Basics of HR

Module 3
Analytics
Predictive
Analytics
Things to
Learn
Origin and Meaning
• The term “analytics” is derived from the Greek word analysis
• Meaning a breaking up, from ana-, “up, throughout,” and lysis, ”a
loosening.”
• In practice, analysis is the isolation and identification of the variables
in a situation for the purpose of better understanding the
phenomenon under consideration.
Reading..
Although analytics is relatively new to human capital measurement, it has been in the business
world since the 1960 launching of American Airlines Sabre reservation system. For the state of
technology at the time, it was an amazing accomplishment. Eventually it linked 350,000 travel
agents and 400 airlines around the world with flight data and reservations. A little later, the
ascendance of Walmart was largely attributed to its inventory management database. Then came
Amazon, which rewrote retailing through the Internet. Google and Facebook are pure data plays.
Google can not only answer search requests in three-tenths of a second, its search system can
predict a query before it is fully typed based on aggregating the billions of searches it processes
every day. In just the past few years, predictability has appeared on the horizon of human capital
management
 “A method for analysing and
HR Analytics predicting the outcome of
various human resources (HR)
investments”
ANALYTICS AND THE NEW WORK
MODEL
• There has been a great deal of interest in the past few years on
workforce planning, competencies, and change management.
• Analytics has a key role to play in this arena.
• Too much attention is being paid to workforce planning as an
industrial-era, gap analysis process that is unsuited for a new work
model.
• The concept of a defined job is dead.
• Jobs are fixed routines that do not at all resemble technical and
professional work in the twenty-first century
SOME BASICS
• There are at least two ways of solving problems.
• The most common one is simply to attack it head on and hope for the
best.
• If your modus operandi is to continually swat problems as they
appear, you waste time and resources by repeatedly dealing with the
same problem.
• You seldom make progress.
• There is another alternative that will avoid costly, redundant
investment of scarce resources.
 • The better way is to invest a
Organizational
little time in analysing the
management problem before you act
• If you gather data on what has happened (descriptive analysis),
analyze it in terms of why it happened and what will likely continue if
untreated (predictive analysis), and then design a treatment for fixing
it, most likely you will eliminate a recurrence of the problem
(prescriptive analysis).
• With this approach, you will have time to focus on building a better
future rather than endlessly repeating the past
WHAT IS ANALYTICS?
• Analytics is a meeting of art and science.
• The arts teach us how to look at the world.
• The sciences teach us how to do something.
• When you say “analytics,” people immediately think of statistics. That
is incorrect.
• Statistics play a major role, but only after we understand something
about the interactions, the relationships, of the problem’s elements.
• Analytics is first a mental framework, a logistical progression, and
second a set of statistical operations.
Human resources (HR) or human
capital analytics
• Primarily a communications device
• It brings together data from different sources, such as surveys,
records, and operations, to paint a unified, actionable picture of
current conditions and likely futures
• An evidence-based approach to making better decisions
• Involves gathering of primarily objective facts and secondarily related
subjective data
Three levels of Analytics
• Descriptive
• Traditional HR metrics are largely efficiency metrics (turnover rate,
time to fill, cost of hire, number hired and trained, etc.)
• The primary focus on cost reduction and process improvement
• Descriptive HR analytics reveal and describe relationships and current
and historical data patterns.
• Foundation of analytics effort.
• Includes, for example, dashboards and scorecards; workforce
segmentation; data mining for basic patterns; and periodic reports.
Three levels of Analytics
• Predictive
• Predictive analysis covers a variety of techniques (statistics, modelling,
data mining) that use current and historical facts to make predictions
about the future.
• It’s about probabilities and potential impact.
• It involves, for example, models used for increasing the probability of
selecting the right people to hire, train, and promote
Three levels of Analytics
• Prescriptive
• Prescriptive analytics goes beyond predictions and outlines decision
options and workforce optimization.
• It is used to analyse complex data to predict outcomes, provide decision
options, and show alternative business impacts.
• Is more pre-emptive in its approach and recommends which of the
possible actions or decisions would most likely lead to the desired
outcome.
• It involves, for example, models used for understanding how alternative
learning investments impact the bottom line (rare in HR).
Predictive vs Prescriptive
• Predictive analytics can help us predict which of our employees are
most likely to quit.
• Prescriptive analytics would prescribe the course of action that is
most likely to succeed in retaining these employees.
• Prescriptive analytics can also be used to generate recommendations
for training strategies that improve employee productivity, strategies
that improve employee engagement, etc.
• if you have reliable and robust data, you can use prescriptive analytics
to empower the HR manager in any of the areas that they deal with
on a day-to-day basis
 The only way
Descriptive to begin to
truly know
analytics what is
happening?

In Why it is
happening
Summary Predictive analytics and where it
is likely to

… lead?

Prescriptive What to do
analytics about it?
As an HR pro, what is it about prescriptive
analytics I need to know? Meaning, do I
need to know how the algorithms work?

• When it comes to solutions generated using prescriptive analytics, the

most important thing that an HR professional needs to understand is
the scope of the solution (i.e. what are the caveats associated with the
solution and in what context and scenarios is this solution applicable.)
• For instance, let us consider the case of Netflix ‘Recommended for You’,
a solution powered by prescriptive analytics. Its scope is limited to
prescribing customized selection of movies, television shows,
documentaries, and other videos for its viewers. But it is not capable of
providing music recommendations. Also, the recommendations
provided for one viewer will not hold good for others. This level of
awareness of the scope and context is sufficient to effectively use
solutions that are powered by prescriptive analytics in the HR realm.
E.g.: Using
analytics to
measure the
impact of
employee
engagement on
financial
performance
• Although financial capital (cash) and economic capital (intangible
assets) are the lifeblood of a business, it is human capital (people)
that apply cash and leverage intangible assets to drive business
performance.
 The purpose of analytics is to find
the best path through a mass of
data to uncover hidden value.

TWO
VALUES
Value comes in two forms

Financial Economic
 • Economic value
• Includes practical, noncash significant items or
processes affecting material resources.
• E.g.: market reputation, customer satisfaction,
best companies to work for, and community
relations.
Two Values • Often referred to as off–balance sheet assets.
• Each of them eventually should turn into
financial value as stockholders invest in
company stock, customers purchase products or
services, high-performing personnel seek
employment with the organization, and
favourable community support ensues.
 • Financial value
• E.g.: Cash and other liquid resources such as
Two Values stock and bonds.
• Are recorded on the income
statement and balance sheet
ANALYTIC CAPABILITIES
• Data can be viewed two ways
• Structured
• Structured data is similar to financial data
• Unstructured
• Unstructured data typically is economic or less tangible data

Since the arrival of the industrial revolution 200 years ago, we have
focused on structured data: costs, process time cycles, and
quantities. Yet, according to IBM, at least 80% of the data currently
being produced is unstructured, nonnumeric images, text, and
audio.
ANALYTIC CAPABILITIES
• As social networking continues its explosive growth, the percentage of
unstructured data will necessarily expand.
• In practice, structured and unstructured data can be merged into a
mixture, or fusion
• Refers to hybrid data.
• While hybrid data will be essential for future analysis, it will also
make the process much more complicated.
• This is precisely why analysis is essential.
When a situation is a complex mixture of objective facts and
subjective beliefs, there is no way other than through logical inquiry
and statistical treatment for us to comprehend what is not readily
apparent.
Example of a Brand that uses Big Data for Targeted
Adverts

Netflix is a good example of a big brand that uses big data analytics for
targeted advertising. With over 100 million subscribers, the company
collects huge data, which is the key to achieving the industry status
Netflix boosts. If you are a subscriber, you are familiar to how they send
you suggestions of the next movie you should watch. Basically, this is
done using your past search and watch data. This data is used to give
them insights on what interests the subscriber most.
ANALYTIC VALUE CHAIN
• Economic and financial values are the rewards gained from a series of linked
activities
• Strategic chain management starts with top executives building their strategic
business plan by asking this basic question:
• How do we make money?
• The answer is generically applicable to all profit-making enterprises, yet unique in practice
to each company.
• For not-for-profits, the question is: How do we service our constituents? Companies that
are successful over time prosper by assessing the market before investing resources.
• That includes customers, competitors, technology, governmental policies, the economy and
labour market, and other macro forces that collectively reveal problems as well as
opportunities.
ANALYTIC VALUE CHAIN
• Management also looks inward at the enterprise’s capabilities.
• These include the enterprise’s vision, leadership, brand, culture, financial
strength, and employee capabilities.
• From this dual assessment, plans are made to produce, sell, and service
the company’s offerings.
• This scenario leads to presumed and/or tested customer responses.
• The next step down the chain is operations, where line managers plan,
design, and manage production systems that presumably will serve and
support those customer responses.
• Operations, in turn, are dependent on human talent.
ANALYTIC VALUE CHAIN
• Now the value chain transforms from the planning side to the execution side.
• The plan is useless without people capable of executing it.
• At this base point, human capital analytics is applied to uncover the most
effective way to manage the workforce to optimize performance and retain
talent.
• The HR function is responsible to play a supporting role with operating
managers in hiring, deploying, paying, developing, engaging, and sustaining a
strong workforce. HR designs the people systems to facilitate operations.
• The workforce is the active ingredient driving operations, turning out products
that will excite customers, who in turn will increase spending with the
company and ultimately generate economic and financial values.
How Your
Company
Makes
Money

Analytic Value Chain

ANALYTIC VALUE CHAIN
• An enterprise sub-optimizes itself if management jumps into the chain
at midpoints without due consideration of the preceding natural
connections.
• Often the connections are ignored by investors, who acquire a
company with the primary objective of quickly turning it around and
selling it at a nice profit.
• This has become such a common practice, and referred as flipping.
• While flippers often achieve their short-term financial goal, they also
leave a vulnerable company whose inherent problems have not been
fully solved
ANALYTICS in Summary …
• A management tool that can be used to identify
• Opportunities
• Solve current problems
• and Predict future returns on investments
ANALYTIC MODEL

How about starting your car and There is a structure to analytics

driving off without a travel plan in that is more than simply running a
mind? statistical analysis
 Data
Analysis
Levels
or
Predictive
modelling
using
statistical
procedures
ANALYSING ANY PROBLEM
• The first step is to ask questions.
• It is important to ask questions that proceed logically from your
current state to your preferred destination
• Identifying the true problem and uncovering the essential conditions
surrounding and driving it
ANALYTIC MODEL - Steps
• Organizing –STEP 1
• Start with collecting and organizing human capital data, which is often a major problem.
• Most corporate databases were designed principally for accounting.
• Later, applications were developed for research and development, production, and
sales/marketing/advertising functions.
• The first HR information systems hit the market around 1970
• In time, applicant tracking systems and compensation and benefits programs came along, and
eventually training and development applications
• but seldom were these isolated programs connected.
• Most of the packages were record keepers only that counted volume and costs.
• Almost none had any predictive capabilities.
• That is why today it is so expensive and time consuming to install a new program.
ANALYTIC MODEL - Steps
ANALYTIC MODEL - Steps
• Displaying – STEP 2
• After collecting basic data, many companies have developed dashboards that
format data for various internal customers.
• The dashboards are capable of reporting degrees of performance that are colour
coded red, yellow, and green.
• This is an extension of descriptive data.
• It shows the current state and trends up to this point.
• Like accounting, it is the basis for reporting the past.
• It does not speak to the future except through vague, unsubstantiated inferences.
• The trends are not extendable because underlying conditions and future
assumptions are not included.
HR dashboard
• A HR dashboard is an advanced analytics tool that displays important
HR metrics using interactive data visualizations.
• It helps the HR department to improve recruiting processes, optimize
the workplace management as well as to enhance the overall
employee performance

View of HR
Dashboard
ANALYTIC MODEL - Steps
ANALYTIC MODEL - Steps
• Relating –STEP 3
• At the relating level, the focus turns outward from the data that has been tracked to its
relationship to other data or phenomena.
• The most common practice today is to benchmark your data to that of other companies.
• The caution with benchmarking is to be certain that the companies to which you compare
yourself are truly similar to you
• If one company is showing superior results in employee development, are the results due to
quality hires, commitment to development, or extraordinary financial investment?
• The only way to answer that is by speaking directly to the benchmarked firm.
• Survey data that compares outcomes across many companies can be meaningless and often
can lead to false assumptions.
 Human

Three
forms of
capital Structural
within an
organizatio
n
Relational
 Three forms of capital within an
organization

Human capital Structural capital Relational capital

Is obviously the employees. Includes the things owned, Is the personal connections
such as facilities, between people within the
equipment, software company and between
programs, patents, and company personnel and
copyrights. external entities.
Can be local community
leaders or governmental
personnel.
 Capital Interactions

In practice, a change in one of the capital asset categories often affects something in the other categories.
ANALYTIC MODEL - Steps
ANALYTIC MODEL - Steps
• Modelling –STEP 4
• At this point you are clearly moving from the descriptive to the predictive
level.
• The descriptive-level data has revealed what has happened to this point.
• But since we need to manage for tomorrow, now we have to build a model of
what we want to change.
ANALYTIC MODEL - Steps
Let’s say that there is an agreement that you need to develop, and you need to do it
as quickly as possible. You may have invested in a formal leadership assessment
program by engaging one of the several assessment vendors. This gives you a clear
idea of the current capabilities of your potential leaders. Now the question becomes:
Potential to lead what? You don’t want to make the mistake of many firms that do
not study what the future market will demand. Peter Drucker put it succinctly when
we heard him ask, “Leadership for what purpose?” This is the beginning of building a
leadership model for the future. It is an example of the confluence of human,
structural, and relational capital. Once you have defined a possible leadership model,
you are ready to test it through analytics.
ANALYTIC MODEL - Steps
ANALYTIC MODEL - Steps
• Evaluating –STEP 5
• Prediction gave you your desired model or outcome for the future.
• What is the best way to attain it?
• In business, the model you developed will probably connect people, policies,
products, and processes to achieve some improvement in performance.
• The model predicted certain patterns or relationships that interact to obtain the
desired outcome.
• When you complete the experiment, you can monitor or measure the amount or
degree of change attained.
• In addition, the model gives you a new routine that should sustain or continue to
improve outcomes.
ANALYTIC MODEL - Steps
TYPICAL APPLICATION
• One of the most common uses of analytics is the study of turnover or
attrition.
• Most of the information needed is already in the HR database
• The employee records contain raw data on date of hire, performance
reviews, any status changes (e.g., promotions, salary increases, or
various jobs held), and date of departure
Example – Application
You might launch your analysis by selecting a job group or groups and looking at the
reasons employees claimed to have left, assuming you had a valid exit interview process.
Then you could apply statistical analysis to uncover combinations of reason and tenure
or reason and position. Let’s assume your analysis revealed a connection between
tenure and position. You might find, as we have, that in some cases, a manager can stay
too long in one job. This can correlate with high employee disengagement or other
operating problems, such as reduction of quality, productivity, or service within the unit.
You can begin to realize that there are many possible links across an organization for any
phenomenon, whether it is attrition, performance, sales revenue, customer retention, or
even market share.
 Preventable
Reasons Why
Employees
Disengage
Predictive Analytics
Prior Reading
Case Continued..
• The Vice President (VP) of human resources (HR) asked you to
produce measures that would help determine the effectiveness of the
Retain & Grow initiative. After several iterations and brainstorming
sessions, you both agreed on a set of metrics and how to display them
for your business leaders.
• This chapter focuses on the actions that happened between the
meetings—in particular, how you gathered the data and began to
analyse it using graphs and dashboards. This chapter also explores
how to link different data sets and how to apply predictive analytics to
create information for stakeholders.
FIRST STEP: DETERMINE THE KEY
PERFORMANCE INDICATORS
Gather data

Segmented the data

into three types of
measures: efficiency,
The key performance effectiveness and
indicators (KPIs) were outcomes
developed through
conversations with
the VP,
FIRST STEP: DETERMINE THE KEY
PERFORMANCE INDICATORS
• Because the process has many measures and many steps, it is useful
to create a tracking tool that contains all of the KPIs and various
pieces of information about the measures that will help you gather
them the first time and in the future.
Tool might include this information
• In which department does the data reside?
• Who is the gatekeeper or owner of the data?
• Is it sensitive information?
• If so, what approvals are necessary to gain access?
• What type of data is it (e.g., nominal, ordinal, interval ratio, qualitative
versus qualitative)?
• What is the format of the data (e.g., HTML, text, comma delimited, etc.)?
• Is there a standard process for requesting the data?
• What is the standard turnaround time for a request?
Data Tracking Tool
FIRST STEP: DETERMINE THE KEY
PERFORMANCE INDICATORS
• Communications
• Because other people in the organization own the data, often it is necessary
to file a formal request for the data.
• Do not be surprised if you have to get one or several approvals before gaining
access to the data.
• Sometimes a formal meeting among all of the relevant stakeholders is all that
is necessary to get approvals.
• Other times it may be necessary to write a formal request, provide a rationale
for the project, and demonstrate how the data will be used once it is analysed
FIRST STEP: DETERMINE THE KEY
PERFORMANCE INDICATORS
Data Request Template
FIRST STEP: DETERMINE THE KEY
PERFORMANCE INDICATORS
• Formatting the Data for Analysis
• Whether the data set is large or small, it often comes in a package that needs a little
unwrapping.
• First, it may be a string of numbers in rows separated by columns, pipes, or some other
delimiters that mark the beginning and end of a column.
• This file is useless unless it can be separated into rows and columns.
• MS Excel and MS Access have import features that allow users to open and align such files.
• Second, the file structure may be a vertical file with only a few columns and many rows or a
cross-tab format
• A vertical file is great for storage and for using a tool like Pivot Tables in MS Excel
• For more advanced analytics, a cross-tab structure is needed, wherein each row is a case
(e.g., person) and each column is a variable (e.g., question on a survey).
FIRST STEP: DETERMINE THE KEY
PERFORMANCE INDICATORS
Delimited Data File
FIRST STEP: DETERMINE THE KEY
PERFORMANCE INDICATORS
Vertical Display of a Data Set
FIRST STEP: DETERMINE THE KEY
PERFORMANCE INDICATORS
Cross-Tab Display of a Data Set
 End of data
FIRST STEP: DETERMINE THE KEY collection
and organising
PERFORMANCE INDICATORS

Shows each measure

and describes
the nature of the data
and how it was
collected
SECOND STEP: ANALYZE AND
REPORT THE DATA

Descriptive
Statistical analysis is
generally divided into two
types
Inferential
SECOND STEP: ANALYZE AND
REPORT THE DATA
• Statistical analysis is generally divided into two types
• Descriptive
• Describe the data by using statistical terms that have become common in day-to-day language,
such as the number of responses, the mean or average, the standard deviation, or a frequency
distribution.
• Descriptive analysis is necessary to understand the data
• Inferential
• Search for relationships among variables using techniques like correlation and regression.
• Inferential statistics also test for differences between groups using t-tests and analysis of variance,
among other techniques results of the statistical test
• The analyst infers that the relationship (or difference) is true for the cases in the sample but also
generalizable to the entire population.
• Predictive analytics is an extension of inferential statistics because inferential techniques are used
to predict future values
Basic Analytics Plan
Basic Analytics Plan (Continued..)
SECOND STEP: ANALYZE AND
REPORT THE DATA
• The majority of the work outlined in this table focuses on descriptive statistics—
computing N counts, averages, or 9-box ratings for each business unit or level.
• These results are necessary to describe the current state of the organization using
the KPIs
• Inferential statistics are also noted. For all cases, the analytic technique is analysis
of variance (ANOVA).
• It is used to compare the averages across groups, such as business units or levels.
• While statistical tests are not always necessary, they can be useful.
• When the results are shared with stakeholders, someone will likely ask, “Is that
difference among groups statistically significant?” If the ANOVA test has been
completed, that question can be answered yes or no with authority.
SECOND STEP: ANALYZE AND REPORT
THE DATA

Once the data set is structured Statistical analysis is generally

into a usable file format, the divided into two types
process of data analysis begins
Descriptive
Inferential
SECOND STEP: ANALYZE AND
REPORT THE DATA
Descriptive statistics Inferential statistics
• Describe the data by using statistical • Search for relationships among variables
terms that have become common in using techniques like correlation and
day-to-day language regression.
• The number of responses • Inferential statistics also test for
• The mean or average differences between groups using t-tests
and analysis of variance
• The standard deviation, or a frequency
distribution • Based on the results of the statistical test,
the analyst infers that the relationship (or
• Necessary to understand the data. difference) is true for the cases in the
sample but also generalizable to the entire
population.
Basic Analytics Plan
Basic Analytics Plan
 Predictive analytics is an extension of
inferential statistics because inferential
techniques are used to predict future values

SECOND
STEP: Graphs should be created to display the
data visually.
ANALYZE
AND The static graphs should be transformed

REPORT into dashboards that will be available for

stakeholders on demand

THE DATA
Will automatically update with new
information
 To complete the Several software The statistical analysis If you do not have the
analysis plan, the packages are available, can be complex. skills, find a business
results should be such as Minitab, MS analyst in your
tested for statistical Excel, SPSS, SAS, and R. organization who does
significance. or hire an external
contractor
 • Finally, information should lead to action.
• When presenting the results to
stakeholders, it is best to highlight
Conclusion meaningful differences that leaders can
act on
• E.g.: Business unit A is significantly
greater than business unit B.”
RELATIONSHIPS, OPTIMIZATION,
AND
PREDICTIVE ANALYTICS
• Based on the analysis performed some simple questions arise
Q Why is business unit A hiring more people than units B and C?
Q Why is it taking longer to hire people in business unit B?
Q Why are salary costs so much higher for business units B and C?
A Business unit A is hiring more people than units B and C because unit A is
developing a new product line and needs more engineers.
A Business unit B is taking longer to hire people because the requisitions are
for senior leaders who are scarce and hard to find.
A For business unit C, salary costs are low because most positions are not
highly specialized technical roles.
RELATIONSHIPS, OPTIMIZATION,
AND
PREDICTIVE ANALYTICS
• Dashboard results often spur stakeholders to ask a what-if set of
questions
Q What if we look outside of our geographic region?
Q What if we lowered the standards in the hiring criteria?
Q What if we offered higher salaries?
Q Would any of these changes lead to improvements in the efficiency
measures?
Q What is the monetized benefit of reducing time to hire?
RELATIONSHIPS, OPTIMIZATION,
AND
PREDICTIVE ANALYTICS
• At this point, when current data spur what-if questions, we begin
venturing into the realm of predictive analytics.
• If X (efficiency) and Y (effectiveness) inputs change, what will happen
to Z (business outcomes)?
• This model applies for all aspects of the business, including HR.
• If Unit B lowers its hiring standards (effectiveness), it will increase its
efficiency (time to hire).
Benefits of predictive analytics
• Isolates the best predictors and eliminates those that have no
influence
• Quantifies the influence of predictors—determines how much the
predictor makes the outcome measure rise or fall
• Provides a mathematical model to describe the current state
• Predicts future values
 Correlation

Analytics Regression
Techniques

Structural equation modelling

Correlation
• Correlation is the simplest of the three techniques.
• It examines the relationship between two variables.
• It answers the question: If X increases in value, what happens to Y?
• If X increases by 1 and Y increases by 1, there is a perfect positive
relationship.
• A correlation is described by the statistic r, which ranges from −1 to + 1.
• A zero value indicates no relationship.
• A −1 indicates that Y decreases proportionally as X increases.
• A +1 indicates that Y increases proportionally as X increases.
Regression
• Examines the correlations among all variables
• Selects the variables that have the strongest relationship with the
outcome variable
• E.g.: productivity or profitability
• Multiple linear regression is a slightly more advanced statistical
technique than correlation
• it also does a much better job of sorting the measures that are the
best predictors of an outcome.
• y = mx1 + mx2+ mx3+ b
Regression Model Showing Factors
that Might Predict ACT/SAT Scores
Structural Equation Modelling
• Structural equation modelling (SEM), is an excellent way to examine
multiple hypotheses at once and determine causal pathways.
• It is based on confirmatory factor analysis and requires large data
sets.
• It is a much more complicated analysis than regression and requires
specialized software, such as Lisrel or AMOS
Hypothetical Structural Equation Model Showing
Possible Predictors of ACT/SAT Scores
INTERPRETING THE RESULTS
• Correlational analysis provides good insight into the relationships
among the key variables in the data set.
• Before we examine the relationships, it’s important to formulate a
hypothesis
• To establish in our own minds our expectations about relationships
among our measures.
 Correlations among Variables
The relationship
between cost to
hire and salary A diagonal
is .999, almost line from the
perfect top left to
the bottom
The correlation right of the
between time figure
to fill and salary shows a
is positive, and value of 1 at
not significant the
intersection
The two values of each
that are measure
underlined are with itself.
nearly significant,
close to the .05
significance level
INTERPRETING THE RESULTS
• The correlational analysis describes which measures are related.
• The next step is to determine which measures are most strongly
related
• Consider the two most important factors: productivity and
profitability.
Rank Ordered
Correlations
for
Productivity

By rank ordering, it becomes

quickly apparent that
performance
ratings are most highly
correlated with productivity
Regression Analysis
• Predicting profitability using software, Statistical Package for Social
Sciences (SPSS)
• Profitability = −28248.06 + 692.06 (Productivity)
• This equation indicates that there is only one statistically significant
predictor of profitability, and it is productivity
 Predicted
Profitability
for Three
New Hires
 Regression
Analysis Predicted Productivity Scores for Five New Hires

• Assume for a moment that

HR wants an early indicator
of the quality of the
candidates who have been
hired.
• Productivity = 83.23 − .30
(Speed to competency) + .
42 (Time to till)
Regression Analysis
• What are the drivers of productivity?
• Can be answered with regression as well.

Productivity = 34.74 + 5.62 (Performance rating at 365 days)

− .12 (Speed to competency) + 2.64 (Sponsor satisfaction)
+ .149 (Time to fill)
STRUCTURAL EQUATION MODELING
• SEM can provide insights beyond other statistical techniques
• It also examines multiple predictive relationships simultaneously
• With regression, we predicted profitability first and productivity
second.
• With SEM, we can create a model to test for causal pathways among
all variables all at once
 Hypothetical
Structural
Equation Model

• The utility of this model

is its simplicity.
• With one picture we
can show the causal
pathways and the
predictors that lead to
profitability
End of Module
3