Week 10 - Forward Steps

Hi everyone!

This week, I got to make major improvements to my process for studying on-campus polling sites that will help me wrap up this stage.

As we discussed last week, my two main sources for this section of the project were Voter Friendly Campus and College Scorecard. The first offers a list of the voter-friendly campuses per state that offer a polling site in two-school-year intervals. The states I chose, which never had Same Day or Automatic Voter Registration, were Arizona, Florida, Kansas, Kentucky, Missouri, and North Carolina. College Scorecard offers comprehensive information on each U.S. college, which I used to determine the most similar pairs of control and treatment colleges.

Now, I increased the number of parameters used to judge whether two colleges were similar, now being able to determine the size of colleges (to avoid comparing very big colleges to small ones) and separate community colleges from 4-year institutions.

When making the new pairs, I included both of their individual silhoutte scores (determining how well they belonged in their own college groups) and their match distance (how closely the two colleges were related to each other). This two-method approach ensures I picked the best pairs. From there, I chose the best from each state.

Now the challenge was one that I had discussed last week, finding the voter registration counts within each college. It’s the best way to see real effects of polling sites. To do this, I was searching the internet for each college’s NSLVE reports, where they disclose information like this.

And while a lot of these pairs can’t be counted because their information isn’t online, the ones that can will be the specific case studies I’ll focus on in my final analysis. It seems like an inevitable limitation of studying on-campus polling sites.

So while finding the data for this wasn’t as straightforward as some other variables I studied, I’m glad I can still move forward.