Abstract
The accuracy of a classification equation (which combined scores on the Drug Problems, Alcohol Problems, and Positive Impression scales from the Personality Assessment Inventory [PAI], Morey, 1991), developed by Fals-Stewart (1996) to identify test-taking response sets among substance-abusing individuals, was evaluated. As in the Fals-Stewart (1996) study, three groups of participants completing the PAI were assessed: (a) substance-abusing patients administered the inventory under standard instructions (n = 25); (b) substance-abusing patients asked to respond defensively (n = 25) combined with a group of individuals suspected of abusing drugs, referred for art evaluation by the criminal justice system, who had reasons to conceal their drug use (n = 25); and (c) non-substance-abusing respondents administered the test under standard instructions (n = 25). Significant validity shrinkage in the classification equation was found when applied to the new sample; only 68% of participants were correctly classified into their respective groups, compared to 82% of participants correctly assigned in the Fals-Stewart (1996) investigation.