Group-individual probability confusion: Implications for suspect prioritization in criminal investigations

Abstract

Purpose

Suspect prioritization is a critical function in criminal investigations suffering from information overload. As this effort involves probability ranking, it is important to avoid confusing group and individual selection probabilities, an ecological fallacy related to Kahneman and Tversky's hit-and-run taxicab color exercise.

Design

We studied this tendency using an experimental survey administered to 1017 university students. The survey involved both individual- and group-level probability questions, presented in randomized order. Participants completed the Subjective Numeracy Scale to assess whether accuracy correlated with SNS self-assessment.

Findings

Subjects averaged 39 % correct responses across scenarios on the individual-level probability question, only half their accuracy on the group-level probability questions (74 %). Interestingly, subjects who did not select the correct response to individual-level questions almost always chose the corresponding group-level answer.

Practical implications

Probability confusion can distort suspect prioritization, an important task in criminal investigations involving information overload. It may also contribute to algorithmic discrimination in the criminal justice system. Our analysis suggests why some people answer incorrectly to the Kahneman and Tversky taxicab exercise.

Originality

This is a novel study. While previous research has explored ecological fallacies, cognitive biases, and common probability errors, this is the first analysis to directly examine group-individual probability confusion.