Abstract
Construct validity problems in contemporary personality disorder (PD) symptoms and diagnoses include poor interrater reliability, high comorbidity, limited discriminant validity, and inadequate criterion validity. These problems are due in part to a limited range of outcome measures and research methods utilized by PD investigators: The vast majority of PD studies use correlational methods to assess relationships among questionnaire- and interview-based self-reports. To strengthen the PD database, researchers must: (1) emphasize behaviorally referenced criterion validation of PD symptoms; and (2) explore PD dynamics using manipulations that alter PD-related psychological processes and produce measurable changes in behavior. Frameworks for conducting behaviorally referenced PD symptom validation studies and experiments are outlined. The advantages of a behaviorally referenced experimental approach to PD research are described, along with possible objections to this approach.