Title of Abstract: Digitizing Maine’s Voting History with a Statistical Analysis of Error Rate
Name of Mentor: Jack O’Brien
Mentor’s Organization or Department: Department of Mathematics, Bowdoin College
Research Abstract: The purpose of my project, completed remotely via Bowdoin Mathematics Summer Research Fellowships, was to digitize Maine voting data and analyze the error rate using three statistical tests. This research was done under the umbrella of my advisor’s project: to create a single online resource for researchers that stores all of Maine’s election data dating to the beginning of voting records. My results suggested that the first test, calculating the p-values of individual rows of election data, identified substantially fewer errors than the second test, summing up individual columns of data and comparing them to the actual totals. Most errors were flagged in the second stage of tests, though a few were identified in the third. I argue that calculating p-values of individual rows of data is not useful in catching outliers on a small scale.