ベイズ自白剤の回答品質改善効果に関する実証的研究

抄録

The Bayesian Truth Serum (BTS; Prelec, 2004) is a game-theoretic scoring rule used in psychological surveys, particularly when addressing sensitive or subjective topics. It aims to promote honest reporting in situations where respondents might otherwise be incentivized to lie or misreport their true preferences. More specifically, it calculates the BTS score, intended to reflect the honesty of responses. By implementing a scoring system that rewards unexpected yet truthful responses with this score, and by providing feedback with incentives for high scores to motivate honest responses, this method has the potential to enhance survey quality. However, there is still insufficient evidence regarding its actual effectiveness. Therefore, this study aimed to examine the replicability of the BTS and evaluate its practical usefulness. We conducted two online experiments consisting of two tasks involving coin flips and dice rolls. Nevertheless, in both tasks, no significant results were found to support the effectiveness of the BTS. These results highlight various challenges, including proper implementation, in using the BTS method.