Exploring and Assessing Project Complexity

Bac Dao, Ph.D., A.M.ASCE 1; Sharareh Kermanshachi, Ph.D., A.M.ASCE 2;

Jennifer Shane, Ph.D., M.ASCE 3; Stuart Anderson, Ph.D., M.ASCE 4; and Eric Hare 5

Abstract: The term project complexity is not well understood in the construction industry by either scholars or practitioners. Project com-
plexity, however, is a critical factor in project management that presents additional challenges to achieving project objectives. Therefore, it is
essential that everyone involved in project management thoroughly understand what project complexity is and how to identify, measure, and
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

manage the various attributes and indicators it comprises. This study provided a constructive approach to identify and assess project complex-
ity. As part of the study, a working definition of project complexity was developed by the research team, and project complexity was described
in terms of project management rather than project physical features. In addition, the research team identified complexity attributes and
developed the indicators that measure those attributes. Project specific data were collected through a survey of companies active in the
construction industry. These data were statistically analyzed to test the significance of complexity indicators in differentiating low-complexity
projects from high-complexity projects. The statistical analysis resulted in 34 significant complexity indicators associated with 22 attributes,
which the research team considers truly representative of project complexity. The research findings can help scholars and practitioners in the
project management field understand critical indicators of project complexity and develop an appropriate strategy to manage project complex-
ity effectively. DOI: 10.1061/(ASCE)CO.1943-7862.0001275. © 2016 American Society of Civil Engineers.
Author keywords: Organizational issues; Project complexity; Complexity attributes; Complexity indicators.

Introduction However, although these two concepts are related to project com-
plexity, they were not the subject of this study.
In the field of project management, both scholars and practitioners There is a need to study complexity as a separate factor influ-
have referred to problems caused by project complexity or particu- encing projects. This includes a need to define project complexity,
lar problems arising from complex projects. References to low study the individual and most important attributes of complexity,
project complexity or high project complexity are commonplace and identify the indicators that truly reflect the complexity of a
across many industry sectors. However, references to low or high project. Most attributes of complexity are constantly changing var-
complexity often seem to be made intuitively and to represent a iables such as project type, project size, project location, project
relative assessment of complexity by comparison to other types team experience, interfaces within a project, logistics/market con-
of projects or to similar projects within an industry sector. There ditions, geo-political and social issues, and permitting and appro-
is no standard definition for complexity that can be applied to vals. Having a better understanding of project complexity in any
all projects. Furthermore, there is no single understanding of com- phase of project development as well as a strategy to manage com-
plexity, what it means, and how to measure it. Project complexity is plexity will influence how efficiently and economically projects are
frequently perceived as a factor related to two project concepts: planned, managed, and executed.
project difficulty and project risk. Project difficulty focuses on The overall goal of the study presented in this paper was to de-
how hard it is to achieve project objectives. Project risk focuses velop a methodology to fully explore and assess project complex-
on uncertainties associated with unknowns around the project. ity. This goal was achieved by attaining two specific objectives:
(1) define complexity and its attributes and (2) identify and test
1
Postdoctoral Research Associate, Durham School of Architectural
the significance of complexity indicators that are the measures
Engineering and Construction, Univ. of Nebraska-Lincoln, 133 Nebraska of project complexity. The resulting assessment of project complex-
Hall, 900 N 16th St., Lincoln, NE 68588 (corresponding author). E-mail: ity adds significant value to the current body of knowledge for
[email protected] scholars and helps practitioners with project resource allocation
2
Assistant Professor, Dept. of Civil Engineering, Univ. of Texas at to complex projects. The research results, from the perspective
Arlington, 425 Nedderman Hall, 416 Yates St., Box 19308, Arlington, of complexity theory and complexity management, make a substan-
TX 76019. E-mail: [email protected] tial contribution to the theoretical basis in the field of project
3
Associate Professor, Dept. of Civil, Construction, and Environmental management.
Engineering, Iowa State Univ., 498 Town Engineering, Ames, IA 50011.
E-mail: [email protected]
4
Professor, Dept. of Civil Engineering, Zachry Chair in Construction Literature Review
Integration, Texas A&M Univ., 3136 TAMU College Station, TX
77843-3136. E-mail: [email protected] To better understand complexity, the researchers reviewed the pub-
5
Ph.D. Candidate, Dept. of Statistics, Iowa State Univ., 1121 Snedecor
lications relevant to complexity theory, project complexity defini-
Hall, Ames, IA 50011. E-mail: [email protected]
Note. This manuscript was submitted on March 4, 2016; approved on tion and identification, impacts of project complexity on project
September 21, 2016; published online on November 16, 2016. Discussion performance, and complexity management practices. Additionally,
period open until April 16, 2017; separate discussions must be submitted relevant models and tools already developed were investigated for
for individual papers. This paper is part of the Journal of Construction insights into different approaches to complexity assessment and
Engineering and Management, © ASCE, ISSN 0733-9364. management.

© ASCE 04016126-1 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126

 Complexity Theory and Management Practice sometimes called a phase transition, or the edge of chaos, in which
they exhibit a sort of bounded instability, that is, the unpredict-
Complexity Theory ability of specific behavior within a predictable general structure
Complexity theory generally defines what a complex system is of behavior.
within a specific area of interest (e.g., natural science, biology,
eco-system, computer science, human society, financial market, Definition of Project Complexity
etc.) and studies the interaction between the elements in that sys-
tem. The existing theoretical issue of complexity theory is that there Through the literature review, more than 30 definitions of complex-
is still no commonly accepted definition of complexity, despite ity were assessed. These 30 definitions went through a screening
there being a large number proposed (Chu et al. 2003). As defined process. This screening process eliminated the definitions that were
by Valle (2000) and Lucas (2000), a complex system is a whole that not related to project management and similar definitions from dif-
consists of several elements interacting with each other in many ferent studies (the most cited definition was selected). In addition,
different ways. Numerous interdependent elements in a complex the definitions that consist of uncertainty elements were also elim-
system continuously interact and spontaneously organize and reor- inated. The literature often uses uncertainty to describe complexity.
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

ganize themselves into increasingly elaborate structures over time. The use of uncertainty has purposely been avoided in this research
Unlike conventional systems (e.g., an aircraft or a computer), a because uncertainty is more often associated with risks rather than
complex system includes elements that do not necessarily have complexity. In this section, 10 definitions of complexity in general
fixed relationships, fixed behaviors, or fixed quantities. Lucas and project complexity in particular are introduced. These defini-
(2000) categorized systems into four levels of complexity: (1) static tions originated from a wide variety of disciplines, and some of
complexity that is related to fixed systems and does not change with them are radically different from the others. The result of the
time; (2) dynamic complexity that is considered to be cyclically screening process for complexity definitions is shown in Table 1.
changed spatially and temporally; (3) evolving complexity refers
to the systems that are changed in spatial and temporal dimensions,
Complexity Attribute
and these systems evolve or alter through time into different com-
plex systems; and (4) self-organizing complexity is the combina- The next step in better understanding complexity is to identify the
tion of the internal constraints of closed systems (like machines) attributes of complexity. Scholars have focused on the identifica-
with the creative evolution of open systems (like people). In “The- tion of complexity attributes more than any other topic in the field
ories of Complexity” (Chu et al. 2003), Chu has considered two of project complexity. Studies in this area have evolved signifi-
fundamental properties generating complexity for a system: radical cantly over the past 20 years. Nassar and Hegab (2006) developed
openness and contextuality. a complexity measure for project schedules. The measure considers
With endeavors of quantifying or modeling complexity theory, the degree of interrelationships between the activities in the proj-
scientists have attempted to apply quantitative techniques to ect’s schedule. The complex schedule of a project may contribute
existing systems or organizations. However, most attempts to to project complexity. However, the complexity measure of a
quantify complexity deal with either the parts (traditional reduc- project schedule is totally different from the complexity measure
tion) or look to simplify the system to a single or few parameters. of a project. Baccarini (1996) identified two major attributes of
Complexity could be simplified by making its features and proper- complexity including organizational complexity and technical
ties reducible, and the modelers can ignore some features without complexity. Organizational complexity reflects the view that a
substantially compromising the validity of the models. However, project is a task containing many interdependent elements. Because
readers and/or users have to be aware of inherent limitations of this dimension of complexity is related to the structure of the
these models and acknowledge that they cannot represent the full project, Williams (1999) refers to this factor as structural complex-
complexity of the system. ity. Technical complexity deals with complexity related to the
transformation processes, which convert inputs (such as project
Complexity Theory and Management Practice resources) into outputs (such as project performance or facility
In the management area, under certain conditions, the systems of quality). Technical complexity is usually divided into facility op-
interest to complexity theory perform in steady, predictable ways. erational requirements, project characteristics, and the level of
Under other conditions, they exhibit behavior in which regularity knowledge required for the project. Examples of technical com-
and predictability is lost. Almost undetectable differences in initial plexity are the sophistication of control system, number of opera-
conditions lead to gradually diverging system reactions until even- tors, location of project, type of work force skills needed, and right
tually the evolution of behavior is quite dissimilar. The systems of type of technical expertise. From an analysis of more than 1,300
interest are dynamic systems with the capability of changing over projects, Puddicombe (2012) proved that technical complexity is
time. Some systems constantly change but do so in a relatively an important characteristic of a project that has distinct effects
regular manner, whereas other systems lack this stability. Unstable on project performance.
systems move further and further away from their starting condi- Though uncertainty is associated with risk rather than complex-
tions until or unless these systems are brought up short by some ity, several scholars still viewed uncertainty as an attribute to
over-riding constraint (Rosenhead 1998). Stable and unstable project complexity. Lebcir (M. Lebcir, “A framework for project
behaviors are part of the traditional range of physical science. In complexity in new product development (NPD) projects,” working
the management field, Stacey (1992) indicates that a system behav- paper, University of Hertfordshire, Hertfordshire, U.K.) introduced
ior may also be divided into two zones: (1) a stable zone where, if uncertainty as one of the factors influencing the complexity of the
the system is disturbed, it returns to its initial state; and (2) an un- project. It has two important dimensions: uncertainty in goals that
stable zone, where a small disturbance leads to movement away means the project is poorly defined at the beginning of its execution
from the starting point, which in turn generates further divergence. and uncertainty in methods that reflects the lack of knowledge on
Which behavior type is exhibited, stable or unstable, depends on how to achieve project goals. Generally, the number of project com-
the organizational conditions. Under some appropriate conditions, ponents, degree of activeness within each component, degree of
systems may operate at the boundary between these zones, interactions between project components (or interfaces within a

© ASCE 04016126-2 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126

 Table 1. Project Complexity Definitions
Number Authors (year) Definition
1 Dictionaries (1) Consisting of many varied interrelated parts; (2) complicated, involved, intricate.
2 Perrow (1965) The complexity of a task is the degree of difficulty and the amount of thinking
time and knowledge required to perform the task.
3 Gidado (1996) Project complexity is the measure of the difficulty of implementing a planned
workflow in relation to the project objectives.
4 Baccarini (1996) Project complexity consists of many varied interrelated parts and can be operationalized
in terms of differentiation and interdependency.
5 Edmonds (1999) Complexity is that property of a model, which makes it difficult to formulate
its overall behavior.
6 Sbragia (2000) The number of elements in the project, intensity of interactions between elements,
and difficulty of cooperation between the functional areas.
7 Brockmann and Girmscheid (2007) The complexity is the degree of manifoldness, interrelatedness, and consequential
impact of a decision field.
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

8 Hass (2008) Complexity is characterized by a complicated or involved arrangement of many

inter-connected elements that it is hard to understand or deal with.
9 Vidal and Marle (2008) Project complexity is the property of a project, which makes it difficult to understand,
foresee, and keep under control its overall behavior.
10 Remington et al. (2009) A complex project demonstrates a number of characteristics to a degree, or level of
severity, that makes it difficult to predict project outcomes or manage project.

project), and interactions of the project with entities outside of the The total CIFTER score is used to categorize each project as either
project were frequently considered as the attributes that create below Global Level 1 (scores less than 12), Global Level 1 (scores
complexity of a project. 12 to 18), or Global Level 2 (score 19 or more). Table 2 shows
influencing factors as well as the method to evaluate them based
Assessing and Measuring Project Complexity on the CIFTER tool. Each of the seven factors in the CIFTER tool
is rated on a point scale of one to four with the total number of
Previous studies explained in this literature review show that for all points across the seven factors determining whether a project is
practical purposes, a project risk framework is not sufficient to Global 1, Global 2, or neither.
identify and measure all the possible positive and negative effects
associated with risk, uncertainty, or complexities related to the
Literature Findings
project. Notably, there is a crucial need for efficient complexity
modeling in order to identify and assess project complexity factors. Several definitions of complexity were found from a wide range of
By conducting a literature search and structured interviewing of disciplines as specifically discussed in each section of the literature
practitioners, Gidado (1996) has defined project complexity and review. For the purposes of this research, the myriad definitions
identified the factors that influence its effect on project success. were consolidated to the following central idea for further
These factors include the inherent complexity, number of technol- discussion: Project complexity is the degree of differentiation of
ogies, rigidity of sequence, overlap of phases, and organizational project elements, interrelatedness between project elements, and
complexity. In addition, the study proposes an approach that mea- consequential impact on project decisions [CII RT 305-11 (CII
sures the complexity of the production process in construction. 2016a)]. Also, factors of complexity suggested by scholars were
The Global Alliance for Project Performance Standards discussed. Although a fair number of papers and books were found
(GAPPS 2007) developed a comprehensive project management around different methods of measuring complexity, it seems that
complexity measurement tool called the Crawford–Ishikura Factor very few scholars have studied project complexity as a separate
Table for Evaluating Roles (CIFTER). CIFTER provides a seven- factor influencing project characteristics in the project management
factor model from which project management complexity of field. A detailed description of project complexity and complexity
projects is assessed. A total project complexity score is created attributes will help scholars understand project complexity and
by adding the scores from all seven factors outlined in CIFTER. study it properly.

Table 2. Crawford–Ishikura Factor Table for Evaluating Roles

Number Project management complexity factor Description and scores
1 Stability of the overall project context Very high (1) High (2) Moderate (3) Low (4)
2 Number of distinct disciplines, methods, or approaches involved in Low (1) Moderate (2) High (3) Very high (4)
performing the project
3 Magnitude of legal, social, or environmental implications from Low (1) Moderate (2) High (3) Very high (4)
performing the project
4 Overall expected financial impact (positive or negative) on the Low (1) Moderate (2) High (3) Very high (4)
project’s stakeholders
5 Strategic importance of the project to the organization or the Very low (1) Low (2) Moderate (3) High (4)
organizations involved
6 Stakeholder cohesion regarding the characteristics of the High (1) Moderate (2) Low (3) Very low (4)
product of the project
7 Number and variety of interfaces between the project and Very low (1) Low (2) Moderate (3) High (4)
other organizational entities

© ASCE 04016126-3 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126

 Research Methodology

Research Approach
For this research, the academic researchers were working closely
with 18 construction industry practitioners to form a research
team. This industry group composed of practicing professional
representatives from owners, contractors, and engineering firms.
The industry practitioners all had at least 10 years of experience
in the construction industry. These practitioners were responsible
for providing research inputs and practical perspectives through
multiple rounds of reviewing and many intensive discussions with
the academic research team members. The research process, the
survey questionnaire, the collected data, and the research results
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

were reviewed and discussed by these industry practitioners in

the validation process.
To fully explore and assess project complexity, a research frame-
work was developed by the research team to support the interaction
of the primary research objectives. The measurement approach of
this research was developed based on literature findings, the results
of prior research, an industry pilot and review process, and the
preliminary results of statistical analysis. For the complexity meas-
urement approach, the term complexity indicator was selected to Fig. 1. Research approach
identify complexity measures based on its common use in the con-
struction industry and academia. The research team members who
were working in the construction industry agreed that using the or for a myriad of other project conditions outside of what is
term complexity indicator provides a common sense descriptor typically perceived as “normal.” Project professionals and teams
for not only construction scholars but also for practitioners when have an intuitive sense of when a project is complex, but the rea-
referring to the measurement of project complexity. The study sons for that complexity may be widely varied and depend on that
started with a literature review to identify potential complexity person’s or project team’s experiences, resource availability, stake-
indicators. The identified indicators were brought to many discus- holder considerations, and many other factors, both objective and
sions with the industry research team members. The complexity subjective. Additionally, the perception of complexity can be com-
indicators were modified and improved upon based on a practical pounded by multiple project factors, which if not managed effec-
perspective to ensure that the survey respondents understand the tively may have a negative impact on the project outcome. To
developed measures and to guarantee that the developed measures address these issues, project complexity is described not in terms
represent an appropriate construct. of a project’s physical features (e.g., types of materials, quantities
The research process was conducted through five tasks. Task 1 of materials, number of systems, and facility technology) but rather
started with the in-depth exploration of the concept of complexity complexity related to managing projects (e.g., internal project team
and its attributes. Contemporary literature on the subject of com- interfaces, site logistics, permitting, etc.). This was necessary to
plexity was reviewed, and background experience was gathered. ensure that the research results could be generalized across con-
Research questions and research objectives were identified to direct struction industry sectors and within a construction industry sector
the research around the focused point. Task 2 reviewed the devel- with different types of projects.
oped definitions of project complexity and described project After substantial consideration of project complexity defini-
complexity by developing a working definition and identifying tions, the initially consolidated definition of project complexity
potential complexity attributes as a basis for this research. Task 3 from the literature review, as presented earlier, was modified to cre-
proposed an approach to develop the complexity indicators deemed ate a final definition as the basis of this research as follows: “Project
to measure the associated complexity attributes and the relevant complexity is the degree of interrelatedness between project attrib-
hypothesis. The data related to testing the hypothesis were col- utes and interfaces, and their consequential impact on predictability
lected through a survey in Task 4. In Task 5, the data were reviewed and functionality” [CII RT 305-1 (CII 2016b)]. This definition at-
and descriptively analyzed to interpret their characterizations, and tempts to capture the essence of how project attributes, such as
then statistical methods were applied to analyze the collected data project scope, team organizational dynamics, project location,
and test the relevant hypothesis to determine which indicators are policies and regulations, unfamiliar technologies, and workforce
truly representative of project complexity. Fig. 1 provides a visual skill sets, interact both within a project and with entities outside
description of the research approach. of the project. Without targeted strategies to manage complexity,
the project’s outcome may be negatively impacted. With proper
management strategies in place to control a diverse set of project
Project Complexity Definition Development attributes and associated interfaces that lead to increased project
Throughout the literature review, several definitions of complexity complexity, the probability that projects can be both successful
were found from a wide range of disciplines. One of the difficulties and predictable is increased.
in addressing the topic of project complexity is that the term is
applied broadly and intuitively. Without a standard definition,
Complexity Attribute and Indicator Development
complexity tends to be a catch-all category that is often used
when project results are unpredictable, when a project has many With the developed definition of project complexity, a methodol-
interacting parts, when details of a project are poorly understood, ogy was developed to identify the level of project complexity. The

© ASCE 04016126-4 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126

 Table 3. Thirty-Five Identified Complexity Attributes
Category Number Attribute
Stakeholder management 1 Clarity of business objectives
2 Level of stakeholder cohesion
3 Public profile
4 Social and political influences surrounding project location
Governance 5 Joint ventures
6 Level of authorizing approvals and duration of receiving proposals
7 Level of control
8 Owner, partnerships
Legal 9 Legal
10 Permitting and regulatory requirements
Fiscal planning 11 Commercial burdens
12 Fiscal planning or financing (funding stream, uncertain political environment)
Interfaces 13 Interfaces within the project
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

14 Number of participants
15 Number of suppliers, subcontractors, contractors
Scope definition 16 Change management (dynamics of market and environment)
17 Clarity of scope definition
Location 18 Climate
19 Local content requirements
20 Logistics
21 Number of locations
22 Physical location
Design and technology 23 Design [number of process steps, health and safety executive (HSE) hazards,
number of recycles, exotic materials]
24 Technology
Project resources 25 Direct field labor management
26 Productivity
27 Resource availability
28 Team experience
29 Turnover
Quality management 30 Quality of suppliers, subcontractors, contractors
31 Unfamiliar procurement
Execution targets 32 Cost targets
33 Schedule targets [schedule targets was defined as the level of schedule
aggressiveness (duration or overlap of engineering and construction)
in relation to industry benchmarks]
34 Schedule (schedule was defined as the number of steps/activities/critical
paths/successors/etc. identified to meet the time objectives)
35 Strategic importance of the project

term “complexity attribute” was used to represent factors that de-

scribe project complexity. Fifty major complexity attributes were
identified using complexity theory variables, the literature review
results, and industry experience. A description for each attribute
was then created that included (1) attribute definition, (2) examples,
(3) measures, (4) impacts, and (5) mitigation strategies. Creating
these descriptions for each complexity attribute aided in eliminat-
ing or combining attributes and enabled the attributes to be ranked
Fig. 2. Complexity measurement hierarchy
in an attempt to reduce the number of attributes to a more mean-
ingful and manageable number because of the similarity or duplic-
ability. The ranking process ultimately resulted in a reduced list of
35 complexity attributes. The complexity attributes were then measuring purposes, each attribute has one or more indicators
grouped into categories to aid in understanding the general nature deemed relevant to measuring it. Ninety-two complexity indicators
of the attributes. Eleven categories were proposed including the were finally identified to measure the 35 associated complexity
following: (1) stakeholder management; (2) project governance; attributes. Fig. 2 presents the complexity measurement hierarchy
(3) legal; (4) fiscal planning; (5) interfaces; (6) scope definition; for a single category. This hierarchical framework is discussed
(7) location; (8) design and technology; (9) project resources; in more detail in the data collection process section that follows.
(10) quality; (11) and execution targets. A category can have a
number of different complexity attributes. Table 3 presents the list
Research Hypothesis
of 35 complexity attributes associated with 11 categories.
The identified complexity attributes presented in Table 3 were To verify the complexity attributes that truly describe project com-
used to develop the complexity indicators thought to measure the plexity, the significance in differentiating low-complexity projects
associated attributes. Each attribute-measuring indicator was from high-complexity projects based on the indicators used to mea-
then converted to one question for data collection. For complexity sure those attributes was tested. If a complexity indicator was not

© ASCE 04016126-5 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126

 statistically significant for describing and measuring project com- improved respondents’ ease of use because the survey was quite
plexity, it was excluded from the list. The significance level (α) for long. Additionally, the survey was administered to CII members,
testing the hypothesis is discussed in the research results section. who understand this scale of measurement. The descriptors on the
The primary research hypothesis was proposed to test the signifi- seven points were modified for each question to be relevant. Ques-
cance of complexity indicators. The research team produced this tions that could not be measured using the seven-point Likert
proposed hypothesis: scale were questions the respondent needed to access specific
Null Hypothesis (H0)—The identified project complexity project data to answer (continuous scale), such as project cost
indicators are not significant in differentiating low-complexity or full-time equivalent project team members. This effort was co-
projects from high-complexity projects. ordinated with the expectation of developing statistically via-
Alternative Hypothesis (H1)—The identified project complex- ble data.
ity indicators are significant in differentiating low-complexity For data collection purposes, each attribute had one or more
projects from high-complexity projects. indicators deemed relevant to measuring it. Each complexity indi-
The assessment of the hypothesis drove the data collection and cator was converted to one question, and the survey was developed
analysis techniques. This hypothesis was quantitatively tested for to collect data on each question in the complexity matrix. The ques-
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

each indicator that was developed to measure the associated com- tionnaire was then refined by the industry practitioners and the CII
plexity attribute. Ninety-two complexity indicators were developed research committee. Many changes were recommended, and it was
to measure 35 complexity attributes presented in Table 3; therefore, decided to use a seven-point scale as a basis for assessing certain
92 corresponding individual hypotheses were tested. Each hypoth- complexity indicators. The research team revised the survey based
esis was tested using statistical methods based on factual historical on these inputs and piloted the survey to ensure it was viable. After
project data collected through the survey. The hypothesis testing the survey, the collected data were analyzed. The set of questions
process resulted in a list of significant complexity indicators that significant in differentiating high-complexity projects from low-
function as the measures of project complexity. complexity projects determined the corresponding indicators that
are significant in measuring project complexity.
Data Collection The survey, containing 106 survey questions and requiring 152
responses, was structured into three sections including the follow-
After developing the project complexity definition, identifying the ing: (1) General Project Description, (2) Project Complexity
complexity attributes, and developing the complexity indicators Metrics, and (3) Best Practice Implementation. Responses were
used to measure each associated attribute, the research team imple- a number, yes/no, or an ordinal scale (e.g., one to seven with
mented the data collection process to collect the data that were one being low and seven being high complexity for a measure).
usable to test the proposed hypothesis and to determine which Section 1, General Project Description, consisted of 16 questions
indicators functioned as measures of project complexity. When that provided general information and project characteristics of the
considering different data collection approaches, the intent of surveyed projects such as project type, project cost/schedule, and
the data collection, the type of data collected, possible analysis design/procurement/construction contract type. Section 2, Project
processes, sample sizes, interpretation, and advantages/disadvan-
Complexity Metrics, consisted of 73 questions with 92 responses.
tages of the approach to derive a set of survey questions were taken
These 92 responses were used to collect data for 92 developed
into account. The questions used to collect the data were based on
complexity indicators that measure the 35 associated complexity
the developed complexity indicators central to the measurement of
attributes. Section 3, Best Practice Implementation, had 13 ques-
complexity attributes.
tions asking the level of best practice implementation for each
The data were collected through a survey to test the relevant
project. At the beginning of the survey, each respondent was asked
hypothesis. Data analysis helped in confirming the definition of
to complete the survey for one high-complexity project and one
complexity including theoretical concepts and complexity’s attrib-
utes, determining different levels of complexity, and providing the low-complexity project based on the survey instructions and their
basis for assessing the impact of project complexity on the use of understanding of complexity. The intent of the survey was to assess
best practices. To support the survey, different ways of categorizing the different complexity indicators based on responses between
the level of complexity based on the proposed complexity defini- low-complexity projects and high-complexity projects. Differences
tion were described. The survey then captured industry project- had to be statistically significant to argue that the indicator was a
focused data on the proposed definition of complexity, levels of true reflection of project complexity.
project complexity, and measures of complexity indicators. The The survey process was conducted online. After the question-
survey was sent to Construction Industry Institute (CII) member naire framework was developed for the survey, a pilot test was
companies, and then follow-up occurred with the survey contact conducted both using hard copies of the survey and the CII online
person (either through e-mails or phone calls) to increase the re- survey software to refine the survey questionnaire, test the appro-
sponse rate. priateness of each question, and ensure that the survey is viable.
The survey questionnaire was developed based upon the com- The survey was pilot tested by several industry research team
plexity measurement hierarchy (Fig. 2). With agreement from the members on 10 projects. The responses for the pilot survey were
industry practitioners on using the term complexity indicator to in- gathered and analyzed to eliminate any questions or information
dicate complexity measures, the academic research team members that was not appropriate to the survey. These pilot surveys helped
developed a set of complexity indicators based on literature find- identify several issues with the survey and helped identify poten-
ings and the inputs from industry practitioners. The measurements tial statistical analyses that would be conducted on the data
for the indicators were revised through several iterations of survey collected from full deployment of the survey. After the question-
development. In the end, the research team decided to use a seven- naire was validated, the survey transmittal memorandum, survey
point Likert scale for many of the indicators. The seven-point Likert instructions, and final questionnaire were uploaded to the CII
scale was decided upon for several reasons, including that this is the online survey system and sent to CII company members. The
scale used for the CII Benchmarking and Metrics (CII 2012), which questionnaire was sent to 140 CII company members to
has been tested and used for many years by CII, and that it collect data.

© ASCE 04016126-6 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126

 Data Analysis and Research Result (yes/no) type data. The x-axis of the bar chart consists of the Likert
scale responses, and the y-axis consists of the observed frequencies
The focus of the statistical analysis was to test the research hypoth- of each of the different possible responses. The side-by-side box-
esis and determine the complexity indicators that are significant in plots and bar charts allow for visual comparison of the data between
differentiating low-complexity projects from high-complexity proj- low-complexity projects and high-complexity projects.
ects. Several different statistical tests were used based on the type of
question response. One of the primary objectives of this research
Inferential Statistics
was to determine how to measure complexity while focusing on a
large array of measures. Survey respondents were asked two sep- Depending on the type of data produced from the survey, the meth-
arate times to evaluate the complexity of their project, once at the ods of analysis varied. This was because there are different assump-
beginning of the survey and once toward the end. The statistical tions and limitations to the statistical analysis tests. Table 4
analysis consistent with the data characteristics was conducted. summarizes the basic formal statistical methods that were used
For example because project complexity was measured by different for data analysis in this research. P-values that indicated the
indicators that describe the associated attributes, the statistical statistical significance of differences between the two groups
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

analysis was implemented to determine which indicators should (low-complexity projects versus high-complexity projects) were
be considered as significant and best reflect project complexity. generated through the relevant tests.
In other words, testing which indicators were statistically signifi- To run the analysis, the Survey Visualization Tool program was
cant in differentiating between low-complexity projects and high- used for the statistical approaches (both formal and descriptive) as
complexity projects helped the researchers understand which described in Table 4. The survey data were initially cleaned, coded,
indicators have an important contribution to describing a project’s and then input into the program. The data analysis results were gen-
complexity. After that, the analysis process focused on these sig- erated regarding the different methods of testing, and the statistical
nificant indicators. The primary question of interest is, “Is there a finding (P-value of each complexity indicator) was presented in
clear difference between low complexity projects and high Table 5.
complexity projects with regards to a specific indicator?” Both ex-
ploratory and inferential statistics were used to determine if this Research Results
difference existed.
General Survey Data Characterizations
and Descriptive Analysis
Exploratory Statistics
Forty-four survey responses were provided from the online survey.
Exploratory statistics in this research were graphical displays in- The survey response rate was 31.4%. Of these responses, 30 re-
cluding boxplots and bar-chart graphics to visualize the data. sponses were for high-complexity projects, and 14 responses were
Side-by-side boxplots were used whenever the data were counts, for low-complexity projects, as subjectively rated by the respondent
dollars, or other numerical values. The boxplots illustrated the dis- at the start of the survey (high complexity and low complexity were
tribution of the data, indicating outliers, minimum and maximum the only possible answers). Before selecting a project for the sur-
values, first quartile and third quartile, and median. Bar charts vey, the respondents were provided the relevant information in a
were used to describe both Likert (seven-point scale) and binary Memorandum, the Survey Request, and the Survey Instruction.

Table 4. Statistical Analysis Methods

Statistical test Null/alternative hypothesis Assumptions
Two-sample t-test (adjusted R2 ):
This test Null hypothesis: The means for high- The two groups (high-complexity projects and
was used where the response is a count or complexity projects and low-complexity low-complexity projects) follow a normal distribution.
numerical value. projects are the same.
Alternative hypothesis: The means for high- Each project was independent from other projects.
complexity projects and low-complexity
projects are different.
Kruskal–Wallis/Wilcoxon test: This test Null hypothesis: The probability that the The two groups follow an identically
was used for Likert data (ordinal seven- median of high-complexity projects is greater scaled distribution.
point scale), where it could not necessarily than the median of low-complexity projects on
be assumed that the data follows a normal this question is 0.5 (the distributions are the
distribution. same).
Alternative hypothesis: The probability that Each project was independent of other projects.
the median of high-complexity projects is
greater than the median of low-complexity
projects on this question is not equal to 0.5 (the
distributions are not the same).
Chi-squared test (Nagelkerke’s R2 ): This Null hypothesis: The observed frequencies of Each project was independent of other projects.
test was used for survey questions with “Yes” and “No” for high-complexity projects
binary responses (“Yes” or “No” are not different from those for low-
response), testing whether the observed complexity projects.
frequencies of “Yes” or “No” are equal for Alternative hypothesis: The observed
both high-complexity projects and low- frequencies of “Yes” and “No” for high-
complexity projects. complexity projects are different from those
for low-complexity projects.

© ASCE 04016126-7 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126

 Table 5. Significant Complexity Indicators and Corresponding P-Values
Category/attribute Complexity indicator (CI) P-value
Stakeholder management
Strategic importance of the project CI-1_Influence of this project on the organization’s overall success (e.g., profitability, 0.0821
Project impact of local social growth, future industry position, public visibility, and internal strategic alignment).
and political groups (stakeholders) CI-2_Impact of required approvals from external stakeholders on the original project 0.0341
execution plan.
CI-3_Impact of required inspection by external (regulatory) agencies/entities on 0.0012
original project execution plan.
Governance
Joint ventures CI-4_Total number of joint-venture partners in this project. 0.0631
Owner, partnerships
Level of authorizing approvals CI-5_Number of executive oversight entities above the project management team who 0.047
and duration of receiving proposals. will have decision-making authority over the project execution plan.
Level of control CI-6_Number of times on this project that a change order will go above the Project 0.0542
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

Manager for approval.

Fiscal planning
Fiscal planning, or financing (funding CI-7_Number of funding phases (gates) from concept to project completion. 0.0756
stream, uncertain political environment)
CI-8_Specific delays or difficulties in securing project funding. 0.025
Quality
Quality of suppliers, subcontractors, etc. CI-9_Quality of bulk materials during project execution. 0.0181
Legal
Permitting and regulatory requirements CI-10_Number of total permits to be required. 0.0761
CI-11_Level of difficulty in obtaining permits. 0.0497
Legal CI-12_Difficulty in obtaining design approvals. 0.0718
CI-13_Impact of external agencies on the project execution plan. 0.039
Interfaces
Interfaces within the project CI-14_Peak number of participants [full-time equivalents (FTE)] on the project 0.0207
Number of participants management team during the detailed engineering/design phase of the project.
CI-15_Peak number of participants [full-time equivalents (FTE)] on the project 0.0313
management team during the procurement phase of the project.
Execution target
Cost targets CI-16_Compare target project funding against industry/internal benchmarks. 0.0118
Schedule targets CI-17_Compare target project schedule against industry/internal benchmarks. 0.0366
Design and technology
Design (number of process steps, CI-18_Difficulty in system design and integration on this project compared to a typical 0.0048
HSE hazards, number of recycles, project for your company.
exotic materials)
Technology CI-19_Company’s degree of familiarity with technologies that will be involved in 0.0138
detailed engineering/design project phase.
CI-20_Company’s degree of familiarity with technologies that will be involved in 0.0065
construction project phase.
CI-21_Company’s degree of familiarity with technologies that will be involved in 0.0106
operating facility project phase.
Location
Number of locations CI-22_Number of execution locations which will be used on this project during 0.0324
detailed engineering/design phase.
Logistics CI-23_Number of execution locations which will be used on the project during 0.0114
fabrication (bulk materials and equipment) phase.
CI-24_Impact of project location on the project execution plan. 0.017
Scope definition
Change management (dynamics CI-25_Identify the percentage of engineering/design completed at the start of 0.0524
of market and environment) construction.
CI-26_Clarity of the change management process to key project team members. 0.0757

CI-27_Impact of the magnitude of change orders on project execution. 0.003

CI-28_Impact of the timing of change orders on project execution. 0.0129

CI-29_ RFIs drive project design changes. 0.0268
Project resources
Direct field labor management CI-30_Percentage of project/construction management staff who will work on the 0.0994
project compared to planned project/construction management staff.
CI-31_Quality issues of skilled field craft labor during project construction. 0.0381
Resource availability CI-32_Frequency of workarounds (work activities out of sequence to continue) 0.0293
because materials are not available when needed to support construction.
Turnover CI-33_Percentage of craft labor turnover. 0.0459
CI-34_Percentage of craft labor sourced locally (within 100-mile radius of job site). 0.0866

© ASCE 04016126-8 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126

 This information included the project complexity definition that not greater than 0.1). This result was reviewed again. Among those
had been developed as a research basis at the initial stage of the 36 statistically significant indicators, two indicators had high cor-
research. The descriptive instructions to preliminarily assess relations with two other indicators. The first indicator was “Average
complexity levels of a project had also been provided based on number of participants during the detailed engineering/design
the preliminarily identified complexity measures such as project phase” that had very high correlation with the indicator “Peak
size, project schedule, number of process steps, or technology number of participants [full-time equivalents (FTE)] on the project
familiarity. Additionally, the respondents were also asked to select management team during the detailed engineering/design phase of
the projects based on their perspective of complexity. The respond- the project.” The correlation coefficient between these two indica-
ents were asked again about the level of complexity for the tors was 0.99 (R ¼ 0.99). The second indicator was, “Average
surveyed project in the final question of the “Project Complexity number of participants during the procurement phase,” which
Metrics” section of the survey. This question asked the respondents had very high correlation with the indicator, “Peak number of par-
to evaluate the project’s complexity on an ordinal seven-point scale ticipants [full-time equivalents (FTE)] on the project management
with one being low complexity and seven being high complexity. team during the procurement phase of the project.” The correlation
The implication of this Likert rating was to verify the initial selec- coefficient between these two indicators was 0.98 (R ¼ 0.98). Stat-
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

tion of complexity level for the surveyed project. The rationale for istically, two or more highly correlated indicators may measure the
asking this question is that, after answering all the survey questions same characteristic of project complexity. As a result, those two
plus their perspective of complexity, the respondents have more indicators as described previously were excluded from the signifi-
information and understanding of project complexity, and then re- cant indicator list. Ultimately, a list of 34 complexity indicators that
assess the complexity level of the selected project on this Likert measure 22 complexity attributes was finalized as presented in
scale. The average scores of the Likert scale question for 14 Table 5. These indicators were considered critical descriptors of
low-complexity projects and 30 high-complexity projects were project complexity and used as a basis for further research. The
2.34 and 5.25 respectively. This confirmed that the perception further relevant research includes developing a project complexity
of low versus high complexity initially classified by the survey re- assessment and management tool, developing project complexity
spondents was consistent with the survey data. management strategies, and developing a project complexity pre-
The survey data set was descriptively analyzed to ensure data dictive model (not covered in scope of this paper).
characteristics were understood. This effort was also used to follow The complexity indicators and complexity attributes listed in
up on the survey responses and to correct any inappropriate or erro- Table 5 are statistically significant in differentiating low-complexity
neous data. Among the 44 projects, there were 30 heavy industrial projects from high-complexity projects. These indicators can be
projects, 3 light industrial projects, 3 building projects, 3 infrastruc- used to describe and measure the complexity of a project. This
ture projects, and 5 projects of other types. The responses ranged finding assists scholars and practitioners in identifying potential
from project costs of $0.4 million to $5,600 million (average problem areas that may arise in the project development and deliv-
$140 million for low-complexity projects and $417 million for ery process. A proactive project management approach can then
high-complexity projects). The total schedule durations for the sur- develop appropriate management strategies to deal with project
vey projects were from 8 months to 70 months (average 25 months complexity. Implementing proper management strategies relevant
for low-complexity projects and 30 months for high-complexity to the identified complexity indicators will help organizations in
projects). reducing the likelihood that the associated attributes will cause poor
project performance.
Testing for Differences between High-Complexity Projects
and Low-Complexity Projects
The primary goal of this research was to develop a methodology to Conclusion
fully explore and assess project complexity by identifying the com-
plexity indicators used to describe and measure project complexity. This research provided a constructive way to assess and measure
In other words, the complexity indicators that were significant in aspects of project complexity. The research contributes to enriching
differentiating low-complexity projects from high-complexity proj- the theoretical basis in the field of project management by
ects were identified and tested. The significance level of 0.05 identifying the root contributors to project complexity defined as
(α ¼ 0.05) was initially chosen to test the significance of each com- complexity indicators. The degree of project complexity was deter-
plexity indicator in differentiating low-complexity projects from mined by identifying and measuring the complexity attributes. This
high-complexity projects. Twenty-four out of 92 complexity knowledge can help project participants develop their competencies
indicators were found significant with the significance level of in managing complex projects in different industry sectors. More-
0.05 as a result of the analysis. This result was then reviewed over, this knowledge can provide project management scholars a
by the research team. Upon reviewing the initial analysis result, basis to study the positive impacts and reduce the negative impacts
the industry members of the research team determined that all of project complexity in managing projects in an effective manner.
aspects of project complexity were not sufficiently described by Practically, the research implications of identifying 22 complexity
these 24 complexity indicators. Based on industry experience attributes and 34 associated complexity indicators were that, when
and discussion, several other complexity indicators that had knowing the true contributors to project complexity, project man-
P-values close to 0.05 were important in measuring project com- agement practitioners would be able to focus on those complexity
plexity and should be included in the list. The significance level indicators, prioritize their activities to the identified complexity
was ultimately increased from 0.05 to 0.1. The research team attributes, and allocate the limited project resources to address
deemed this an acceptable significance level based on the industry those attributes. Understanding the complexity indicators also helps
and research standards and intended use for the results. project practitioners develop an appropriate management strategy
With the significance level of 0.1, there were 36 complexity to minimize the impact of the identified complexity indicators.
indicators (CIs) associated with 22 complexity attributes that were One limitation of this study is that the research team chose not to
statistically significant in differentiating low-complexity projects describe complexity primarily in terms of a project’s physical fea-
from high-complexity projects (the indicators that have P-value tures but rather to address complexity as it is related to managing

© ASCE 04016126-9 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126

 projects. With this focus, the study did not consider complexity CII (Construction Industry Institute). (2016a). “Measuring project
measures that are specifically related to the physical features of complexity and its impact: Research Rep. No. 305-11.” CII RT
a project. This limitation could be addressed through future re- 305–11, Univ. of Texas at Austin, Austin, TX.
search that studies project complexity focusing on projects in a spe- CII (Construction Industry Institute). (2016b). “Measuring project
complexity and its impact: Research Summary 305-1.” CII RT 305–1,
cific industry sector with their unique physical features. Projects in
Univ. of Texas at Austin, Austin, TX.
different sectors have different physical features that more or less Edmonds, B. (1999). “Syntactic measures of complexity.” Ph.D. thesis,
contribute to project complexity. Therefore, focusing more on Univ. of Manchester, Manchester, U.K.
physical features of a project can help in exploring other specific GAPPS (Global Alliance for Project Performance Standards). (2007). “A
complexity attributes in the corresponding sector. This result could framework for performance based competency standards for global
further enhance the theoretical basis and understanding of project level 1 and 2 project managers.” Sydney, NSW, Australia.
complexity and its impact on project development and delivery. Gidado, K. (1996). “Project complexity: The focal point of construction
In this research, a set of complexity management strategies was production planning.” Constr. Manage. Econ., 14(3), 213–225.
developed. Each indicator has a number of management strategies Hass, K. B. (2008). “Introducing the new project complexity model.”
that can be applied to minimize the complexity impact. One limi- Manag. Concepts, 22–31.
Downloaded from ascelibrary.org by Zhejiang University on 07/21/24. Copyright ASCE. For personal use only; all rights reserved.

tation to this paper is that this separate, second contribution of the Lucas, C. (2000). “Quantifying complexity theory.” 〈http://www.calresco.org
/lucas/quantify.htm〉.
research project, the process of developing complexity manage-
Nassar, K., and Hegab, M. (2006). “Developing a complexity measure for
ment strategy, will be published in another paper. project schedules.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-
9364(2006)132:6(554), 554–561.
Perrow, C. (1965). “Hospitals: Technology, structure, and goals.” Hand-
Acknowledgments book of organizations, Rand McNally, Chicago, 910–971.
Puddicombe, M. (2012). “Novelty and technical complexity: Critical con-
The study described in this paper was supported by the Construc- structs in capital projects.” J. Constr. Eng. Manage., 10.1061/(ASCE)
tion Industry Institute (CII RT 305 Research Project). This paper CO.1943-7862.0000459, 613–620.
forms a part of the research project titled “Measuring Project Remington, K., Zolin, R., and Turner, R. (2009). “A model of project com-
Complexity and Its Impact,” from which other deliverables have plexity: Distinguishing dimensions of complexity from severity.” Proc.,
been produced with common background and methodology. The 9th Int. Research Network of Project Management Conf., International
authors also acknowledge the contributions of other CII RT 305 Research Network on Organizing by Projects (IRNOP), Technische
Research Team members for providing significant inputs to Universität Berlin, Berlin.
complete this study. Rosenhead, J. (1998). “Complexity theory and management practice.”
LSEOR 98.25, Dept. of Operational Research, London School of
Economics and Political Science, London.
References Sbragia, R. (2000). “The interface between project managers and functional
managers in matrix organized product development projects.” 9th Int.
Baccarini, D. (1996). “The concept of project complexity—A review.” Conf. on Management of Technology, International Association for
Int. J. Project Manage., 14(4), 201–204. Management of Technology, Univ. of Miami, Miami.
Brockmann, D. I. C., and Girmscheid, I. G. (2007). “Complexity of Stacey, R. D. (1992). Managing the unknowable: Strategic boundaries be-
megaprojects.” CIB World Building Congress 2007, International tween order and chaos in organizations, Jossey-Bass, San Francisco.
Council for Building (CIB), Rotterdam, Netherlands, 219–230. Valle, V., Jr. (2000). Chaos, complexity and deterrence, National War
Chu, D., Strand, R., and Fjelland, R. F. (2003). “Theories of College, Washington, DC.
complexity: Common denominators in complex systems.” Complexity, Vidal, L. A., and Marle, F. (2008). “Understanding project complexity: Im-
8(3), 19–30. plications on project management.” Kybernetes, 37(8), 1094–1110.
CII (Construction Industry Institute). (2012). “Benchmarking and metrics: Williams, T. M. (1999). “The need for new paradigms for complex
Project level survey, version 11.” Univ. of Texas at Austin, Austin, TX. projects.” Int. J. Project Manage., 17(5), 269–273.

© ASCE 04016126-10 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(5): 04016126