Bayesian truth serum

Projects involve testing and theoretically extending the Bayesian truth serum (BTS) scoring method.

BTS is a scoring system for eliciting and evaluating subjective opinions from a group of respondents, in situations where the user of the method has no independent means of evaluating respondents’ honesty or their ability. It leverages respondents’ predictions about how other respondents will answer the same questions. Through these predictions, respondents reveal their meta-knowledge, which is knowledge of what other people know.

Current projects apply BTS to artistic judgments, chess, medical diagnoses, and forecasting of political events and developments. Research in the forecasting category is linked to a larger Draper-MIT forecasting project, funded by Intelligence Advanced Research Projects Activity (IARPA), under the Aggregative Contingent Estimation program (ACE).

Key publications

Prelec, D. “A Bayesian truth serum for subjective data.” Science, 2004, 306, 462-466.

Other Publications

Weaver, R. and Prelec, D. “Creating truthtelling incentives with the Bayesian Truth Serum,” Journal of Marketing Research (in press).

John, L., Loewenstein, G., and D. Prelec. “Measuring the Prevalence of Questionable Research Practices with Incentives for Truth-telling,” Psychological Science (March 2012). 

Back to top