The Behavior of Judges under the Gambler's Fallacy

Previous studies show that when making sequential decisions, current judgments are affected by proceeding ones and people may exhibit negative autocorrelation. I investigated the behavior of judges when making sequential performance evaluations in competitions to determine if there is a negative autocorrelation that is independent of the performers' merits. I checked its compatibility with the interpretation of the gambler's fallacy, where individuals underestimate the occurrence probability of the sequential streaks. I examined two empirical examples: synchronized swimming competition, and hip hop dance championship, leveraging data from the 2019 FINA World Aquatics Championships, and the 2019 Hip Hop Dance Championships, respectively. The findings suggest that generally, negative autocorrelation is significant and is consistent with the gambler's fallacy, even after procedures such as dropping extreme scores and taking average scores. In these cases, the spontaneous adjustments backfire the goal of fairness. Thus, further improvements such as adjusting the scoring procedures or correcting judges' erroneous beliefs are needed to mitigate the effect.