Abstract
The Rutgers Alcohol Problem Index (RAPI), a popular measure of alcohol-related problems in adolescents, varies with many theoretically-relevant measures of individual differences, including sex. The sex differences in RAPI scores fit many models of alcohol problems but could also arise from biased items. In addition, a short form could increase the scale's utility. The current study examined RAPI scores, an additional inventory of problem drinking, and measures of alcohol consumption in over 2000 college student drinkers. Analyses revealed items that functioned differentially for men and women. Dropping these items created a shorter scale with almost identical psychometric properties but less potential for bias. Correlations with drinking habits and drinking problems were the same as those for the full scale, and the size of the effect for the difference between men and women's responses remained essentially the same. These results confirm previous work using different analytic approaches, and suggest that a short form of the RAPI could prove helpful in future research. In addition, these data suggest that analyses of differential item functioning in other scales can reveal important information about the measurement of drug problems.