Bayesian truth serum for true performance diagnostics

Abstract

Purpose. The purpose of this study is to investigate the potential of the Bayesian Truth Serum mechanism for identifying hidden causes of changes in employee productivity using corporate survey data. Methodology. The study applies the Bayesian Truth Serum mechanism to extract more reliable information from corporate survey responses. The method collects two signals from each respondent: a personal answer to a survey question and an estimate of how other participants are expected to respond. This structure allows analysis not only of individual opinions but also of respondents’ perceptions of the collective distribution of answers. The analytical procedure compares the empirical frequency of each response option with the predicted frequency derived from respondents’ expectations. Based on this comparison, an information score is calculated, reflecting the difference between observed and expected response distributions. Options that occur more frequently than predicted may indicate underestimated factors influencing productivity. To demonstrate the approach, a model corporate survey aimed at identifying causes of productivity changes was analyzed. Aggregated responses were used to calculate empirical frequencies, predicted frequencies, and information scores, allowing identification of patterns that diverge from expected collective perceptions. Results. The results demonstrate that the Bayesian Truth Serum mechanism allows identification of responses that significantly differ from expected collective evaluations and therefore contain additional information about real factors influencing productivity. The approach reduces the impact of socially desirable responses and other systematic distortions commonly present in traditional surveys. Scientific novelty. The scientific novelty of the study lies in the application of the Bayesian Truth Serum mechanism for analyzing hidden drivers of productivity changes within organizational environments. Practical value. The proposed approach can be used to improve the reliability of corporate surveys and support more accurate identification of factors influencing employee productivity, which may assist managers in making betterinformed organizational decisions.