Differences between Black, Mexican-American, and White probationers on the revised MacAndrew Alcoholism Scale of the MMPI-2

Abstract

Revised MacAndrew Alcoholism Scale (MAC-R) scores of 270 Black, Mexican-American, and White adult male misdemeanor probationers were compared to determine if ethnicity affects the scale's validity in differentiating problem drinkers from nonproblem drinkers. Subjects were problem drinkers on probation for two or more DWI offenses, and nonproblem drinkers were those on probation for offenses other than DWI with no prior problem drinking or substance abuse history. Examination of ethnicity and drinking status effects using two-way analysis of covariance indicated no significant difference in MAC-R mean scores among the three ethnic groups; however, problem drinkers obtained significantly higher MAC-R mean scores than nonproblem drinkers. Evaluation to determine the MAC's ability in classifying problem drinkers from nonproblem drinkers was conducted by dividing the sample into an estimation sample (n = 180) and a cross-validation sample (n = 90). The estimation sample was used to examine classification accuracy of various cutting scores by determining overall hit rates, sensitivity, specificity, positive predictive power (PPV), and negative predictive power (NPV). Modest discriminative validity was obtained on cutting scores from 23 to 27; however, a cutting score of 26 achieved the best hit rate (62.2%) with optimal balance between sensitivity, specificity, PPV, and NPV. Application of the five cutting scores from the estimation sample to the cross-validation sample resulted in relatively little validity shrinkage, but because only modest validity was obtained in the estimation sample, the decline was of sufficient magnitude that none of the cutting scores achieved hit rates greater than chance. Item analysis of the 49 MAC-R items revealed that significant ethnic differences in response patterns occurred on Items 387, 407, and 422 for problem drinkers and Items 69, 82, 202, 299, and 407 for nonproblem drinkers. Overall findings indicated that the MAC-R has at best moderate validity in differentiating problem and nonproblem drinkers in a misdemeanor population, and that while ethnicity effects were not significant at the group mean score level, differences in responses occurred at the item level.