MGMT 408 HR Analytics

This project will help you take a real world set of data, develop hypotheses and research questions you are interested in, and then test your hypotheses and research questions. For this project we will be using data collected from Federal Employees. The following guidelines provide an overview of how to complete the project. Be sure to fully review this document as well as the Project Rubric, and ask any questions you have about the project.

Step 1: Review the Codebook of Available Data

The first step in this project is to review the “codebook” for the data you will be analyzing. The codebook indicates what questions were asked and tells you what the label is for each question. In the codebook, information is separated onto different tabs, but in the dataset all information will be on one sheet.

Step 2: Selecting a subgroup. This data comes from all federal employees and has over 600,000 rows of data. Both R and Excel can handle this amount of data. However, you might want to reduce the overall sample size or focus in on specific groups. Important: Which departments or groups you select might play an important role in the Hypotheses/Research questions you want to investigate. There are at least two reasons you might want to conduct all of your analyses within a specific group or set of groups. 

1. From a practical standpoint, the data will just be easier to work with and the results will run a bit faster if less data is being input into the analyses. There is also less likelihood for Excel to freeze up when using a smaller dataset.

2. Second, comparing results across all departments might not make the most sense. If you were working in HR at the Department of Education, what is happening with employees in the US Air Force, is probably not as practically relevant. Based on that, it can make more sense to use only 1 department or include 2-3 departments so that you can make comparisons between them. 

3. In addition to department, you might want to reduce your sample to include only a specific employee group (e.g., employees with less than 10 years of experience, male or female employees, minority group member employees, etc.). If that is the case, you could filter your data to only include those employees. For example, you might want to conduct your analyses on only female employees in the US Army, and US Navy. Then you would take this large dataset and select only female employees in those two departments. Caution: Be sure your total sample size is at least 100 employees. If you filter your data too specifically, you’ll end up with very few responses, and your analyses will not be very meaningful. Also, if you take this approach (let’s say you only included women in the US Airforce and Navy), you won’t be able to compare between men and women once you remove the men from your dataset. 

Step 3: Background Research & Develop Hypotheses and Research Questions

As a group, develop 4 Hypotheses and/or Research Questions. You must have at least 1 Hypothesis and 1 Research question. However, you could split these up as A) 2 Hypotheses and 2 Research Questions, B) 3 research questions, and 1 hypothesis, or C) 1 research question and 3 hypotheses. Your hypotheses should be developed based on A) Your interests, B) The topics you think would be most important to your specific group/subgroup, C) the data available (based on the codebook), and D) Any background research you’ve done, you now want to develop your  hypotheses/research questions (4 total). Note, you are not required to do background research prior to developing your hypotheses/research questions, but it can help with developing hypotheses. For example, if you are interested in job satisfaction, you might look at peer-reviewed journal articles to see what factors are known to be related to job satisfaction. See the video for how to find peer-reviewed journal articles in Blackboard under the Group Project Folder.