2017-18 projects

Forecasting NHL Hat Tricks

Sam Forstner '18

In this paper, I explain the frequency of hat tricks in the National Hockey League (NHL). Assuming goal scoring follows a Poisson process, I propose models intended to generate an accurate number of hat tricks across multiple seasons. Additionally, I examine whether the players who score hat tricks are typically the ones we would expect to do so, given rates of goal scoring. I use ice time and goal scoring data from HockeyReference.com from the 2000-2001 through 2016-2017 seasons and construct three models for my analysis—one that pools over players and seasons, one that allows for season-to-season changes, and one that allows for different scoring rates by player. Finally, I estimate the expected number of hat tricks scored by the highest scoring players of the years 2000-2017 and compare estimates to the observed numbers scored.

PDF iconFull text

Evaluating the Effects of Redlining on American Cities

Annika Roise '18

In the 1930s, the Home Owners’ Loan Corporation (HOLC) produced maps that graded the neighborhoods of hundreds of American cities according to lending risk, creating a legal basis for the racial and socioeconomic separation of residents. Today, over 80 years later, cities remain divided by similar factors and along similar borders. Given the persistently unequal nature of urban neighborhoods since the early 20th century, this paper explores the extent to which ratings from Depression-era redlining maps correspond with contemporary racial and socioeconomic characteristics. I match digitized HOLC maps from the 1930s with 2010 tract-level Census data to study modern neighborhood composition by historical rating in six U.S. cities. I perform significance tests using two treatment classifications to demonstrate the distinctly isolated nature of particularly high- and low-rated neighborhoods and affirm that urban demographics continue to reflect the impact of historical exclusionary practices.

Measuring Temporal Effects and Evidence of Racial Prejudice in Illinois Traffic Stop Data, 2004 - 2015

Gabrielle Kirlew '18

Abstract: Is a driver more likely to be searched at certain times of day? How does the driver’s race affect this probability? This paper will explore the temporal effects of traffic stops and how it impacts evidence of racial prejudice. Temporal effects are measured in three ways: time of day, proximity of stop to end of the officer’s shift, and whether the stop occurred at night. Using traffic stop data from Illinois acquired through the Stanford Policing Project, I will perform an OLS regression to predict searches and the probability of finding contraband upon search. I find that compared to whites, Blacks and Hispanics are more likely to be searched as the day progresses, and that probability of finding contraband decreases over the course of the day as compared to whites.