Predictive validity despite social desirability: evidence for the robustness of self-report among offenders JEREMY F. MILLS 1 , WAGDY LOZA 2 AND DARYL G. KRONER 3 , 1 Bath Instit ution, Bath, Canada and Carleton Uni versit y, Ottawa, Canada; Kingston Penitentiar y, Kingston, Canada; 2 Uni v ersities of Carleton (Psychology) and Queen’s (Psychiatry), Canada; 3 Pittsburgh Institution, Pittsburgh, Ontario, Canada ABSTRACT Introduction Many professionals believe that self-report questionnaires used to predict recidivism have a low validity. The aim of the present study was to investigate the assumption that the validity of self-report is vulnerable to self-presentation biases in offender samples. Method The participants consisted of 124 male offenders who volunteered to complete the Self-Appraisal Questionnaire (SAQ). Results Lower scores on measures of social desirability were significantly associated with higher levels of risk (as measured by self-report and a rated actuarial instrument) and a higher likelihood to re-offend. Further, stepwise regression analysis revealed that social desirability added significantly unique variance in the prediction of violent recidivism. Discussion The authors propose that impression management may be an enduring person-based characteristic within an offender sample rather than a situationally determined response style. The variance associated with this characterological infor- mation is proposed to be the source of the unique predictive variance. Introduction Many forensic/correctional professionals believe that self-report questionnaires are either not valid when used to predict offender recidivism or that they have inferior validity as compared to professionally rated measures (Kroner and Loza, 2001). Specific concerns regarding self-report measures are their vulnera- bilit y to l y ing, manip u lation and self-presentation biases. The lack of confidence in self-report measures in predicting recidivism may be the result of a generali z ation of poor res ults obtained from research projects using the 140 Criminal Behaviour and Mental Health, 13, 140–150 2003 © Whurr Publishers Ltd