Data Into Insights (Spring 2021)/Purdue WP Case

From CommunityData
Revision as of 21:30, 28 January 2021 by Jdfoote (talk | contribs) (→‎Counterfactual thinking)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

The Setup[edit]

Imagine that you work for Purdue's marketing department. The goal of your group is to increase the number and caliber of undergraduate students who apply to Purdue.

You have been given an assignment to assess the name and brand recognition of Purdue. You start by looking at pageview data from Wikipedia, and comparing how often Purdue is viewed compared to some of our peer institutions in the Big Ten.

Brainstorming[edit]

  • What are some potential insights that your team might be interested in from this data? Think of 2-3.
    • These should be causal stories that are worth investigating - i.e., interest in Purdue spiked in 2019 because X

Counterfactual thinking[edit]

For example, it looks like interest in Purdue has not been growing as quickly as interest in some of the other schools in recent months. You are a big Purdue basketball fan, and want to make the argument that fielding a good basketball team is key to recruiting undergrads. You suspect that our recent drop in interest is because our basketball team hasn't been doing that well.

This is the strong version of that counterfactual claim: If our basketball team had performed better, we would have had more and better undergraduate applicants.

  • What are some reasons that this claim might not be true?
  • What kind of evidence would help you to test if this claim is true (or to show that it is true)?
  • Is there other pageview data that you could use to help figure out if you are right?

Ethics[edit]

You need to put together a report next week outlining what you have learned and what suggestions you have for the marketing department.

  • What are some of the ethical considerations you should make as you prepare your report?
  • What other ethical difficulties could come up when doing data analysis like this?