Alternatives to Cronbach's Alpha: Bias, Sensitivity to Mismatch, Scale Strength, Scale Length, and Sample Size

• In Event: AERA Virtual Poster Session 8
  In Poster Session: Poster Session 1

Abstract

Cronbach’s alpha is the most widely used reliability indicator of scale quality. In two Monte Carlo simulation studies, we compared how alpha and its potential alternatives (omega total, omega RT, omega h, GLB, and coefficient H) were affected by factor loading, scale length, sample size, and mismatched items. Results indicated that alpha was still a good choice with small bias, and GLB and coefficient H should not be used with small sample sizes for unidimensional scales. For scales with mismatched items, omega RT and GLB performed better than others with smaller bias and better recovery efficiency. One golden reliability standard for all conditions is impossible. Other supplementary information from confirmatory factor analyses and item-total correlations is also useful.

Authors

• Leifeng XIAO, The Chinese University of Hong Kong

• Dan Wang, The Chinese University of Hong Kong

• Luyang Guo, The Chinese University of Hong Kong

• Kit-Tai Hau, Chinese University of Hong Kong