Personality disorders are not granted the same clinical attention accorded Axis I disorders despite their instrumental role in treatment and outcome. Even when standardized assessments are used, their clinical utility may be limited by an overly favorable self-presentation. The current study focused on defensiveness, the intentional denial of symptomatology, by examining individuals’ ability to minimize their presentation on personality disorder diagnostic measures. Using a within-subjects simulation design, dually diagnosed inpatients were assessed under both honest and defensive conditions. The study used self-report (Structured Clinical Interview for DSM-IV – Axis II – Personality Questionnaire, SCID-II-PQ) and interview-based (Structured Interview for DSM-IV Personality, SIDP-IV) diagnostic measures and a self-report measure of favorable self-presentation (Paulhus Deception Scales, PDS). The inpatients were quite capable of hiding maladaptive personality traits on diagnostic measures, with similarly large effect sizes on both the SCID-II-PQ and SIDP-IV. In addition to the PDS, two new detection strategies for identifying defensiveness showed promise.

The current study explored the effects of simulated adjustment domains (i.e., partial denial and social desirability) on specialized substance abuse measures in a sample of 104 inpatient offenders with substance use disorders. Results indicated nearly two thirds (57.9%) of offenders successfully escaped detection on the SASSI-4 and InDUC-2R. Further, subtle approaches employed by the SASSI-4 failed to detect unacknowledged substance use by simulators, and the remaining decision rules were redundant and most likely contributed to the false positive rates for non SU offenders. More promisingly, the current study developed an empirically based validity scale to bolster the SASSI-4's effectiveness in detecting response styles that showed initial promise. Results, a review of detection strategies for SU validity scales, and implications for assessment are further explored.