Introduction

Men who abuse their intimate partners respond differently than nonviolent men to relationship conflict. Intimate partner abusers attribute negative intent to wives’ behaviors that elicit jealousy, rejection, abandonment, and potential public embarrassment (Dutton and Browning 1988; Holtzworth-Munroe and Hutchinson 1993). Intimate partner abusers have also been shown to articulate more general irrational beliefs and cognitive distortions than nonviolent groups when angered via an articulated thoughts in simulated situations (ATSS) paradigm (Barbour et al. 1998; Eckhardt et al. 1998). Given these differences, Eckhardt et al. (1998) suggest that research examine whether articulated cognitive distortions differ among different subtypes of intimate partner abusers.

The Holtzworth-Munroe and Stuart (1994) typology categorizes intimate partner abusers along three dimensions: severity of relationship abuse, use of abuse in and outside the relationship, and psychopathology as evidenced by antisocial and borderline traits. Empirical clustering analysis of these variables yields three (Waltz et al. 2000) or four subtypes of abusers (Holtzworth-Munroe et al. 2000). The content of articulated cognitive distortions may vary for specific subtypes of intimate partner abusers. For example, among those with borderline personality features, cognitive distortions depicting jealousy and fears of abandonment may be articulated, especially in scenarios that elicit abandonment fears. On the other hand, among those with antisocial or psychopathic features, cognitive distortions reflecting anger, verbally abusive controlling behavior, and rigid sex-roles may emerge. No research to date has examined angry articulations of specific cognitive distortions in intimate partner abusers with personality disorder features.

Individuals with antisocial and borderline features might differ in terms of their articulated thoughts, especially in response to socially relevant stressors. Intimate partner abusers with antisocial features are thought to have the most rigid attitudes about sex-roles as assessed with questionnaires (Holtzworth-Munroe and Stuart 1994). They may use violence in order to maintain power and control (Babcock et al. 2000). They may also use emotionally aggressive communication styles while engaging in a conflict discussion with their partners (Gottman et al. 1995). Thus, intimate partner abusers with antisocial features may articulate sexist beliefs and verbally aggressive statements reflecting power and control in response to the ATSS.

Individuals with borderline features have mood instability, with periods of dysphoria and anger. They may avoid real or imagined abandonment that can result in a dependency upon others manifested as jealousy (Heard and Linehan 1993). Thus, intimate partner abusers with borderline features may articulate anger and dysphoria, as well as jealousy and fear of abandonment as compared to those with antisocial personality features.

However, borderline and antisocial personality disorders are highly comorbid (Widiger and Corbitt 1997), making it difficult to make differential predictions as to their correlates. Using psychopathy as a measure may elucidate differences between borderline and antisocial personality disorders. Psychopathy is comprised of two distinct factors: interpersonal features and psychopathic conduct (Cleckley 1941). The interpersonal features reflect callous/unemotional disregard, which are in sharp contrast to the dependency and emotional lability of individuals with borderline features. Individuals with the psychopathic interpersonal features are likely to not show anger and dysphoria. The other factor, psychopathic conduct is comprised of criminality, similar to symptoms of antisocial personality disorder (APA 2000).

The current study explored specific interpersonal and affective cognitive distortions that may be differentially relevant to intimate partner abusers with personality disorder features. The current investigation employed an interpersonally relevant, standardized anger induction task using the ATSS. The use of such a task permitted a naturalistic examination of cognitive distortions in two specific contexts (flirtation and criticism) shown to elicit anger (Eckhardt et al. 1998). Our research questions were two-fold: Do intimate partner abusers articulate different cognitive distortions in response to anger inducing relationship scenarios as compared to nonviolent men? Does the articulation of certain cognitive distortions correlate with borderline and antisocial personality disorder features and other variables relevant to the Holtzworth-Munroe and Stuart (1994) typology?

The current study predicted that intimate partner abusers would articulate greater cognitive and affective distortions as compared to distressed and satisfied/nonviolent men during the ATSS, replicating Eckhardt and colleagues’ work. To extend their findings, the variables relevant to the Holtzworth-Munroe and Stuart (1994) typology were examined dimensionally, correlated with affective and cognitive distortions. Specifically, it was expected that borderline features would positively correlate with anger, dysphoria, jealousy, and abandonment fears in response to a flirtation scenario. Antisocial features were expected to positively correlate with anger, power and control, and sexism in reaction to the criticism scenario. The interpersonal features of psychopathy were expected to correlate negatively with articulated anger and dyshporia. By contrast, psychopathic conduct was predicted to function the same way as antisocial personality features. General violence and intimate partner abuse frequency were also hypothesized to be positively correlated with articulations of power and control, sexism, and anger.

Method

Participants

A sample of 184 couples from the Houston community was recruited to participate as part of two larger studies. Men were classified as intimate partner abusive (IPA; n = 130) on the basis of either partner reporting an act of male to female abuse in the past year on the Conflict Tactics Scale-2 (CTS-2; Straus et al. 1996). Couples were classified as nonviolent (n = 48) if both partners reported no act of male-to-female physical abuse on the CTS-2 during the past year and no serious violence ever. Additionally, nonviolent men who scored less than 115 on the Dyadic Adjustment Scale (DAS; Spanier 1976), were classified as Distressed/Nonviolent (DNV; n = 27). Nonviolent men who scored greater than 115 on the DAS (n = 21) were classified as Satisfied/Nonviolent (SNV).

Men’s average age was 31 (SD = 9.78). Average family income was approximately $35,277 (SD = $19,033). Average education level attained was graduating from high school. A one-way MANOVA revealed no significant differences on demographics, F(6,171 = 1.90, n.s. There were, as expected, differences on relationship dissatisfaction F(2,175) = 22.38, p < 0.001, with the IPA group (M = 101.91, SD = 18.02) scoring lower on the DAS than the SNV group (M = 127.39, SD = 7.29), t(175) = 6.79, p < .0001. The IPA and the DNV couples did not differ in terms of relationship distress, t(175) = 0.43, n.s.

Measures

Conflict Tactics Scale-2 (CTS-2)

The physical assault subscale of the CTS-2 was used to determine final group membership and as a correlate of articulated themes. Higher scores on the physical assault subscale represent a greater frequency of physical abuse. Internal consistency of the physical assault subscale was.89 in our dataset.

Generality of Violence

A ten question self-report measure adapted from Gottman et al. (1995) was utilized to assess for antisocial behavior including criminal activity and the use of aggression outside of the intimate relationship. Higher scores represent a greater breadth of criminal conduct and physical aggression towards individuals outside of the intimate relationship. Cronbach’s alpha in our dataset was 0.70.

Millon Clinical Multiaxial Inventory-III (MCMI-III)

The antisocial subscale of the MCMI-III (Millon 1994) and the borderline subscale were administered to assess for psychopathology. Items were weighted as outlined in the manual (Millon 1994). Reliability of the antisocial scale was α = 0.83 and α = 0.76 for the borderline scale.

Self-Report of Psychopathy

The Self Report of Psychopathy-II (SRP-II) (Hare et al. 1989) is a 59-item self-report derived from Hare’s (1991) Psychopathy Checklist-Revised (PCL-R). The SRP-II captures both the interpersonal and antisocial features of psychopathy. Interpersonal features include the personality characteristics of psychopaths such as callousness (psychopathy Factor 1). Psychopathic conduct includes social deviance such as impulsivity (psychopathy Factor 2). Reliability of the SRP-II Interpersonal features in the current dataset was α = 0.44 and α = 0.78 for Psychopathic Conduct. Because the internal consistency was unacceptable for the Interpersonal features scale, three items were excluded, raising the internal consistency to a marginal α = 0.61.

State-Trait Anger Expression Inventory (STAXI)

Ten items composing the state anger scale of the STAXI (Spielberger 1988) was administered to measure the intensity of currently experienced angry feelings. Higher scores represent greater state anger. A computerized version of the state anger scale was completed both pre- and post-anger induction. In attempt to conceal the purpose of the anger induction, 18 items were administered in a random order, interspersed with eight detractor statements assessing emotions besides anger. Reliabilities for state anger averaged α = 0.82.

Imagination Experience Questionnaire

Participants were administered a two-item, project designed questionnaire about the ability to mentally engage in the ATSS. Participants were asked to rate on a four-point scale the degree to which they could imagine themselves in scenarios. Cronbach’s alpha in the current study was 0.63.

Dyadic Adjustment Scale (DAS)

The DAS is a 32-item self-report that assesses marital satisfaction. Internal consistency on the DAS was 0.57 in the current dataset.

Articulated Thoughts in Simulated Situations Paradigm (ATSS)

The present study used a modified version of the ATSS procedure (Davidson et al. 1983) in order to reduce confounds of frequent speaking and listening demands on psychophysiological recording. Thus, participants articulated aloud for one minute, only at the end of each scenario. Participants were asked to imagine that their partner was the protagonist in the situations being described. At the end of each scenario, participants were asked to say aloud their thoughts and feelings. They were first presented with the flirtation scenario depicting the man overhearing a conversation between his imagined female partner and a male acquaintance subtly flirting. Participants were next presented with the criticism scenario depicting their imagined partner and her female friend criticizing his job, performance in bed, family and friends, and intelligence.

Anger Rating Dial

Men were asked to use a dial to indicate their anger while listening to the ATSS scenarios. Responses were continuous, with anchor points of neutral, angry, very angry, enraged, and violently angry marked on the dial face.

Coding System for Articulations

Transcriptions were coded using a project-designed coding manual (Costa 2004), developed in order to determine the thematic content of interpersonal cognitive distortions thought by several theorists (e.g. Dutton 1995; Pence and Paymar 1993; Sonkin et al. 1985; Wexler 2000) to be relevant to intimate partner violent men. Transcribed articulations were rated dimensionally on 14 items along five-point scales, which were summed into six categories: dysphoria, anger, sexism, jealousy, power and control, and abandonment fears. Items were rated with 0 indicating “not present” and 5 meaning “a great amount.” Five undergraduates were trained to reliability. Reliability was established by double-coding 25% (n = 44) of the scenarios, and calculating intraclass correlation coefficients (ICC) between the coders and the developer of the coding scheme.

The dysphoria subscale is comprised of articulations indicating sadness and disappointment. ICC for dysphoria averaged 1.00. Anger was coded for articulations and nonverbal behavior communicating being “pissed off,” furious, or enraged. ICC for the anger subscale averaged 0.88. Sexism assessed for perceptions of being an authority figure and perceived ownership (ICC = 0.53). Power and control comprised a range of verbal abuse (ICC = 1.0). Abandonment Fear was coded for dependency and feelings of helplessness or hopelessness resulting from partner abandonment (ICC = 1.00). Jealousy assessed for perceiving the protagonist as wanting attention or assuming infidelity (ICC = 0.94).

Procedure

Participants were recruited through local newspaper advertisements and flyers requesting “couples experiencing conflict” or “couples needed.” Couples had to be 18 years of age, married or living together as if married for at least 6 months, heterosexual, and able to speak and write English fluently. The study was conducted as part of two larger research projects in which subjects (N = 184) were administered a series of questionnaires and engaged in a conflict discussion and anger induction task (Babcock et al. 2005; Babcock et al. 2008; Costa et al. 2007). Couples were paid $10 per person per hour for their participation. There were no differences between the two samples on demographics. As a result, the two studies were merged for data analytic purposes.

Results

Manipulation Checks

A 3 × 2 (Group X Time) repeated measures, mixed ANOVA was conducted on state anger pre- and post-anger induction. There were no significant differences between the groups on state anger pre-anger induction, F(2,162) = 1.61, n.s. As expected, the manipulation was successful in increasing state anger after the anger induction paradigm, F(2,162) = 175.90, p < .0001. The group × time interaction was significant, [F(2,162) = 3.77, p < 0.05] with the SNV group reporting a greater increase in state anger following the anger induction than the DNV group, t(162) = 1.28, p < 0.05 (see Fig. 1).

Fig. 1
figure 1

Mean ratings of state anger at two time points for satisfied nonviolent, distressed nonviolent, and intimate partner abusers

Full size image

A two-way MANOVA (Group × Scenario) was conducted analyzing the Imagination Questionnaire to ensure that group differences and correlates are not attributable to differences in men’s abilities to mentally engage in the ATSS. Overall, there were no main effects due to scenario [F(1,112) = 0.39, n.s.] or group [F(2, 112) = 2.27, n.s.]. The scenario by group interaction was also not significant, F(2,112) = 0.00, n.s. Thus, the manipulation was successful, as most participants agreed somewhat that they could imagine themselves in the situations.

To examine concurrent validity, peak ratings of anger on the rating dial were correlated with anger coded from the articulations. Peak anger was positively correlated with anger articulated at the end of the flirtation, (r = 0.36, p < 0.001) and criticism scenario, r = 0.27, p < 0.001.

Analyses of Psychopathology and Cognitive Distortions

As a preliminary step, a one-way MANOVA compared the violent and nonviolent groups on the variables pertinent to the Holtzworth-Munroe and Stuart (1994) typology (see Table 1). As expected, the groups differed on all of the variables comprising the typology (all F’s significant p < 0.05). To control for Type I error, a Bonferroni correction of 0.0167 was applied to all planned comparisons. IPA scored higher than the SNV group on borderline [t(175) = −4.10, p < 0.001] and antisocial features, t(175) = −4.62, p < 0.001. Surprisingly, the SNV group reported more interpersonal psychopathic features than the IPA, t(175) = 3.72, p < 0.001. IPA reported more acts of general violence than the SNV group, t(175) = −3.07, p < 0.01. By definition, IPA had greater reports of male to female intimate partner violence than the SNV [t(175) = −3.08, p < 0.01] and DNV groups [t(175) = −3.43, p < 0.001).

Table 1 Differences between intimate partner abusers and nonviolent men on variables relevant to the Holtzworth-Munroe and Stuart (1994) typology
Full size table

Table 2 presents a 3 × 2 (Group × Time) repeated measure, mixed model MANOVA comparing IPA, DNV, and SNV men on articulated cognitive distortions. Although IPA men tended to express more anger in response to both scenarios as compared to SNV and DNV men [F(2,175) = 3.01, p < 0.05], the overall main effect due to group was not significant, F(10, 342) = 0.10, n.s. However, there was a small to moderate effect size obtained for anger; with the largest difference being between the SNV and IPA men, d = 0.56. Except for anger expression, the groups did not differ on frequency of coded articulated thoughts. However, there were significant main effects due to scenario, F(5, 171) = 20.85, p < 0.001. Men expressed more sexism, anger, and jealousy in response to the flirtation scenario whereas they articulated more power and control and dysphoria in response to the criticism scenario. There was little expression of jealousy in response to the criticism scenario. The Group × Scenario interaction was significant, F(10, 342) = 2.67, p < 0.05. Specifically, IPA expressed very little dysphoria in response to either scenario (d = 0.49) whereas, SNV and DNV men expressed more dysphoria than IPA in response to the criticism scenario, d = 0.65.

Table 2 Differences between intimate partner abusers and nonviolent men on articulated thoughts
Full size table

Because the IPA, DNV and SNV groups did not differ on cognitive distortions, Pearson correlations were analyzed across the entire sample of all groups. For the flirtation scenario, articulation of anger was positively correlated with borderline features, antisocial features, and psychopathic conduct. Anger was negatively correlated with interpersonal psychopathic features, r = −0.18, p < 0.01. Abandonment fears were negatively correlated with psychopathic conduct, r = −0.13, p < 0.05. Jealousy was positively correlated with borderline features (r = 0.13, p < 0.05) and negatively correlated with interpersonal psychopathic features (r = −0.16, p < 0.05). For the criticism scenario, jealousy was not examined due to range restriction. Anger was positively correlated with psychopathic conduct (r = 0.13, p < 0.05) but negatively correlated with interpersonal psychopathic features (r = −0.16, p < 0.05; see Table 3).

Table 3 Correlates of questionnaire measures and coded articulations on the entire sample
Full size table

Discussion

The current study extends the work of Eckhardt and colleagues (Barbour et al. 1998; Eckhardt et al. 1998) by examining specific cognitions of intimate partner abusers when angered. Although we expected intimate partner abuse to be related to the articulation of cognitive distortions, psychopathology, rather than violence, was related to cognitive distortions. Overall, violent men did not express any theme more frequently than either relationship distressed or satisfied nonviolent men. This suggests that personality disorder features, rather than intimate partner violence and relationship dissatisfaction account for differences in intimate partner abuser’s cognitive distortions when angered.

Although we expected to replicate previous findings that partner violent men responded differently than distressed and satisfied nonviolent men in response to both scenarios, overall intimate partner violence was unrelated to the degree to which men articulated anger, dysphoria, sexist beliefs, power and control, abandonment fears, or jealousy. However, there were differences in response to the criticism scenario. Whereas nonviolent men expressed sadness in reaction to the criticism scenario, IPA were unlikely to articulate such feelings and tended to express anger instead. It is unclear whether this is because partner violent men truly did not experience sadness or because they were reluctant to express it aloud in the laboratory. Sadness could be the underlying felt emotion, expressed in terms of the secondary emotion of anger (Greenberg and Safran 1989), as is common among men, especially those with externalizing problem behaviors (Greenberg 1993), such as intimate partner violence.

Because there is heterogeneity within intimate partner abusers (Holtzworth-Munroe et al. 2000), broad categorical analyses may have obscured important differences. Examined dimensionally, the content of the articulations was related in some anticipated ways to the psychopathology components of the Holtzworth-Munroe and Stuart (1994) typology. As expected, anger was positively correlated with psychopathology with the exception of the interpersonal features of psychopathy (Factor 1). Anger and jealousy were negatively correlated with interpersonal psychopathic features. Thus, men with cold and callous interpersonal traits may not experience or express anger and jealousy, even though they may be violent. As predicted, borderline personality features were positively correlated with expressing jealousy, but only in response to the flirtation scenario, which was designed to provoke feelings of jealousy and anger. Although we expected abandonment fears to also be related to borderline features, abandonment fears were not commonly articulated among our sample. If anything, antisocial and psychopathic features seemed to suppress the expression of abandonment fears. Perhaps due to this low base rate, abandonment fears were not correlated with borderline features. Like dysphoria, it is possible that abandonment fears are experienced but are not articulated among partner violent men. While abandonment fears may be the primary causal emotion, it, like dysphoria, may be expressed only in terms of secondary emotions of jealousy and anger (Greenberg and Safran 1989).

As expected, intimate partner abusers were characterized as having the most personality disorder features. Intimate partner abusers were the most likely to exhibit borderline features, antisocial features, psychopathic conduct, and engage in violence both in and outside of the relationship. However, the SNV group scored highest on interpersonal psychopathic features. This suggests that there may be measurement problems in this subscale (Hare et al. 1989). Although it was designed to measure callous, unemotional disregard for others, it may actually be assessing calmness and composure.

Limitations

Although the overall sample size was large, the nonviolent groups were small. Significant relationships between psychopathology and cognitive distortions were of a small magnitude. In addition, measurement of psychopathology was limited to self-report inventories, one of which the study revealed psychometric problems with. Future studies may wish to use multiple measures to assess for psychopathology. Though the study utilized a novel approach for measuring articulated thoughts and feelings, the coding system is in need of further validation.

Conclusions about anger as it relates to intimate partner violence should be made cautiously, as results differ depending whether anger is self-reported or spontaneously articulated. Surprisingly, the SNV group self-reported that they were the most angered. However, the IPA group tended to articulate more anger. Others have found similar contradictory findings when different methods are employed to assess anger (Eckhardt et al. 1998). Intimate partner abusers and nonviolent men may differ in how they identify anger. Intimate partner abusers may deny that they feel angry when asked directly yet show anger during expressive tasks. Given the inconsistency in labeling versus expressing anger, these findings may cast doubt on the utility of self-report measures of anger among intimate partner abusers. The ATSS procedure makes “faking good” difficult, given that there is a relatively small amount of time allotted for responding.

Clinical Implications

Intimate partner violent men appear to have difficulty recognizing and labeling their anger. Reluctant to express primary feelings of sadness and dysphoria, they may express anger instead (Stosny 1995). Anger management, especially for violent individuals with antisocial or psychopathic features, may be useful adjuncts to battering intervention programs (Babcock et al. 2007). Therapeutic approaches that target emotional regulation, such as Dialectical Behavior Therapy (DBT) in the context of imagined infidelity would be most appropriate for individuals with borderline features. DBT may aid in teaching them to manage their anger and regulate their jealousy. The efficacy of battering interventions may be improved by tailoring emotion-focused interventions to specific batterers based on their personality features and corresponding cognitive distortions.