Steps to Analyze Variables in a Dataset

Examine the variables in your dataset

1. Examine the variables in your dataset to ensure that you understand what each variable is measuring.

2. Which variables are categorical or quantitative? Discrete or continuous?

3. How do you interpret the units of measurement?

4. Brainstorm potential relationships that you might expect to see between the variables in your dataset.

5. Which relationships might be important or useful to discover?

6. Do you have any guesses about what type of relationships might exist?

7. If you could accurately model the relationship between certain variables, how could you use that information to apply to business/policy/management decisions?

8. Use methods of descriptive statistics to summarize individual variables in your dataset.

9. Present your summary in tables and/or graphs using methods practiced in this course.

10. You do not have to summarize every aspect of every variable.

11. Some variables will end up being more important to your model(s); you will want to include relatively more summary information about these variables.

12. Use methods of descriptive statistics to summarize important relationships in your dataset.

13. Present your summary in tables (e.g. cross-tabulations) and/or graphs using methods practiced in this course.

14. Remember that how you investigate relationships will depend on the types of variables (categorical v. categorical, categorical v. quantitative, quantitative v.

quantitative).

15. Develop a set of single linear regression models (at least 2) to further investigate the relationships between key variables. Estimate the coefficients in these models.

16. Which relationship(s) is/are valuable to investigate at this level? What do you learn from your regression estimation(s) that you couldn’t learn from the descriptive methods in step 4?

17. Develop a set of multiple regression models (at least 1) to further investigate the relationships between key variables. Estimate the coefficients in these models.

18. Carefully consider which variables to include in your model(s) based on the guidelines discussed in this course.

19. What do you learn from your multiple regression estimation(s) that you couldn’t learn from the single regression estimation(s) in step 5?

20. Compile a report of your analysis and findings—including potential applications of your findings—into a business memo.

21. Do your findings match your guesses about potential relationships?

22. What might explain the relationships that you observe (or fail to observe)?

23. How might your findings be applied to business/policy.

In order to further analyze the type of relationships in your dataset, what other data/variables would you like to have and what would you hypothetically be able to do with that new data? You will work with your partner to analyze the data in your dataset using the tools that we have learned in this course.

The end goal of this project is to develop a meaningful model of an interesting or important relationship between the variables available in your dataset. This process will follow the steps outlined below. You will likely complete some analysis (e.g., create some graphs, estimate some regressions and other analysis) that will not show up in your final project. Think carefully during each step about which information is valuable to include in your final report and which is not. In other words, don’t include everything you can do just because you can do it. Ultimately, you are striving to develop a coherent story based on your analysis and clearly conveying that story is as important as completing the analysis correctly. You will likely need to iterate through many of the assignment steps multiple times.