A Reliability Generalization Meta-Analysis of Runco Ideational Behavior Scale

Abstract

ABSTRACT The purpose of this study was to estimate the overall reliability values for the scores produced by Runco Ideational Behavior Scale (RIBS) and explore the variability of RIBS score reliability across studies. To achieve this, a reliability generalization meta-analysis was carried out using the 86 Cronbach’s alpha estimates obtained from 77 studies that met the inclusion criteria. The analyses were conducted with the random-effects model using the transformed coefficient values from the Bonett’s method. The average transformed coefficient alpha value was estimated to be .898 (95% CI: .887, .907) and found to be significant (p < .001) under the random-effects model. Analog to the ANOVA and meta-regression analyses were also conducted using the sample descriptors to explore the variability of the Cronbach’s alpha estimates. Results revealed that education level (university vs non-university) and test version (19-item vs 23-item) significantly affected reliability estimates, suggesting that the alphas differed across the subcategories. The average reliability estimates for non-university samples (α = .925) and 23-item version (α = .913) were found to be higher than university samples (α = .886) and 19-item version (α = .865). Results also indicated that higher number of items, mean, and standard deviation of total scale score were associated with increased internal consistency in the RIBS. Suggestions are proposed for substantive research using the RIBS and for future psychometric research on the instrument.