Cronbach’s Alpha: Genesis, Issues and Alternatives

Abstract

The most widely used metric for assessing a scale’s reliability is Cronbach’s alpha. Since its introduction in 1951 by Lee Cronbach, it has evolved into the accepted benchmark for scale reliability. A quick search for ‘Cronbach’s alpha’ in Google Scholar yields more than 8 hundred thousand results. By any measure, this is enormous, and the usage of alpha appears to be present in practically every domain of academics. The fact that alpha has been so successful is surprising, as researchers have consistently criticised and have pointed out a plethora of problems with it. For instance, is that Cronbach never suggested alpha as a reliability metric—rather he proposed alpha as an alternative measure for equivalency in a test-retest context. Another issue Cronbach never suggested the lower bound of 0.70 as a benchmark of reliability. However, the lower bound of 0.70 has become the holy grail of scale reliability. Alpha continues to lead all measures of scale reliability despite its numerous issues. This article examines the history of alpha, its derivations based on classical test theory, and its limitations. It then suggests three alternative measures, along with software procedures to calculate them: alpha with confidence interval, omega and greatest lower bound. The purpose of this article is to enable researchers to have a better idea about the limitations of Cronbach’s alpha, and to make them aware of other measures of reliability which are available. It is hoped that this article will help researchers report better and more accurate reliability measures in their research works alongside the alpha.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 25 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.