Internal Consistency of the DELTA Scale Assessing Psychotic-Like Experiences

Abstract

This study’s objective was to conduct a reliability generalization meta-analysis of the DELTA scale capturing Disintegration, i.e., proneness to psychotic-like experiences and behaviors. Eligible studies applied various versions of the DELTA scale; 573 α coefficients nested in 76 reports were identified. The pooled effect of all included α coefficients was calculated by applying a 4-level meta-analytic random-effects model. Test length, sample type, demographic sample characteristics (mean age, share of females), language, the proportion of reverse-keyed items, publication status, documentation quality of reliability estimates (reported vs. post hoc computed reliabilities), and DELTA score M and SDs were tested as moderators. The model yielded an average Cronbach’s α for DELTA and its facets of .86 (95% CI [.84, .88]), with medium and long DELTA versions reaching .96 (95% CI [.94, .96]) and .97 (95% CI [.95, .98]), respectively. Moderators explaining the heterogeneity among α coefficients were as follows: test length (expectedly, higher reliability estimates for longer tests), mean scores and SD of the DELTA scores (lower α estimates for larger means and lower SDs), and language (highest α estimates for the German language). Overall, the DELTA scales used in past research showed high-reliability estimates.