نوع المستند : المقالة الأصلية
المؤلف
قسم علم النفس التربوي كلية التربية جامعة بني سويف
المستخلص
الكلمات الرئيسية
الموضوعات الرئيسية
عنوان المقالة [English]
المؤلف [English]
Estimating the reliability of measurement instruments is a fundamental step in assessing the quality of research data. Although multiple methods exist for calculating reliability, Cronbach’s Alpha remains the most widely used in psychological and educational research. However, it is often applied without verifying its underlying statistical assumptions, which leads to inaccurate reliability estimates. Cronbach’s Alpha is based on the tau-equivalent model, which assumes equal factor loadings for all items a condition that is rarely met in practice. To address these limitations, alternative coefficients have been proposed, such as ordinal alpha, designed for ordinal data using tetrachoric correlation matrices, and stratified alpha, applied to multidimensional instruments. Nevertheless, both alternatives also rely on the tau-equivalent model. In contrast, the congeneric model provides greater flexibility by allowing variability in factor loadings and error variances. McDonald’s Omega, also known as composite reliability, is based on this model and can be applied to both unidimensional instruments (Omega total) and multidimensional instruments (Omega hierarchical). Despite its advantages, Omega has faced methodological critiques, which have led to the development of Weighted Omega. This coefficient assigns each item a weight proportional to its contribution relative to error variance, rather than relying solely on its loading on the latent construct. Moreover, reporting reliability estimates alongside confidence intervals has become essential for improving the precision and interpretability of results. This study concludes with a strong recommendation to adopt Weighted Omega as a more robust alternative to both Cronbach’s Alpha and traditional Omega. Stratified Alpha and ordinal Alpha should be used only when their assumptions are met. Furthermore, the study emphasizes the importance of promoting awareness of Weighted Omega, integrating confidence intervals into reliability reporting, and utilizing advanced statistical software such as R and JASP to implement these modern approaches.
الكلمات الرئيسية [English]
)
View on SCiNiTO