Psychometric Reliability of ERN and Pe Across Flanker, Stroop, and Go/No-Go Tasks: A Direct and Conceptual Replication.

Abstract

The effectiveness of error-related negativity (ERN) in assessing individual differences hinges on its psychometric reliability. Despite evidence that the task used to record ERN moderates internal consistency, this moderation is rarely examined within the same sample, risking inaccurate generalizations of psychometrics. A direct and conceptual replication of Meyer et al. (2013, Psychophysiology) was conducted in 182 participants to assess the internal consistency of ERN from flanker, go/no-go, and Stroop tasks as a function of increasing trials. Analyses were extended to include error positivity (Pe) and difference scores (ΔERN, ΔPe), and generalizability theory and multilevel models were used to statistically compare internal consistency across tasks. Overall, data supported the internal consistency of results across three tasks in a healthy undergraduate sample, with values ranging from 0.70 to 0.97 when examining all data. However, estimates were in part outside the confidence intervals of the original study, and ERN scores showed lower internal consistency than previously reported for a flanker task and higher internal consistency than previously reported for a Stroop task. Pe score internal consistency was similar across tasks when examining the average number of error trials. These findings underscore the importance of examining reliability in each study rather than relying on universal trial cutoffs. Overall, a flanker task may be better suited for studies of ERN due to the higher internal consistency of ERN scores when including data from all error trials. However, exclusively using a single task is discouraged because understanding the functional significance of ERN and Pe requires considering task-specific nuances and the varying contributions of cognitive processes, such as cognitive control or response inhibition.