An evaluation of the validity of the Crowne–Marlowe need for approval scale

Over the past four decades, research has consistently documented negative correlations between the Crowne?CMarlowe (CM) social desirability trait scale and numerous measures of sensitive behaviors, conditions, and opinions. These findings have been interpreted as evidence that persons with self-presentation concerns tend to under-report negative information. In contrast to this classic social desirability interpretation, a second perspective, labeled the true-behavior hypothesis, maintains that these correlations in fact reflect accurate reporting of both sets of variables, suggesting that the CM scale is not a sensitive indicator of the social desirability trait. We test these alternative interpretations by examining data from a community survey that collected both self-reports of cocaine use and the biological specimens necessary to validate the self-reports. In bivariate analyses, the CM scale was found to be associated with the concordance of cocaine use reporting and biological assays in a manner consistent with the classic social desirability hypothesis. The CM scale was not found to be associated with actual cocaine use, as measured by drug test assays, a finding inconsistent with the true-behavior hypothesis. After adjusting for other known correlates of substance use in logistic regression models, the CM scale was not associated with cocaine use under-reporting, nor with actual cocaine use behavior. Until further evidence is available, we conclude that the CM may be a questionable indicator of socially desirable reporting behavior in social surveys.