Regression to the Mean: Experimental Evidence and a Proposed Heuristic

The literature on judgment and decision-making suggests that people don't recognize the implications of regression to the mean. Explanations have involved ambiguous feedback and the failure to recognize the role of chance in outcomes. And yet, there is a lack of systematic experimental evidence on how, when, and why the phenomenon occurs. We address this issue by conducting four experiments. Three experiments involved predictions and one a diagnosis. We explored the role of chance by asking participants to provide explanations for their judgments, but our tasks did not allow for the investigation of feedback. On aggregate, predictions were mainly consistent with regression to the mean and content analysis of explanations revealed the use of causal reasoning. Thus, although participants' judgments were in agreement with statistical principles, it is not clear that this was always for the correct reasons. To improve the ability to recognize the effects of regression to the mean, we advocate specific training. To make more accurate predictions, we propose a 50/50 heuristic that gives equal weight to the cue and the mean of the target variable and comes close to achieving optimal performance. In diagnostic tasks, the same rule can be used to generate counterfactual statistical benchmarks to challenge possible causal explanations.