Cognitive functioning in criminal offenders: A multi-level meta-analysis

Cognitive functioning is increasingly recognized as an important factor in understanding offending behavior, yet findings across cognitive domains remain fragmented and inconsistent. This multi-level meta-analysis investigates the relationship between criminal offending and cognitive functioning. A search of PubMed, PSYCINFO, and Web of Science (January 2000–March 2025) yielded 101 studies (N participants = 12,789, N effect sizes = 1757). Multi-level meta-analyses were conducted for each cognitive domain, comparing neuropsychological test performance between offenders and non-offending comparisons (non-offending clinical groups and general population groups). The meta-analyses confirmed weaker cognitive test performance in offenders compared to non-offending control groups. The average weighted effect size ( g = 0.44) falls within the small to medium range. Significant differences were observed between offenders and comparisons across all DSM-5 neurocognitive domains: attention ( g = 0.33); executive functions ( g = 0.40); language ( g = 0.53); learning/memory ( g = 0.36); perceptual-motor functions ( g = 0.24); and social cognition ( g = 0.56). Tests differed in their ability to differentiate between offenders and comparison groups. Moderator analyses showed that observed differences in performance were largely absent when offenders were compared to clinical comparison groups, suggesting that difficulties may reflect general clinical dysfunction rather than offender-specific patterns. However, such difficulties could still play an indirect role in offending by reducing treatment responsivity or increasing the likelihood of dropout, both of which are known risk factors for reoffending. • Offenders show weaker cognitive test performance than non-offending comparison groups (g = 0.44, small to medium effect). • Significant differences found across all DSM-5 cognitive domains, with the largest effects in social cognition (g = 0.56) and language (g = 0.53). • Differences between offenders and non-offending comparisons were minimal when clinical comparison groups were used. • Findings suggest weaker test performance may reflect general clinical dysfunction rather than offender-specific deficits.