ORGB 330 People Analytics

Final Project Guidelines and Suggestions

Overview. You will conduct an interview with someone in an organization and develop a people analytics
proposal for the organization covering both statistical and organizational considerations.

Who to interview. You may interview someone at any type of organization who deals with managing people
(employees, volunteers, athletes). The project should not be focused on analyzing customer behavior. You
may also interview a company that produces a people analytics product. Contact me if you need help finding
an appropriate organization.

Why? The purpose of the assignment is for students to:

• Gain an appreciation for people analytics in its current state in the real world
• Demonstrate knowledge of statistical methods by developing a detailed and nuanced plan, including:
o selecting methods that are appropriate for a particular organizational question
o specifying the decisions and data that would have to be figured out in order to implement this
method
o explaining how to interpret and use the results
o discussing limitations or key assumptions/weaknesses of the method and/or data
• Develop skills in explaining statistical concepts to lay audiences
• Show an understanding of HR and management practices and how data can be used to improve these
practices
• Demonstrate an understanding of the organizational context of a people analytics project and how to
select projects, choose methods, use the results, and frame the purpose and results of the analysis to
gain organizational buy-in from both higher-level decision-makers and employees.
• Show a consideration of the ethical issues involved in conducting people analytics

Conducting the interview

See separate documents with information about conducting the interview as well as the short interview
report.

Final report details

The final report will be about 3000-5000 words and should cover the points listed below. You do not have to
follow a specific format, any approach that covers the main ideas is fine.

Grading rubric:
Presents the information in a clear and organized way 15%
Statistical methods selected are appropriate for the question and the 15%
proposal explains their appropriateness well
The proposal demonstrates an understanding of the methods, how they 15%
work, and their strengths/weaknesses
The proposal covers all of the important statistical details one would need to 10%
consider in implementing this project

1
 The proposal is appropriate for the organizational context and these aspects 15%
are explained in the proposal
The proposal addresses how to gain organizational buy-in in a detailed and 15%
nuanced way
The proposal addresses the ethical questions involved in this analysis, or 15%
explain why ethical issues are not a concern in this case

Note that individual grades may be adjusted if the peer evaluations show a consistent story of contributions
that are significantly greater or less than expectations.

Notes about content and format

Key points to cover in the content
• Why you chose this question and approach
o Organizational reasons
o Statistical reasons
• Describe the details of the proposed analysis, including:
o Where the data would come from
o how you would collect it and/or any considerations regarding data quality
o The variables you would use and why
o The statistical approach you would use and why it is appropriate
o Any cautions, considerations, or limitations of this analytic approach
o How you would interpret and use the results
o In the conclusion you might also discuss some ideas for next steps, how you might build upon
these results
• Discuss applicable organizational and ethical considerations
o How would you get buy-in from top decision-makers?
o How would you get buy-in from employees?
o How would you address any ethical or privacy concerns? (if you believe that there are no
ethical concerns in this particular case, explain why)

Additional comments/details:
• Make the proposal as realistic and appropriate as possible. If there are barriers to actually
implementing a project and these barriers are really limiting, though, it’s ok for you to make certain
assumptions that loosen the constraints a bit, for instance “we are going to assume that they decide to
hire an analyst and this is our recommendation for a first project” or “we are going to assume that the
current director retires and they have a new director that is more enthusiastic about analytics” or even
“we are going to assume that they are actually a larger organization.” Similarly, you should
acknowledge any challenges they might face in obtaining or cleaning the data, but if necessary it is fine
to then say “assuming that they overcome these challenges, here is what we propose that they do with
the data…” Just be clear about the assumptions that you are making and why you had to make them.
And don’t assume away all of the constraints, still try to make some aspects of the proposal realistic.
• It is important that you delve into the details of at least a major chunk of your proposed project. It is
possible, however, that what you are proposing a quite a large project with multiple stages and it’s just
not possible to go into every single detail. If this is the case, it is ok for you to provide less details on
some parts and focus most of your efforts on one stage. Just be very clear in the proposal that this is
what you are doing. For instance, perhaps you are suggesting a project that would require linking

2
 together several datasets, then conducting an analysis, and then running an experiment. You could
decide to focus primarily on the second stage of conducting the analysis. In this case, you would want
to mention that linking the datasets is an important first step and perhaps mention a few key
considerations for this stage, but you don’t necessarily have to delve into all of the details. Same with
the experiment after the analysis. You might just mention this as the next step and make a few
comments. Alternatively, you could decide to only focus on the experiment. In this case, you still want
to briefly discuss the previous stages and their key considerations. It’s also ok in this case to say that
you are going to assume that the second stage ended up with certain results, and you are proposing
your third stage based upon your assumed results.
• Don’t be too idealistic and assume that everything will be wonderful and easy, discuss the limitations
and potential problems that this project might encounter and the limitations of the results.
• I generally expect that you will propose statistical techniques covered in class. If you want to suggest
using a technique not discussed in class, though, that’s fine as long as you explain why it’s
useful/necessary.
• It is fine to use “we” (e.g. “we believe that this is the best approach because….”, “we would
recommend”)
• It is a good idea to use section headings to organize your work. You do not have to follow the order
that I listed above, you may organize your paper as you see fit.
• If you use outside resources (which is not required), please cite them. You may use either APA or MLA
styles

3
 Topics to consider
In this list I am going through all of the class sessions and listing key topics/questions that come to mind that
you may want to consider for your project. Note that this is just a partial list, and you should spend some time
yourself going through the class topics to make sure you have covered everything. Just because I don’t list it
here does not mean that I won’t expect you to cover a topic if it is particularly relevant to your project.

Obviously not all topics will apply to all projects.

• Collecting data
o Does the organization already collect the needed data?
o If not, how would you go about collecting the data?
§ How would you motivate people to provide the data? Deal with privacy concerns, etc?
§ How would you ensure that the data is good quality?
o Are you sure that the data is good quality? Why or why not?
o Is it possible that the data will have missing values, bias, random noise, skew, small sample
sizes? If so how might this affect the results?
• Performance:
o If your analysis involves performance measures, how does this organization measure
performance?
o What might be some issues/concerns with measuring performance here?
o Why is performance important to consider here?
o How easy is it to assess performance?
o How much is performance affected by random factors, Matthew Effect, autocorrelation?
o Is regression to the mean a concern?
o Is performance mostly an individual or team outcome?
• Other outcomes
o Why is this outcome important to consider here?
o What data is collected about this outcome and how is it collected?
o What does this outcome measure and what does it miss?
• Types of analyses
o Will your analysis use descriptive, observational, machine learning, social networks,
experiment, or natural experiment approaches? Why?
o Are there causal issues to consider?
o Are there confounding variables that need to be taken into account? Or that can’t be taken into
account in which case how should we be cautious about the results?
• Regression/logistic regression/survival analysis
o Why are these models appropriate?
§ What are the strengths and weaknesses of these models?
o What will be your dependent variable, your independent variables and why?
o Will you use fixed effects? Interactions? Why would they be appropriate?
o What aspects of the results will be the most useful? How would you use the results?
o Do you propose estimating multiple models and seeing the effects of adding controls? If so,
which variables will be each model, why, and how do you interpret the coefficients these
results?
o What confounding (anteceding) and mediating variables would you want to consider?
• Machine learning
o Why is machine learning the best approach here?
4
 o What type of machine learning would you use and why? (you don’t have to get too detailed
about the specific method)
o How would you use the results of this analysis?
o Are the concerns about machine learning applicable here? Why or why not?
o Is machine learning appropriate for this organization?
o What steps will the organization need to take to successfully implement this tool?
• Experiments
o How would you implement an experiment? How would you get buy-in from the participants?
Why would an experiment be appropriate?
o Is there an opportunity for a natural experiment?
• Social networks
o Why is social network analysis appropriate here?
o What is the goal in doing this analysis and how would the results be used?
o Where will the social network data come from? What are the strengths and weaknesses of this
data source?
o What measures would you use and why would they be useful?
§ How would you interpret the results?
§ What are their strengths/weaknesses?
• Organizational buy-in
o Why is the project you proposed appropriate for the current organizational context?
o How will it improve the organization?
o Is it important to choose a project that will establish the legitimacy of people analytics in the
organization? If so, how will your project achieve this?
o Did the organizational context influence your choice of statistical approaches? If so make this
clear and why the statistical approach is more appropriate for this organization.
o Does the project require participation of employees (e.g. collecting new data)? If so, how do
you recommend pitching the project to employees to get their participation? Is the project
appropriate for these employees?
o Does the project involve using data or methods that might raise concerns among employees? If
so, how would you suggest addressing those concerns?
o How would you suggest using the results and how will this affect organizational buy-in?
o Are there ways to design your project so that it not only answers the question at hand, but also
increases the enthusiasm around people analytics in the organization at all levels?