Gender bias in diagnostic criteria for personality disorders : an item response theory analysis

Some personality disorders are diagnosed more frequently in men and others more frequently in women. It is important to determine if more men or more women truly experience these disorders, or if bias in the diagnostic criteria are responsible for producing illusory gender differences. Participants represent a subset of people drawn from the Peer Assessment of Personality Traits and Pathology Project, specifically the 599 individuals who were interviewed. The sample includes 433 Air Force recruits, 256 men and 177 women ranging in age from 17 to 35 (mean= 19.8 sd = 2.4 years) and 166 undergraduates, 59 men and 107, women most of whom were 18 years of age. Interviews were conducted using the Structured Interview for DSM-IV Personality. Item Response Theory (IRT) analysis using differential item functioning (DIF) was applied to the items. Six of the diagnostic criteria are biased, meaning that at each latent trait level the endorsement of the criteria are more likely to predict gender than personality pathology. Four of the criteria are biased against men, these occurred in paranoid and antisocial personality disorder. Two of the criteria are biased against women, these occurred in schizoid personality disorder. The criteria for schizotypal, borderline, histrionic, narcissistic, avoidant, dependent and obsessive-compulsive personality disorders do not demonstrate bias. Gender differences do exist in personality disorder diagnoses, some of which are affected by DIF. In general, diagnostic criteria that emphasize aggressive action are more likely to be biased against men and criteria involving solitude are more likely biased against women. More research on the application of IRT methods to personality measures is needed before conclusions are drawn about the DSM-IV criteria.