Publication Date

2019

Document Type

Honors Project

Degree Name

Bachelor of Arts

Department

Statistical and Data Sciences

Advisors

Ben Baumer

Keywords

National Football League, Sentiment analysis, Natural language processing, R, Python, Twitter, Multilevel modeling, Data ethics, Reproducible research

Abstract

Following the actions of Colin Kapernick in August of 2016, incidents involving racially charged language towards NFL players became widespread across many forms of social media. Previous studies offer qualitative evidence of racially disproportionate negative language and blame in sports contexts. We believe that this phenomenon is measureable on a larger scale. This thesis seeks to quantify previous research by first accessing the Twitter API then running a sentiment analysis to determine the percentage of negative words for a subset of 24 NFL players at 5 time points. We fit a multilevel regression model to test our hypothesis that during a game that was lost, sentiments will be more negative toward black players, whereas this difference will be smaller, or zero, in games that were won. Additional variables including yards and position were added to our model. Our results demonstrate that on average, the percentage of negative words during a loss decreases by 7.22 percentage points for white players relative to black players, holding position and yards constant. During a win, there is almost no difference in the observed sentiment between white and black players.

Rights

©2019 Julianna Grace Alvord Access limited to the Smith College community and other researchers while on campus. Smith College community members also may access from off-campus using a Smith College log-in. Other off-campus researchers may request a copy through Interlibrary Loan for personal use.

Language

English

Comments

97 pages : charts (some color). Includes bibliographical references (pages [91]-97)

Share

COinS