In many contexts, informed decisions rely on the correct perception and processing of distributional information. For example, to understand how we and others rank regarding wealth, attractivity or intelligence, we must correctly process how these variables are distributed in others. Despite its importance for decision making, research shows that distribution cognition is often flawed, but the reasons why are not yet completely understood. Our research therefore aims at furthering our knowledge about the underlying processes of distribution cognition.
In several laboratory and online studies, we presented subjects with distributions either in the form of numbers or variables like income or age. Values were presented sequentially, and distributions varied along several dimensions, including symmetry, amount of skew, or value range. After the presentation, participants were asked to give different estimations about the distribution, like the lowest numerical value or the mean of the four quarters. Quantitative model fitting via the APE method (Accumulated one-step-ahead Prediction Error) was used to test how accurate different cognitive models were able to predict participants’ estimations. Results suggest that Frequency Regression as well as End Point Anchoring have a strong and reliable influence across distributions and settings. These results are of great importance to better understand how people process distributional information, for example regarding economic inequality.