Estimating Internal Consistency: Did We Choose the Right Coefficient?

Internal consistency is a concept extensively used in academic discourse, yet its definition remains debated. In the context of validation studies, it is noteworthy that, although internal consistency is commonly assessed, some studies could benefit from employing more accurate estimators that are better suited to the underlying factorial structure. These editorial addresses the various recommended estimators for calculating internal consistency based on the characteristics of the studied model. We explore one-dimensional measures, identifying when estimators such as α are suitable, particularly for tau-equivalent models. For congeneric measurement models, coefficient ω is recommended. We also discuss complex models incorporating multidimensional structures, including essential unidimensionality, scales with multiple correlated or uncorrelated factors, and higher-order models. Researchers should avoid reporting the total internal consistency of the instrument unless unidimensionality or a higher-order factor structure has been demonstrated. When data are approximately unidimensional, measures are congeneric with moderate factor loadings, and sample sizes are large, it is reasonable to report both α and ω.