio p-Value
Errors
e6 0.255 0.025 10.215 ***
e7 0.241 0.024 9.977 ***
e8 0.261 0.027 9.777 ***
Leadership e9 0.299 0.029 10.329 ***
Commitment (LC) e10 0.202 0.019 10.443 ***
e11 0.268 0.027 9.976 ***
e13 0.347 0.034 10.264 ***
e17 0.526 0.051 10.362 ***
e12 0.390 0.039 10.053 ***
Chemical Exposure e14 0.363 0.037 9.943 ***
Management (CEM) e15 0.265 0.037 7.125 ***
e16 0.375 0.05 7.508 ***
e23 0.676 0.065 10.323 ***
e25 0.547 0.056 9.818 ***
Health and Safety e27 0.704 0.071 9.946 ***
Risk Assessment e32 0.702 0.072 9.771 ***
(HSRA) e37 0.398 0.052 7.609 ***
e38 0.687 0.068 10.032 ***
e39 0.838 0.082 10.279 ***
Process Hazard e20 0.167 0.028 6.037 ***
Analysis (PHA) e21 0.121 0.024 5.078 ***
e28 0.217 0.024 8.98 ***
Permit to Work e29 0.095 0.012 7.791 ***
(PTW) e31 0.366 0.037 9.799 ***
e36 0.303 0.032 9.506 ***
Note: *** Sig (p) value is infinitesimally small (close to 0) hence cannot be reported.

Critically reviewing all the variances estimated by the model, Table 6.12, proposes that the model is
acceptable in this respect. Moreover, to ensure the adequacy of the final research model the
appropriateness of the standard error should be assessed due to the fact that it supports the accuracy of
the parameter estimates. Extremely large value of standard error indicates a poor model fit. In addition,

170
the statistical significance of the parameter estimates should be checked as all the estimates should be
significant in the model.

Table 6.13 Parameter Estimated and Standard Errors

Parameter Estimates and Standard Errors
Observed Variable Estimate Standard Errors Critical Ratio P-Value
LCH6 ← LC 0.928 0.089 10.37 ***
LCH7 ← LC 1
LCH8 ← LC 1.114 0.099 11.297 ***
LCH9 ← LC 0.969 0.095 10.197 ***
LCH10 ← LC 0.752 0.076 9.866 ***
LCH11 ← LC 1.062 0.097 10.996 ***
LCH13 ← LC 1.076 0.104 10.368 ***
LCH17 ← LC 1.252 0.125 10.039 ***
CEMH12 ← CEM 1
CEMH14 ← CEM 0.998 0.129 7.752 ***
CEMH15 ← CEM 1.454 0.16 9.112 ***
CEMH16 ← CEM 1.649 0.182 9.055 ***
HSRAH23 ← HSRA 1
HSRAH25 ← HSRA 1.102 0.126 8.71 ***
HSRAH27 ← HSRA 1.204 0.14 8.594 ***
HSRAH32 ← HSRA 1.281 0.145 8.81 ***
HSRAH37 ← HSRA 1.481 0.15 9.856 ***
HSRAH38 ← HSRA 1.152 0.136 8.483 ***
HSRAH39 ← HSRA 1.147 0.142 8.104 ***
PHAH20 ← PHA 1.046 0.073 14.355 ***
PHAH21 ← PHA 1
PTWH28 ← PTW 1
PTWH29 ← PTW 0.813 0.081 10.025 ***
PTWH31 ← PTW 1.053 0.125 8.444 ***
PTWH36 ← PTW 1.044 0.118 8.831 ***

Note: *** Sig (p) value is infinitesimally small (close to 0) hence cannot be reported.

6.7 Expert Validation

The aim of this section was to further ascertain external validation by experts using literature review.

6.7.1 Leadership Commitment

Having senior managers who take a proactive interest in establishing a health and safety culture has
been considered to be a key influence on organisational health and safety performance (Hardy, 2013).
According to Okoye et al. (2016) a vital element in health and safety management system is visible
health and safety commitment from leadership and managers. Perceptions of the commitment of

171
leadership towards health and safety rather than just the intentions have a strong bearing on the actual
behaviours and performance of the people in the organisation (Holstvoogd et al., 2006).

Management commitment provides the motivation force and resources for health and safety activities
within the organisation and creating an environment of continuous improvement belongs to all levels
of management (Okoye et al., 2016). According to Hardy (2013) it is recognised that leadership is
important in the creation of a culture that supports and promotes a strong health and safety performance
of an organisation.

According to Elssayed et al. (2012) management actions related to health and safety should be adequate
and be prioritised to improve health and safety quality efficiently. Health and safety issues cannot be
tackled effectively without interference of employers with a particular pattern of behaviour as important
criteria needed to change employee’s behaviours (Zin, 2012).

6.7.2 Chemical Exposure Management

According to Vitharana et al. (2015) health hazards are properties of a chemical that have the potential
to cause adverse health effects and exposure usually occurs through inhalation, skin contact or ingestion.
The major consequences of chemical disasters include impact on livestock, flora/fauna, the environment
(air, soil, and water) and losses to industry (Reddy and Yarrakula, 2016). Chemical process hazards at
a chemical plant can give rise to accidents that affect both workers inside the plant and members of the
public who reside nearby (Chen, 2016).

According to International Labour Organisation (2017) some of the duties of the employer are to
identify risk of physical or chemical reaction of hazardous chemical and ensuring the stability of
hazardous chemicals, ensure that exposure standards are not exceeded. Process health and safety
management is critical in the chemical process industry and improving organisation knowledge and
knowledge management capabilities is an important means to prevent chemical accidents and improve
organisations’ health and safety level (Chen, 2016).

6.7.3 Health and Safety Risk Assessment

According to Eyayo (2014) globally, there are 2.9 billion workers who are exposed to hazardous risks
at their work place and annually there are two million deaths that are attributable to occupational
diseases and injuries while 4% of gross domestic product is lost due to occupational diseases and
injuries. According to Dabup (2012) the risk assessment provides a systematic approach for the
identification, management and reduction of the risk to an acceptable level.

172
Health and safety risk assessment is the evaluation of hazards to determine their potential to cause an
accident and identifying health and safety hazards in order to prevent and control them is very
imperative to the health and well-being of the workers (Eyayo, 2014). According to Albert, Hallowell
and Kleiner (2014) a critical component in health and safety risk management is to adequately identify
hazards and mitigate its associated risk using safety program elements.

The management of health and safety at work require employers to make a suitable and sufficient
assessment of the health and safety risks to employees and non-employees, arising from their work
activities (Kumar et al., 2017). Health and safety risk analysis remains an important tool in safety
management for prioritizing actions and plans, in order to commit top health and safety management
systems (Elssayed et al., 2012).

6.7.4 Process Hazard Analysis

The process hazard analysis is vital to the health and safety efforts as it provides information to assist
management and operations team to improve health and safety by reducing risk (Hardy, 2013). The
hazards may not be obvious and it is imperative that the assessments must be done by a qualified person
familiar with the work to be done (Karthika, 2013). A process hazard analysis must be conducted by a
team with expertise in engineering and process operations, including at least one employee who has
experience and knowledge on the system (Department of Transport [USA], 2000). The employers
should ensure the safe use, handling and storage of hazardous substances (Vitharana et al., 2015).

6.7.5 Permit to Work

Hazards are identified, the risk to health and safety assessed, an appropriate prevention or control
strategy gets done, but many times people still get killed in the petrochemical industry because of
misunderstandings like entering confined space without permission (Karthika, 2013). Effective
implementation of a comprehensive permitting program certainly helps preventing several undesirable
incidents. However, deficiencies in implementing a permit to work system have been a contributing
factor in several catastrophic incidents (Reddy and Reddy, 2015). According to Karthika (2013)
normally all maintenance, repair, construction work shall be carried out with a proper work permit.

6.8 Chapter Summary

This chapter evaluated the research model further and validated it. Internal reliability and validity tests
were carried out through model fitness tests and hypotheses testing involving a statistical test of
significance. The internal reliability and validity test results were used in this chapter. The normal

173
distribution and multicollinearity of the data was checked, and the validity of the model was assessed
via tests of the research hypotheses. The next chapter discusses the results of the model fit and the
hypotheses in relation to existing literature.

174
 CHAPTER SEVEN
Discussion
___________________________________________________________
7.1 Introduction

This chapter discusses the findings of the study in relation to the research objectives. The main
objectives were, to identify the main process health and safety management deficiencies that require
senior management’s attention, to identify the main drivers that could be used to improve health and
safety to reach generative process health and safety culture level five and to develop a model of
effectively managing hazardous chemical substance exposure in the petrochemical industry. The
discussion and results emphasis is on model fit in relation to existing literature. The chapter also
highlights the observed variables that were eliminated from the research and the latent variables that
were eliminated from the research.

7.2 Discussion of Findings

The study has investigated the process health and safety management systems, the ten latent variables
were assessed with observed variables. The ten latent variables were reduced to five latent variables
after principal component analysis and then structural equation modelling was employed. This
advanced method was employed to test the statistical adequacy of the proposed research model in order
to confirm whether or not the hypothesised relationships between the latent variables towards generative
health and safety culture were true. The analysis result statistically proved that the five latent variables,
namely, leadership commitment, chemical exposure management, health and safety risk assessments,
process hazard analysis and permit to work collectively influenced a generative health and safety
culture.

Path models and cognitive mapping were used to describe and explain the observed and latent variables
that significantly influenced generative health and safety culture. The reliability and validity of the five
latent constructs in this study are adequate. The goodness-of-fit indices of the structural research model
are satisfactory. The hypothesised relationship in the structural research model was tested by AMOS
version 25. The strength of the relationship between the latent constructs was represented by the
statistical significance of the path coefficient. Maximum likelihood method was used to estimate path
coefficients.

In testing the conceptual model, after identifying the underlying structure based on the existing
literature, the five latent constructs, namely, leadership commitment, chemical exposure management,

175
health and safety risk assessments, process hazard analysis and permit to work were retained in the final
model. The other five latent constructs, namely, Training and Competency, Process Health and Safety
Information, Control of Confined Space Entry, Operating Procedure and Control of Ignition Source in
the original conceptual model were eliminated from the final research model.

There were three observed variables that were eliminated since their loadings were less than 0.5 and
they are namely, observed variable 18 - The organisation communicates effectively all lessons learned
after the occupational health and safety incidents. Observed variable 19 - The organisation closes all
corrective action items effectively after the root cause analysis for all incidents happening onsite).
Observed variable 34 - My organisation has all management systems in place to manage substance
misuse. Other Observed variables were allocated to different latent variables after principal component
analysis.

Leadership commitment is the latent construct that had nine observed variables and only LCH40 was
eliminated, LCH40 - Audit compliance is an excellent practice to prevent most of health and safety
incidents in the petrochemical industry. Under chemical exposure management there were originally
five observed variables and only CEMH30 was eliminated, CEMH - All employees are aware that when
you handling hazardous chemicals you need to use prescribed personal protective equipment.

Health and safety risk assessment is the latent construct that had nine observed variables and two
HSRAH33 and HSRAH35 were eliminated, HSRAH33- My organization diligently manages fatigue in
both permanent employees and contractors, and HSRAH35 - Most of the health and safety incidents
are due to human error in my organization.

Process hazard analysis is the latent construct that had five observed variables, namely, PHAH20 – In
my organization all engineering changes undergo a comprehensive management of change. PHAH21 –
The organization does comprehensive process hazard analysis before engineering changes are made.
PHAH22 – The organization has all process health and safety information available to all employees.
PHAH24 – In my organization we have a comprehensive pre-activity start up review and pre-activity
shutdown review. PHAH26 – In my organization all work activities have a detailed operating procedure
or work instruction. The final model eliminated PHAH22, PHAH24 and PHAH26 and remained with
only PHAH20 and PHAH21.

Ten observed variables were eliminated during the model refinement/modification for the final model
to achieve the best fit, namely, TCH18, PTW19, TCH34, LCH40, CEMH30, HSRAH33, HSRAH35,
PHAH22, PHAH24 and PHAH26.

176
7.2.1 Findings related to Leadership Commitment

Table 7.1 Leadership Commitment Hypotheses Ranking

Leadership Commitment Mean Std. Deviation Ranking

Senior Management communicates Health and Safety policy
1.40 0.591 1
to all employees (LCH10).
Senior Management prioritises Health and Safety as in my
1.48 0.707 2
organisation (LCH7).
My organization has excellent chemical exposure
1.62 0.692 3
management systems (LCH6).
Senior Management allocates enough time to address Health
1.63 0.748 4
and Safety concerns (LCH11).
Senior Management has an open door policy on Health and
1.64 0.763 5
Safety issues (LCH8).
There are effective noise exposure management systems in
1.70 0.737 6
my organization (LCH9).
Employees undergo comprehensive training on health and
1.78 0.805 7
safety in my organization (LCH13).
Senior management prioritizes mechanical/asset integrity of
1.98 0.966 8
our processing plant (LCH17).

Table 7.1 shows the responses to statements about leadership commitment. It is evident that LCH10
(Senior Management communicates Health and Safety policy to all employees) with a mean of 1.40
ranked highest out of the eight statements presented to the participants. Further, LCH 7 (Senior
Management prioritises health and safety in my organisation) with a mean of 1.48 ranked second
highest.

LCH6 (My organization has excellent chemical exposure management systems), LCH11 (Senior
Management allocates enough time to address Health and Safety concerns), LCH8 (Senior Management
has an open door policy on health and safety issues), LCH9 (There are effective noise exposure
management systems in my organization), LCH13 Employees undergo comprehensive training on
health and safety in my organization, and LCH17 (Senior Management prioritises mechanical/asset
integrity of our process plant), ranked 3rd to 8th within the leadership commitment latent variable.

This research indicates that senior management must effectively communicate health and safety policy
to all employees. It is revealed in this study the senior management must prioritise health and safety in
the organisation at all levels and demonstrate that by felt leadership.

177
According to South African Bitumen Association (2010) leadership and commitment create and sustain
an organisational culture that supports effective HSE management through appropriate personal
behaviour of leaders at all levels.

 Management demonstrates strong commitment, accountability and visible leadership to HSE

through measurable actions.
 Management ensures HSE management system expectations are communicated, understood and
implemented at the appropriate levels.
 Management insists on compliance with applicable laws and regulations and the requirements of
HSE management system, and takes appropriate action to correct deficiencies.

7.2.2 Findings related to Chemical Exposure Management

Table 7.2 Chemical Exposure Management Hypotheses Ranking

Chemical Exposure Management Mean Std. Deviation Ranking

Most employees know how to handle hazardous chemicals
1.75 0.766 1
in the work place (CEMH14).
Most employees are aware of hazardous chemicals in their
1.75 0.783 2
work environment (CEMH12).
Contractor’s on boarding appreciates all hazardous
2.22 0.858 3
chemicals in my organization (CEMH15).
Most contractors know how to handle hazardous chemicals
2.59 0.991 4
in my organization (CEMH16).

Table 7.2 shows the responses to statements about chemical exposure management. It is evident that
CEMH14 (Most permanent employees know how to handle hazardous chemicals in the work place.),
with a mean of 1.35 ranked highest out of the four statements presented to the participants. Further,
CEMH12 (Most employees are aware of hazardous chemicals in their work environment.), with a mean
of 1.75 ranked second highest. CEMH15 (Contractor’s on boarding appreciates all hazardous chemicals
in my organisation.), and CEMH16 (Most contractors know how to handle hazardous chemicals in my
organisation.) ranked 3rd to 4th within the chemical exposure management latent variable.

This research revealed that all employees are aware that when handling hazardous chemicals one need
to use prescribed personal protective equipment. It has been highly indicated in this research that
employees should know how to handle hazardous chemicals to prevent process health and safety
incidents and improve the culture towards generative health and safety culture. In this research there
was only one observed variable that became a management deficiency, and that had to do with
contractors knowing how to handle hazardous chemicals.

178
7.2.3 Findings related to Health and Safety Risk Assessment

Table 7.3 Health and Safety Risk Assessment Hypotheses Ranking

Health and Safety Risk Assessment Mean Std. Deviation Ranking

Poor health and safety risk assessments are responsible for
most of health and safety incidents in the petrochemical 2.79 1.169 1
industry (HSRAH39).
Most of the health and safety incidents in the petrochemical
industry are due to not verifying energy isolation before you 2.83 1.166 2
start working on equipment (HSRAH32).
Most of the health and safety incidents in petrochemical
industry are due to poor controls of source of ignition 2.99 1.133 3
(HSRAH27).
Poor housekeeping in my organization is the cause for many
3.02 1.103 4
health and safety incidents (HSRAH38).
Most of the health and safety incidents in petrochemical
industry are due to poor controls when working at heights 3.15 1.018 5
(HSRAH25).
Most of the health and safety incidents are due to poor
controls in place when working with suspended loads 3.16 1.129 6
(HSRAH37).
Most of the health and safety incidents are due to poor
3.21 1.038 7
engineering design integrity (HSRAH23).

Table 7.3 shows the responses to statements about health and safety risk assessment. It is evident that
HSRAH39 (Poor health and safety risk assessments are responsible for most of health and safety
incidents in the petrochemical industry.), with a mean of 2.79 and standard deviation of 1.169 ranked
highest out of the nine statements presented to the participants. Further, HSRAH32 (Most of health and
safety incidents in the petrochemical industry are due to not verifying energy isolation before you start
working on equipment.) with a mean of 2.89 and standard deviation of 1.166 ranked second highest.

HSRAH27 (Most of the health and safety incidents in petrochemical industry are due to poor controls
of source of ignition), HSRAH38 (Poor housekeeping in my organization is the cause for many health
and safety incidents), HSRAH25 (Most of the health and safety incidents in petrochemical industry are
due to poor controls when working at heights), HSRAH37 (Most of the health and safety incidents are
due to poor controls in place when working with suspended loads), HSRAH23 (Most of the health and
safety incidents are due to poor engineering design integrity) ranked 3rd to 7th within the health and
safety risk assessment latent variable.

According to Dabup (2012) the risk assessment provides a systematic approach for the identification,
management and reduction of the risk to an acceptable level. Risk assessment is a critical step in risk
management. If done correctly, it determines the minimum level of preparedness in order to respond
effectively. It involves applying qualitative or quantitative techniques to potential risks. It reduces the
uncertainties in measuring risk and it usually involves frequency and severity.

179
This research revealed that 90% of the management deficiencies are in health and safety risk
assessment. The composition of the team members that do health and safety risk assessments should
have highly skilled and competent employees. It is imperative that the health and safety risk assessments
should not be just a paper exercise but must be treated with high respect from senior management.
According to Albert, Hallowell and Kleiner (2014) a critical component in health and safety risk
management is to adequately identify hazards and mitigate its associated risk using safety program
elements.

7.2.4 Findings related to Process Hazard Analysis

Table 7.4 Process Hazard Analysis Hypotheses Ranking

Process Hazard Analysis Mean Std. Deviation Ranking

The organization does comprehensive process hazard
1.70 0.732 1
analysis before engineering changes are made (PHAH21).
In my organization all engineering changes undergo a
1.72 0.787 2
comprehensive management of change (PHAH20).

Table 7.4 shows the responses to statements about process hazard analysis. It is evident that PHAH21
(The organisation does comprehensive process hazard analysis before engineering changes are made.)
with a mean of 1.49 ranked highest out of the five statements presented to the participants. Further,
PHAH20 (In my organisation all engineering changes undergo a comprehensive management of
change.) with a mean of 1.56 ranked second highest.

In this research comprehensive pre-activity start up review and pre-activity shutdown review was
identified as top priority under process hazard analysis latent construct. It was also revealed that detailed
operating procedure or work instructions are critical to reduce health and safety incidents and improve
culture towards generative health and safety culture in organisations. It is imperative that employees
comply with all operating procedure or work instructions and intervene immediately when they identify
non-compliance.

According to Hardy (2013) operating procedures describe the tasks that must be performed, data to be
recorded, and operating conditions to be maintained. The procedures also identify the health and safety
precautions, operating procedures must be clear, concise, accurate and consistent with process safety
information derived from the process hazard analysis. The procedures should be formally reviewed and
updated as necessary to assure that they are consistent with existing processes.

180
7.2.5 Findings related to Permit to Work

Table 7.5 Permit to Work Hypotheses Ranking

Permit to Work Mean Std. Deviation Ranking

In my organization before you start excavation or entering a

1.20 0.478 1
trench you need to obtain authorization (PTWH29).
All the work activities in my organization are done after a valid
1.34 0.648 2
permit to work has been signed by the authorities (PTWH28).
In my organization all safety critical equipment is disabled with
1.45 0.770 3
permission from the authorities (PTWH31).
My organization has effective management systems to manage
1.46 0.723 4
working in confined space (PTWH36).

Table 7.5 shows the responses to statements about permit to work. It is evident that PTWH29 (In my
organisation before you start excavation or entering a trench you need to obtain authorisation.)
with a mean of 1.20 ranked highest out of the four statements presented to the participants. Further,
PTWH28 (All the work activities in my organisation are done after a valid permit to work has
been approved by the authorities.) with a mean of 1.34 ranked second highest.

PTWH31 (In my organisation all safety critical equipment is disabled with permission from
the authorities.), PTWH36 (My organization has effective management systems to manage
working in confined space) ranked 3rd to 4th within the permit to work latent variable.

This research revealed that it is imperative that excavation or entering a trench must be done after
obtaining permission from the correct authorities. It is recommended that before a worker enters any
excavation over 1.2m (4 ft) in depth shoring must be in place and before a worker approaches closer to
the side or bank, the excavation must be sloped, benched, shored or supported. Excavating or trenching
work can be highly dangerous and may lead to death or severe injuries if not carried out safely. The
hazards presented by this work include burial, falls from height or being struck by objects falling into
the excavation, drowning, as well as asphyxiation or poisoning caused by fumes entering the
excavation.

It was revealed in this research that a valid permit to work is critical and plays a vital role in reducing
health and safety incidents in the work place. According to Reddy and Reddy (2015) a permit to work
is a document which specifies the task to be performed, associated foreseeable hazards and the safety
measures. The permit to work system has been widely used to ensure safety during maintenance and/or
construction activities in almost every major hazard industry worldwide.

181
7.3 Research Questions

The research questions that responses were sought for were:

 What deficiencies exist in process health and safety management systems when dealing with
hazardous chemicals?
 How should the major deficiencies be prioritised by top management to prevent process health
and safety incidents when handling hazardous chemicals?
 What are the critical drivers to achieve generative health and safety culture?

These questions were aimed at illustrating the significance of health and safety to organisations.

7.3.1 Deficiencies relative to Hazardous Chemical Exposure

There were 32 statements from the original 35 statements in the questionnaire that had to be rated in
order to identify the deficiencies relative to hazardous chemical exposure. The three statements that
were eliminated had a loading that was less than 0.5 after principal component analysis. The key process
health and safety management deficiencies for an effective process health and safety management
system therefore are, namely;

 Poor engineering design integrity.

 Poor controls when working with suspended loads.
 Poor controls when working at heights.
 Poor housekeeping.
 Poor controls of source of ignition.
 Verifying energy isolation before start working on equipment.
 Poor health and safety risk assessments.
 Effective handling of hazardous chemicals.
 Human error.
 Fatigue management for both permanent employees and contractors.

Training of employees should be emphasised and done all the time to minimize human error that could
result in process health and safety incident. All the employees should be competent in the tasks they do.
The human element of the system has one of the biggest potentials for either causing or preventing an
accident. Safe job performance by operating, maintenance personnel and contractors has a tremendous
positive impact on health and safety (Karthika, 2013). Human error is the starting point of an

182
investigation, not the end point. To do something about error, one must look at the system in which
people work and focus must extend past “what occurred” to why do it occur (Bridger, 2015). Applying
the hierarchy of controls – eliminate the risk and reduce the chance of human error.

This study has highlighted that organisations should manage fatigue for both permanent and contractors
diligently to prevent health and safety incidents. Fatigue is an experience of physical or mental
weariness that results in reduced alertness. For most people, the major cause of fatigue is not having
obtained adequate rest or recovery from previous activities. In simple terms, fatigue largely results from
inadequate quantity or quality of sleep. Both the quality (how good) and the quantity (how long) of
sleep are important for recovery from fatigue and maintaining normal alertness and performance
(Canada Aviation Industry, 2007).

7.3.2 Priority Deficiencies

Table 7.6 Means and Standard Deviations of Deficiencies

Observed Variables Mean Std. Deviation Ranking
Most of the health and safety incidents are due to poor
3.21 1.038 1
engineering design integrity.
Most of the health and safety incidents are due to poor controls
3.16 1.129 2
in place when working with suspended loads.
Most of the health and safety incidents in petrochemical
3.15 1.018 3
industry are due to poor controls when working at heights.
Poor housekeeping in my organization is the cause for many
3.02 1.103 4
health and safety incidents.
Most of the health and safety incidents in petrochemical
2.99 1.133 5
industry are due to poor controls of source of ignition.
Most of the health and safety incidents in the petrochemical
industry are due to not verifying energy isolation before you 2.83 1.166 6
start working on equipment.
Poor health and safety risk assessments are responsible for most
2.79 1.169 7
of health and safety incidents in the petrochemical industry.
Most contractors know how to handle hazardous chemicals in
2.59 0.991 8
my organization.
Most of the health and safety incidents are due to human error
2.45 0.970 9
in my organization.
My organization diligently manages fatigue in both permanent
2.41 1.092 10
employees and contractors.

Table 7.6 shows the list of 10 process health and safety deficiencies that require senior management’s
attention. It is evident that most of the health and safety incidents are due to poor engineering design
integrity with a mean of 3.21 and standard deviation of 1.038 ranked highest out of the ten statements
identified as process health and safety deficiencies. Further, it is revealed that most of the health and
safety incidents are due to poor controls in place when working with suspended loads with a mean of
3.16 and standard deviation of 1.129 ranked second highest.

183
Poor controls when working at heights, Poor housekeeping, Poor controls of source of ignition,
Verifying energy isolation before start working on equipment, Poor health and safety risk assessments,
Effective handling of hazardous chemicals, Human error and Fatigue management for both permanent
employees and contractors were ranked 3rd to 10th within the key process health and safety management
deficiencies.

7.3.3 Critical Drivers towards Generative Health and Safety Culture

The key process health and safety drivers to be prioritized for generative process health and safety
culture are, namely;

 Leadership Commitment with nine enablers.

 Chemical Exposure Management with five enablers.
 Health and Safety Risk Assessment with nine enablers.
 Process Hazard Analysis with two enablers.
 Permit to Work with only four enablers.

7.3.4 Leadership Commitment Enablers

According to Fernandez (2011) cited in Elssayed et al. (2012) employers and employees with good
safety behaviour are particularly playing a significant role in achievement of safety compliance to
occupational, safety and health improvement in industry. Safety behaviour depends directly on the
communication that exists in the firm and indirectly on management’s commitment

Maintaining a sense of vulnerability – complacency built on past success is the enemy. Leadership has
to ensure continuous improvement in environment, equipment, strategy and management systems.
Senior management must address culture and leadership through objective assessment and
improvement plans where required. Organisations must providing courageous leadership in leading
change and holding individuals accountable for safety and fatality free production. Increasing the focus
on high potential near miss events reporting and investigating them at high level is critical. High
potential incidents are any incident that, under different circumstances, would have caused more severe
consequences leading to a major incident like fatality or lost time injury.

This research revealed that senior management should communicate health and safety policy to all
employees effectively. It is imperative to prioritise health and safety over production in every
organisation. To improve health and safety organisations should have excellent chemical exposure

184
management systems. This research also revealed that senior management should allocate enough time
to address all health and safety concerns raised by employees and contractors. It is encouraged that
senior management should have open door policy on health and safety issues. The organisations are
motivated to have effective noise exposure management systems. All employees should undergo a
comprehensive training on health and safety and may be refresher training be done after every three
years. Under leadership commitment it is vital that senior management priorities mechanical/asset
integrity for all equipment.

7.3.5 Chemical Exposure Management Enablers

Chemicals can be classified on the base of hazards. The Globally Harmonised System divides hazardous
chemicals in the workplace into different categories; physical hazards, health hazards and
environmental hazards (Naafs, 2018).

This study encourages that all employees should know how to handle hazardous chemicals to prevent
unnecessary incidents and also employees should be aware of hazardous chemicals in their work
environment. The hazardous effect of chemicals comes through three ways, namely: fire, explosion and
toxicity. The first essential step towards greater plant safety is being aware of the potentially dangerous
properties of the substances, i.e. whether they are flammable, explosive or toxic (Almanssoor, 2008).

This research emphasised the importance of contractor’s on boarding to appreciate all hazardous
chemicals contractors will be exposed to during the project and that contractors should know how to
handle hazardous chemicals.

7.3.6 Health and Safety Risk Assessment Enablers

It is critical to maintain alertness to increase and unexpected risks during abnormal operating conditions.
According to Kumar et al. (2017) health and safety risk assessment is a management tool that allows
the workplace to comply with her occupational policy,

 Helps the workers do their jobs without damage to their health,

 Enables the workplace to meet her legal responsibilities,
 Enables the workplace show due diligence in the protection and promotion of health and safety of
the workers,
 Provides an auditable platform and involves the work force in protecting the health and safety of
the workers.

185
According to South African Bitumen Association (2010), company must provide structured support for
a systematic approach to manage HSE risks. High-risk work must be assessed and approved by the
high-risk working committee including client’s senior management and contractor’s senior
management team. It is also recommended that contractor’s senior management team do regular health
and safety talks while the task is being executed. Health and safety risk assessments should be done by
competent employees, it should not be just a paper exercise but these assessments should be respected.

It is highlighted in this study that verifying energy isolation before one can start working on equipment
is vital to prevent health and safety incidents. Isolation is a safety critical activity and is an integral part
of effective control of work. All stored energy sources must be identified, isolated and proven safe prior
to any work commencing. All isolations must be carried out by approved and competent personnel.

Poor controls of ignition source was identified as one of the items that should be treated with respect
by senior management and all employees. According to Puttick (2008) fire and explosion hazard
assessment flammable and potentially flammable atmospheres must be identified and compared with
the potential ignition sources present. Avoidance of ignition sources can be a useful and reliable basis
of safety in certain circumstances provided that it is restricted to the inside of chemical plants, and
certain well defined charging and discharging areas.

After every job is completed, housekeeping should be done diligently to remove all tripping hazardous
and unnecessary scrap that could be hazard in the work place. It is very important to keep the work area
tidy from debris and from the materials and substances that are part of the everyday work process. One
of the most common findings in workplaces is poor housekeeping i.e. untidiness, disorder, poor storage
of materials and stock (Ness, 2015).

This study highlights that there should be excellent controls when working at heights, fall arresters,
safety harness, three point contact must be respected at the time. The inspection of fall arrester, safety
harness should be done all the time before the safety equipment is used. Work at heights means work
in any place where, if there were no precautions in place, a person could fall a distance liable to cause
personal injury. Employers and those in control of any work at height activity must make sure work is
properly planned, supervised and carried out by competent employees.

The most obvious danger is that what goes up usually comes down and not always in the fashion we
plan it. If a load falls, it can quickly break and split, becoming a series of injurious and deadly
projectiles. It is discouraged in this study that employees can position themselves under suspended
loads. This is considered as one of the root causes for serious health and safety incidents.

186
Engineers and all technical employees should improve design integrity in all new projects in the work
place. Design integrity is assurance and verification function that ensure a product, process, or system
meets its appropriate and intended requirements under stated operating conditions (Hardy, 2013).

7.3.7 Process Hazard Analysis Enablers

According to Hardy (2013) process hazard analysis is defined as a systematic approach for identifying,
evaluating and controlling the hazards of processes involving highly hazardous chemicals. This study
also encourages organisations to have comprehensive process hazard analysis before engineering
changes are made to the existing facility and that these changes should go through a comprehensive
management of change.

7.3.8 Permit to Work Enablers

According to Karthika (2013) normally all maintenance, repair, construction work shall be carried out
with a proper work permit. Jobs where work permit is required include but not limited to following,
namely, major and minor maintenance work, inspection, construction, alteration, any hot work
(including use of normal battery driven equipment in operating areas), entry into confined space,
excavation, vehicle entry into process areas, work at height, handling of materials using mechanised
means in operating areas, erecting and dismantling of scaffolding and radiography.

This research reveals that before excavation or entering a trench one should obtain authorisation and
that all work activities should not proceed if there is no valid permit to work required for the job signed
by the correct authority. It is emphasised in this research that all safety critical equipment should be
disabled or put on override with permission from the authorities. It is vital that all organisations should
have effective management system to manage working in confined space and that permission should be
given by the correct authorities.

7.4 Summary

This chapter discussed the results of the model fit and the hypotheses in relation to existing literature.
The next chapter summaries the conclusion on the key findings of the research study in relation to the
research objectives. The next chapter will also highlight the key contributions to knowledge and the
implications for research and petrochemical industry practices. The limitations of the research study
will be outline in the next chapter. It will also share the recommendations to guide the senior

187
management team of petrochemical industries as to which factors should be prioritised to improve
health and safety towards a generative culture.

188
 CHAPTER EIGHT
Recommendations and Conclusion
___________________________________________________________
8.1 Introduction

This study adopted a positivism paradigm in order to achieve the research objective by carrying out an
extensive literature review and questionnaires were distributed, collected and analysed via SPSS version
25 and AMOS using path modelling. The chapter concludes the research effort by analysing how each
chapter has contributed towards addressing the research questions. This chapter summarises the key
findings of the study, extrapolates from the research findings, draws conclusions and makes
recommendations for future research. The limitations of the study were outlined and recommendations
were put forward to guide petrochemical industry senior management.

8.2 Key Research Findings and Conclusions

The findings and conclusions for each research questions are as follows:

1. What deficiencies exist in process health and safety management systems when dealing with
hazardous chemicals?

This study has explored the process health and safety management systems to identify the deficiencies
that require senior management’s attention. The key findings in relation to the research questions are as
follows. The key process health and safety management deficiencies to be prioritized for an effective
process health and safety management system are, namely;

 Poor engineering design integrity.

 Poor controls when working with suspended loads.
 Poor controls when working at heights.
 Poor housekeeping.
 Poor controls of source of ignition.
 Verifying energy isolation before start working on equipment.
 Poor health and safety risk assessments.
 Effective handling of hazardous chemicals.
 Human error.
 Fatigue management for both permanent employees and contractors.

189
The objectives addressed in this section are to identify health and safety hazards and describe the
awareness of occupational health and safety hazards to the employees in the petrochemical industry, to
test the effectiveness of the existing process health and safety management systems in petrochemical
industry and to review the process health and safety management systems within industries where
hazardous chemical substances are being used and identify the deficiencies.

2. How should the major deficiencies be prioritised by top management to prevent process health
and safety incidents when handling hazardous chemicals?

It is evident in Table 8.1 that engineering design integrity ranked highest out of the ten statements
identified as process health and safety deficiencies. Further, it is revealed that working with suspended
loads ranked second highest. working at heights, housekeeping, controls of source of ignition, energy
isolation before start working on equipment, health and safety risk assessments, contractors knowing
how to handle of hazardous chemicals, human error and fatigue management were ranked 3rd to 10th
within the key process health and safety management deficiencies.

Table 8.1 Deficiencies Ranking

Process Health and Safety Deficiencies Ranking

Engineering design integrity 1
Working with suspended loads 2
Controls when working at heights 3
Housekeeping 4
Controls of source of ignition 5
Energy isolation before you start working on equipment 6
Health and safety risk assessments 7
Contractors knowing how to handle hazardous chemicals 8
Human error 9
Fatigue Management 10

The objective addressed in this section is the one to assess the existing process health and safety
management systems deficiencies and prioritise the major deficiencies that need urgent senior
management’s attention.

3. What are the critical drivers to achieve generative health and safety culture?

This study has explored the process health and safety management systems to identify the drivers
towards generative process health and safety culture. The key process health and safety drivers to be
prioritized for generative process health and safety culture are, namely;

190
 Leadership Commitment
 Chemical Exposure Management
 Health and Safety Risk Assessment
 Process Hazard Analysis
 Permit to Work

The objective addressed is to identify critical drivers to achieve generative process health and safety
culture;

Table 8.2 Generative Process Health and Safety Culture Toolkit (JX Nyawera Toolkit)
Generative Process Health and Safety Culture Toolkit
Latent Variable Observed Variable %
Do we have excellent chemical exposure management systems?
Is senior management prioritising health and safety?
Leadership Do we have open door policy on health and safety issues?
Commitment Do we have effective noise exposure management systems?
(Each Score > Is senior management communicating health and safety policy to all employees?
70% and Average Is senior management allocating enough time to address health and safety
Score > 80%) concerns?
Are employees undergoing a comprehensive health and safety training?
Is senior management prioritising asset integrity of the plant?
Are employees aware of all hazardous chemicals in their work environment?
Chemical
Do all employees know how to handle hazardous chemicals in their work
Exposure
environment?
Management
(Each Score > Do we have effective contractors on boarding management system that
80% and Average appreciates all hazardous chemicals?
Score > 90%) Do all contractors know how to handle hazardous chemicals in their work
environment?
Do we do effective health and safety risk assessments at engineering design
stage?
Health and Do we do effective health and safety risk assessments when employees are going
Safety Risk to work at heights?
Assessment (Each Do we have effective controls to manage source of ignition?
Score > 85% and Do we have effective lock out tag out management systems?
Average Score > Do we have effective management systems for lifting objects?
95%) Do we have effective housekeeping management system?
Do we have effective health and safety risk assessments done by competent
people?
Process Hazard Do we do comprehensive management of change (organisation and engineering
Analysis (Each changes)?
Score > 80% and
Average Score > Do we do comprehensive process hazard analysis before engineering changes are
90%) made?
Permit to Work Do we have an effective permit to work management system?
(Each Score > Do we have an effective excavation management system?
85% and Average Do we have an effective safety critical protective device management system?
Score > 95%) Do we have an effective confined space entry management system?
Total Average Total Average Score is recommended > 90% for the organisation to be considered to
Score have a Generative Health and Safety Culture

191
Table 8.2 proposes a management tool that could be used as a ‘thermometer’ for organisations to assess
the state of their health and safety management systems with an intention to gauge themselves in terms
of how close they are to a generative health and safety culture. The percentages are merely suggestive
of what could be used to give an indication of just where the organization might be at a given point in
time and the particular areas that were deficient and needed attention to improve the prevailing health
and safety culture within the organization. The generative process health and safety culture toolkit is
recommended to be used by experienced employees with more than 15 years in the industry that would
appreciate what was being examined within the organization and its processes. The same tool would be
used to evaluate the effectiveness of interventions introduced as a result of a previous assessment. In
this way the tool would be part of an iterative and evaluative process of continuous improvement.

8.3 Proposed Generative Process Health and Safety Culture Model

The proposed process health and safety model to improve organizational culture towards a generative
process health and safety culture is shown in Figure 8.1.

Leadership
Commitment (LC)
8 Observed Variables
Chemical
Health and Safety
Exposure
Risk Assessment
Management
(HSRA)
(CEM)
7 Observed Variables
4 Observed Variables

Process Hazard Generative Permit to Work

Analysis (PHA) Health and (PTW)
Safety
2 Observed Variables 4 Observed Variables
Culture

Figure 8.1 Generative Process Health and Safety Culture Model (JX Nyawera Model)

Leadership commitment had eight enablers, namely;

 Communication of Health and Safety policy to all employees;

 Senior Management to prioritise Health and Safety;
 Excellent chemical exposure management systems;
 Senior management to allocate enough time to address Health and Safety concerns;

192
  Senior Management has an open door policy on Health and Safety issues;
 Effective noise exposure management systems;
 Comprehensive training on health and safety; and
 Senior management to prioritize mechanical/asset integrity.

Chemical exposure management had four enablers, namely;

 Employee’s knowledge of how to handle hazardous chemicals;

 Awareness of hazardous chemicals;
 Contractor’s on boarding to appreciate all hazardous chemicals; and
 Contractor’s knowledge of how to handle hazardous chemicals.

Health and safety risk assessment had seven enablers, namely;

 Engineering design integrity;

 Controls when working with suspended loads;
 Controls when working at heights;
 Housekeeping;
 Controls of source of ignition;
 Verifying energy isolation before start working on equipment; and
 Health and safety risk assessments.

Process hazard analysis had two enablers, namely;

 Comprehensive process hazard analysis before engineering changes are made; and
 All engineering changes undergo a comprehensive management of change.

Permit to work latent construct had four enablers, namely;

 Obtain authorization before you start excavation or entering a trench;

 A valid permit to work has been signed by the authorities;
 All safety critical equipment is disabled with permission from the authorities; and
 Effective management systems to manage working in confined space.

193
8.4 Contribution of the Research

This section focuses on the contributions of this thesis. The section is divided into three subsections
that address its contribution to theory, methodology and practice.

8.4.1 Theoretical Contribution

The study offers an innovative analytical and methodological approach to assessment of process health
and safety culture. The thesis is part of the larger discussion of increasing importance in health and
safety policymaking. This study aims at contributing to the literature in the field of health and safety by
incorporating management deficiencies that require senior management’s attention and the drivers
towards a generative process health and safety culture. It offers an innovative methodology in assessing
petrochemical industry performance in health and safety.

8.4.2 Methodological Contribution

Methodological contribution lies in the experience gained through the application of a positivist
approach and techniques applied for data collection. The other methodological contribution relates to
the appropriateness of applying theoretical concepts and theories developed in other contexts. The
successful use of these theories in this study contributes towards providing examples of the positivism
research approach. The study contributes methodologically the five drivers and ten main deficiencies
that require senior management’s undivided attention. The research employed reliability measures and
validity to ensure that the research instruments were consistent and valid.

8.4.3 Practical Contribution

One of the practical contributions of this research is the comprehensive awareness provided by the
review of the literature as part of this study. The literature review revealed that senior management
needed to acquire new skills in improving the health and safety culture in the petrochemical industry.
The contribution of this research is to understand, based on theoretical assumptions, how the health and
safety improvement could be institutionalised in an organisation and how it might contribute to ‘happy’
employees. The developed model can be used as a practical tool.

The research proposed a path model to improve health and safety culture towards generative health and
safety where the whole organisation appreciates that health and safety systems is the way that business
is and should be done in the organization. The path model in this research supported five latent
constructs that were found to be drivers towards a generative process health and safety culture. There

194
were ten deficiencies in the existing management system that required the undivided attention of senior
management to improve health and safety towards the goal of a generative process health and safety
culture.

8.5 Limitations of the Research

There are limitations to this study. The proposed conceptual model has been tested and validated by
gathering data and information from the key petrochemical industry stakeholders that are responsible
for production, technical services, maintenance, health, safety and environment, turnarounds and
engineering projects. Due to self-reported methods of data collection, there is a probability of bias
existing in the results of the study. The limitations associated with self-reporting include honesty in
response or social desirability, introspective ability of participants, question understanding,
interpretation of the rating scale and respondent response bias. It is possible that the research sample
may differ significantly from the general population of interest even though there was no evidence
found to suggest so.

The sample was based on a convenient sample. The research has some limitations on external validity
and so may not be entirely generalizable beyond the research sample. The research is based on survey
instrument which was designed by the authors and so have not been broadly validated. The reliability
and validity of the research instrument may be questioned as it was not adequately assessed for face
validity through a panel of experts. Predictive and constancy were not assessed.

Common method bias might have occurred due to the data gathered from the questionnaire being self-
assessed even though considerations were given to avoid bias when devising the questionnaire.
Different participants might have had different interpretations about the statements assessed. In spite of
this limitation, this study provides empirical support for the conceptual research model showing the
relationships between observed variables, latent variables and generative health and safety culture. This
builds a foundation for further studies.

The scope of information gathering for this research was limited to South Africa KwaZulu-Natal
province and data collection is from the petrochemical industry. This research was limited to only
petrochemical industry. However, the theoretical principles and methodologies on which the study was
based are general and can be applied to other industries. The potential limitations of this study and
hence the suggestions for future studies need to be addressed. The data set may be a good representation
in this context only. Although the study attempts to encompass as much about process health and safety
management systems, potential limitations do exist due to not assessing all health and safety
management systems.

195
This study was based entirely on quantitative research data. A mixed methods approach would have
provided a check on the validity of the findings by comparing results from a qualitative data source.
The proposed generative health and safety model was not validated by testing too establish its
effectiveness when used.

8.6 Suggestions for Future Research

The following suggestions are made for possible future studies.

 It is not clear from literature how the organisations are assessed for placement in the five levels
in the health and safety culture ladder. Therefore, a standard check list assessment tool will be
a great contribution to health and safety culture.
 Using the findings of this study as a starting point, future studies could repeat the research with
broader populations, which would assist in generalisability of the findings.
 The research only considered petrochemical industry. Further study should increase the scope
to other construction industries.
 This study was positivist. Future research studies could use mixed methods in order to obtain a
greater perspective on the topic. A different methodology may be used to validate results from
this study.
 Further research could be conducted on the hypothesized relationships that were rejected.
 This research could be used as a basis for organisations to improve the health and safety culture
to generative culture.
 This research could be used where the application of the model is assessed for generative health
and safety culture.

8.7 Conclusion

A generative process health and safety culture model was developed from an extensive literature review.
It was theorised that five drivers, namely, Leadership Commitment, Chemical Exposure Management,
Health and Safety Risk Assessment, Process Hazard Analysis and Permit to Work had a positive
influence towards improving process health and safety culture. It was also theorised that the main
deficiencies that require senior management’s attention to improve health and safety culture towards
generative culture level five are, namely, Poor engineering design integrity, Poor controls when working
with suspended loads, Poor controls when working at heights, Poor housekeeping, Poor controls of

196
source of ignition, Verifying energy isolation before start working on equipment, Poor health and safety
risk assessments, Effective handling of hazardous chemicals, Human error and Fatigue management for
both permanent employees and contractors.

The petrochemical industry is still considered as major hazardous installation that needs senior
management team to advance health and safety culture by prioritising it at all cost. Security,
affordability and sustainability remain the most fundamental trilemma for individuals, organisations,
nations and the world.

197
 REFERENCES

Abdelhamid, T.S. & Everett, J.G. (2000). Identifying Root Causes of Construction Accidents. Journal
of Construction Engineering and Management.

Agumba, J.N. & Haupt, T.C. (2018). The influence of health and safety practices on health and safety
performance outcomes in small and medium enterprise projects in the South African
construction industry. Journal of the South African institution of civil engineers. Vol. 60, No.3,
pp. 61-72.

Aggarwal, C.C. (2016). Outlier Analysis. Second Edition. IBM T.J. Watson Research Center.

Aibinu, A. A. & Al-Lawati, A. M. (2010). Using PLS- SEM technique to model construction
organization’s willingness to participate in bidding. Automation in Construction, 19(6), 714-
724.

Albert, A., Hallowell, M.R. & Kleiner, B.M. (2014). Emerging Strategies for Construction Safety &
Health Hazard Recognition. Journal of Safety, Health and Environmental Research. Vol. 10,
No.2, pp. 152 -161.

Anderson, J. C. & Gerbing, D. W. (1988). Structural Equation Modelling in Practice: A

Review and Recommended Two-Step Approach. Psychological Bulletin, Vol. 103, pp. 411-
423.

Almanssoor, A.A. (2008). Planning of Petrochemical Industry under Environmental Risk and Safety
Considerations. University of Waterloo.

Alshetewi, S., Karim, F., Goodwin, R. & de Vries, D. (2015). A structural equation model of
governing factors influencing the implementation of T-Government. International journal of
advanced computer science and applications. Vol. 6, No.11, pp 119-125.

Alvi, M. (2016). A manual for selecting sampling techniques in research. Munich Personal Repec
Archive. University of Karachi.

Andrews, T. (2012). Grounded Theory Review. What is Social Constructionism? Vol. 11, Issue 1.

198
Angrosino, M. (2007). Doing ethnographic and observational research. Thousand Oaks,
CA: Sage.

Arbuckle, J. L. (2015). IBM SPSS AMOS 23 user's guide. New York: IBM.

Arthur, A. L., Hoffmann, A. A. & Umina, P. A. (2014). Estimating Densities of the Pest
Halotydeus destructor in Canola. Journal of Economic Entomology, Vol. 107, Issue 6, pp.
2204-2212.

Azen, R. & Sass, D. A. (2008). Comparing the squared multiple correlation coefficients of non-
nested models: An examination of confidence intervals and hypothesis testing. British Journal
of Mathematical and Statistical Psychology, Vol. 61, pp. 163 - 178.

Bagozzi, R. P. (2010). Structural equation models are modelling tools with many ambiguities:
comments acknowledging the need for caution and humility in their use. Journal of Consumer
Psychology, Vol. 20, pp. 208 - 241.

Bagozzi, R. P. & Yi, Y. (2012). Specification, evaluation, and interpretation of structural equation
models. Journal of the Academy of Marketing Science, Vol. 40, pp 8 - 34.

Barrett, P. (2007). Structural equation modelling: adjudging model fit. Personality and Individual
Differences, Vol. 42, pp. 815 - 824.

Bentler, P. M. & Bonett, D. G. (1980). Significance tests and goodness of fit in the analysis of
covariance structures. Psychological Bulletin, Vol. 88, pp. 588 - 606.

Beriha, G.S., Patnaik, B., Mahapatra, S.S. & Padhee S. (2012). Assessment of safety
performance in Indian industries using fuzzy approach, Elsevier, Vol. 39 No. 3, pp. 3311-3323.

Blaikie, N. (2007). Approaches to social enquiry. Cambridge: Polity Press.

Bond, J. (2007). A safety culture with justice: a way to improve safety performance, IchemE
Symposium series no. 153, pp. 1-6.

Boone, H.N. & Boone, D.A. (2012). Analysing Likert Data. Journal of Extension. Vol. 50, No.2.

199
Bridger, R.S. (2015). Human Factors and Human behaviour in Safety Management and
Accident Investigation, Human Factors Department, October 16, Institute of Naval
Medicine, pp.1-13.

Bridges, W. & Revonda, T. (2010). Human Factors Elements missing from process safety
management, 6th Global Congress on Process Safety, 44th Annual Loss Prevention Symposium,
Texas, March 2010, Process Improvement Institute, pp. 1-24.

Broadribb, M.P. (2007). Lessons from Texas City a Case History, IChemE Symposium series No.153,
pp. 1-9.

Bryman, A., & Bell, E. (2011). Business Research Methods (3rd ed.). New York: Oxford
University Press.

Bryman, A. & Cramer, D. (2005). Quantitative Data Analysis with SPSS 12 and 13 – A guide for
Social Scientists. London: Routledge, Talyor & Francis Group.

Bryman, A. (2016). Social Research Methods. The nature and process of social research. Fifth edition.

Byrne, B.M. (2010). Structural Equation Modelling with AMOS — Basic Concepts, Applications and
Programming. London: Routledge, Taylor and Francis.

Byrne, B.M. (2012). Structural Equation Modelling with Mplus. Basic Concepts, Applications, and
Programming. Multivariate Applications Series.

Campbell, J.M. (2008). Safety hazard and risk identification in infrastructure management (Doctoral
dissertation). NDT and Construction Management Research Group.

Campbell, D. & Campbell, S. (2008). Introduction to Regression and Data Analysis. StatLab
Workshop Series.

Canada Aviation Industry, (2007). Fatigue Management Strategies for Employees. Fatigue Risk
Management System for the Canadian Aviation Industry.

Charmaz, K. & Mitchell, R.G. (2001). Grounded Theory in Ethnography. In: P. Atkinson, A.
Coffey, S. Delamont, J. Lofland and L. Lofland, eds, Handbook of Ethnography. pp.
160-176.

200
Charmaz, K. (2006). Constructing Grounded Theory. London: Sage.

Chen, M. (2016). Process Safety Knowledge Management in the Chemical Process Industry. American
Journal of Chemical Engineering. Vol. 4, No.5, pp. 131-138.

Chin, W.W. (1998). Issues and opinion on structural equation modelling. MIS Quarterly, Vol. 22, No.
1, pp. 7-16.

Cohen, L., Manion, L. & Morrison, K. (2007). Research methods in education. 6th ed.
London: Routledge.

Cornish, R. (2006). An introduction to sample size calculations. Mathematics Learning Support Centre.

Cooper, D. (1998). Improving Safety Culture, A practical Guide. Applied Behavioural Science.

Cooper, D.R. & Schindler, P.S. (2005). Business research methods. 5th ed. New York:
McGraw-Hill.

Creswell, J.W. (2005). Educational Research, Planning, conducting, and evaluating

quantitative and qualitative research. Upper saddle river, Pearson Merrill prentice hall.

Creswell, J.W. & Plano Clark, V.L. (2007). Designing and conducting mixed methods
research. Thousand Oaks, CA: Sage Publications.

Creswell, J.W. (2009). Research design: Qualitative and mixed methods approaches. London:
Sage Publications.

Crotty, M. (1998). The foundations of social research. London: Sage Publications.

Dabup, N.L. (2012). Health, Safety and Environmental Implications in Nigeria’s Oil and Gas Industry.
Nelson Mandela Metropolitan University.

Dahiru, T., Aliyu, A. & Kene, T.S. (2006). Statistics in medical research: misuse of sampling and
sample size determination. Annals of African Medicine. Vol. 5, No.3, pp. 158-161.

DeCarlo, L. T. (1997). On the meaning and use of kurtosis. Psychological Methods, 2, 292-307.

201
Denzin, N.K. & Lincoln, Y.S. (2011). The Sage Handbook of qualitative research. Washington D.C.:
SAGE Publications.

Department of Labour [USA], (2016). Process Safety Management Guidelines for Small
Business Compliance. Government of United States of America.

Department of Transport [USA], (2000). Federal Aviation Administration System Safety Handbook,
Chapter 8, Safety Analysis/Hazard Analysis Tasks.

Diamantopoulous, A. & Sigaw, J.A. (2000). Introduction LISREL. London: Sage Publications:

Dillman, D. A. (2007). Mail and internet surveys: the tailored design. New York: John Wiley & Sons.

Driscoll, D.L., Yeboah, A.A., Salib, P. & Rupert. D.J. (2007). Merging qualitative and
quantitative data in mixed methods research: How to and why not. Ecological and
environmental anthropology. Vol.3, No.1, pp. 19-28.

Duguid, I.M. (2008). Analysis of Past Incidents in the Process Industries. Symposium Series No.154.

Dunbar, M. (2014). Hazard Identification and Risk Assessment. Publication of the Association for Iron
& Steel Technology.

Easterby-Smith, M., R., T., & Lowe, A. (2002). Management research: an introduction (2nd ed.): Sage.

Elssayed, M.Y., Hassan, H.M. & Hosny, G. (2012). Assessment of Occupational Level of Awareness
for Health and Safety in Sidi Kerir Petrochemical Company. Egyptian Journal of Occupational
Medicine. Vol. 36, Issue 2, pp. 191-203.

Eyayo, F. (2014). Evaluation of occupational health hazards among oil industry workers: A
case study of refinery workers. Journal of environmental science, toxicology and food
technology. Vol. 8, issue 12, pp. 22-53.

Ezejiofor, T.I.D. (2014). Risk Assessment: Reappraisals for Potential Hazards in the Operational
Environment and Facilities of Petroleum Refining and Distribution Industry in Nigeria –
Research and Review. Occupational Medicine and Health Affairs. Vol. 2, Issue 4, pp. 1-19.

Fetterman, D. M. (1998). Ethnography: Step by step (2nd edition). Thousand Oaks, CA: Sage.

202
Flin, R. (2015). Safety Intelligence and Chronic Unease in Senior Managers. Developing
improvement programmes for safety culture.

Ford, M.A., Haskins, M.A. & Wade, C. (2014). Weight management through motivational counselling
in the workplace. The journal of safety, health and environment research. Vol. 10, issue 2, pp.
178-183.

Fornell, C. & Larcker, D. F. (1981). Evaluating structural equation models with unobservable
variables and measurement error. Journal of Marketing Research, 18(1), 39-50.

Fox, N., Hunn, A. & Mathers, N. (2009). Sampling and Sample Size Calculation. National Institute for
Health Research.

Fuller, C.W. & Vassie, L.H. (2005). Health and Safety Management: Principles and Best
Practice. London: FT Prentice Hall.

Gainey, T. W. and Klaas, B. S. (2003). The outsourcing of training and development: Factors
impacting client satisfaction. Journal of Management, 29(2), 207-229.

Galvani, S. (2015). Alcohol and other Drug Use: The Roles and Capabilities of Social Workers.
Manchester Metropolitan University.

Gauri, P. & Gronhaug, K. (2002). Research methods in business studies. New York: Pearson
Prentice Hall.

Gelo, O., Braakmann, D. & Benetka, G. (2008). Quantitative and qualitative research: Beyond the
debate. Integrative Psychological and Behavioural Science. Vol. 42, Issue 3, pp. 266-290.

George, A.L. & Bennet, A. (2005). Case studies and theory development in the social
sciences. Cambridge, MA: The MIT Press.

Gogtay, N.J. (2010). Principles of sample size calculation. Indian Ophthalmol. Vol. 58, pp. 517-518.

Grajo, L. C., Candler, C., Bowyer, P., Schultz, S., Thomson, J. and Fong, K. (2016). Determining the
internal validity of the inventory of reading occupations: an assessment tool of children’s
reading participation. American Journal of Occupational Therapy, Vol. 70, No.3, pp. 1-9.

203
Green, T. (2016). A methodological review of structural equation modelling in higher education
research. Studies in Higher Education, Vol 41, No.12, pp. 2125-2155.

Grix, J. (2004). The foundations of research. London: Palgrave Macmillan.

Groanland, E. & Stalpers, J. (2012). Structural equation modelling-A verbal approach. Nyenrode
Research Paper, Vol. 12, No. 2, pp. 1-39.

Guetterman, T.C. (2015). Descriptions of Sampling Practices within five approaches to qualitative
research in education and health sciences. Forum Qualitative Social Research. Vol. 16, No.2,
Art.25.

Hafeez, K., Keoy, K. H. and Hanneman, R. (2006). E-business capabilities model validation and
comparison between adopter and non-adopter of e-business companies in UK. Journal of
Manufacturing Technology Management, Vol. 17, No. 6, pp. 806 - 828.

Hair, J. F., Black, W.C., Babin, B. J. & Anderson, R. E. (2010). Multivariate data analysis. (7th ed.).
Upper saddle River, NJ: Prentice Hall.

Hardy, T.L. (2013). Elements of Process Safety Management, Great Circle Analysis, Denver,
CO, USA. pp. 1-12.

Harinarain, N. (2013). Effective HIV and AIDS Management: A South African Construction Sector
Model. University of KwaZulu-Natal.

Hooper, D., Coughlan, J. & Mullen, M. R. (2008). Structural Equation Modelling:

Guidelines for Determining Model Fit Electronic Journal of Business Research Methods, Vol.
6, No.1, pp. 53 - 60.

Holstvoogd, R., Van der Graaf G., Bryden R., Zijlker, V. & Husdon, P. (2006). Hearts and
Minds Programmes the road map to improved HSE culture, Symposium Series No. 151, Shell
Global Solutions International, pp. 1-13.

Hox, J.J. & Boeije, H.R. (2005). Data Collection, Primary vs. Secondary. Encyclopaedia of social
Measurement. Vol. 5, pp. 593-599.

204
Hox, J. J. and Kleiboer, A. M. (2007). Retrospective questions or a diary method? A two level
multitrait-multimethod analysis. Structural Equation Modelling, Vol. 14, pp. 311-325.

Hoyle, R. H. (1995). Structural equation modelling: concepts, issues, and applications. Thousand
Oaks, CA: Sage Publications.

Hsu, C. & Sandford, B.A. (2007). The Delphi Technique: Making Sense of Consensus.
Practical Assessment Research & Evaluation, Vol.12, No.10, pp. 1-8.

Hu, L. & Bentler, P. M. (1999). Cut-off criteria for fit indexes in covariance structure analysis:
conventional criteria versus new alternatives. Structural Equation Modelling, Vol. 6, No.1,
pp.1-55.

Hughes, P. & Ferret, E. (2007). An Introduction to Health and Safety in Construction.

The Handbook for the NEBOSH (National Examination Board in Occupational Safety and
Health).

Hwang, H. (2009). Regularized generalized structured component analysis. Psychometrika, Vol. 74,
No. 3, pp. 517- 530.

Iacobucci, D. (2010). Structural equations modelling: Fit indices, sample size, and advanced topics.
Journal of Consumer Psychology, Vol. 20, pp. 90-98.

International Labour Organisation, (2017). Occupational safety and health in the oil and gas industry in
selected sub-Saharan African countries. Sectoral Policies Department. First Edition.

Jha, S.K., Dwivedi, A.K. & Tiwari, A. (2011). Necessity of Goodness of Fit Tests in Research and
Development. International Journal of Computer Science and Technology. Vol. 2, Issue 3, pp.
135-141.

Johnson, M.P. (2014). Secondary Data Analysis: A Method of which the time has come. Qualitative
and Quantitative Methods in Libraries. Vol. 3, pp. 619-626.

Joshi, A., Kale, S., Chadel, S. & Pal, D.K. (2015). Likert Scale: Explored and Explained. British
Journal of Applied Science and Technology. Vol. 7, No.4, pp. 396-403.

Joubert, D. & Bates, G. (2008). Occupational heat exposure. Occupational health southern Africa.

205
Jung Kim, H. (2017). Regression Analysis – Basics.

Karthika, S. (2013). Accident Prevention by using Hazop Study and Work Permit System in Boiler.
International Journal of Advanced Engineering Research and Studies. Vol. 2, Issue 2, pp. 125-
129.

Kelava, A., Moosbrugger, H., Dimitruk, P. & Schermelleh-Engel, K. (2008). Multicollinearity and
missing constraints a comparison of three approaches for the analysis of latent nonlinear effects.
Methodology, Vol. 4, No.2, pp. 51-66.

Kim, I. (2016). Ergonomic involvement for occupational safety and health improvements in
the oil and gas industry, Journal of Ergonomics, Vol. 6, Issue 3, pp. 1-3.

Kimberlin, C.L. &Winterstein, A.L. (2008). Validity and Reliability of measurement instruments used
in research. Research Fundamental. Vol. 65, pp. 2276-2284.

Kline, R.B. (2011). Principles and practice of structural equation modelling. 3rd ed. New York:
Palgrave-Macmillan.

Kulkarni, S.J. (2017). Safety and Hazards in Petroleum Industries: Research, Studies and
Surveys. International Journal of Petroleum and Petrochemical Engineering, Vol. 3, Issue 1,
pp. 31-34.

Kumar, R.M., Karthick, R.B., Bhuvaneswari, V. & Nandhini, N. (2017). Study on

occupational health and diseases in oil industry, International research journal of engineering
and technology, Vol. 4, Issue 12, pp. 954-958.

Langkulsen, U., Vichit-Vadakan, N. & Taptagaporn, S. (2011). Safety and health in the Petrochemical
Industry in Map Ta Phut, Thailand. Journal of Occupational Health. Vol.53, pp. 384-392.

Lauridsen, J. & Mur, J. (2006). Multicollinearity in cross-sectional regressions. Journal of

Geographic Systems, Vol. 8, No. 4, pp. 317-333.

Leedy, P.D. & Ormrod. J.E. (2005). Practical research, planning and design. 8th ed. New York: Pearson
Prentice Hall.

206
Leedy, P.D. & Ormrod. J.E. (2010). Practical research, planning and design. 9th ed. Upper Saddle
River, New Jersey: Pearson Education Inc.

Leeuw, E.D., Hox, J. & Huisman, M. (2003). Prevention and Treatment of Item Nonresponse. Journal
of Official Statistics. Vol. 19, No.2, pp. 153-176.

Lerman, S.E., Flower, D.L., Benjamin, G. & Hursh, S.R. (2012). Fatigue Risk Management in the
Workplace. Vol. 54, No. 2, pp. 231- 258.

Lin, W. & Green, D.P. (2016). Standard Operating Procedures: A Safety Net for Pre-Analysis Plans.
American Political Science Association.

Linn, R. (1989). Educational measurement. Washington, DC: Council on Education.

Lipsey, A.S.M. (2009). Generalisability: the role of meta-analysis. National research council,
Washington DC.

Malhotra, G. (2017). Strategies in Research. International Journal of Advance Research and

Development. Vol. 2, Issue 5, pp.172-180.

Marsh, H.W., Ludtke, O., Nagengast, B. & Morin, A.J.S. (2013). Why item Parcels are (almost) never
appropriate: Two wrongs do not make a right – camouflaging misspecification with item
parcels in CFA models. American psychological association. Vol. 18, No.3, pp. 257-284.

Maydeu-Olivares, A. & Garcia-Forero, C. (2010). Goodness of Fit Testing. International Encyclopaedia

of Education. Vol. 7, pp. 190-196.

McKenzie, C. (2007). Lessons Learned from the BP Refinery Explosion and other Incidents, 10th
Annual Symposium: The prevention of Serious Accidents. U.S. Chemical Safety and Hazard
Investigation Board.

Mona, E. & Martin, B. (2016). Factor structure and construct validity of the worker role self-
assessment (WRS) when used for people with psychiatric disabilities in Sweden. Evaluation
& the Health Professions, Vol. 39, No.3, pp. 299-316.

207
Molenaar, K., Washington, S. & Diekmann, J. (2000). Structural equation model of construction
contract dispute potential. Journal of Construction Engineering and Management, Vol. 126,
No.4, pp. 268-277.

Molwus, J. J., Erdogan, B. & Ogunlana, S. (2017). Using structural equation modelling (SEM) to
understand the relationships among critical success factors (CSFs) for stakeholder management
in construction. Engineering, Construction and Architectural Management, Vol. 24, No.3, pp.
426-450.

Motawa, I. & Oladokun, M. G. (2015). Structural equation modelling of energy consumption in

buildings. International Journal of Energy, Vol. 9, No.4, pp. 435-45

Mouton, J. (1996). Understanding Social Research. Pretoria: Van Schaik.

Murray, T. R. (2003). Blending Qualitative and Quantitative Research Methods in Theses and
Dissertations. California: Corwin Press, Inc.

Naafs, M.A.B. (2018). Occupational Diseases in the Petrochemical Sector and Offshore Upstream
Petrochemical Industry. Progress in Petrochemical Science. Vol.2, Issue 2, pp. 1-6.

Navadiya, R. (2017). Practice of Permit to work in industries and its challenges. International journal
of scientific research. Vol.6, Issue 6.

Ness, A. (2015). Lessons Learned from recent process safety incidents. Centre for Chemical Process
Safety. American institute of chemical engineers, pp. 23-29.

Neuman, L. (2014). Social Research Methods: Qualitative and Quantitative Approaches. Pearson New
International Edition. Seventh Edition.

Nwachukwu, A.E. & Nabofa, O.E. (2014). Assessment of Petrochemical Industry work-related stress:
A case study of Port Harcourt City Nigeria. International Journal of Business and Finance
Management Research.

Okoye, P.U., Ezeokonkwo, J.U. & Ezeokoli, F.O. (2016). Building Construction Workers’ Health and
Safety Knowledge and Compliance on Site. Journal of Safety Engineering. Vol 5, Issue 1, pp.
17-26.

208
OSHAcademy, (2017). Introduction to Process Safety Management. OSHAcademy course 736 study
guide. Occupational Safety & Health Training.

Ostbye, T., Dement, J.M. & Krause, K.M. (2007). Obesity and workers’s Compensation. Results from
the duke Health and Safety Surveillance System. Vol. 167, pp. 766-773.

Othman, N.A., Jabar, J., Murad, M.A. & Kamarudin M.F. (2014). Factors influencing safety
management systems in petrochemical processing plants. Journal of technology management
and technopreneurship.Vol.2, pp. 27-39.

Ozgur, C., Dou, M., Li, Y. & Rogers, G. (2017). Selection of statistical software for data scientists and
and teachers. Journal of modern applied statistical methods. Vol.16, Issue 1, pp. 753-774.

Pallant, J. (2010). SPSS Survival Manual: A Step by Step guide to Data Analysis using SPSS. (4th ed.).
England: McGraw Hill.

Patino, C.M. & Ferreira, J.C. (2016). What is the importance of calculating sample size? J Bras
Pneumol. Vol. 42, No.2, pp. 162-162.

Peterson, M.J. (2009). Bhopal Plant Disaster – Situation Summary. International Dimensions of Ethics
Education in Science and Engineering Case Study Series.

Puttick, S. (2008). Avoidance of Ignition Sources as a basis of Safety- Limitations and

Challenges, IchemE, Symposium series no. 154, pp. 1-9.

Purohit, D.P., Siddiqui, N.A., Nandan, A. & Yadav, B.P. (2018). Hazard Identification and Risk
Assessment in Construction Industry. International Journal of Applied Engineering Research.
Vol. 13, No. 10, pp. 7639-7667.

Rahman, I. A., Memon, A. H., Azis, A. A. A. & Abdullah, N. H. (2013). Modelling causes of cost
overrun in large construction projects with partial least square-SEM approach: contractor's
perspective. Research Journal of Applied Sciences, Engineering and Technology, Vol. 5, Issue
6, pp. 1963-1972

Ramasamy, J. & Yusof, S.M. (2015). A literature review of subsea asset integrity framework for
project execution phase. Industrial Engineering and Service Science. Procedia Manufacturing.
Vol. 4, pp. 79 - 88.

209
Raykov, T. (2006). Interval Estimation of Individual Optimal Scores from Multiple-Component
Measuring Instruments. Structural Equation Modelling, Vol. 13, pp. 252-261.

Reddy, V. & Reddy I. (2015). Study of Electronic Work Permit System in Oil and Gas Industry-
Kuwait. International Journal of Innovative Science, Engineering & Technology. Vol. 2, Issue
4, pp. 533-537.

Reddy, K. & Yarrakula, K. (2016). Analysis of Accidents in Chemical Process Industries in the period
1998 – 2015. International Journal of ChemTech Research. Vol. 9, No.04, pp. 177-191.

Rigdon, E. E. (2016). Choosing PLS path modelling as analytical method in European management
research: A realist perspective. European Management Journal, Vol. 34, pp. 598-605.

Robinson, S. (2013). Conceptual Modelling for Simulation, Winter Simulation Conference, School of
Business and Economics.

Robinson, J. P., Shaver, P. R. & Wrightsman, L. S. (1991). Measures of personality and social
psychological attitudes. San Diego, CA: Academic Press.

Robson, C. (2002). Real World Research. 2nd ed. Oxford: Blackwell.

Rolfe, G. (2006). Validity, trustworthiness and rigour: Quality and the idea of qualitative
research. Journal of Advanced Nursing, Vol. 53, No.3, pp. 304-310.

Rousseeuw, P.J. & Hubert, M. (2011). Robust statistics for outlier detection. John Wiley and Sons.
Vol. 1, pp. 73-79.

Safety Association for Canada’s Upstream Oil and Gas Industry, (2012). Controlling Chemical
Hazards in the Oil and Gas Industry.

Sameer, A., Karim, F., Goodwin, R. & De Vries, D. (2015). A structural Equation Model of governing
factors influencing the implementation of T-Government. International Journal of Advanced
Computer Science and Applications. Vol.6, No.11, pp. 119-125.

Saunders, M., Lewis, P. & Thornhill, A. (2003). Research Methods for Business Students. 3rd ed. New
York: Pearson Prentice Hall.

210
Saunders, M., Lewis, P. & Thornhill, A. (2012). Research Methods for Business Students (6th
ed.). Edinburgh Gate, England: Pearson Education Limited.

Schick, J. T. & Vaughn, L. (2002). How to think about weird things: Critical thinking for a new age.
Boston: McGraw-Hill.

Schumacker, R.E. & Lomax, R.G. (1996). A beginner’s guide to structural equation modelling.
Hillsdale, NJ: Erlbaum.

Schumacker, R.E. & Lomax, R.G. (2010). A beginner’s guide to structural equation modelling.
Hillsdale, NJ: Erlbaum.

Schreiber, J.B., Stage, F.K., King, J., Nora, A. & Barlow, E.A. (2006). Reporting Structural
Equation Modelling and Confirmatory Factor Analysis Results: A Review. The journal of
educational research. Vol. 99, No. 6, pp. 323-337.

Scotland, J. (2012). Exploring the philosophical underpinnings of research: Relating Ontology

and Epistemology to the Methodology and Methods of the Scientific, Interpretive, and Critical
Research Paradigms. English Language Teaching, Vol. 5, No. 9, pp. 9-16.

Sekaran, U. & Bougie, R. (2010). Research Methods of Business: A Skill Building Approach (5th
ed.). West Sussex: John Wiley and Sons.

Sharma, A., Sharma, P., Sharma, A., Tyagi, R. & Dixit, A. (2017). Hazardous effects of Petrochemical
Industry: A Review. Recent advances in Petrochemical Science. Vol. 3, Issue 2, pp. 1-3.

Shafique, F. & Mahmood, K. (2010). Model Development as a research tool: An example of PAK-
NISEA. Library Philosophy and Practice (e-journal).

Shanmugapriya, S. & Subramanian, K. (2016). Developing a PLS path model to investigate the
factors influencing safety performance improvement in construction organizations. KSCE
Journal of Civil Engineering, Vol. 20, No.4, pp. 1138-1150.

Soley-Bori, M. (2013). Dealing with missing data: Key assumptions and methods for applied analysis.
Department of Health Policy and Management, Boston University. Technical Report No.4.

South African Bitumen Association, (2010). Health, Safety, and Environment Management System.

211
Spillane, J. & Oyedele, L. (2015). Strategies for Effective Management of Health and Safety in
Confined Site Construction. Australasian Journal of Construction Economics and Building.
Vol. 13, Issue 4, pp. 50-64.

Sterba, S.K. & MacCallum R.C. (2010). Variability in Parameter Estimates and Model Fit Across
Repeated Allocations of items to Parcels. Vol. 45, pp. 322-358.

Stojkovic, A. (2013). Occupational Safety in Hazardous Confined Space. Safety Engineering.

Tabachnick, B. G. & Fidell, L. S. (2013). Using Multivariate Statistics: Pearson Education.

Teddlie, C. & Tashakkori, A. (2009). Foundations of mixed methods research: Integrating

quantitative and qualitative approaches in the social and behavioural sciences. Thousand Oaks
CA: Sage Publications.

Temme, D., Kreis, H. & Hildebrandt, L. (2006). PLS path modelling—A software review. Berlin,
Germany: Humboldt University.

Thakur, D. (1993). Research Methodology in Social Sciences. New Delhi: Deep and Deep
Publications.

Thomas, B. J. & Babu, J. (2014). Design of Safety System and Management in Petrochemical
Industry. International Journal for Innovative Research in Science and Technology. Vol. 1,
Issue 4, pp. 22-28.

Todres, L. & Holloway, I. (2006). Phenomenological research. In K. Gerrish & A. Lacey

(Eds.), the research process in nursing, pp. 177-187.

Toepoel, V. & Schonlau, M. (2017). Dealing with nonresponse: Strategies to increase participation
and methods for postsurvey adjustments. Mathematical Population Studies. Vol. 24, No.2, pp.
79-83.

Trochim, W.M.K. (2006). Research methods knowledge base. Retrieved on April, 2018
from http://www.socialresearchmethods.net

212
Tzou, T.L., Hankinson, G., Edwards, D. & Chung, P. (2004). Evaluating safety information
management performance- a key to preventing disaster Bhopal and its effects on process. Safety
International Conference on the 20th anniversary of the Bhopal gas tragedy at Indian institute
of technology, Kanpur, India Dec 1-3.

Vandenplas, C., Joye, D., Staehli, M.E. & Pollien, A. (2015). Identifying pertinent variables for non-
response follow-up surveys: lessons learned from four cases in Switzerland. European Survey
Research Association. Vol. 9, No. 3, pp. 141-158.

Vannette, D.L. (2014). Questionnaire Design: Theory and Best Practices. Maximizing the reliability
and validity of survey data collection.

Vitharana, V.H.P., De Silva, G.H.M.J. & De Silva S. (2015). Health Hazards Risk and Safety Practices
in Construction Sites – A Review Study. The Institution of Engineers, Sri Lanka. Vol. XLVIII,
No. 3, pp. 35-44.

Wachter, J. & Yorio, P. (2014). Investigation Accident Investigation Characteristics and

Organisational Safety Performance. Journal of Safety, Health and Environment Research. Vol.
10, No. 2, pp. 169-177.

Waters, D. (2001). Quantitative methods for business. 3rd ed. London: Pearson Education.

Westland, J. C. (2010). Lower bounds on sample size in structural equation modelling. Electronic
Commerce Research and Applications, Vol. 9, pp. 476-487.

Wong, K. (2016). TECHNICAL NOTE: Mediation analysis, categorical moderation analysis, and
higher-order constructs modelling in Partial Least Squares Structural Equation Modelling
(PLS-SEM): A B2B example using Smart PLS. Marketing Bulletin, Vol. 26, pp. 1-22.

Williams, J. (2011). Research paradigm and philosophy.

Williamson, K. (2006). Research in Constructivist Frameworks Using Ethnographic

Techniques. Library Trends, Vol. 55, No.1, pp. 83-101.

Wilson, H.A. (1989). Organizational behavior and safety management in the construction industry.
Construction Management and Economics, Vol. 7, Issue 4, pp. 303-319.

213
Yilmaz, K. (2013). Comparison of quantitative and qualitative research traditions: Epistemological,
theoretical, and methodological differences. European Journal of Education. Vol. 48, No. 2, pp.
1-16.

Yin, R. K. (2003). Case Study Research. Design and Methods. Sage

Publications, Newbury Park, pp. 17- 46.

Yin, R. K. (2009). Doing case study research. 4th ed. Thousand Oaks, CA: Sage Publications.

Yong, A.G. & Pearce, S. (2013). A Beginner’s Guide to Factor Analysis: Focusing on Exploratory
Factor Analysis. Tutorials in Quantitative Methods for Psychology. Vol. 9, No. 2, pp. 79-94.

Youngshin, S., Young-Jung, S. & Doonam, O. (2015). Methodological Issues in Questionnaire

Design. Korean Society of Nursing Science. Vol. 45, No. 3, pp. 323-328.

Zait, A. & Bertea, P.E. (2011). Methods for testing discriminant validity. Management and
Marketing. Vol. 9, Issue 2, pp. 218 - 224.

Zikmund, W.G. (2003). Business research methods. 7th ed. Ohio: Thomson South-Western.

Zin, S.M. (2012). Employers’ Behavioural Safety Compliance Factors Toward Occupational Safety and
Health Improvement in the Construction Industry, Procedia. Social and Behavioural Sciences
Vol. 36, pp. 742-751.

Zohrabi, M. (2013). Mixed Methods Research: Instruments, Validity, Reliability and Reporting
Findings. Theory and Practice in Language Studies. Vol. 3, No. 2, pp. 254-262.

Zhang, W., Lou, I., Ung, K.W., Kong, Y. & Mok, K. (2013). Spatio-temporal variations of
phytoplankton structure and water quality in the eutrophic freshwater reservoir of Macau.
Desalination and water treatment.

Zhao, L.L. (2017). Building development cost drivers in the New Zealand construction industry: a
multilevel analysis of the causal relationships. School of Engineering and Advanced
Technology.

214
 APPENDICES

Appendix 1 (Ethical Clearance Approval)

_____________________________________________________________________
 Appendix 2 (Research Instrument)
_____________________________________________________________________

University of KwaZulu-Natal

COLLEGE OF AGRICULTURE, ENGINEERING AND SCIENCE

Dear, Sir/Madam

Researcher: James Xolani Nyawera (Cell No. 082 5577 802)

Emails: [email protected] and [email protected]

Supervisor: Prof Theo C Haupt (Phone No. 0312602712)

Emails: [email protected], [email protected]

My name is James Xolani Nyawera, a Ph.D. Candidate at the College of Agriculture, Engineering and
Science in the University of KwaZulu-Natal, Howard College. My contact details and my supervisor’s
details are declared above. You are being invited to consider participating in a research study that is entitled
“the study of process health and safety management deficiencies relative to hazardous chemical
exposure”.

The aim and purpose of this research is to review the process health and safety management systems within
industries where hazardous chemical substances are being used and identify the deficiencies. The study will
test the effectiveness of the existing process health and safety management systems in this industry and
prioritize the major deficiencies that need urgent senior management’s attention. The study is expected to
enroll about 150 participants. The research will be conducted in a systematic way in Petrochemical Industry
and will be limited to Durban employees in KwaZulu-Natal. Sample population will be selected randomly
and will focus on the private sector fulltime employees. The study will be used to answer the research
questions using this questionnaire as the research instrument. Research approach is quantitative
methodology, and a model will be developed and validated. The duration of your participation if you choose
to enroll and remain in the study is expected to be only to answer the research questionnaire (20 - 30
minutes). If you have any questions or concerns about completing the questionnaire or about participating in
this study, you may contact me or my supervisor at the numbers listed above. We hope you will take the time
to complete this survey.

There is absolutely no risk and or discomforts. There are no direct benefits to the participants/respondents,
however the petrochemical industry will benefit in understanding management deficiencies better and close
the gaps, employees will be far less exposed to process health and safety incidents. This research may
consider interviews incase potential respondents/participants are not comfortable in answering research
questionnaire.

ii
The final thesis will not be allowed to be on loan from the library for a period of 5 years. The thesis will be
store in the university for 5 years with access control. There will be no organization and individual names
mentioned in the research documents. The research questionnaire answers will be shredded after 5 years of
the final report approval. There will be no feedback to the research participants, unless it is requested, an
electronic copy will be sent to those that ask for it, after all approvals from authorities have been granted.

This study has been ethically reviewed and approved by the UKZN Humanities and Social Sciences
Research Ethics Committee (Approval Number). In the event of any problems or concerns/questions you
may contact the researcher at provided contact details above or the UKZN Humanities and Social Sciences
Research Ethics Committee, contact details are as follows:

HUMANITIES & SOCIAL SCIENCES RESEARCH ETHICS ADMINISTRATION

Research Office, Westville Campus
Govan Mbeki Building
Private Bag X 54001
Durban
4000
KwaZulu-Natal, SOUTH AFRICA
Tel: 27 31 2604557- Fax: 27 31 2604609
Email: [email protected]

Sincerely
James Xolani Nyawera
Signature__________________________________________Date_________________

CONSENT
I _________________________________________________________________ the undersigned have
read and understand the above information. I hereby consent to participate in the study outlined in this
document. I understand that participation is voluntary and that I may withdraw at any stage of the process. I
will not incur any penalties or loss of treatment. I will verbally notify the researcher in case I decide to
withdraw and that will be enough. I will incur no incentives or reimbursements as a result of participation in
the study. If I have any further questions/concerns or queries related to the study I understand that I may
contact the researcher: James Xolani Nyawera (Cell No. 082 5577 802 or [email protected]). If I have
any questions or concerns about my rights as a study participant or if I am concerned about an aspect of the
study or the researchers then I may contact: HUMANITIES & SOCIAL SCIENCES RESEARCH ETHICS
ADMINISTRATION in the address provided above.

Participant’s Signature____________________________________ Date________________

 Questionnaire

Respondent’s Particulars Questions

1. Gender.
Male Female

2. Age.

3. Marital Status.

4. Years of service in a Petrochemical Industry.

5. Department
Health, Safety Others (Specify Below)
Operations Maintenance Technical
and Environment

Please indicate on scale where 1 = Strongly Agree; 2 = Agree; 3 = Neutral; 4 = Disagree and 5 = Strongly
Disagree, the extent to which you agree with the following statements:

Statements 1 2 3 4 5

6. My organization has excellent chemical exposure management systems.

7. Senior Management prioritises Health and Safety in my organisation.

8. Senior Management has an open door policy on Health and Safety issues.

9. There are effective noise exposure management systems in my organization.

10. Senior Management communicates Health and Safety policy to all
employees.
11. Senior Management allocates enough time to address Health and Safety
concerns.
12. Most employees are aware of hazardous chemicals in their work
environment.
13. Employees undergo comprehensive training on health and safety in my
organization.
14. Most permanent employees know how to handle hazardous chemicals in
the work place.
15. Contractor’s on boarding appreciates all hazardous chemicals in my
organization.
16. Most contractors know how to handle hazardous chemicals in my
organization.
17. Senior management prioritizes mechanical/asset integrity of our processing
plant.

iv
Please indicate on scale where 1 = Strongly Agree; 2 = Agree; 3 = Neutral; 4 = Disagree and 5 = Strongly
Disagree, the extent to which you agree with the following statements:

Statements 1 2 3 4 5
18. The organization communicates effectively all lessons learned after the
occupational health and safety incidents.
19. The organization closes all corrective action items effectively after the
root causes analysis for all incidents happening onsite.
20. In my organization all engineering changes undergo a comprehensive
management of change.
21. The organization does comprehensive process hazard analysis before
engineering changes are made.
22. The organization has all process health and safety information available
to all employees.
23. Most of the health and safety incidents are due to poor engineering design
integrity.
24. In my organization we have a comprehensive pre-activity start up review
and pre-activity shutdown review.
25. Most of the health and safety incidents in petrochemical industry are due
to poor controls when working at heights.
26. In my organization all work activities have a detailed operating procedure
or work instruction.
27. Most of the health and safety incidents in petrochemical industry are due
to poor controls of source of ignition.
28. All the work activities in my organization are done after a valid permit to
work has been approved by the authorities.
29. In my organization before you start excavation or entering a trench you
need to obtain authorization.
30. All employees are aware that when you handling hazardous chemicals
you need to use prescribed personal protective equipment.
31. In my organization all safety critical equipment is disabled with
permission from the authorities.
32. Most of the health and safety incidents in the petrochemical industry are
due to not verifying energy isolation before you start working on equipment.
33. My organization diligently manages fatigue in both permanent employees
and contractors.
34. My organization has all management systems in place to manage
substance misuse.
35. Most of the health and safety incidents are due to human error in my
organization.
36. My organization has effective management systems to manage working
in confined space.
37. Most of the health and safety incidents are due to poor controls in place
when working with suspended loads.
38. Poor housekeeping in my organization is the cause for many health and
safety incidents.
39. Poor health and safety risk assessments are responsible for most of health
and safety incidents in the petrochemical industry.
40. Audit compliance is an excellent practice to prevent most of health and
safety incidents in the petrochemical industry.
 Appendix 3 (Publication 1)
_____________________________________________________________________

ASOCSA2018 - ASOCSA 12 – 46

Process Health and Safety Management Elements: An initial review of literature

Author: Mr. James Xolani Nyawera, BSc Eng (Chemical), Pr.Eng, MBA.

Ph.D. Candidate: University of KwaZulu-Natal (College of Agriculture, Engineering and Science)

Email: [email protected] and [email protected]

Office: +27 314603540, Mobile: +27 825577802

Co-Author: Prof. Theo C. Haupt, Ph.D., M.Phil.,

Research Professor: Engineering, Mangosuthu University of Technology

Emails: [email protected], [email protected] and [email protected]

Office: +27 31 819-9294 and +27 31 260-2712, Mobile: +27 82 686-3457

Orcid.org/0000-0002-2531-3789

vi
ABSTRACT AND KEYWORDS

Purpose: There are management systems to prevent health and safety incident from taking place in the
petrochemical industry but incidents still happen, people get injured, assets get damaged and environment is
polluted. The aim of the study was to assess the existing process health and safety management systems and
identify elements that require senior management’s vigilant attention. Design/Methodology/Approach: This
research used literature review and the methodology is qualitative. The research approach is inductive.
Findings/Observations: The ten key process health and safety management elements from literature to be
prioritized for an effective process health and safety management system are, namely Leadership
Commitment, Training and Competency, Chemical Exposure Management, Health and Safety Risk
Assessment, Process Hazard Analysis, Process Health and Safety Information, Operating Procedure, Control
of Ignition Source, Control of Confined Space Entry and Permit to Work. Research Limitations: Only
petrochemical industry was considered in this literature review. The next phase in this study will be to
determine the extent of the application of the key elements outlined in the literature within the petrochemical
industry in KZN province. Value of Paper: Paper presents ten key process health and safety management
elements to the petrochemical industry.

Key words: Process Health and Safety Management Elements.

1. INTRODUCTION

Process health and safety management systems are a proactive management and engineering approach to
control risk of failures and errors by reducing chance of human error and protecting employees, contractor
employees, environments and assets from the risks associated with hazardous chemicals, but incidents still
happen, people get injured, assets get damaged and environment is polluted. According to Cooper (1998) an
organisation that has a good health and safety culture consists of a strong senior management commitment,
leadership and involvement in health and safety, better communications between all organisational levels,
greater hazard control, good induction and follow up on health and safety training and an ongoing health and
safety schemes reinforcing the importance of health and safety, including near miss reporting. Material
substances, processes or circumstances which pose threat to health and well-being of workers in any
occupation are termed as occupational hazards. The occupational and hazard is major issue in oil and gas
extraction industry (Kulkarni, 2017). Occupational and process health and safety accidents can be reduced
through effective preventative measures by investing on health and safety equipment, training and educating
the employees, process design and machinery. In order to develop a good health and safety culture, attitude
of the workers needs to be reoriented by applying best practices, good housekeeping, change in work culture,
and work practices (Beriha et al., 2012). The most important indicator of a positive health and safety culture
is the extent to which employees are actively involved in health and safety on a daily basis (Cooper, 1998).

The present study was conducted by reviewing petrochemical industry health and safety literature with the
aim to assess the existing process health and safety management systems and identify the top ten elements
that require senior management’s further vigilant attention to reduce process health and safety incidents in
the petrochemical industry.

2. LITERATURE REVIEW

The petrochemical industry as the name implies is based upon the production of chemicals from petroleum
such as natural gas and crude oil. The occupational health and safety plays central role in industry as it
protects all workers from health and safety related issues in their working environment. It is vital to protect
workers from injuries on a social level, but there is also a positive economic impact in reducing health and
safety hazards. Health and safety is without doubt, the most crucial investment and the question is not what it
costs, but what it saves (Hughes and Ferret, 2007).

2.1 Leadership Elements

There are five leadership process health and safety elements namely leadership and commitment, training
and competency, contractor management, asset integrity and effective communication. It is recognised that
leadership is important in the creation of a culture that supports and promotes a strong health and safety
performance of an organisation. According to Fuller and Vassie (2005) having senior managers who take a
proactive interest in establishing a health and safety culture has been considered to be a key influence on
organisational health and safety performance. It is imperative that leadership ensures that each employee is
trained in an overview of the process and in the operating procedures, emphasis on the specific safety and
health hazards, emergency operations including shutdown, and safe work practices applicable to the
employee's job tasks. The other element that requires leadership’s attention is contractor management,
contractors working in petrochemical industry face a great risk during maintenance work. They are exposed
to a number of inevitable hazards as most workers employed by contractors are unfamiliar with hazardous
materials and more work performed under high pressure (Othman et al., 2014).

viii
It is the duty of the leadership team to ensure assets are maintained effectively and are safe to be operated at
all times. According to Hardy (2013) mechanical equipment must be maintained to ensure that it will
continue to operate correctly and safely. Process health and safety management standards require that an
organisation establish and implement written procedures for maintaining equipment such pressure vessels,
storage tanks, piping systems, pressure safety valves, emergency shutdown systems, and controls
(monitoring devices, sensors, alarms and interlocks). An effective quality assurance program must be
implemented to assure conformance to standards and codes, identify and record deficiencies and confirm that
deficiencies have been corrected (Hardy, 2013). The last element is effective communication from all levels
in an organisation. It is vital that information feedback is available to the right people at the right time and
that means an effective communication system needs to be in place (Cooper, 1998).

2.2 Health and Safety Elements

Process health and safety elements include chemical exposure management, health and safety risk
assessment, incident investigation, emergency response and audit compliance. The management of health
and safety at work requires employers to make a suitable and sufficient assessment of the health and safety
risks to employees and non-employees, arising from their work activities. It must be noted though that health
and safety incidents still happen in the work place and it is important to investigate those incidents and
appreciate root causes so that corrective actions can be assigned to employees that will close them. The
intention is to prevent the similar incidents from happening again. According to Bond (2007) the sole
objective of the investigation of an accident or incident under regulations shall be the prevention of accidents
and incidents. Such incidents include those that could be called a near miss and incident or problem
investigation should be factored back into the hazard analysis. According to Hardy (2013) organisations must
plan for an emergency and be prepared to respond. As an absolute minimum, employers must develop an
emergency action plan that includes evacuation and shelter in place instructions and training in the use of
personal protective equipment. Audit compliance is a proactive attempt to identify gaps to comply and
intervene when non-compliance is ascertained. Compliance audits provide a means for assuring that the
procedures and practices in process health and safety management systems are being followed and are
adequate (Hardy, 2013). The other potential topics to be considered under process health and safety when
dealing with hazardous chemicals are noise exposure management, substance misuse and fatigue
management.
 2.3 Technical Elements

Four technical elements are important, namely, management of change, process hazard analysis, process
health and safety information and human factor. To prevent or minimize process health and safety incidents
in a petrochemical industry process hazard analysis is conducted by a team with expertise in engineering and
process operations, including at least one employee who has experience and knowledge on the system. These
process hazard analyses require process health and safety information to be clear and understood by all team
members involved. According to Tzou et al. (2004) managing health and safety related information
inadequately has been cited as a significant factor to industrial accidents. Accidents can take place even
where process health and safety management systems exist and the probability of such occurrence increase if
documentation is deficient.

Health and safety issues cannot be tackled effectively without interference of employers with a particular
pattern of behaviours as important criteria needed to change employee’s behaviours (Zin, 2012). Employers
and employees with good safety behaviour are particularly playing a significant role in achievement of safety
compliance to occupational health and safety improvement in industry. According to Bond (2007) it is clear
that in the vast majority of accidents or near misses employees were acting with the best intentions and did
not expect a serious event to occur from their actions. It has to be accepted that to err is human but one can
learn from these events and share with others the lesson learnt. Human error is not the sole cause of failure,
but it is a symptom of a deeper trouble. Human error is the starting point of an investigation, not the end
point. To do something about error, one must look at the system in which people work and focus must extend
past “what occurred” to why do it occur (Bridger, 2015). Many accidents have resulted from small changes
that did not appear to have an effect on health and safety prior to the incident (Hardy, 2013). These changes
must be thoroughly evaluated to assure that health and safety in the process industry is maintained.

2.4 Operational Elements

Operational elements include pre-activity and shutdown reviews, operating procedure, control of ignition
source, control of confined space entry and permit to work. Often accidents occur in the transition between
operational phases, rather than when the process system is up and running in “steady state” mode. Start up
and shutting down new and existing process systems can be hazardous because changes to design or
operations may be made in real time to meet schedule temperatures and pressures, potentially introducing
new hazards (Hardy, 2013). The other high risk operational activity in the petrochemical industry is the
confined space entry. A confined space is an enclosed or partially enclosed area that is big enough for a

x
worker to enter. It is not designed for someone to work in regularly, but workers may need to enter the
confined space for tasks such as inspection, cleaning, maintenance, and repair. Confined spaces may contain
hazardous atmospheres, including insufficient oxygen, toxic air, and an explosive atmosphere (Stojkovic’,
2013).

In a petrochemical industry control of ignition sources is vital. Fire and explosion hazards prevention
involves dealing with the elements of the fire triangle. They fall into three categories, namely absence of
flammable atmospheres achieved by limited fuel quantities, control/avoidance of ignition sources and
absence of flammable atmospheres by limiting the quantities of oxidant present (Puttick, 2008). A permit to
work system is a formal written system used to control work that is potentially hazardous. According to Iliffe
et al. (1998), many incidents in the petrochemical industry workplace are associated with maintenance
works, which are typical controlled by permit to work. There are many other operational elements that may
be considered when dealing with hazardous chemicals, such as working at heights, energy isolation,
suspended loads, excavation and entering a trench, disabling safety critical equipment and use of correct
personal protective equipment when handling hazardous chemicals.

The hearts and minds paper published by Shell Global Solutions International symposium in 2016, suggests
that there is a health and safety culture ladder with five levels, namely pathological (level 1), reactive (level
2), calculative (level 3), proactive (level 4) and generative (level 5). Pathological culture level 1 is where
nobody cares to understand why accidents happen and how they can be prevented and at generative culture
level 5, health and safety is seen as a profit centre and it is how business is done. Reactive culture level 2
says safety is important, and people do a lot when there has been an accident while the proactive culture is
where employees work safely because they intrinsically motivated to do the right things naturally. Level 3 is
calculative where an organisation says there are management systems in place to manage all hazards.
Figure 2.1 Health and Safety Culture Ladder
Adapted from Holstvoogd et al., 2006.

The health and safety culture ladder in Figure 2.1 shows the various levels of cultural maturity, and the
change process required to achieve a lasting change in personal and organisational culture (Holstvoogd et al.,
2006).

3. RESEARCH DESIGN AND METHODOLOGY

According to Crotty (1998) methodology is the strategy or plan of action which lies behind the choice and
use of particular methods. Methodology is concerned with why, what, from where, when and how data is
collected and analysed. Methodology asks the question “how can the researcher go about finding out
whatever they believe can be known (Guba and Lincon, 1994)? Methods are the techniques and procedures
used to collect and analyse data. The data collected will either be qualitative or quantitative. All paradigms
can use both qualitative and quantitative data (Crotty, 1998). However, this paper has reviewed literature to
determine the important elements to be considered in an effective process health and safety management
system. The research methodology is qualitative and the research approach is inductive.

4. KEY OBSERVATIONS

The key ten process health and safety management elements from literature to be prioritized for an effective
process health and safety management system are, namely Leadership Commitment, Training and
Competency, Chemical Exposure Management, Health and Safety Risk Assessment, Process Hazard
Analysis, Process Health and Safety Information, Operating Procedure, Control of Ignition Source, Control
of Confined Space Entry and Permit to Work. It is not clear from literature how the organisations are
assessed for placement in the five levels in the health and safety culture ladder hence a standard check list
assessment tool will be a great contribution to health and safety culture.

5. CONCLUSION

Health and safety standards and regulations reasonably cannot cover all the possible cases for different types
of works in a variety of hazards in a petrochemical industry. It is the duty of health and safety experts to
share as much as they can on this subject of health and safety in the petrochemical industry that can expose

xii
employees and contractors to hazardous chemicals. The next phase in this study will be to determine the
extent of the application of the key elements outlined in the literature within the petrochemical industry in
KZN province.

6. REFERENCES

Beriha, G.S., Patnaik, B., Mahapatra, S.S., & Padhee S., 2012. Assessment of safety performance in Indian
industries using fuzzy approach, Elsevier, Vol. 39 No. 3, pp. 3311-3323.
Bond, J., 2007. A safety culture with justice: a way to improve safety performance, IchemE Symposium
series no. 153, pp. 1-6.
Bridger, R.S., 2015. Human Factors and Human behaviour in Safety Management and Accident
Investigation, Human Factors Department, October 16, Institute of Naval Medicine, pp.1-13.
Cooper, D., 1998. Improving Safety Culture, A practical Guide. Applied Behavioural Science.
Crotty, M., 1998. The foundations of social research. London: Sage Publications.
Eyayo, F., 2014. Evaluation of occupational health hazards among oil industry workers: A case study of
refinery workers. Journal of environmental science, toxicology and food technology. Vol 8 , issue
12, pp. 22-53.
Fuller, C.W., & Vassie, L.H., 2005. Health and Safety Management: Principles and Best Practice. London:
FT Prentice Hall.
Guba, E. G., & Lincoln, Y. S., 1994. Competing paradigms in qualitative research. Handbook of qualitative
research (pp. 105-117).
Hardy, T.L., 2013. Elements of Process Safety Management, Great Circle Analysis, Denver, CO, USA.
February 20, pp. 1-12.
Holstvoogd, R., Van der Graaf G., Bryden R., Zijlker, V., & Husdon, P., 2006. Hearts and Minds
Programmes the road map to improved HSE culture, Symposium Series No. 151, Shell Global
Solutions International, pp. 1-13.
Hughes, P., & Ferret, E., 2007. An Introduction to Health and Safety in Construction. The Handbook for the
NEBOSH (National Examination Board in Occupational Safety and Health).
Iliffe, R.E., Chung, P.W.H., & Kletz, T.A., 1998. More effective permit to work systems, IchemE
Symposium Series no. 144, pp. 181-194.
Kulkarni, S.J., 2017. Safety and Hazards in Petroleum Industries: Research, Studies and Surveys.
International Journal of Petroleum and Petrochemical Engineering, Vol. 3, Issue 1. pp. 31-34.
Othman, N.A., Jabar, J., Murad, M.A. & Kamarudin M.F., 2014. Factors influencing safety management
systems in petrochemical processing plants. Journal of technology management and
technopreneurship.Vol.2, Dec 2014. pp. 27-39.
Puttick, S., 2008. Avoidance of Ignition Sources as a basis of Safety- Limitations and Challenges, IchemE
Symposium Series no. 154, pp. 1-9.
Stojkovic’, A., 2013. Occupational Safety in Hazardous Confined Space, Safety Engineering, vol 3, No. 3,
pp. 137-144.
Tzou, T.L., Hankinson, G., Edwards, D., Chung, P., 2004. Evaluating safety information management
performance - a key to preventing disaster Bhopal and its effects on process. Safety International
Conference on the 20th anniversary of the Bhopal gas tragedy at indian institute of technology,
Kanpur, India Dec 1-3.
Zin, S.M., 2012. Employers’ Behavioral Safety Compliance Factors toward Occupational Safety and Health
Improvement in the Construction Industry, Procedia. Social and Behavioral Sciences 36: pp. 742-
751.

xiv
 Appendix 4 (Publication 2)
_____________________________________________________________________

ASOCSA2019 - ASOCSA 13 – 59

Process Health and Safety Management Deficiencies

James Xolani Nyawera1, Theo C. Haupt2

[email protected], [email protected]
1, 2
School of Engineering
University of KwaZulu-Natal
Durban, South Africa

orcid.org/0000-0001-7346-3194
orcid.org/0000-0002-2531-3789

ABSTRACT AND KEYWORDS

Purpose: The study aims to investigate existing process health and safety systems to identify deficiencies
that require senior management’s attention to improve process health and safety to a generative culture level.
Design/Methodology/Approach: A quantitative research methodology and deductive research approach was
used, and questionnaire survey to collect the data. The targeted population was 800 employees in one major
petrochemical enterprise in the KwaZulu-Natal province of South Africa. The study was conducted by
distributing 400 questionnaires manually to the randomly selected potential participants of which 259 were
returned duly completed and used. They were statistically analysed using descriptive statistics (frequency of
responses) in SPSS version 25. Ethical clearance to conduct the study was obtained from the University of
KwaZulu - Natal Humanities and Social Sciences Research Ethics Committee (HSS/1094/018D). Findings:
The results showed that the key process health and safety focus areas for an effective process health and
safety management system are, namely, Effective handling of hazardous chemicals, Poor engineering design
integrity, Poor controls when working at heights, Verifying energy isolation before start working on
equipment, Fatigue management for both permanent employees and contractors, Poor controls when working
with suspended loads, Poor housekeeping, Human error, Poor health and safety risk assessments, and Poor
controls of source of ignition. Research Limitations: The study was conducted in South Africa (KwaZulu-
Natal Province) and only the petrochemical industry was considered. Value of Paper: This study will assist
senior management with a framework to reduce process health and safety incidents in the petrochemical
industry. Response to the Conference Theme: Paper responds to Health and Safety.

Key words: Process Health and Safety Management Deficiencies.

 1. INTRODUCTION

Many studies have been conducted on health and safety, attempting to improve health and safety in
organisations. The workers in the oil and gas industry are exposed to many hazards, namely, physical
hazards, chemical hazards, ergonomic hazards, psychosocial hazards and radiological hazards (Kim, 2016).
The evolution of petroleum refining from simple distillation to today’s sophisticated processes has created a
need for health and safety management procedures and safe work practices (Kumar et al., 2017). The
International Labour Organisation (ILO) has estimated that 2 million workers die each year from work
related injury and illness. According to Ezejiofor (2014) difficulties are encountered in obtaining information
concerning occupational diseases and injuries in developing countries due to lack of comprehensive and
harmonious data collecting systems. In 2002, in Sub-Saharan Africa alone, the ILO estimated more than
257,000 total work-related fatalities, including about 50,000 injuries. Previous studies have shown that the
fatality rate in the petroleum industry is 2.5 times more than construction industry and 7 times more than
general industry (Kulkarni, 2017).

The study aims to investigate the deficiencies in the existing process health and safety management systems
that require senior management attention for the improvement of process health and safety systems in the
petroleum industry. The research objective was to review the process health and safety management systems
where hazardous chemical substances are being used and identify the major deficiencies.

2. COMPONENTS OF PROCESS HEALTH AND SAFETY MANAGEMENT SYSTEMS

There are several components that comprise the health and safety management systems used in the
petrochemical industry. These components include leadership commitment and chemical exposure
management, health and safety risk assessment and process hazard analysis, permit to work and operating
procedures, training and competency, process health and safety information, control of confined space entry
and ignition source. These are briefly discussed in the following sub-sections.

2.1 Leadership Commitment and Chemical Exposure Management.

According to Hardy (2013) it is recognised that leadership is important in the creation of a culture that
supports and promotes a strong health and safety performance of an organisation Chemical process hazards
at a chemical plant can give rise to accidents that affect both workers inside the plant and members of the
public who reside nearby (Chen, 2016). The hazardous effect of chemicals comes through three ways,

xvi
namely: fire, explosion and toxicity. The first essential step towards greater plant safety is being aware of
the potentially dangerous properties of the substances, i.e. whether they are flammable, explosive or toxic
(Almanssoor, 2008).

2.2 Health and Safety Risk Assessment and Process Hazard Analysis.

Risk assessment is the evaluation of hazards to determine their potential to cause an accident. According to
Albert et al., (2014) a critical component in safety risk management is to adequately identify hazards and
mitigate its associated risk using safety program elements. According to Dabup (2012) the risk assessment
provides a systematic approach for the identification, management and reduction of the risk to an acceptable
level. According to Hardy (2013) process hazard analysis is defined as a systematic approach for identifying,
evaluating and controlling the hazards of processes involving highly hazardous chemicals. A process hazard
analysis must be conducted by a team with expertise in engineering and process operations, including at least
one employee who has experience and knowledge on the system (Department of Labour [USA], 2016).

2.3 Permit to Work and Operating Procedures

The permit to work system has been widely used to ensure safety during maintenance and/or construction
activities in almost every major hazard industry worldwide (Reddy and Reddy, 2015). According to
Navadiya (2017) design of permit to work is very significant but most key thing is definition of roles and
responsibilities of involved employees in procedure part and preparing checklist which is to be covered in
synchronize way. The procedures should be formally reviewed and updated as necessary to assure that they
are consistent with existing processes. Training must accompany these operating procedures, with an
emphasis on what employees should do in case of emergency (Hardy, 2013).

2.4 Training and Competency

Improving organisational knowledge and knowledge management capabilities is an important means to

prevent chemical accidents and improve organisations safety level (Chen, 2016). According to Hardy (2013),
training provides employees with the knowledge and tools to fully understand the risks in working with
hazardous chemicals. Health and safety knowledge encompasses awareness of occupational health and safety
risks, including an evaluation of occupational health and safety programmes in an organisation (Okoye et al.,
2016).
 2.5 Process Health and Safety Information

According to Tzou et al., (2004) managing safety related information inadequately has been cited as a
significant factor to industrial accidents. Awareness on possible risk factors and knowledge on how to reduce
these risk factors among workers and contractors will enhance site safety (Vitharana et al., 2015).

2.6 Control of Confined Space Entry and Ignition Source

Confined spaces may contain hazardous atmospheres, including insufficient oxygen, toxic air, and an
explosive atmosphere (Stojkovic’, 2013). According to Karthika (2013) even though accidents can never be
eliminated completely, employers can prevent many of the injuries and fatalities that occur each year.
According to Puttick (2008) fire and explosion hazard assessment flammable and potentially flammable
atmospheres must be identified and compared with the potential ignition sources present.

3. RESEARCH DESIGN AND METHODOLOGY

A quantitative research methodology and the deductive research approach in the form of questionnaire
instrument was used to collect the data. According to Trochim (2006) quantitative research often translates
into the use of statistical analysis to make the connection between what is known and what can be learned
through research. The major advantage of this method is that it allows one to measure the responses of
number of participants to a limited set of questions, thereby facilitating comparison and statistical
aggregation of the data (Yilmaz, 2013). The deductive approach is concerned with developing a hypothesis
(or hypotheses) based on existing theory, and the designing a research strategy to test the hypothesis. The
study was conducted by distributing 400 questionnaires manually to the randomly selected potential
participants within a single petrochemical organization and therefore considered to be a convenience sample.
The targeted population was 800 employees. The employees to whom questionnaires were handed during
health and safety talks and production meetings were then requested to remain behind for an explanation.
The duly completed questionnaires were 259, and were analysed using descriptive statistics (frequency of
responses) in SPSS version 25. The response rate was computed to be 64.75%. The observed variables was
measure through a Likert scale where 1=Strongly Agree, 2=Agree, 3=Neutral, 4=Disagree and 5=Strongly
Disagree.

xviii
 4. FINDINGS

4.1 Demographics of Participants

Table 4.1 Age and Years of Service

Minimum Maximum Median

Age 22 66 38
Years of Service 1 46 11

From Table 4.1 it is evident that the median age of participants was 38 with a minimum age of 22 years and a
maximum age of 66 years. Further, the median number of years of service was 11 years with a minimum of 1
year and a maximum of 46 years. The participants were matured with considerable years of experience in the
petrochemical industry. This aspect increases the reliability of the responses received from the participants in
terms of their accuracy and completeness.

Table 4.2 Gender, Marital Status and Department

Gender
Percent
Male 80,6
Female 19,4
Total 100,0
Marital Status
Single 37,2
Married 61,6
Divorced 1,2
Total 100,0
Department
Health, Safety and Environment 8,2
Operations 50,6
Maintenance 24,1
Technical 12,5
Others 4,7
Total 100,0

Table 4.2 indicates that 80.6% of the participants were males, 61.6% were married and most of the
respondents were from the Operations Department (50.6%), followed by the Maintenance Department
(24.1%).The results show that the petrochemical industry in the case of the sample organization is still male
dominated. Operations and maintenance generally have more employees that are exposed to health and
safety risks in petrochemical industry.
 Table 4.3 Latent Variables and Observed Variables Frequency Table

(Str Agree + (Str Disagree +

Latent Variables and Observed Variables Neutral%
Agree)% Disagree)%
Leadership Commitment (LCH7) - Senior
Management prioritises health and safety in my 93.4 4.3 2.3
organisation.
LCH8 - Senior Management has an open door policy on
88.4 9.3 2.3
health and safety issues.
LCH10 - Senior Management communicates Health and
96.9 1.9 1.2
Safety policy to all employees.
LCH11 - Senior Management allocates enough time to
88.4 9.3 2.3
address Health and Safety concerns.
LCH17 - Senior Management prioritises
73.6 18.6 7.8
mechanical/asset integrity of our process plant
LCH38 - Poor housekeeping in my organisation is the
33.7 31.0 35.3
cause for many health and safety incidents.
LCH40 - Audit compliance is an excellent practice to
prevent most of health and safety incidents in the 87.3 9.3 3.4
petrochemical industry.
Chemical Exposure Management (CEMH6) - My
organisation has excellent chemical exposure 92.2 6.2 1.6
management systems.
CEMH12 - Most employees are aware of hazardous
86.5 9.6 3.9
chemicals in their work environment.
CEMH14 - Most permanent employees know how to
84.9 12.8 2.3
handle hazardous chemicals in the work place.
CEMH15 - Contractor’s on boarding appreciates all
62.4 32.1 5.5
hazardous chemicals in my organisation.
CEMH16 - Most contractors know how to handle
43.8 40.3 15.9
hazardous chemicals in my organisation.
CEMH30 - All employees are aware that when you
handling hazardous chemicals you need to use prescribed 95.8 2.7 1.5
personal protective equipment.
Health and Safety Risk Assessment (HSRAH9) - There
are effective noise exposure management systems in my 88.8 8.5 2.7
organisation.
HSRAH32 - Most of health and safety incidents in the
petrochemical industry are due to not verifying energy 40.5 27.8 31.7
isolation before you start working on equipment.
HSRAH33 - My organisation diligently manages fatigue
55.4 29.5 15.1
in both permanent employees and contractors.
HSRAH34 - My organisation has all management
88.0 8.5 3.5
systems in place to manage substance misuse.
HSRAH39 - Poor health and safety risk assessments are
responsible for most of health and safety incidents in the 46.1 22.1 31.8
petrochemical industry

xx
 (Str Agree + (Str Disagree +
Latent Variables and Observed Variables Neutral%
Agree)% Disagree)%
Process Hazard Analysis (PHAH20) - In my organisation
all engineering changes undergo a comprehensive 87.2 3.5 9.3
management of change.
PHAH21 - The organisation does comprehensive process
88.8 9.3 1.9
hazard analysis before engineering changes are made.
PHAH23 - Most of the health and safety incidents are due to
20.6 38.5 40.9
poor engineering design integrity.
PHAH24 - In my organisation we have a comprehensive pre-
93.8 4.6 1.6
activity start up review and pre-activity shutdown review.
Permit to Work (PTWH25) - Most of the health and safety
incidents in petrochemical industry are due to poor controls 24.9 36.6 38.5
when working at heights.
PTWH28 - All the work activities in my organisation are
done after a valid permit to work has been approved by the 95.7 2.7 1.6
authorities.
PTWH29 - In my organisation before you start excavation or
98.4 1.2 0.4
entering a trench you need to obtain authorisation.
PTWH31 - In my organisation all safety critical equipment is
90.3 6.6 3.1
disabled with permission from the authorities.
Training and Competency (TCH13) - Employees undergo
comprehensive training on health and safety in my 86.1 9.7 4.2
organisation.
TCH19 - The organisation closes all corrective action items
effectively after the root cause analysis for all incidents 80.7 15.4 3.9
happening onsite.
TCH35 - Most of the health and safety incidents are due to
55.0 32.2 12.8
human error in my organisation.
Process Health and Safety Information (PHSIH18) - The
organisation communicates effectively all lessons learned 88.0 6.6 5.4
after the occupational health and safety incidents
PHSIH22 - The organisation has all process health and safety
90.0 8.1 1.9
information available to all employees.
Control of Confined Space Entry (CCSEH36) - My
organisation has effective management systems to manage 93.0 4.3 2.7
working in confined space.
CCSEH37 - Most of the health and safety incidents are due to
28.7 29.8 41.5
poor controls in place when working with suspended loads.
Operating Procedure (OPH26) - In my organisation all
work activities have a detailed operating procedure or work 91.5 5.8 2.7
instruction.
Control of Ignition Source (CISH27) - Most of the health
and safety incidents in petrochemical industry are due to poor 34.1 31.4 34.5
controls of source of ignition.

The frequency of responses is shown in Table 4.3. These were used to determine the major deficiencies
existing in process health and safety management systems when dealing with hazardous chemicals. The
following formula was used to decide on main deficiencies (Strongly Agreed + Agreed < 60% , Neutral >
20% = Main Deficiencies). The cumulative percentages of both strongly agree plus agree and disagree plus
strongly disagree were used to simplify the assessment of observed variables.

The key process health and safety management focus areas from Table 4.3 that have to be prioritized to
minimise health and safety incidents are, namely, CEMH16 - Effective handling of hazardous chemicals
(43.8%, 40.3), PHAH23 -Poor engineering design integrity (20.6%, 38.5%), PTWH25 - Poor controls when
working at heights (24.9%, 36.6%), HSRAH32 - Verifying energy isolation before start working on
equipment (40.5%, 27.8%), HSRAH33 - Fatigue management for both permanent employees and contractors
(55.4%, 29.5%), CCSEH37 - Poor controls when working with suspended loads (28.7%, 29.8%),. LCH38 -
Poor housekeeping (33.7%, 31.0%), TCH35 - Human error (55.0%, 32.2%), HSRAH39 - Poor health and
safety risk assessments (46.1%, 22.1%), and CISH27 - Poor controls of source of ignition (34.1%, 31.4%).
Participants in this study reported low levels of satisfaction with these observed variables of health and
safety in the petrochemical industry.

According Zohrabi (2013) one of the main requirements of any research process is the reliability of the data
and findings. The reliability and validity of the data are vital for a research study. The Cronbach’s Alpha is a
high-quality test widely used for reliability testing and an essential test for evaluating a questionnaire
instrument. Cronbach’s Alpha test is a widely-used method to test the internal consistency of measurement
indicators, based on the correlations between indicators.

Table 4.4 Reliability Test for Constructs

Latent Construct No of Observed Variables Cronbach's Alpha
7 (LCH7, LCH8, LCH10, LCH11, LCH17,
Leadership Commitment (LC) 0.819
LCH38, LCH40)
6 (CEMH6, CEMH12, CEMH14,
Chemical Exposure Management (CEM) 0.797
CEMH15, CEMH16, CEMH30)
5 (HSRAH9, HSRAH32, HSRAH33,
Health and Safety Risk Assessment (HSRA) 0.435
HSRAH34, HSRAH39)
4 (PHAH20, PHAH21, PHAH23,
Process Hazard Analysis (PHA) 0.776
PHAH24)
4 (PTWH25, PTWH28, PTWH29,
Permit to Work (PTW) 0.718
PTWH31)
Training and Competency (TC) 3 (TCH13, TCH19, TCH35) 0.481
Process Health and Safety Information (PHSI) 2 (PHSIH18, PHSIH22) 0.581
Control of Confined Space Entry (CCSE) 2 (CCSEH36, CCSEH37) 0.105
Operating Procedure (OP) 1 (OPH26) N/A
Control of Ignition Source (CIS) 1 (CISH27) N/A

Table 4.4 outlines the ten latent constructs with observed variables. Reliability test shows four latent
constructs with acceptable Cronbach’s alpha > 0.7 and four latent constructs with Cronbach’s alpha < 0.7
and two latent constructs that were not computed due to one observed variable.

4.2 Principal Component Analysis

Confirmatory factor analysis was carried out to test whether the measure of latent construct correspond with
the study of the nature of the individual factor. At this stage, the indicator elimination and model re-

xxii
specification are performed for each latent construct. A common acceptable threshold value for a good
indicator is having a loading higher than 0.5 (Rahman et al., 2013).

Table 4.5 Principal Component Analysis

Rotated Component Matrixa
Latent Variables
Observed Variables
PTW PHA LC HSRA CEM
Permit to Work (PTWH29) 0,779
Permit to Work (PTWH28) 0,687
Permit to Work (PTWH31) 0,661
Permit to Work (PTWH36) 0,649
Process Hazard Analysis (PHAH24) 0,630
Process Hazard Analysis (PHAH22) 0,587
Process Hazard Analysis (PHAH20) 0,564
Process Hazard Analysis (PHAH21) 0,504
Process Hazard Analysis (PHAH26) 0,586
Leadership Commitment (LCH8) 0,726
Leadership Commitment (LC10) 0,725
Leadership Commitment (LCH11) 0,698
Leadership Commitment (LCH7) 0,660
Leadership Commitment (LCH13) 0,547
Leadership Commitment (LCH9) 0,524
Leadership Commitment (LCH17) 0,519
Leadership Commitment (LCH6) 0,604
Leadership Commitment (LCH40) 0,587
Health and Safety Risk Assessment 0,833
(HSRAH37)
Health and Safety Risk Assessment 0,788
(HSRAH32)
Health and Safety Risk Assessment 0,785
(HSRAH25)
Health and Safety Risk Assessment 0,739
(HSRAH27)
Health and Safety Risk Assessment 0,690
(HSRAH23)
Health and Safety Risk Assessment 0,598
(HSRAH38)
Health and Safety Risk Assessment 0,573
(HSRAH39)
Health and Safety Risk Assessment 0,637
(HSRAH33)
Health and Safety Risk Assessment 0,769
(HSRAH35)
Chemical Exposure Management (CEMH30) 0,735
Chemical Exposure Management (CEMH16) 0,763
Chemical Exposure Management (CEMH15) 0,731
Chemical Exposure Management (CEMH14) 0,695
Chemical Exposure Management (CEMH12) 0,654
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.a
a. Rotation converged in 6 iterations.
Structural equation modelling combines multiple regression analysis and factor analysis together to analyse
the relationship between measured variables and latent constructs or factors (Raykov, 2006). It provides a
quantitative method to test a hypothesised model (Byrne, 2016). Structural equation modelling can be
employed to capture complex relationships between one or more dependent variables that can be sourced
from qualitative or quantitative data (Hox and Kleiboer, 2007).

According to Schreiber et al., (2006) Confirmatory Factor Analysis and Structural Equation Modelling are
statistical techniques that one can use to reduce the number of observed variables into a smaller number of
latent variables by examining the covariation among the observed variables. In this research paper, there
were three observed variables (OV) that were eliminated since the loading was less than 0.5 and they are
namely, PHSIH18 - The organisation communicates effectively all lessons learned after the occupational
health and safety incidents. TCH19 - The organisation closes all corrective action items effectively after the
root cause analysis for all incidents happening onsite). HSRAH34 - My organisation has all management
systems in place to manage substance misuse. Other observed variables were allocated to other latent
variables after principal component analysis as indicated in Table 4.5.

Table 4.6 Reliability Test after Principal Component Analysis for Constructs
Latent Construct No of Observed Variables Cronbach's Alpha
9 (LCH6, LCH7, LCH8, LCH9, LCH10, LCH11,
Leadership Commitment (LC) 0.867
LCH13, LCH17, LCH40)
Chemical Exposure Management 5 (CEMH12, CEMH14, CEMH15, CEMH16,
0.781
(CEM) CEMH30)
9 (HSRAH23, HSRAH25, HSRAH27,
Health and Safety Risk Assessment
HSRAH32, HSRAH33, HSRAH35, HSRAH37, 0.829
(HSRA)
HSRAH38, HSRAH39)
5 (PHAH20, PHAH21, PHAH22, PHAH24,
Process Hazard Analysis (PHA) 0.810
PHAH26)
Permit to Work (PTW) 4 (PTWH28, PTWH29, PTWH31, PTWH36) 0.749

The five latent constructs for process health and safety culture in this study are, namely, Leadership
Commitment (0.867), Chemical Exposure Management (0.781), Health and Safety Risk Assessment (0.829),
Process Hazard Analysis (0.810) and Permit to Work (0.749), they all have Cronbach’s Alpha > 0.7, and that
confirms reliability of the data.

5. CONCLUSION AND RECOMMENDATIONS

The study investigated existing process health and safety systems by surveying a sample of employees in a
petrochemical business enterprise in KZN province to identify deficiencies that required senior
management’s attention to improve process health and safety to a generative culture. It can be concluded that

xxiv
senior management has to increase attentiveness to handling of hazardous chemicals, engineering design
integrity, controls when working at heights, verification of energy isolation before start working on
equipment, fatigue management, controls when working with suspended loads, housekeeping, human error,
health and safety risk assessments and controls of ignition sources to improve health and safety culture. It is
recommended that industry develop process health and safety elements from the focus areas and to enforce
compliance and intervention timely when there is non-compliance. It is recommended that the next phase in
this study be to determine the key drivers to generative health and safety culture within the petrochemical
industry.

6. REFERENCES

Albert, A., Hallowell, M.R., & Kleiner, B.M., 2014. Emerging Strategies for Construction Safety & Health
Hazard Recognition. Journal of Safety, Health and Environmental Research. Vol. 10, No.2, pp. 152 -
161.
Almanssoor, A.A., 2008. Planning of Petrochemical Industry under Environmental Risk and Safety
Considerations. University of Waterloo.
Byrne, B.M., 2012. Structural Equation Modelling with Mplus. Basic Concepts, Applications, and
Programming. Multivariate Applications Series.
Chen, M., 2016. Process Safety Knowledge Management in the Chemical Process Industry. American
Journal of Chemical Engineering. Vol. 4, No.5, pp. 131-138.
Dabup, N.L., 2012. Health, Safety and Environmental Implications in Nigeria’s Oil and Gas Industry. Nelson
Mandela Metropolitan University.
Department of Labour [USA], 2016. Process Safety Management Guidelines for Small Business
Compliance. Government of United States of America.
Ezejiofor, T.I.D., 2014. Risk Assessment: Re:appraisals for Potential Hazards in the Operational
Environment and Facilities of Petroleum Refining and Distribution Industry in Nigeria – Research
and Review. Occupational Medicine and Health Affairs. Vol. 2, Issue 4, pp. 1-19.
Hardy, T.L., 2013. Elements of Process Safety Management, Great Circle Analysis, Denver, CO, USA.
February 20, pp. 1-12.
Hox, J.J. & Boeije, H.R., 2005. Data Collection, Primary vs. Secondary. Encyclopedia of social
Measurement. Vol. 5, pp. 593-599.
Karthika, S., 2013. Accident Prevention by using Hazop Study and Work Permit System in Boiler.
International Journal of Advanced Engineering Research and Studies. Vol. 2, Issue 2, pp. 125-129.
Kim, I., 2016. Ergonomic involvement for occupational safety and health improvements in the oil and gas
industry, Journal of Ergonomics, Vol. 6, Issue 3, pp. 1-3.
Kulkarni, S.J., 2017. Safety and Hazards in Petroleum Industries: Research, Studies and Surveys.
International Journal of Petroleum and Petrochemical Engineering, Vol. 3, Issue 1, pp. 31-34.
Kumar, R.M., Karthick, R,B., Bhuvaneswari, V. & Nandhini, N., 2017. Study on occupational health and
diseases in oil industry, International research journal of engineering and technology, Vol. 4, Issue
12, pp. 954-958.
Navadiya, R., 2017. Practice of Permit to work in industries and its challenges. International journal of
scientific research. Vol.6, Issue6.
Okoye, P.U., Ezeokonkwo, J.U. & Ezeokoli, F.O., 2016. Building Construction Workers’ Health and Safety
Knowledge and Compliance on Site. Journal of Safety Engineering. Vol 5, Issue 1, pp. 17-26.
Puttick, S., 2008. Avoidance of Ignition Sources as a basis of Safety- Limitations and Challenges, IchemE,
Symposium series no. 154, pp. 1-9.
Rahman, I. A., Memon, A. H., Azis, A. A. A. and Abdullah, N. H., 2013. Modeling causes of cost overrun in
large construction projects with partial least square-SEM approach: contractor's perspective.
Research Journal of Applied Sciences, Engineering and Technology, 5(6), 1963-1972.
Raykov, T. (2006). Interval Estimation of Individual Optimal Scores from Multiple-Component Measuring
Instruments. Structural Equation Modeling, Vol. 13, pp. 252-261.
Reddy, V. & Reddy I., 2015. Study of Electronic Work Permit System in Oil and Gas Industry- Kuwait.
International Journal of Innovative Science, Engineering & Technology. Vol. 2, Issue 4, pp. 533-
537.
Schreiber, J.B., Stage, F.K., King, J., Nora, A. & Barlow, E.A., 2006. Reporting Structural Equation
Modelling and Confirmatory Factor Analysis Results: A Review. The journal of educational
research. Vol. 99, No. 6, pp. 323-337.
Stojkovic, A., 2013. Occupational Safety in Hazardous Confined Space. Safety Engineering.
Trochim, W.M.K., 2006. Research methods knowledge base. Retrieved on April, 2018 from
http://www.socialresearchmethods.net
Tzou, T.L., Hankinson, G., Edwards, D. & Chung, P., 2004. Evaluating safety information management
performance- a key to preventing disaster Bhopal and its effects on process. Safety International
Conference on the 20th anniversary of the Bhopal gas tragedy at indian institute of technology,
Kanpur, India Dec 1-3.
Vitharana, V.H.P., De Silva, G.H.M.J. & De Silva S., 2015. Health Hazards Risk and Safety Practices in
Construction Sites – A Review Study. The Institution of Engineers, Sri Lanka. Vol. XLVIII, No. 3,
pp. 35-44.
Yilmaz, K., 2013. Comparison of quantitative and qualitative research traditions: Epistemological,
theoretical, and methodological differences. European Journal of Education. Vol. 48, No. 2, pp. 1-
16.
Zohrabi, M., 2013. Mixed Methods Research: Instruments, Validity, Reliability and Reporting Findings.
Theory and Practice in Language Studies. Vol. 3, No. 2, pp. 254-262.

xxvi