Interpersonal Assessment of Borderline Personality Disorder: Preliminary Findings

Abstract

We examined the reliability and validity of scores on an interpersonal measure of borderline personality disorder (BPD). Ratings on the Interpersonal Measure of Borderline Personality Disorder (IM–B) were based on nonverbal behaviors and interpersonal interactions occurring during clinical interviews with 276 adults. Scores on the IM–B exhibited good reliability. IM–B scores also displayed expected patterns of associations with scores on other measures of BPD, as well as with scores on measures of affective dysfunction, interpersonal pathology, and behavioral impairment associated with BPD, including indexes of maladaptive emotion regulation, interpersonal sensitivity, and self-harm. The pattern of associations for IM–B scores was quite similar to what would be expected for a dimensional measure of BPD symptoms. Scores on the IM–B were also associated with symptoms of disorders generally comorbid with BPD. Finally, IM–B scores contributed incrementally to the prediction of interpersonal dysfunction and suicidal ideation and behavior. Discussion focuses on implications for the assessment of BPD.

Acknowledgments

We wish to thank Heather Hall, Gregg Adamo, Moria Smoski, and Jennifer Cheavens for help with data collection during this project. We are grateful to the following individuals who made suggestions about nonverbal behaviors and interpersonal interactions thought to be characteristic of individuals with BPD: Risto Antikainen, Donald Black, Jen Cheavens, Jill Compton, Janine Flory, Leslie Horton, Lorne Korman, Valerie Porr, Thomas Widiger, Mary Zanarini, and others who attended the New Directions in Borderline Personality Disorder workshop sponsored by the National Institute of Mental Health. We also wish to thank the students who agreed to score the Interpersonal Measure of Borderline Personality Disorder from videotaped interviews to assist in our evaluation of interrater agreement—Brittany Beel, Yessie Castillo, Whitney Dickinson, Billy Dwight, Hermes Hernandez, Lily House, Brian Levine, and Coreena Taylor—and Marcus Rodriguez, who assisted with the tabulation of internal consistencies.

Notes

¹ A t test indicated that the subset of participants that completed the BAI (n = 101) and the subset of participants that completed the PDS and PSS (n = 52) were younger on average than the subset that did not complete these measures, t(250.84) = 2.14, p =.03, t(272) = 2.29, p =.02, respectively. (The degrees of freedom for the BAI analysis were corrected due to a significant Levene test.) Although these subsets of participants did not differ in the proportion of males and females, the proportion of African Americans versus European Americans, or in years of education completed, and the effect sizes for these age differences were relatively small, d =.25, d =.35, these differences raise the possibility that the age of participants might have influenced relationships involving these measures. Consequently, we computed partial correlations between IM–B ratings and scores on these variables, controlling for age. The resulting partial correlations were very similar to the zero-order correlations reported in Table 3: for BAI = partial r =.45; for PDS Impact score, partial r =.56; for PDS Severity score, partial r =.56; for PSS total score, partial r =.45, all ps <.001.