Laypeople’s interpretations of ‘high confidence’

Abstract

ABSTRACT High confidence has been associated with high accuracy under certain conditions. Yet, how researchers operationalize ‘high confidence’ varies across publications and depends on who is asked. In this study, we collected numeric interpretations to determine thresholds for high confidence. Layperson participants provided a minimum, best, and maximum estimate for ‘high confidence’ in an eyewitness lineup decision on a scale of 0-100. The distribution of best estimates peaked at 90.90%. The peak value for the minimum estimate was 83.80%. Critically, the distributions of responses were highly variable: 68.27% of participants (one standard deviation around the mean) provided best estimates between 79% and 97% and minimum estimates between 60% and 93%. This variability in laypeople’s perceptions implies there is likely to be considerable variability in how jurors and practitioners interpret confidence. Research and practice would benefit from a standardized definition of what constitutes ‘high confidence.’