What is Wrong with Cronbach’s Alpha and What to Use Instead

Abstract:  When interested in estimating the proportion of scale variance due to a latent variable common to all of a scale’s indicators, the vast majority of applied researchers believe that Cronbach’s alpha is the index of choice. Whereas many methodologists are aware of problems with using Cronbach’s alpha for this purpose, few seem to be aware that there is a better alternative to Cronbach’s alpha.  In this talk, I will discuss the importance of the proportion of scale variance due to a latent variable common to all of a scale’s indicators and the central problem with Cronbach’s alpha for estimating this important parameter.  In addition, an alternative to Cronbach’s alpha - coefficient omegahierarchical – will be introduced and shown to overcome the positive bias often inherent in Cronbach’s alpha.  Finally, I will present the results of simulations testing the accuracy of several different methods for estimating omegahierarchical and conclude with recommendations regarding how to estimate omegahierarchical.