Times are displayed in (UTC-08:00) Pacific Time (US & Canada) Change
Previous literature explains sports fan behaviors with reference-dependent preference with loss aversion; fans have a reference or expectation before going to the game and their utility depends on the combination of the reference and the actual outcome. Empirical evidence on this topic reports that upset result triggers fans’ emotions and leads to the next behavior such as violent crime or more tips. However, previous evidence regards observed behaviors held by sports fans in the region with aggregated data. In this context, this study tests whether the National Football League (NFL) fans are loss averse and upset results alter their comment-posting behavior using comments from YouTube game highlight videos. For the recent two seasons, we collect all comments of every regular season game highlight via the YouTube application programming interface (API) and approximately over 800,000 comments are collected. Afterwards, we keep comments posted by users who posted more than 10 comments, to figure out their supporting team and we regard a user as a fan of one team if the user posted the most comments on that team game. From this classification, we extracted comments supporting one team in a game, excluding neutral fans' comment, which results about 160,000 comments remaining. Using these comments, we performed sentiment analysis to estimate sentiment scores of each comment using pre-trained deep learning model with eXtreme Gradient Boosting (XGBoost). For an empirical analysis, least square dummy variable regressions are applied to test the impact of an upset result on sentiment score of comments. The results suggest that NFL fans post more negative comments when they see the upset loss and post more positive comments for the upset win. The absolute value of coefficient on upset loss is greater than the one on upset win, as a reference-dependent preference with loss aversion predicts. Also, the results from the same regression analysis with the interaction terms between upset results and the difference between two teams' win probability confirms that upset results triggers more emotional reactions when upset results are less likely.
Presenter(s)
Hyunwoong Pyun, Sungkyunkwan University
Non-Presenting Authors
Yoonji Ryu, Sungkyunkwan University
Eunsoo Park, Sungkyunkwan University
Wonseok Jang, Sungkyunkwan University
Gyemin Lee, Seoul National University of Science and Technology
Are Fans Loss Averse? Text Analysis of Football Highlight Comments
Category
Organized Session Abstract Submission
Description
Session: [198] WORKSHOP FOR INTERNATIONAL JOURNAL OF EMPIRICAL ECONOMICS
Date: 7/4/2023
Time: 4:30 PM to 6:15 PM
Date: 7/4/2023
Time: 4:30 PM to 6:15 PM