New meta-analyses slash GMA's predictive power for job performance to ~0.2 correlation

rasbt · x · 2026-11-19

Sebastian Raschka comments on an open-access 2026 meta-analysis in the Journal of Business and Psychology, calling a correlation of 0.2 "pretty meh." For decades, general mental ability (GMA) was described as the single best predictor of job performance with corrected validities near .50 — but that rested on liberal range-restriction corrections and pre-1980 data. Three recent meta-analyses covering 1990–2022 report far lower estimates between .16 and .23, and this paper integrates all three datasets for a comprehensive new estimate. The findings challenge the long-standing hiring dogma that general intelligence is the best proxy for job performance.

Related event: Meta-Analyses Find Intelligence Predicts Job Performance at Only ~0.2 Correlation(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →