AGGRESSION IN DIGITAL INTERACTIONS: THE EFFECT OF TOXICITY IN ONLINE GAMING COMMUNICATION

This study analyzes the computer mediated text communication of non-professional video game players. The purpose of this study is to identify the impact of player communication toxicity on team performance. The dataset comprises 42,720 matches played between November 5 and November 18,2015, including game statistics and chat messages. We use a BERT-model to classify toxicity of messages. Then regression analysis was used to estimate the impact of toxic behavior on winning probability and game satisfaction. The results indicate that the highest winning probability of a team corresponds to a low number of toxic messages in team communication, low average toxicity of communication, but high variation in toxicity levels. Furthermore, low levels of toxicity in communication can enhance overall game satisfaction. Keywords: toxic behavior, team performance, logistic regression, Dota 2, esport.