Confidence Intervals for Cronbach's Coefficient Alpha Values

Coefficient Alpha, which is widely used in empirical research, estimates the reliability of a test consisting of parallel items. In practice it is difficult to compare values of alpha across studies as it depends on the number of items used. In this paper we provide a simple solution, which amounts to computing the confidence intervals of an alpha, as these intervals automatically account for differences across the numbers of items. We also give appropriate statistics to test for significant differences of alpha values across studies.