A psychometric evaluation of the DSM-IV personality disorder criteria

Abstract

In this study, we examined the psychometric quality of the recently revised DSM-IV Personality Disorders (PD). A nationally drawn sample of mental health professionals rated 280 patients known to them, using a symptom checklist containing all the DSM-IV PD criteria. From this symptom level data we determined internal consistency, convergent validity, and discriminant validity for each DSM-IV PD criteria. Overall, the findings were encouraging. The internal consistency--a measure of reliability--of the PD criteria sets appears to be substantially improved. The median coefficient alpha for the DSM-IV PDs was .73. Convergent validity was generally adequate for the DSM-IV PD criteria sets. However, discriminant validity, a measure of the specificity of a criterion's relationship to its parent scale, continues to be a problem of the PDs. The findings suggest that the DSM-IV PDs may have better reliability than did their DSM-III-R predecessors. However, the weak discriminant validity found for the DSM-IV PD criteria sets indicates that the Axis II system will continue to show high levels of comorbidity. The usefulness of psychometric procedures in the development and refinement of the DSM PDs is also highlighted.