A Reliability Generalization of the Torrance Tests of Creative Thinking-Figural

Abstract: The Torrance Tests of Creative Thinking-Figural (TTCT-F) is a well-known measure of creativity, yet its reliability evidence is hardly reported. In this meta-analysis, we provide such evidence by performing parameter-based meta-analytic structural equation modeling on empirical studies that reported the correlations among the TTCT-F indices. Examining the factor structure of the TTCT-F showed that a two-factor structure (Innovative and Adaptive) fitted the primary study data better than a previously assumed single-factor structure. We conducted a reliability generalization study with a sample of 44 correlation matrices from 38 studies ( N = 13,108). We obtained McDonald’s Omega coefficients for the two factors and the composite reliability. Our analyses indicated that the composite reliability was acceptable (ω = .81). Furthermore, we found higher reliability coefficients for the Innovative (INNOV) factor (ω = .85) as compared to the Adaptive (ADAPT) factor (ω = .62). Moderator analyses showed that reliability estimates for the INNOV factor were higher with Form A and when there is evidence of discriminant evidence violation. Between the two forms, Form A had higher reliability estimates than Form B. Our findings suggest that composite scores and INNOV scores are more precise than the ADAPT factor scores, supporting their use in research and practical settings.