Abstract
Borderline and antisocial personality features relate to multiple externalizing behaviors, including intimate partner violence (IPV). However, not all individuals with borderline and antisocial traits perpetrate IPV. The strength of the personality-IPV link may be related to problematic substance use. We examined borderline and antisocial personality features, problematic substance use, and IPV in a community sample of couples. Positive relations between both borderline and antisocial features and IPV were stronger in conditions of high problematic alcohol use relative to low problematic alcohol use. Alcohol misuse may be an important factor to consider for IPV reduction in men with these personality features.
Intimate partner violence (IPV) is a serious public health problem that affects millions of individuals around the world. IPV is defined as physical, sexual, or psychological harm perpetrated by a current or former romantic partner (Saltzman, Fanslow, McMahon, & Shelley, 2002). In the United States alone, nearly one in four women (22.3%) have been victims of severe physical violence perpetrated by a romantic partner in their lifetime (Breiding et al., 2011). Physical IPV victimization is associated with increased risk of generally poor health, depressive symptoms, substance use, and developing a chronic disease, chronic mental illness, and injury (Coker et al., 2002). Studies have found that borderline and antisocial personality features (Costa & Babcock, 2008) and substance use are important risk factors for IPV perpetration (Caetano, McGrath, Ramisetty-Mikler, & Field, 2005; Leonard, 1993; Stuart, Temple, & Moore, 2007). Problematic substance use and borderline and antisocial personality features may have an additive or synergistic impact on IPV perpetration. As such, the goal of the present study was to examine the extent to which alcohol and drug problems serve as potential moderators of the respective relations between both borderline and antisocial personality features and physical IPV.
Personality Features and IPV
Borderline and antisocial features are thought to characterize subtypes of men who are violent toward intimate partners (Holtzworth-Munroe & Stuart, 1994) and men with borderline and/or antisocial personality traits are likely to commit more serious violence and cause more injuries to their partners (Ehrensaft, Cohen, & Johnson, 2006; Ross & Babcock, 2009). Borderline personality disorder (BPD) and antisocial personality disorder (ASPD) both involve erratic behaviors that violate social norms and are impulsive (American Psychiatric Association, 2013). BPD is characterized by a lack of control of anger, intense and frequent mood changes, impulsive acts, disturbed interpersonal relationships, and life-threatening behaviors (American Psychiatric Association, 2013). Research suggests a link between BPD and externalizing behaviors, especially IPV (Costa & Babcock, 2008; Hines, 2008; Látalová & Praško, 2010; Liu et al., 2012; Sansone & Sansone, 2012). Upon real or imagined threats of abandonment, intimate partner abusers with borderline features may go to great lengths, such as physical abuse, to prevent their partners from leaving them (Ross & Babcock, 2009). These men may physically lash out at their partners in an attempt to regulate negative emotions when they become distressed (Keltner & Kring, 1998). ASPD is characterized by manipulation of others for personal gain and persistent disregard for and aggression toward others (American Psychiatric Association, 2013). Unlike the IPV perpetrators with borderline features, the violence committed by men with antisocial features may be due to deficiencies in empathy as opposed to abandonment fears (Bovasso, Alterman, Cacciola, & Rutherford, 2002). Intimate partner violent men with antisocial features may use violence as a means to resolve conflict with their partners and to maintain power and control in their relationship (Babcock, Jacobson, Gottman, & Yerington, 2000). Thus, these personality disorder features may not simply be correlates of, but rather causally related to men’s perpetration of IPV (Ehrensaft et al., 2006).
Mediators and Moderators of the Personality-IPV Link
Although the general connections between borderline and antisocial personality features and IPV are well established (Ross & Babcock, 2009), research on the specific mechanisms that explain or influence the strength of these relations is limited. One study found that diminished sensitivity to their female partners’ facial expressions of happiness mediated the relation between borderline personality features and IPV perpetration in men (Marshall & Holtzworth-Munroe, 2010). Biased social processing may underlie both the emotional dysregulation and interpersonal dysfunction evidenced in individuals with elevated borderline personality features (Jackson, Sippel, Mota, Whalen, & Schumacher, 2015). In fact, both emotional dysregulation and interpersonal dysfunction were found to mediate the borderline-IPV link in multiple longitudinal studies (Jackson et al., 2015; Newhill, Eack, & Mulvey, 2012; Stepp, Smith, Morse, Hallquist, & Pilkonis, 2012).
Beyond emotion and interpersonal variables, there is also evidence to suggest that substance misuse may be another important factor in the personality-IPV link (Trull, Sher, Minks-Brown, Durbin, & Burr, 2000). Intimate partner violent men who used alcohol or drugs were found to evidence both more borderline personality features and commit more severe IPV compared with those who did not use alcohol or drugs (Thomas, Bennett, & Stoops, 2013). In addition, there have been some investigations of the interplay between antisocial personality features and substance misuse insofar as how their combination relates to IPV. A multithreshold model has been set forth, which combines effects of alcohol use and antisocial personality characteristics to predict severe and nonsevere types of IPV. The model posits that IPV occurs when a perpetrator’s aggression threshold is exceeded, which occurs when the strength of the aggressive motivations exceeds the strength of the inhibitions. Aggressive inhibitions are presumed to be higher for severe violence than for minor aggression and also lower for men with ASPD relative to men without ASPD. Alcohol intoxication is hypothesized to lower aggression thresholds by impairing inhibitions through its effect on cognitive processing abilities (Stappenbeck & Fromme, 2010).
Klostermann, Mignone, and Chen (2009) conducted a latent class analysis on a sample of men who were entering an alcohol treatment program and perpetrated at least one prior incident of IPV. This analysis yielded two “classes” of male IPV perpetrators: Men in Class 1 were more likely to perpetrate violence more rapidly and severely relative to those in Class 2. Findings from this analysis also related to ASPD specifically, as Class 1 had more men diagnosed with ASPD than did Class 2 (Klostermann et al., 2009). Fals-Stewart, Leonard, and Birchler (2005) found that alcohol consumption increased the likelihood of severe violence, but not nonsevere violence, for men with ASPD. These studies provide a solid foundation for examining potential interactions between ASPD and alcohol misuse, but additional studies are warranted to analyze these relations more specifically (Jackson et al., 2015), as well as in the context of drug misuse.
Substance Use and IPV
Data are more abundant, however, on the direct link between substance use and IPV. Men are eight to 11 times more likely to perpetrate IPV on a day during which they have been drinking (Fals-Stewart, 2003) and the severity of alcohol problems predicts more severe IPV perpetration (Stuart et al., 2006). In a multiwave national study, IPV perpetration was found to be longitudinally predictive of increased risk of binge drinking and alcohol misuse among Caucasian men. However, among African American men, alcohol misuse was found to be longitudinally predictive of increased risk of IPV perpetration (Lipsky, Kernic, Qiu, Wright, & Hasin, 2014). Leonard and colleagues hypothesized that alcohol use is an important driving force underlying IPV and that drinking patterns influence IPV over the long term, while acute intoxication may influence individual violent events (Leonard, 1993, 2001; Leonard & Senchak, 1996).
There are several theories why individuals who drink may be more prone to violence. The disinhibition hypothesis is one way to account for the propensity for people who drink to commit violence (Gustafson, 1994). This theory asserts that alcohol weakens brain mechanisms that normally restrain impulsive behaviors, including inappropriate aggression, and impairs information processing. Thus, alcohol can also lead a person to misjudge social cues, and therefore overreact to a perceived threat (Gustafson, 1994). Alternatively, the self-medication theory suggests that individuals drink to offset the negative effects associated with major traumatic events, such as IPV perpetration (Testa, Livingston, & Leonard, 2003). The alcohol myopia theory (AMT) posits that alcohol’s pharmacological properties limit attention focus and reduce individuals’ abilities to process meaning from information they are able to perceive (Steele & Josephs, 1990). The attention-allocation model (AAM), one subcomponent of AMT, theorizes that as a result of the impairing effects of alcohol, intoxicated individuals allocate their attention in accordance with observing and processing only the most noticeable cues of situations, such as an insult from one’s partner, and exclude the less noticeable inhibitory cues, such as consequences of IPV (Steele & Josephs, 1988). Finkel’s (2007) I3 Theory, which involves (a) instigating, (b) impelling, and (c) inhibiting factors in aggression, posits that individuals will only perpetrate violence when the violence-impelling forces they experience at the time exceed the violence-inhibiting forces. Personality disorders that involve impulse control problems can be seen as an impelling factor and alcohol as an inhibiting factor, which set the stage for IPV perpetration. The aforementioned theories serve as possible frameworks through which the relation between alcohol misuse and violence, particularly IPV, can be conceptualized.
There are conflicting findings about the role of problematic drug use in IPV. While some studies found that problematic drug use is a stronger predictor of IPV than problematic alcohol use (Stuart et al., 2008), others found no relation between drug use and IPV perpetration (Cunradi, Caetano, & Schafer, 2002). Although Stuart et al. (2008) found a high rate of illicit substance use in men arrested for domestic violence, these findings appear to depend on the type of drug. Problematic marijuana use was not associated with IPV perpetration, but stimulant use was (Stalans & Ritchie, 2008). A meta-analysis on the relation between use of specific substances and IPV perpetration found that, regarding physical IPV perpetration, significant effects were observed for cocaine and opiates. However, the effects for sedatives, marijuana, stimulants, and hallucinogens were not statistically significant (Moore et al., 2008).
Substance Use and Personality Features
Both borderline and antisocial personality features commonly co-occur with substance misuse. A large nationally representative survey found that 50.7% of individuals with a lifetime diagnosis of BPD also met criteria for a substance use disorder over the previous 12 months (Grant et al., 2008). Lifetime alcohol misuse was observed among 49.2% of individuals, in a large nationally representative sample, meeting criteria for ASPD (Trull, Jahng, Tomko, Wood, & Sher, 2010). In addition, researchers have found that 23.4% of individuals who met criteria for ASPD also had a co-occurring drug use disorder (Trull et al., 2010). The common co-occurrence of both borderline and antisocial features and substance misuse could be due to overlapping symptoms, shared risk factors, or a causal connection (Robins, 1998; Trull et al., 2000). Personality features have been found to precede substance misuse (Trull, Waudby, & Sher, 2004) and alcohol use is associated with increased variability in negative affect among individuals diagnosed with BPD (Jahng et al., 2011). Moreover, some research suggests that substance misuse may relate to one’s attempts to regulate negative emotions (Tragesser, Sher, Trull, & Park, 2007). As both BPD and ASPD involve the lack of ability to regulate one’s emotions, problematic substance use may interact with these personality disorder features to influence IPV perpetration.
The Present Study
The present study examined alcohol problems and drug problems as potential moderators, such that problematic substance use would moderate the relation between borderline personality traits and IPV. Specifically, we predicted that in conditions of high alcohol problems and drug problems, there would be a stronger positive relation between borderline personality traits and IPV than in conditions of low alcohol problems and drug problems. In addition, it was hypothesized that problematic substance use would moderate the relation between antisocial personality traits and IPV. We predicted that in conditions of high alcohol and drug problems, there would be a stronger positive relation between antisocial personality traits and IPV than in conditions of low alcohol problems and drug problems. Previous studies in this area have referenced both diagnostic categories (e.g., BPD and ASPD) and dimensional borderline and antisocial personality features, and therefore, our literature review included both types of conceptualizations. However, we chose to measure borderline and antisocial personality features dimensionally and continuously in the present study. This procedure allows for examining the full range of personality symptomatology and increases statistical power. Categorizing participants based on the presence or absence of stringent, all-or-nothing personality disorder diagnoses would preclude such benefits, especially in a community sample not specifically recruited for personality disorder diagnoses.
Method
Participants
An ethnically diverse community sample of couples was recruited via local newspaper advertisements in a large southwestern United States city. The advertisements asked for “couples experiencing conflict.” Participants were recruited as part of a larger project investigating emotional reactivity and IPV. There were several inclusion criteria to participate. The couples were required to be married or living together as if married for at least 6 months prior to participation, at least 18 years of age, in a heterosexual relationship, and able to read and write English fluently. Interested couples were contacted by phone by trained undergraduate research assistants to be admitted into the study. Potential female participants were administered the physical assault subscale of the Revised Conflict Tactics Scale-2 (CTS2; Straus, Hamby, Boney-McCoy, & Sugarman, 1996) over the phone. For a couple to be eligible for participation, the female must have reported at least two acts of male-to-female physical violence in the preceding year. A sample of 135 male participants and their female partners was used in the present study. Average age for male participants was 32.0 years (SD = 10.2) and average age for female participants was 29.9 years (SD = 9.4). Participants reported an average annual family income of US$20,344 (SD = 34,079). The ethnic breakdown of the sample was as follows: 54.2% African American, 26.7% Caucasian, 14.5% Hispanic, 0.8% Asian, 0.8% Native American, and 3.1% Other.
Procedure
Data were collected over two separate 3-hr assessment periods. During the first session, male participants came into the laboratory to fill out paper-and-pencil questionnaires. During the second session, both male and female participants filled out paper-and-pencil questionnaires separately and were later reunited for a series of structured relationship interviews and marital interaction tasks. At the conclusion of participation, male and female participants were debriefed and received payment separately. Couples were paid US$90 to US$100 for their participation in both assessments.
Measures
IPV
Female participants reported on their male partners’ type, severity, and frequency of IPV by completing the CTS2 (Straus et al., 1996). The CTS2 is a 78-item questionnaire that measures instances of both male-to-female and female-to-male physical, psychological, and sexual aggression in the preceding year. Female participants responded to questions on five subscales: Negotiation, Psychological Aggression, Physical Assault, Sexual Coercion, and Injury. For the purposes of the present study, only the Physical Assault subscale was used as a measure of male-to-female IPV. Female participants responded to items assessing the frequency of physical assault on a 7-point Likert-type scale ranging from 0 (never) to 6 (20 times in the past year). Scores on items were summed to form a composite physical IPV score. Internal consistencies for the full CTS2 range from .79 to .95 and from .78 to .92 for the Physical Assault subscale specifically (Connelly, Newton, & Aarons, 2005; Straus et al., 1996). In the present study, Cronbach’s alpha for the Physical Assault subscale was .74.
Borderline personality traits
Levels of borderline personality traits were assessed using the 24-item borderline scale of the Personality Assessment Inventory (PAI-BOR; Morey, 1991). This scale measures the degree of borderline features evident in individuals. The PAI-BOR contains four subscales: Affective Instability, Negative Relationships, Identity Problems, and Self-Harm. Male participants rated the degree to which certain statements were true of them on a 4-point Likert-type scale ranging from 1 (false, not at all true) to 4 (very true). Scores on items were summed to form a composite PAI-BOR score. Raw scores of 60 or higher (T scores of 92 or higher) on the PAI-BOR are considered clinically significant. The PAI-BOR maintains adequate internal consistency: Cronbach’s alpha = .93 in a community sample (Gardner & Qualter, 2009; Morey, 1991). In the present study, Cronbach’s alpha for the PAI-BOR was .78.
Antisocial personality traits
Levels of antisocial personality traits were assessed using the 24-item antisocial scale (PAI-ANT) of the PAI (Morey, 1991). This scale measures the degree of antisocial features evident in individuals. The PAI-ANT contains three subscales: Antisocial Behaviors, Stimulus-Seeking, and Egocentricity. Male participants rated the degree to which certain statements were true of them on a 4-point Likert-type scale ranging from 1 (false, not at all true) to 4 (very true). Scores on items were summed to form a composite PAI-ANT score. Raw scores of 42 or higher (T scores of 82 or higher) on the PAI-ANT are considered clinically significant. The PAI-ANT maintains adequate internal consistency: Cronbach’s alpha = .84 for men (Morey, 1991). In the present study, Cronbach’s alpha for the PAI-ANT was .74.
Alcohol problems
Levels of alcohol problems were assessed using the 12-item alcohol problems scale (PAI-ALC) of the PAI (Morey, 1991). This scale measures the degree of alcohol problems within an individual as a function of behaviors and consequences associated with the use, abuse, and dependence on alcohol (Morey, 2007). Scores on items were summed to form a composite PAI-ALC score. The PAI-ALC has been validated against other robust measures of alcohol misuse, including the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association, 1994) Axis I Disorders, Clinician Version (SCID-IV; First, Spitzer, Gibbon, & Williams, 1997; Magyar, Edens, Lilienfeld, Douglas, & Poythress, 2011). In the present study, Cronbach’s alpha for the PAI-ALC was .80.
Drug problems
Levels of drug problems were assessed using the 12-item drug problems scale (PAI-DRG) of the PAI (Morey, 1991). This scale measures the degree of drug problems within an individual as a function of behaviors and consequences associated with the use, abuse, and dependence on drugs (Morey, 2007). Scores on items were summed to form a composite PAI-DRG score. The PAI-DRG has been shown to moderately to highly correlate with other robust measures of drug misuse, including the Drug Abuse Screen Test (Kellogg et al., 2002) and the Addiction Severity Index (Parker, Daleiden, & Simpson, 1999; Magyar et al., 2011). In the present study, Cronbach’s alpha for the PAI-DRG was .68.
Personality disorder
Although the present study analyzed dimensional and continuous borderline and antisocial personality traits, the BPD and ASPD sections of the Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II; First, Gibbon, Spitzer, Williams, & Benjamin, 1997) were also administered by graduate students trained in SCID assessments to all male participants as part of the larger project on emotional reactivity and IPV. A second trained coder independently scored 25% of the interviews from videotapes. Interrater agreement between the scorers was 80% for BPD and 82% for ASPD diagnoses.
Data Analytic Strategy
To test the respective moderating effects of alcohol problems and drug problems on the links between both borderline personality traits and antisocial personality traits and IPV, four separate multiple regression analyses were performed. All predictor (borderline personality and antisocial personality) and moderator (alcohol problems and drug problems) variables were centered by subtracting their means. Subsequently, interaction terms were created and steps for establishing moderation were followed (Frazier, Tix, & Barron, 2004). All regressions were conducted by entering predictor variables followed by the appropriate interaction variable (Borderline Personality × Alcohol Problems, Borderline Personality × Drug Problems, Antisocial Personality × Alcohol Problems, or Antisocial Personality × Drug Problems). All interaction terms were explained graphically by computing a predicted value of IPV for those who were one SD above and below the mean on the predictor and moderator variables (Frazier et al., 2004).
Results
Means and standard deviations for all main study variables are presented in Table 1. Preliminary results indicated that females reported on average 16.24 (SD = 23.05) acts of physical violence from their male partners in the past year. The physical assault score distribution demonstrated a high standard deviation relative to the mean, suggesting skewness. For this reason, an outlier analysis was conducted for this variable, according to the recommendations of Field (2013). The physical assault variable was first standardized, and z-scores were examined. Although originally intended for normally distributed data, this procedure also remains robust with nonnormal distributions. According to Field (2013), a data point is considered an outlier if the z-score value exceeds an absolute value of 3.29. In the present study, two participants had standardized physical assault scores that exceeded this absolute value (3.37 and 3.63, respectively). One of these participants met criteria for BPD according to the SCID-II, and demonstrated subclinical elevations on the PAI-ANT. The other participant met criteria for both BPD and ASPD on the SCID-II and demonstrated subclinical levels on the PAI-BOR and PAI-ANT. However, further analyses in the present study excluded these outliers, leading to a final sample of 133 couples. Men reported a mean borderline personality features (PAI-BOR) score of 25.41 (SD = 10.70) and a mean antisocial personality features (PAI-ANT) score of 23.67 (SD = 10.78). In terms of the moderator variables in this study, men reported a mean alcohol problems (PAI-ALC) score of 7.95 (SD = 6.67) and a mean drug problems (PAI-DRG) score of 9.83 (SD = 7.13). See Table 1. Examination of the SCID-II revealed that 4.51% (n = 6) met criteria for BPD only, 13.64% (n = 17) met criteria for ASPD only, and 10.53% (n = 14) met criteria for both BPD and ASPD. None of the six participants who met SCID-II criteria for BPD were clinically elevated on the PAI-BOR. However, two out of six (33.33%) had scores in the subclinical range. Similarly, none of the 17 participants who met SCID-II criteria for ASPD were clinically elevated on the PAI-ANT. However, four out of 16 (23.53%) had scores in the subclinical range.
Descriptives and Correlations of Main Study Variables.
Note. IPV = Physical Assault subscale of CTS2; N = 133.
p < .05. **p < .01.
The bivariate correlations among all main study variables are presented in Table 1. Correlations between IPV and borderline personality (r = .19), alcohol problems (r = .20), and drug problems (r = .24) were statistically significant. Both borderline personality and antisocial personality were significantly related to alcohol problems (r = .38 for borderline and r = .39 for antisocial), drug problems (r = .49 for borderline and r = .49 for antisocial), and were significantly related to one another (r = .62). In addition, alcohol problems and drug problems were significantly related to one another (r = .51), further supporting the high instance of co-morbidity. Analyses revealed that IPV was not significantly related to demographic variables, including age, race, education level, and income. However, we controlled for race by entering it into the multiple regressions as a covariate, given the aforementioned differential effects among Caucasian versus African American men in terms of IPV perpetration suggested in Lipsky et al. (2014). In the present study, results of a one-way analysis of variance (ANOVA) revealed that there were no significant differences in IPV scores between racial groups of male participants, F(5, 128) = .60, p = .70.
Multiple Regression Analyses
Alcohol Problems was proposed to moderate the relation between borderline personality traits and IPV. Specifically, we hypothesized that conditions of high alcohol problems would predict a higher frequency of IPV than conditions of low alcohol problems. Results of the first multiple regression suggested that there was a significant interaction between borderline personality traits and alcohol problems in predicting frequency of IPV (β = .21, p < .05), see Table 2. A graphical representation comparing those who scored one SD above and below the mean on alcohol problems suggested that when alcohol problems is high, borderline personality traits are more strongly related to IPV than when alcohol problems is low (see Figure 1). Drug Problems was proposed to moderate the relation between borderline personality traits and IPV. Specifically, we hypothesized that conditions of high drug problems would predict a higher frequency of IPV than conditions of low drug problems. Results of the second multiple regression suggested that there was not a significant interaction between borderline personality traits and drug problems in predicting frequency of IPV (β = .10, p = .35).
Four Multiple Regressions Examining Relations Between Predictor and Moderator Variables.
p < .05.

Alcohol problems and borderline personality interaction on frequency of IPV.
Alcohol Problems was proposed to moderate the relation between antisocial personality traits and IPV. Specifically, we hypothesized that conditions of high alcohol problems would predict a higher frequency of IPV than conditions of low alcohol problems. Results of the third multiple regression suggested that there was a significant interaction between antisocial personality traits and alcohol problems in predicting frequency of IPV (β = .23, p < .05), see Table 2. A graphical representation comparing those who scored one SD above and below the mean on alcohol problems suggested that when alcohol problems is high, antisocial personality traits are more strongly related to IPV than when alcohol problems is low (see Figure 2). Drug Problems was proposed to moderate the relation between antisocial personality traits and IPV. Specifically, we hypothesized that conditions of high drug problems would predict a higher frequency of IPV than conditions of low drug problems. Results of the second multiple regression suggested that there was not a significant interaction between antisocial personality traits and drug problems in predicting frequency of IPV (β = .20, p = .06).

Alcohol problems and antisocial personality interaction on frequency of IPV.
Due to the inconsistency in SCID-II BPD and ASPD diagnoses and respective clinical elevations on the PAI evidenced in the present study, we also conducted the same four multiple regressions as above, but with BPD and ASPD diagnoses as statistical predictors instead of PAI-BOR and PAI-ANT scores. There were no significant interactions between BPD or ASPD and problematic alcohol or drug use. Although the presence of BPD and ASPD diagnoses were significantly positively correlated with frequency of IPV (r = .19 for BPD and r = .30 for ASPD), these relations were not moderated by alcohol or drug problems.
Discussion
The present study was designed to increase our understanding of the role alcohol and drug problems have in the relation between borderline and antisocial personality features and IPV. The key findings from this study suggest that problematic alcohol but not drug use appears to moderate the personality features and IPV link. Problematic alcohol use was found to be a moderator of the relation between borderline personality features and IPV perpetration. In addition, problematic alcohol use was found to be a moderator of the relation between antisocial personality features and IPV perpetration. Considering participants in the present study were screened for at least two acts of male-to-female IPV, results suggest that the combination of problematic alcohol use and borderline and antisocial features may increase the severity of physical IPV in those already experiencing some physical IPV. Contrary to what we hypothesized, problematic drug use was not found to be a moderator of the relations between borderline and antisocial personality features and IPV perpetration. While some studies have found significant relations between drug misuse and IPV perpetration (Moore et al., 2008; Stuart et al., 2008), others have not (Cunradi et al., 2002). Violence risk depends on the class of drug (Stuart et al., 2008) and drug type was not assessed in the present study. In addition, the lack of a significant effect for problematic drug use could be due to the limited number of illicit drug users in the sample of the present study. However, there were no significant moderating effects of problematic alcohol use in the relation between BPD or ASPD diagnoses and IPV perpetration. Considering participants in the present study were recruited from the community and not an inpatient setting, in which greater severity of psychopathology would be expected, variability in individuals meeting full criteria for BPD or ASPD was limited. It is also likely that more statistical power is present when examining a moderator variable in relation to a continuous variable (e.g., PAI scores), rather than a categorical one (e.g., SCID-II diagnoses).
Alcohol may facilitate violence especially in men with borderline and/or antisocial personality features because it disinhibits their tendencies toward erratic and impulsive behavior (Peterson, Rothfleisch, Zelazo, & Pihl, 1990). One of the most commonly cited impairments in BPD is the uncontrolled use of aggression (Látalová & Praško, 2010). Applying the I3 Theory (Finkel, 2007), an individual with borderline personality features will tend to exhibit deficits in controlling aggression (impelling factor), and the addition of alcohol (inhibiting factor) may influence IPV perpetration. This IPV perpetration may occur since the individual’s violence-impelling factor of uncontrolled use of aggression exceeds the violence-inhibiting forces, as alcohol would likely reduce inhibition of aggression. I3 Theory provides context for the strengthening the influence alcohol misuse may have in the relation between borderline personality features and IPV.
Furthermore, individuals with antisocial personality features already experience deficits in higher reasoning or executive brain functions (Giancola & Moss, 1998), thereby suggesting that they may be more vulnerable to the aggression-inducing effects of alcohol. In addition, AMT posits that rather than creating disinhibiting effects, alcohol produces a myopia effect, which causes intoxicated individuals to respond almost completely to their immediate environment rather than thinking about future consequences or regulating their reactive impulses. Alcohol can cause changes in brain chemistry, such as altering activities of neurotransmitters including gamma-aminobutyric acid (GABA) and serotonin. Both serotonin and GABA have been linked to aggressive behaviors (Moeller & Dougherty, 2001). Individuals with borderline personality and/or antisocial personality features may already be predisposed to have abnormalities in GABA and serotonin levels. Alcohol can exacerbate these problems and potentially lead individuals with these personality features to engage in aggressive behavior (Moeller & Dougherty, 2001). Although the present study could not address such causal mechanisms responsible for violence, results nonetheless suggest there is reason to believe the interaction of borderline and antisocial personality features and problematic alcohol use is important in the context of IPV.
Understanding moderators of violence risk may be useful in developing targeted interventions for IPV. Existing interventions for IPV stop violence recidivism by an additional 5% over and above the effects of arrest alone (Babcock, Green, & Robie, 2004). Given the limited effectiveness demonstrated by current approaches for IPV intervention (Babcock et al., 2004; Eckhardt et al., 2013), it is critical to identify other problematic traits or behaviors, which may be more amenable to specialized intervention strategies and can potentially affect IPV perpetration beyond the effects of personality features alone. Men with emotional dysregulation or impulsivity problems may be prone to violence, especially against an intimate partner, when intoxicated. As such, problematic alcohol use may deserve greater emphasis in current treatment protocols. Especially if perpetrators show elevated signs of borderline and/or antisocial personality features, addressing alcohol reduction may be an even more compelling source for intervention. Considering the relative stability of the presence of personality features, alcohol consumption may be a more modifiable target. Even if alcohol use is not associated with IPV perpetration for a specific individual, less problematic use will allow perpetrators to more thoroughly apply other skills learned in treatment to improve their relationship functioning and reduce violence.
Limitations and Future Directions
Although results suggest vital points of emphasis for future research and clinical work, the present study is not without limitations. First, the data used in this study were obtained cross-sectionally and based exclusively on self-report (with the exception of female report of male violence). Therefore, no causal relations between borderline and antisocial personality features, problematic substance use, and IPV could be noted. Future studies would benefit from examining the relations among these variables longitudinally. Second, we did not corroborate the participants’ reports of their own substance use or personality traits, so the data are limited by the accuracy of the perpetrators’ reports. Individuals might have not been honest with regard to their alcohol or drug use. Third, all illicit drugs were classified together in the assessments, and therefore, we did not look at the differences between different drug classes and how different types of drugs may differentially affect the relations between borderline and antisocial personality features and IPV. Fourth, although the PAI is a widely used and well-validated measure with adequate psychometric properties, all personality and problematic substance use variables in the present study were measured using this instrument. As such, there is a possibility that our results are at least partially accounted for by within-measure variance. Future studies would benefit from measuring these personality features and problematic substance use with a greater variety of measurement instruments.
Finally, the rates of men who met criteria for a formal diagnosis of BPD, ASPD, or both in the community sample used in the present study were relatively low (5.30%, 13.64%, and 11.36%, respectively). As such, using personality disorder diagnoses, as opposed to dimensional traits, as predictor variables in the present study was likely not feasible due to a floor effect. However, the moderating effects of alcohol and/or drug problems may be stronger in samples in which severe psychopathology is more common (e.g., inpatient, forensic, court-mandated). Furthermore, there was little consistency between BPD and ASPD diagnoses and corresponding clinical elevations on the PAI in the present study. Although the PAI subscales are not meant to directly map onto psychiatric diagnoses, the discrepancy between self-report and structured clinical interview measures is a concern.
Despite these limitations, the data from this study lend support to the idea that there are important moderators to take into account between certain personality features and IPV perpetration frequency. Overall, it was found that the connections between borderline and antisocial features and IPV were strengthened by problematic alcohol use. The results from the present study suggest that IPV treatment programs may benefit from greater attention to substance misuse and personality features than currently exists. Education about how using alcohol to regulate negative emotions may backfire, particularly for individuals with borderline or antisocial personality features, may be useful to incorporate into battering intervention programs. In summary, the results highlight the importance of taking a nuanced view of risk factors for IPV to expand evidence-based treatment options.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was funded by Grant R03 MH066943-01A1 from the National Institutes of Health and by the University of Houston.
