The MacAndrew Alcoholism Scale was derived from the Minnesota Multiphasic Personality Inventory in 1965. Abuses in its use in both research and applied (personnel selection) settings arise from the assumption that the scale has sufficient construct and predictive validities. We critically examine this assumption, informed by a review of the 74 papers published between 1976 and 1987 containing empirical data. After examining the origins of the scale, we evaluate its construct and predictive validities by attending to sensitivity, specificity, positive predictive power (PPP), and negative predictive power (NPP) with adult males and females, adolescent males and females, various ethnic groups, and with alcoholics, psychiatric patients, and normals, and with various cutting scores. The PPP with a cutting score of 24 in the general population is only 15%; thus, only 15% of individuals identified by the scale as alcoholics are correctly classified, while 85% of those called "alcoholic" are actually not affected. The data are so clear that we call for a suspension of the use of the MacAndrew Alcoholism Scale outside of research settings. Future efforts to derive such scales must take into account the demonstrated genetic and cultural etiologual heterogeneities of alcohol abuse.
Bibliographical noteFunding Information:
We wish to acknowledge support for our work from a Fellowship to the Center for Advanced Study in the Behavioral Sciences, Stanford, CA., and the John D. and Catherine T. MacArthur Foundation. We thank two anonymous reviewers for their helpful comments on an earlier version of this manuscript.
Copyright 2014 Elsevier B.V., All rights reserved.