The capability to align estimated performance with actual performance: A meta-analysis


Background: Within daily tasks (e.g., studying, driving), error in self-assessment of task capability can lead to significant implications and consequences including compromised task performance (e.g., exams) and risk of personal injury (e.g., accident). Objectives: This study sought to determine the overall capability to accurately align estimated and actual performance, identify the pervasiveness of estimation error (i.e., the Dunning-Kruger Effect - DKE), and identify moderating variables on self-assessment accuracy. Methods: A meta-analysis of systematically reviewed studies assessed estimation error in tasks, pooling correlations, mean differences and binary moderating variables. Results: Across studies, a small-moderate pooled correlation between estimated and actual performance (r = .32, CI = 0.29-0.35, p < .001) was identified. Further, the lowest quartile of performers consistently overestimated their performance (i.e., 37.44 percentile points; g = 2.17, CI = 1.74-2.60, p < .001), while the highest quartile performers underestimated (by -19.96 percentile points, g = -1.22, CI = -1.43- -1.01, p < .001). Pooled correlations were moderated by study methodological (e.g., timing of estimation) and task (e.g., physical context) factors, but not participant related factors. Conclusions: Findings indicate that estimation error is prevalent across social-demographics, and was associated with task competency. The DKE was persistent across study contexts. The influence of specific moderators on estimation error, supports the notion of psychological mechanisms (e.g., meta-cognitive skill) as accounting for self-assessment error. Investigation into the consequences of persistent long-term estimation error, and identification of effective interventions capable of error prevention and mitigating are valuable future directions.