Anchoring and subjective belief distributions

We investigate how the anchoring effect-a well-established cognitive bias-influences the full distribution of subjective beliefs. While prior research extensively examines the impact of anchoring and other biases on point estimates, their effect on higher moments of the distribution remains unexplored. Through a pre-registered online experiment (N=732), we find that anchoring impacts the mean, variance, and skewness of belief distributions. Notably, the anchoring effect diminishes when eliciting distributions rather than means. Furthermore, presenting anchors prior to eliciting beliefs reduces the variance in belief distributions compared to when elicited without anchors. Our study shows that cognitive biases may have important impacts beyond point estimates.