A Bayesian latent-mixture model analysis shows that informative samples reduce base-rate neglect

We examined the conditions under which sampling information from different probability distributions reduces base-rate neglect in intuitive probability judgments. To assess the impact of our manipulations, we employed a novel Bayesian latent-mixture model analysis that allowed us to quantify evidence for base-rate neglect. Experience with samples from the posterior distribution in the form of sequential sampling and a descriptive summary tally both markedly reduced base-rate neglect relative to baseline, and the summary tally improved performance over sequential sampling. Experience with samples from the prior distribution reduced base-rate neglect when conveyed as a descriptive summary, but not when sequentially sampled over time. The results indicate that (a) a summary of sample information can be more beneficial to judgment performance than sequentially sampling the same information, and (b) the benefits of sampling experience are more likely to be realized when the contents of the sample are perceived as directly relevant to the judgment problem. These findings help to clarify when and how sampling experience facilitates intuitive probability judgment