The Role of Defense Mechanisms in Borderline and Antisocial Personalities

Abstract

We examined whether borderline personality disorder (BPD) and antisocial personality disorder (APD) could be differentiated based on defense mechanisms as measured by observer (Defense-Q; MacGregor, Olson, Presniak, & Davidson, 2008) and self-report (Defense Style Questionnaire; Andrews, Singh, & Bond, 1993) measures. We conducted 2 studies whereby nonclinical participants were divided into borderline and antisocial groups based on scores from the Personality Assessment Inventory (Morey, 1991). Multivariate analysis of variance results revealed significant overall group differences in defense use. Univariate analyses further showed group differences on several individual defenses (e.g., acting out, denial, and turning against self). Together, the findings suggest that in BPD, the defenses may emphasize interpersonal dependency and a tendency to direct aggression toward the self; whereas in APD, the defenses may emphasize egocentricity, interpersonal exploitation, and a tendency to direct aggression toward others. Overall, this study demonstrates important differences in defense use between borderline and antisocial personality groups across both observer and self-report measures.

Acknowledgment

This study was conducted with the support of a Canada Graduate Scholarship awarded to M. D. Presniak by the Social Sciences and Humanities Research Council of Canada.

Notes

^a n = 34.

^b n = 38.

^c Reliability calculated using Cronbach's alpha. Reliability for rationalization is not calculated because it consists of one item.

¹After examining the variables for normal distribution, one variable (Projection) was significantly skewed (z > 3.33). We conducted a square root transformation, which resulted in a normal distribution. Because the results with the transformed score confirmed the results with the nontransformed score, the latter results were reported.

∗ p <.05.

∗∗ p <.001.

²Three variables (denial, projection, and splitting) were significantly skewed (z > 3.33) and we conducted transformations, resulting in a normal distribution. Because the results with the transformed scores confirmed the results with the nontransformed scores, the latter results were reported.

^a n = 34.

^b n = 49.

^c Defense-Q reliability calculated using one-way random effects model intraclass correlation coefficients.

^a n = 34.

^b n = 49.

^c Reliability calculated using Cronbach's alpha. Reliability for rationalization is not calculated because it consists of one item.

³For a correlation table of the convergence between the DSQ and Defense-Q defenses from this study, please contact M. D. Presniak.