Uses and abuses of coefficient alpha.

Abstract

The article addresses some concerns about how coefficient alpha is reported and used. It also shows that alpha is not a measure of homogeneity or unidimensionality. This fact and the finding that test length is related to reliability may cause significant misinterpretations of measures when alpha is used as evidence that a measure is unidimensional. For multidimensional measures, use of alpha as the basis for corrections for attenuation causes overestimates of true correlation. Satisfactory levels of alpha depend on test use and interpretation. Even relatively low (e.g., .50) levels of criterion reliability do not seriously attenuate validity coefficients. When reporting intercorrelations among measures that should be discriminable, it is important to present observed correlations, appropriate measures of reliability, and correlations corrected for unreliability. Presentation of coefficient alpha (hereinafter alpha; Cronbach, 1951 ) as an index of the internal consistency or reliability of psychological measures has become routine practice in virtually all psychological and social science research in which multiple-item measures of a construct are used. In this article I describe four ways in which researchers' use of alpha to convey information about the operationalization of a construct or constructs can represent a lack of understanding or can convey less information than is actually required to evaluate the degree to which measurement problems are or are not a concern in the interpretation of the research results. In each instance, I will also indicate which additional or supplementary information is necessary to evaluate the measurements used in the research.