Investigating gender-based differential item functioning on the McLean Screening Instrument for Borderline Personality Disorder (MSI-BPD): An item response theory analysis.

Abstract

The McLean Screening Instrument for Borderline Personality Disorder (MSI-BPD) is a popular screening tool for identifying people who may have borderline personality disorder (BPD). However, because women are more frequently diagnosed with the disorder than men, it is possible that the MSI-BPD differs in its ability to identify BPD as a function of gender identity. Using item response theory (IRT), we sought to determine if components of the MSI-BPD would demonstrate differential item functioning (DIF), such that one gender identity would be more likely to endorse certain items. Twenty-two thousand thirty-five college undergraduates (14,305 women) aged 18-55 years (M = 18.77, SD = 1.75) were assessed using the MSI-BPD as part of a subject pool screening between 2008 and 2019. The MSI-BPD contains 10 items that are measured dichotomously, and the authors recommend a cut-off of 7 of 10 items endorsed to maximize sensitivity and specificity to BPD. Results suggested that a two-parameter model was the best fit to the data and that unidimensionality and local independence assumptions were met. The following items demonstrated DIF: self-harm/suicidality, affective lability, abandonment, impulsivity, and anger. At equal levels of the latent construct of BPD, women were more likely to endorse self-harm/suicidality, affective lability, and abandonment. Women were more likely to endorse impulsivity at higher levels of BPD, and men were more likely to endorse anger at lower levels of BPD. Ultimately, the effect sizes of these differences were small, however, and likely do not impact the individual's overall outcome on the measure. Public Significance StatementThis study evaluates potential measurement bias in the MSI-BPD, a common screening measure for BPD, on the basis of gender identity, using IRT. Results indicate that although some items demonstrate bias toward men and women, the overall effect is small and the tool should be considered equally valid for men and women.