Obviously, a number of issues present in the study, but I can appreciate they tried to account for some of them. Lots of variables are unaccounted for like time of day, win/loss, character, what the researcher said in chat, etc.

Similarly, the opening preamble points out that often the worst harassment is potentially related to when players hear female voices. Obviously there is a different way you could do this study, which would be to invite ~300 players of various skill levels to play a single game and record the results for analysis. That type of study would have it's own problems, but you could at least be more likely to control for some of the problems in this study.