Measuring impulsivity differently yields different answers: A three-level meta-analysis of associations between impulsivity and adolescent non-suicidal self-injury.

BACKGROUND: Impulsivity is frequently examined in relation to adolescent non-suicidal self-injury (NSSI), yet the strength of this association remains unclear because findings vary across studies. This study quantified the association and examined potential sources of heterogeneity. METHODS: A three-level random-effects meta-analysis of 37 studies (N = 69,572) accounted for hierarchical dependencies among effect sizes. Potential moderators included assessment modality (self-report scale, semi-structured interview, or neurocognitive task), specific measurement instrument, impulsivity domain, sample source, and psychiatric comorbidity. RESULTS: Impulsivity was significantly associated with NSSI severity (d = 0.46) and was higher among adolescents with NSSI than controls (d = 0.37). Effect sizes varied significantly across specific impulsivity measurement instruments after FDR correction. In contrast, the omnibus test of broader assessment modalities was nonsignificant. Separately, self-report scales yielded a significant pooled estimate (d = 0.48), whereas neurocognitive tasks yielded a small and nonsignificant estimate (d = 0.11); these subgroup estimates do not establish modality differences. Neither sample source nor psychiatric comorbidity significantly moderated the association. Trait and symptom-level impulsivity yielded significant pooled estimates (d = 0.48 and 0.40, respectively). Negative urgency showed a relatively large pooled estimate (d = 0.60) based on only four effect sizes. LIMITATIONS: Several moderator estimates were based on few effect sizes, limiting their precision. CONCLUSIONS: Impulsivity is an important but nonspecific correlate of adolescent NSSI. Heterogeneity was more consistently associated with specific measurement instruments, highlighting measurement specificity when interpreting the impulsivity-NSSI literature. Self-report and neurocognitive measures may capture partly different aspects of impulsivity, warranting further multimethod research.