Ostil

Clinical Psychology Review 25 (2005) 119 152
Anger, hostility, and male perpetrators of intimate partner violence: A meta-analytic review
Bradley Norlandera, Christopher Eckhardtb,*
b
Department of Psychology, University of North Texas, Denton, Texas, United States Department of Psychological Sciences, Purdue University, West Lafayette, IN 47907-2081, United States Received 21 May 2004; received in revised form 6 August 2004; accepted 8 October 2004
Abstract There has been significant interest in, and controversy about, whether anger and hostility problems are meaningfully related to male-to-female intimate partner violence (IPV). In this meta-analytic review, we empirically evaluated whether the constructs of anger and hostility discriminated between IPV perpetrators and nonviolent comparison males. Thirty-three studies reporting data from 28 independent samples were included for analysis. IPV perpetrators consistently reported moderately higher levels of anger and hostility than nonviolent men across assessment methods (i.e., self-report, observational, and spouse-specific). In prior reviews, relationship distress has been proposed as a moderating variable between relationship distress and IPV. In this review, IPV perpetrators also consistently reported moderately higher levels of anger and hostility than relationship-discordant nonviolent men. Additionally, comparisons of subtypes of IPV perpetrators found that men in moderate-high severity IPV subtypes reported higher levels of anger and hostility than low-moderate IPV subtypes. While the pattern of results in this review suggests that elevated anger and hostility are distinguishing characteristics of IPV perpetrators, empirically based conclusions regarding the functional and contextual relationship between anger, hostility, and IPV remain elusive. The implications and limitations suggested by this review are discussed in the context of emerging models of anger and IPV and treatment programs for abusive men. D 2004 Elsevier Ltd. All rights reserved.
Keywords: Anger; Hostility; Intimate partner violence; Assessment; Meta-analysis
* Corresponding author. E-mail address: eckhardt@purdue.edu (C. Eckhardt). 0272-7358/$ - see front matter D 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.cpr.2004.10.001
120
B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152
1. Anger, hostility, and male perpetrators of intimate partner violence: a meta-analytic review Attempts to understand the correlates and causes of intimate partner violence (IPV) have increasingly focused on the role of individual factors specific to the abusive male. Several reviews of this research have consistently concluded that in addition to childhood precursors, societal influences, and interpersonal contexts, male IPV perpetrators are characterized by problems related to psychopathology (Holtzworth-Munroe & Stuart, 1994; Kessler, Molnar, Feurer, & Appelbaum, 2001), cognitive distortions and deficiencies (Eckhardt & Dye, 2000), and social skills deficiencies (Anglin & Holtzworth-Munroe, 1997). While few, if any, researchers in this area would endorse a purely intrapersonal model of IPV, more careful consideration of the characteristics internal to the abuser may assist in the elucidation of etiological models of IPV and in the development of more targeted intervention programs for IPV than currently exist. The present review is focused on a particular internal characteristic of abusive males that would seem to have much potential as an aid to understanding IPV etiology, treatment, and prevention: anger arousal. In this paper, we focus on why anger might be relevant to IPV, discuss relevant etiological models that may explain how problematic anger may relate to IPV, and provide a quantitative review of the evidence regarding whether male IPV perpetrators can be distinguished from nonviolent men in terms of anger and hostility. 1.1. Setting the stage: why should disturbed anger relate to IPV? On the surface, disturbed anger and hostility would appear to be obvious risk factors for partner violence; does it not follow that aggressive people also tend to be angry people? The scenario wherein an enraged husband explodes with verbal and physical aggression directed toward his spouse appears to have ample face validity and frequent representation in media reports of IPV incidents (Purdy & Nickle, 1981; Sonkin, Martin, & Walker, 1985; Walker, 1979). Clinically focused authors have long advocated the usage of anger deescalating treatments for men who abuse their partners (e.g., Deschner, McNeil, & Moore, 1986; Hamberger, 1997). Likewise, the accumulated data suggest that problems relating to anger arousal are at least moderately consistent in discriminating domestically violent from nonviolent males (Eckhardt, Barbour, & Stuart, 1997; Schumacher, Feldbau-Kohn, Slep, & Heyman, 2001). There are several problems, however, that seriously limit enthusiasm among both researchers and practitioners concerning the status of anger as a risk factor for IPV. First, despite being a central feature of the human emotional experience, we know little about the anger construct relative to the accumulated knowledge concerning other negative emotions such as depression or anxiety. Anger is underresearched and infrequently defined in the general clinical research literature and in IPV research in particular (DiGiuseppe, Tafrate, & Eckhardt, 1994; Eckhardt et al., 1997; Eckhardt & Deffenbacher, 1995). Researchers in this area infrequently define their constructs, especially in terms of addressing conceptual distinctions between anger and hostility, resulting in a hodgepodge of loosely connected results that defy integration. Indeed, Berkowitz (1994) suggested that, b[a]ny really close and thorough examination of the psychological research into the origins of anger and emotional aggression must leave the thoughtful reader somewhat dissatisfied. The literature presents us with occasional inconsistencies and unexpected findings that most of the investigators seem not to have noticed. . .Q (p. 35). In a review of two dozen studies examining anger and the perpetration of marital violence, Eckhardt et al. (1997) reported that there was substantial inconsistency in findings related to anger, hostility, and IPV due in part to researchers inconsistent attention to the basic theoretical and
121
conceptual nuances of the anger and hostility constructs. In a later section of this paper, we will more specifically address these issues. A second related factor concerns a diffuse confusion about how anger should be assessed and whether existing anger assessment methods are psychometrically adequate (Biaggio, Supplee, & Curtis, 1981; Eckhardt et al., 1997; Eckhardt, Norlander, & Deffenbacher, 2004). Thus, the issue of whether high levels of anger and hostility are unique risk factors for IPV may be dependent upon, and attenuated by, the methods of assessment used. Given the degree of conceptual confusion about the very nature of anger and hostility as well as the functional issue of how they may relate to aggressive behavior, it would seem likely that methods of assessing these constructs are similarly unclear. Given the close relationship existing between the quality of assessment of a particular construct and the development of etiological models of that construct (e.g., Loevinger, 1957), the conceptual status of the angerIPV relationship will continue to be limited by the inadequacies surrounding anger assessment. Finally, and perhaps most importantly, there is a persistent bias against the mere notion of anger as a correlate of IPV among many professionals in the domestic violence field. Gondolf and Russell (1986) made ba case against anger control treatment programs,Q noting that such programs not only may be ineffective treatments but may actually put victims at risk for being the recipient of future violence. The argument raised by these and other authors (Pence & Paymar, 1993; Tolman & Saunders, 1988) is that by attributing the cause of battering to anger disturbances rather than the societal supports for domestic violence and mens choice to engage in that behavior, anger management interventions may overlook other significant causes of violence. In approximately one-half of states, guidelines for batterer programs explicitly prohibit anger management programs as recognized interventions for IPV perpetrators (Healey, Smith, & OSullivan, 1998), with adherence to these guidelines a requirement for programs to receive state and county funding. The net result of this stance has not only been a resistance toward anger-based interventions, but a steadfast dismissal of anger as a potential risk factor for violent behavior. Ultimately, however, all of these concerns must be answered empirically rather than ideologically. That is, before making decisions in favor of, or against, anger-focused treatments, it would first be prudent to investigate whether problems relating to anger arousal indeed relate to increased risk of male-to-female IPV perpetration. Thus, regardless of ones theoretical perspective concerning the etiological role of factors internal to the perpetrator, it is hoped that the available evidence will provide the most persuasive answer to these questions. We commence this review by first outlining several theoretical models addressing the potential linkage between anger arousal and IPV perpetration. 1.2. Etiological models of anger and aggression Does being angry make one aggressive? As noted by Tavris (1989), the anger-leads-to-aggression hypothesis exists primarily as a social myth and is not well supported by existing data, as most studies utilizing community samples of adults do not support a simple linear relation between the two. For example, Averill (1982) reported that in a sample of community adults, aggressive behaviors occurred in conjunction with anger arousal in only about 10% of angry episodes. Verbally aggressive responses to anger arousal were significantly more common, occurring 49% of the time. Other investigators have suggested that the likelihood of aggressive behavior following episodes of anger covary with the individuals enduring disposition to experience anger arousal cross-situationally (Deffenbacher et al., 1996; Tafrate & Kassinove, 2002). Thus, individuals scoring high on measures of trait anger typically report more aggressive behaviors in the context of their anger episodes than those scoring low on such
122
measures. But at this point, we should emphasize the sizable gap in the literature, both generally and within the IPV area in particular, concerning anger arousal as a causal factor in episodes of violence. Studies that have examined anger as an important etiological factor in IPV are scarce, and there are surprisingly few causal models of IPV that directly incorporate anger-related factors as etiological components. Disturbances in anger arousal are therefore likely to be related to aggressive behavior only in the context of other factors of theoretical and practical relevance. Thus, intimate partner violence is a complex set of behaviors that is multidetermined and resistant to univariate predictions such as the anger-leads-to-aggression hypothesis. In order to understand the range of the factors theoretically related to aggressive outcomes, we turn to a variety of recent data-driven theoretical models from the general aggression literature that may provide useful heuristics for understanding whether, and how, anger arousal may be related to intimate partner violence. More traditional social learning theory models (e.g., Bandura, 1973) have suggested that aggressive behavior can be learned in the same manner as other complex behaviors, either via direct conditioning or observational learning. Indeed, one of the more accepted causal explanations of the origins of domestic violence is based on the social learning theory-based notion that witnessing parental aggression as a child may set the stage for partner violence perpetration as an adult (Clarke, Stein, Sobota, Marisi, & Lucy, 1999). While this more general model provides the topography for the contextual, individual, and historical factors that may give rise to a complex behavior such as aggression, more detailed causal models are needed to explain why particular contexts give rise to aggression. The models reviewed in the next section, therefore, advance the role of cognitive factors in the onset and maintenance of aggressive behavior. Beck (1999) has recently suggested that aggression can be understood through a cognitive vulnerability model that focuses on particular style of cognitive processing termed bprimal thinking.Q According to this perspective, adverse childhood experiences produce a tendency to experience situations egocentrically. Thus, individuals with this frame may overinterpret situations in terms of their own self-interest, especially in regard to a preoccupation with perceived present and past injustices, and perceived threats to the self. This tendency of the aggressive individual to overperceive others behaviors as unjust and deliberate attempts to block his/her goals in a personally threatening manner sets the stage for the individual to take corrective actions, such as punishing acts of aggressive behavior. This process is reinforced by a characteristic set of tacit, automatic cognitive biases, such as overgeneralization (establishing inflexible rules and conclusions that apply to all situations), dichotomous thinking (viewing events or people in all-or-nothing abstractions), personalization (inferring the self to be critically affected by otherwise impersonal events), causal thinking (establishing inferences and conclusions in the absence of supporting evidence), and demandingness (absolutistic and inflexible demands that others act and events occur in accordance with the individuals desires). Together, this model presumes that following some environmental event, which need not be aversive, the aggressive individual is prone to perceive the event as indicative of loss and a threat to the self, which then leads to the experience of distress and the conclusion that the event is an unjustified violation of a strongly held personal rule. This awareness of a rule violation activates memories of prior violations in similar contexts, which may further intensify affective arousal. The quality and degree of anger eventually experienced is generated by the final meaning placed on the event by the perceiver, which then sets the stage for the relative acceptability of aggressive responding. However, while this model establishes the cognitive contexts surrounding intense anger in aggressive individuals, it is less clear in its delineation of how contexts come to be associated with
123
particular cognitive processing styles, and is similarly unclear about the intervening steps between angry experiences and aggressive behavior. A more comprehensive account of human aggression was provided by Berkowitz (1989, 1993, 2001), who noted that a variety of aversive external (loud noises, unpleasant smells, object classes) stimuli have the ability to produce an unpleasant and undifferentiated negative affective state in the perceiver. This negative affect, in turn, gives rise to automatically associated memories, images, cognitive processes, physiological processes, and behavioral intentions and produces either a tendency toward a bfightQ or bflightQ response. If a fight response is activated rudimentary feelings of anger are experienced, which become more fully differentiated based upon the specific contextual cues present in the environment, as well as the specific nature of the automatic context-activated thoughts, memories, and images. In keeping with a spreading activation model of memory and mood (Bargh, 1994; Blaney, 1996; Bower, 1981; Collins & Loftus, 1975), contexts automatically give rise to semantically related concepts stored in memory networks, with the activation spreading to other parts of the network with conceptually related meanings. As one part of the system is activated (e.g., a man perceiving a critical statement by ones wife), it automatically activates contiguous elements of the system (memories of similar statements from parents; inferences about being disrespected; feelings of anger), further solidifying the associative connections existing between the various interlinkages within the system. Thus, depending upon the individuals learning history and range of associative experiences, a variety of cues can serve to initiate the anger arousal process. Higher-level cognitions, such as cognitive biases (demands that others treat him fairly) or causal attributions (concluding that his wife says these things because she is a disrespectful nag), may be experienced in the midst of this ongoing anger reaction to further refine the quality and intensity of the angry affect, and other social problem solving cognitions (the man believing it is his role to keep her bin lineQ) may then guide possible aggressive response options. These general models of aggression, as well as more recent integrative models combining these approaches (e.g., Anderson & Bushman, 2002), suggest that anger may play a significant role in determining aggressive behavior, although this role may not necessarily be causal. Based upon the evidence generated by these theoretical models to date, Anderson and Bushman (2002) concluded that anger may influence aggression in three ways. First, anger may reduce prohibitions against aggression, either by justifying the aggressive response (for a review of the literature on attributions and IPV, see Eckhardt & Dye, 2000) or by disrupting normal cognitive processes that would otherwise suppress aggressive retaliations. For example, maritally violent men induced to feel angry in the laboratory articulated fewer anger reducing thoughts than similarly angered nonviolent comparison husbands, suggesting that such disruption is especially likely at higher levels of anger arousal (Eckhardt, Barbour, & Davison, 1998). Second, each episode of anger keeps individuals focused on aggressive motivations over time by serving as an information cue and as a prime for aggressive bscripts.Q To the extent that certain thoughts, images, and memories are closely linked to anger experience, each new episode of anger arousal will serve to activate those same processes and enhance the motivational set that promotes aggressive responding. Thus, as an individual experiences intense anger, this reaction automatically activates a set of anger relevant scripts (Abelson, 1981; Huesmann, 1988) that serve to give meaning to the situation and guide the behaviors necessary to successfully manage that situation. Aggressive individuals enact aggressive behaviors in part because they have effortlessly and tacitly accessed a highly routinized script that dictates how that individual should think (bIve been disrespectedQ), feel (bIm furious!Q), and respond (bNo one gets away with that; Im gonna make them pay!Q) to that situation. This hypothesis has
124
been supported in a number of studies investigating the social information processing of martially violent men (Eckhardt et al., 1998; Holtzworth-Munroe & Hutchinson, 1993). Third, anger energizes behavior by increasing arousal levels. Excitation-transfer models (Zillman, 1979) have shown that individuals experiencing increased physiological arousal from one source will transfer that arousal to a temporally related second source and misattribute the cause of the arousal to the second source. If the present context is conflict- or anger-related, the already-energized individual will exhibit higher levels of anger arousal than if there was no prior excitation. Thus, the data clearly support social stress and marital conflict as reliable risk factors for IPV perpetration (e.g., Kessler et al., 2001; OLeary et al., 1989). To the extent that individuals bring a high level of physiological arousal stemming from these stressors into new conflict situations, that arousal may be misattributed to the partner and in turn motivate aggressive responding. While there are no laboratory studies that provide direct evidence of this causal sequence, clinically focused authors have long discussed the idea that IPV may be exacerbated by ongoing social, financial, and lifestyle stressors (e.g., see Cano & Vivian, 2001). In summary, available theoretical models imply a relationship between anger arousal disturbances and aggressive behavior in close relationships. Indeed, several authors in the IPV area have also constructed theoretical models that suggest anger may at least be indirectly related to IPV (Dutton, 1995; Holtzworth-Munroe & Stuart, 1994; OLeary, 1993). However, anger does not cause aggression, and all aggressive individuals are not angry, but for some individuals in certain situations with a characteristic style of processing social information, anger hyperarousal may be a variable of critical importance to understanding aggressive responding. In the next section, we consider what previously published reviews of this literature have concluded as a prelude to the meta-analytic focus of the present paper. 1.3. Prior literature reviews Numerous reviews examining the characteristics of IPV perpetrators have appeared in the past 20 years (Edleson, Eiskovits, & Guttman, 1985; Holtzworth-Munroe, Smutzler, & Bates, 1997; HoltzworthMunroe, Bates, Smutzler, & Sandin, 1997; Hotaling & Sugarman, 1986; Rosenbaum & OLeary, 1981; Tolman & Bennett, 1990). Most had little to say about the role of anger and hostility until the 1990s, and few differentiated between these two constructs. Qualitative reviews of the literature by Tolman and Bennett (1990) and Holtzworth-Munroe et al. (1997) suggested that high levels of anger were related to partner violence perpetration although this relationships was somewhat inconsistent, with Tolman and Bennett concluding that bmen who batter are more hostile and angrier than nonviolent controls. However, they do not seem to differ from generally violent menQ (p. 93). Eckhardt et al.s (1997) qualitative review reported similar conclusions, and suggested that measurement problems and inconsistencies surrounding the definitions of hypothetical constructs were the foundation for these inconsistencies (rather than the angerIPV relationship at the construct level), subjects we will return to in a subsequent section. In the only meta-analytic review examining anger as a discriminating characteristic of IPV perpetrators, Schumacher et al. (2001) reviewed five studies and concluded that banger and hostility are consistent predictors of [IPV], but the effects range from relatively small (r =0.18) to large (r =0.52) depending on the study and the measures usedQ (p. 327). However, these authors suggested that these specific relationships may be mediated by marital distress, noting that three of the five studies did not use a relationship-distressed nonviolent sample of IPV perpetrators, and a fourth study indeed found a hostilitymarital distressIPV mediational relationship. Only one study reviewed by Schumacher et al.
125
(2001) suggested that anger/hostility was higher in IPV perpetrators relative to a distressed nonviolent sample of males (Barbour, Eckhardt, Davison, & Kassinove, 1998). While the Schumacher et al. (2001) meta-analysis was a timely and critically important quantitative review of the prior two and a half decades of research on all risk factors for IPV, the two-paragraph section on anger and hostility falls short in several important respects. First, the authors reviewed only a small portion of the published research on anger, hostility, and male IPV perpetrators. In the present review, for example, we include 33 separate studies that investigated these factors. Second, the authors did not separate effect sizes according to the construct assessed or the method of assessment used. Eckhardt et al. (1997) noted that the terms anger and hostility have been used interchangeably in the IPV literature, with little attention paid to important construct-level differences between these two traits. To the extent that researchers provide conceptually clear definitions of anger and hostility, resulting models will more accurately describe and predict the phenomenon of interest. Similarly, definitional murkiness will directly impact the quality of assessment in this area as researchers utilize measures that assess the bwrongQ construct or that possess weak psychometric properties for a particular assessment goal (for a more general review of anger assessment, see Eckhardt et al., 2004). These problems, in turn, will only serve to attenuate or otherwise produce spurious relationships between anger-related variables and partner violence, and will continue to thwart more general models of IPV. Thus, these issues point to the need for an additional and more comprehensive quantitative analysis of the empirical literature on anger, hostility, and male-to-female IPV perpetration. 1.4. Definitions In the previous section, we indicated the importance of distinguishing between the constructs of anger and hostility. Historically, hostility has been regarded as an attitudinal construct involving the dislike and negative evaluation of others (Berkowitz, 1993; Buss, 1961), or as a cognitive trait that indicates ba devaluation of the worth and motives of others, an expectation that others are likely sources of wrongdoing, a relational view of being in opposition toward others, and a desire to inflict harm or see others harmedQ (Smith, 1994, p. 26). While other researchers view hostility as being a more multifaceted construct involving affect and expressive behavior in addition to negative attitudes (e.g., Barefoot, 1992; Barefoot & Lipkus, 1994), the hostility construct primarily involves the cognitive variables of cynicism (believing that others are selfishly motivated), mistrust (an overgeneralization that others will be hurtful and intentionally provoking), and denigration (evaluating others as dishonest, ugly, and mean) (Miller, Smith, Turner, Guijarro, & Hallet, 1996). Most researchers assume that the negative cognitive set that is hostility in turn predisposes a tendency to produce frequent episodes of anger and to motivate aggressive and often vindictive behavior (Spielberger, 1988). Attempts to define anger have focused on narrow indices such as physiological arousal (Ax, 1953), subjective phenomenology (Spielberger, Jacobs, Russell, & Crane, 1983), or social constructivism (Averill, 1982). However, these definitions do not sufficiently address the constellation of events that occur during the experience of anger. Thus, a given angry episode involves a myriad of physiological alarm responses, escape and attack behaviors, subjective labels of internal feeling states, and transgression-related cognitions that are experienced simultaneously as an episode of anger. Not surprisingly, therefore, more current definitions regard anger as a multidimensional construct consisting of physiological (general sympathetic arousal, hormone/neurotransmitter function), cognitive (irrational beliefs, automatic thoughts, inflammatory imagery), phenomenological (subjective awareness and
126
labeling of angry feelings), and behavioral (facial expressions, verbal/behavioral anger expression strategies) variables (Berkowitz, 1993; Deffenbacher, 1994; Eckhardt & Deffenbacher, 1995; Kassinove & Sukhodolsky, 1995). While it is important to conceptually distinguish between anger and hostility, most researchers would support a more functional relationship between the two constructs. For example, Suarez and Williams (1989) reported that individuals high and low in trait hostility did not differ on various physiological measurements at rest. When confronted with an anger-arousing laboratory task, however, significant differences between the two groups were noted with individuals high in hostility showing significant excesses in blood pressure and heart rate. Similarly, Fredrickson et al. (2000) found that individuals high in interview-assessed hostility had larger and longer-lasting blood pressure reactivity following laboratory anger induction. Thus, a continued theoretical and measurement-based separation of the constructs of anger and hostility may actually assist researchers in advancing our understanding of how these constructs interact conceptually and practically. In light of equivocal research findings from previous qualitative reviews, and limitations of a previous quantitative review, the purpose of the present study was to quantitatively examine the existing research literature on anger, hostility, and IPV using a more comprehensive search of the literature and a more careful consideration of potential moderating factors. It was hoped that this review would clarify the equivocal results obtained this far and provide much-needed insight concerning the ability of and hostility to differentiate IPV perpetrators from their nonviolent counterparts.
2. Method 2.1. Procedure 2.1.1. Literature search Empirical studies were systematically collected using several search strategies (Rosenthal, 1991). First, computerized database searches of PsycINFO, Medline, and Educational Resources (ERIC) were conducted using the following key words: Partner abuse, domestic violence, perpetrator, abusive men, batterer, partner assault, and wife assault. These words were paired separately with the keywords anger and hostility. The searches were conducted to exclude studies using the following keywords: Sex*, child*, or wom*n (an asterisk indicates that the search contained, but was not limited to that word or word fragment). Second, the reference sections of previously conducted literature reviews were searched for related studies. The third strategy involved searching the references of collected articles, books, and book chapters obtained in the previous two search methods. Fourth, authors in the field were asked to identify any related articles that were currently in press. Overall, 174 abstracts were identified for review. 2.1.2. Criteria for inclusion In order to identify a relevant sample of studies, several criteria were applied to determine eligibility for inclusion (Johnson, 1989). Each study must have: (a) been published in a research journal or book and reported in English; (b) reported data from measures of anger or hostility (not aggression); (c) included samples of men residing in North America; (d) participant data reported independently rather than consisting of dyad composite scores; (e) data from independent samples unless the assessment of different constructs, or measures used, were reported for each publication; (f) referred to their sample
127
using some form or variant of intimate partner-assault other than dating relationships; (g) reported data that included a group of identified partner-assaultive men.1 After applying these criteria, 122 studies were excluded with 52 studies to be coded. 2.1.3. Coding of studies1 The following categories of variables were coded from each study by the first author and a research assistant: (a) publication characteristics, (b) demographics, (c) recruitment procedures, (d) participant selection procedures, (e) construct(s) being measured, (f) assessment method(s), and (g) measurement data. A variance components analysis was used to evaluate the level of agreement between coders for each study characteristic. Across variables, the average variance attributed to coders was 0.12 (0.88 explained by other components). The identification of discrepancies required coders to independently recode these variables. Subsequent re-evaluation identified no occurrences of disagreement. The coding process resulted in the further exclusion of 19 studies2 due to the duplication of data across multiple publications, data reported in such a way that an appropriate effect size could not be calculated (e.g., no data reported even though the construct was measured), or a similar construct to anger or hostility was measured but not defined as such. In all, 33 studies, derived from 28 independent sample populations, were included in this analysis (see Table 1 for a summary of these studies). For the comparison of intimate partner violent and nonviolent men 28 studies, from 23 independent sample populations, were identified. Of those 28 studies, 14 were identified for comparisons between relationship-discordant nonviolent men and intimate partner assaultive men, and five were identified for comparisons of subtypes of intimate partner assaultive men. An additional five studies that provided data for comparisons of subtypes of intimate partner assaultive men were also included. An initial presentation of meta-analytic procedures (Glass, 1976) suggested that features associated with study quality should be evaluated by correlating them with study outcomes. Several strategies for utilizing study quality indices in quantitative reviews have been developed, including using study quality data as criteria for inclusion, as a characteristic that is subjectively coded, and/or as an index used to weight the calculation of effect sizes to favor studies determined to be of higher quality. A fourth strategy, and the one adopted in this study, is to allow the data obtained from the studies to determine if study qualities influenced outcomes (Cooper & Hedges, 1994). The variables that are typically used to evaluate study quality (research design and assessment strategies) are of particular interest and will be coded and incorporated into categorical model testing. 2.1.4. Selection and calculation of effect size estimates The meta-analytic software program D-STAT (Johnson, 1989) was used for calculation of Cohens d effect size estimate (Cohen, 1977), which represents a standardized score of the difference between two group means, divided by the pooled standard deviation of the groups. Cooper (1989) suggested that small effect sizes ranging from 0.2 to 0.5 represented weak relationships, effect sizes ranging from 0.5 to 0.8 represented moderate relationships, and that effect sizes above 0.8 represented strong relationships. In this study, effect sizes were treated as dependent variables and study characteristics were examined as independent variables. The most frequent types of data reported were (in order)
1 2
A list of studies are available upon request to the first author. A list of variables are available upon request to the first author.
128
Table 1 Studies coded for meta-analysis Author Barbour et al. (1998) Population Characteristics Description of Results Sample Size Clinical; N =88; 31 IPV, 23 DNV, 34 SNV Using the STAXI (d =0.58, p b0.009), IPV scored higher than both NV groups on all subscales (State: d =0.40, p b0.06; Trait, d =0.68, p b0.002; Anger In, d =0.48, p b0.03; Anger Out, d =0.70, p b0.002; Anger Control, d =0.01). Significant differences were not found between IPV and NV using the ATSS paradigm (d =0.26, ns) to assess anger expression. DNV scored lower than IPV men across all scales (d =0.28, p =0.30). IPV scored significantly higher than all NV using the BDHI (d =0.37, p b0.012). IPV scored higher than DNV (d =0.39, p b0.07). Using the STAS, IPV scored higher than NV on both the State (d =0.96, p b0.01) and Trait (d =0.59, p =0.01) anger scales. Across measures of general and relationship-specific anger and hostility, IPV scored higher than NV (STAS-T, d =0.43, p b0.01; MAI-H, d =0.22, p =0.06; SS Anger rating, d =0.50, p b0.01; Modified BDHI, d =0.69, p b0.01). IPV comparisons with DNV were inconsistent (Modified BDHI, d =0.84, p b0.01; STAS-T, d =0.42, p b0.01; MAI-H, d =0.21, p =0.19; SS Self-Rating, d =0.01, ns) as were comparisons between LMIPV and MSIPV (SS Self-Rating, d =0.53, p b0.01; STAS-T, d =0.24, p =0.11; Modified BDHI, d =0.16, ns; MAI-H, d =0.12, ns). Using a modified SPAFF, IPV (described as Physically Assaultive, Verbally Assaultive, and Withdrawing) scored higher than NV (described as Non Distressed) on a composite rating of anger/contempt (d =0.48, p b0.06). LMIPV scored significantly higher than MSIPV (d =0.74, p =0.02). Using the BDHI, LMIPV scored significantly lower than MSIPV (d =0.40, p b0.01). IPV men reported significantly higher scores than the normative sample (d =0.51, p b0.01). Using the STAS-T, IPV (Family Only) scored significantly higher than NV (convicted of nonviolent crimes; d =1.47, p b0.001).
Barnett et al. (1991)
Beasley and Stoltenberg (1992) Boyle and Vivian (1996)
Clinical; N =186; 93 IPV, 42 VOa, 43 DNV, 50 SNV Clinical; N =84; 49 IPV, 35 NV Clinical/Community; N =312; 69 LMIPV, 100 MSIPV, 94 DNV, 49 SNV
Burman, Margolin, and John (1993)
Community; N =65; 33 LMIPV, 17 MSIPV IPV, 15 NV
Cadsky and Crawford (1988) Date and Ronan (2000)
Clinical; N =172; 106 LMIPV, 66 MSIPV Clinical; N =59; 20 IPV, 19 VOa, 20 NV Incarcerated Sample Clinical/Community; N =60; 24 IPV, 18 DNV, 18 CCG Clinical/Community; N =75; 32 MSIPV, 26 IPV (incarcerated), 17 CCG Clinical/Community; N =176; 132 IPV, 44 CCG
Dutton and Browning (1988) Dutton and Starzomski (1993) Dutton (1995)
Dutton et al. (1994) Dutton, Starzomski, and Ryan (1996)
Using self-ratings of anger, compared IPV with verbally abusive DNV and a community comparison group (CCG; d =0.59, p b0.03). The CCG reported some levels of IPA. Assessed anger using the MAI in an incarcerated and clinical sample of IPV and a CCG (reported some levels of IPA; d =0.16, p =0.50). Using the MAI, IPV scored significantly higher than a CCG reporting some levels of partner assaultive behavior (d =0.36, p b0.02). Clinical/Community; Using the MAI, IPV scored significantly higher than CCG N =160; 120 IPV, 40 CCG (d =0.31, p =0.051). Clinical/Community, Using the MAI, NV men scored significantly lower than N =185; 140 IPV, 45 CCG court-referred and self-referred men (d =0.41, p b0.01). Some individuals in the CCG reported levels of IPA
B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152 Table 1 (continued ) Author Eckhardt and Kassinove (1998) Eckhardt et al. (1998) Population Characteristics Description of Results Sample Size Community; N =40; 20 IPV, 20 NV Community; N =88; 31 IPV, 23 DNV, 34 NV
129
Eckhardt et al. (in press) Else, Wonderlich, Beatty, Christie, and Stanton (1993) Gavazzi, Julian, and McKenry (1996)
Clinical; N =199; 61 LMIPV, 138 MSIPV Clinical; N =42; 21 IPV, 21 NV Clinical/Community; N =152; 52 IPV, 100 NV
Using the ATSS paradigm to assess anger compared CMIPV and relationally satisfied NV (d =0.75, p b0.02). Using the ATSS paradigm, IPV articulated significantly more hostile attribution biases (d =1.04, p b0.001) than DNV. IPV articulated fewer anger control statements (d =0.19, ns) than DNV. Using the STAXI, MSIPV scored significantly higher than LMIPV (d =1.10, p b0.001). Using the HDHA , IPV not significantly different from NV (d =0.11, ns).
Includes the sample in McKenry, Julian, and Gavazzi (1995). Using the BSI-Hostility subscale, IPV scored significantly higher than NV (d =0.67, p b0.001). Hamberger, Lohr, Bonge, Clinical; N =695; Using the NAS, MSIPV scored significantly higher than LMIPV and Tolin (1996) 335 LMIPV, 180 SMIPV (d =0.64, p b0.001). Hastings and Clinical; N =168; Using the NAS, IPV with alcohol problems (n =47) and IPV Hamberger (1988) 125 IPV, 43 DNV without alcohol problems (n =78) both scored significantly lower than DNV. Overall, IPV scored significantly lower than DNV (d =0.45, p b0.01). Hershorn and Clinical; N =41; Undercontrolled-hostile IPV men scored higher than Rosenbaum (1991) 24 LMIPV, 17 MSIPV overcontrolled-hostile IPV men (d =0.85, p b0.02). Holtzworth-Munroe Community; N =163; Using general and relationship specific measures of anger and et al. (2000) 37 LMIPV, 665 MSIPV, hostility, Overall IPV scored higher than NV (d +=0.52, p b0.01). 23 DNV, 38 SNV Overall LMIPV scored significantly lower than MSIPV (d +=0.75, p b0.001), but significantly higher than DNV (d +=0.71, p b0.009). Using the Biglan rating system, IPV scored marginally higher on Holtzworth-Munroe Clinical/Community; anger (d =0.34, p =0.10) and hostility (d =0.40, p =0.05) than both and Smutzler (1996) N =97; 25 MSIPV, NV groups in response to written and videotaped stimuli. Overall, 21 LMIPV, 23 DNV, CLIPV scored higher than CMIPV (d =0.68, p b0.02). 28 SNV Leonard and Blane (1992) Community sample of Using a composite measure of the TAS and BDHI-Assault, IPV newlyweds; N =340 reported significantly higher levels of hostility than NV (d =0.56, p b0.001). Leonard and Senchak Community sample of IPV scored significantly higher than NV on TAS-BDHI-Assault (1993) newlyweds; N =607 composite measure (d =0.60, p b0.001). Maiuro, Cahn, and Clinical; N =106; 77 IPV, Using the BDHI, IPV scored significantly higher than NV (d =0.92, Vitalliano (1986) 29 NV p b0.001). Maiuro, Vitaliano, and Clinical; N =119; 30 Using the BAAQ, IPV scored higher than NV (d =0.66, p b0.01). Cahn (1987) LMIPV, 37 MSIPV, A comparison of LMIPV and MSIVP (d =0.08, p =0.76) was not 26 VOa, 26 NV significant. IPV scored higher than NV on the BDHI (d =1.02, p b0.001). LMIPV Maiuro, Cahn, Vitalliano, Clinical; N =129; did not differ from MSIPV (d =0.04, p =0.75) on the BDHI but 39 LMIPV, 58 MSIPV, Wagner, and Zegree scored significantly higher than NV (d =1.09, p b0.001). 29 NV (1988) Margolin (1988) Community; N =45 Using the NAS IPV (described as Physically Assaultive (PA), 25 IPV, 20 NV Verbally Assaultive VA, and Withdrawing (WI)) scored approximately the same as NV (described as Non Distressed (ND); d =0.02, p =0.92). (continued on next page)
130 Table 1 (continued ) Author Margolin, John, and Gleberman (1988)
Population Characteristics Description of Results Sample Size Community; N =77 59 IPV, 18 NV Using an affective coding system IPV (described as Physically Assaultive (PA), Verbally Assaultive (VA), and Withdrawing (WI)) scored higher than NV (described as Non Distressed (ND); d =0.82, p =0.001). Using the BDHI IPV (Described as Physically and Emotionally Abusive (EPA) scored significantly higher than NV (Described as Emotionally Abusive (EA) and Non Abusive (NA); d =0.45, p b0.004). MSIPV scored significantly higher on the NAS than the LMIPV (d =0.71, p b0.001). Using the MAI, IPV men scored higher than CCG (d =0.44, p =0.21).
Margolin, John, and Foo (1998)
Community, N =171 31; IPV, 140 NV
Clinical; N =165; 86 LMIPV, 40 MSIPV Tweed and Dutton (1998) Clinical/Community; N =123; 70 IPV (32 Instrumental, 38 Impulsive), 44 CCG Waltz et al. (2000) Community; N =107; 75 IPV (18 Generally Violent, 17 Pathological, 40 Family Only), 32 DNV
Saunders (1992)
IPV scored higher on SPAFF anger (d =0.29, p =0.14) and SPAFF Contempta (d =0.45, p =0.25)than NV. LMIPV scored higher than MSIPV on the SPAFF anger (d =0.20, p =0.42), and SPAFF Contempta (d =0.24, p =0.33) scales. A comparison between IPV and DNV (d =0.20, p =0.31) was not significant.
IPA=Intimate Partner Assault; IPV=Intimate Partner Violent; CLIPV=Clinical Intimate Partner Violent; CMIPV=Community Intimate Partner Violent; NV=Non-Intimate Partner Violent; DNV=Maritally Distressed Non-Intimate Partner Violent; SNV=Maritally Satisfied Non-Intimate Partner Violent; LMIPV=Low-Moderate Intimate Partner Violence Subtype; MSIPV=Severely Intimate Partner Violence Subtype; VO=Violent Outside of Intimate Relationships Only; INPA=Incarcerated; CCG=Community Comparison Group. a Excluded from analysis.
group means, standard deviations, t -scores, F -scores, v 2, and correlation coefficients. In one instance (Tweed & Dutton, 1998), the effect was computed from a p -value of 0.05, as these were the only data reported. The present study used a fixed effects meta-analytic research design and all effects reported in this analysis have been weighted by the inverse of their variance to allow for the weighting of effects based upon sample size. An effect size was calculated for each reported measurement of anger and hostility. Since some studies only reported data for the individual subscales of a measure, separate effect sizes were calculated for each subscale and then averaged to produce a single effect size for that measure. When the same measure was used to assess both the constructs of anger and hostility within a given study, an individual effect size was calculated for each construct. The calculation of effect sizes for each instance of measurement allowed for the examination of differences due to variations in research designs and assessment strategies. Using the effect sizes calculated at the measurement level, an overall effect size for both constructs of anger and hostility were calculated by averaging the measurement level effect sizes for each construct within each study. In order to obtain an overall effect size for each study, these construct level effect sizes were then averaged. Effect sizes derived from studies that reported on the same sample were also averaged. These overall effect sizes from each independent sample were combined to obtain a grand effect size (Johnson, 1989).
131
Research attempting to differentiate IPV men from their nonviolent counterparts has been criticized for failing to include relevant comparison groups (see Schumacher et al., 2001). Specifically, researchers have suggested that consideration of the level of marital satisfaction in comparison groups would provide a more meaningful level of analysis. Several researchers have incorporated a comparison group including men with low levels of marital satisfaction but no history of IPV, thus identifying a relationship-discordant, nonviolent comparison group (Holtzworth-Munroe, Jacobson, Fehrenbach, & Fruzetti, 1992). Thus, effect sizes were calculated between relationship discordant, nonviolent men, and IPV perpetrators (see Table 3). Several researchers have attempted to identify different subtypes of partner-assaultive men according to a variety of psychological variables. Common to each categorization was the consideration of the severity of assault. Based upon study results and researcher descriptions, subtypes were classified as engaging in low-moderate IPV or moderate-severe IPV and then compared to produce effect sizes based on this categorization. 2.2. Description of measures Researchers have used several assessment strategies to investigate the relationship between anger, hostility, and IPV. These strategies include traditional endorsement/paper-and-pencil rating scales as well as alternate assessment methods such as observational ratings of men engaged in laboratory-based communication tasks with their wives, and verbal recordings of participants thoughts and feelings while undergoing a laboratory-based mood induction method. Researchers have modified some of these measures in an attempt to obtain situation specific assessments of anger directed toward intimate partners, and are therefore referred to as bspouse specific.Q 2.2.1. Novaco Provocation Inventory (NPI; Novaco, 1975, 1977) The NPI consists of 90 potentially anger provoking situations, and respondents rate on a five-point Likert scale the degree of anger arousal they would feel if they were placed in that situation. A revised 80-item version was subsequently published (Novaco, 1977). Both versions have been shown to have excellent internal consistency estimates above 0.90 (Novaco, 1975, 1977) and testretest correlations in the 0.780.91 range (Mills, Kroner, & Forth, 1998). Novaco (1977) found the instrument was effective in differentiating between an anger prone psychiatric sample and a normal control group. Overall, the NPI may provide useful data on antecedent or activating situations associated with anger arousal. 2.2.2. Multidimensional Anger Inventory (MAI; Siegel, 1985, 1986) The MAI was rationally constructed to assess anger as a multidimensional construct. The measure consists of 38 items on a five-point Likert scale to assess anger frequency, duration, magnitude, mode of expression, hostile outlook, and range of anger eliciting situations. Factor analyses have produced differing results ranging from the authors reported five-factor solution (Siegel, 1986), to single-factor solutions reported with a community population of non-clinical adults (Riley & Treiber, 1989) and incarcerated males (Kroner, Reddon, & Serin, 1992) and two-factor solution (anger arousal/anger in and situations/hostile outlook) in a sample of patients with coronary heart disease (Kneip et al., 1993). A series of studies by Dutton et al. (Dutton, 1995; Dutton, Saunders, Starzomski, & Bartholomew, 1994; Dutton & Starzomski, 1993, 1994) involving IPV males offered support for the MAI Total score and certain subscales (Anger In, Anger Out, Anger Magnitude, Anger Frequency) as valid measures for
132
assessing anger in men with anger-related problems. In summary, however, the accumulated evidence concerning the psychometric properties of the MAI, especially its dimensional structure, warrants caution in its usage and interpretation. 2.2.3. StateTrait Anger Expression Inventory (STAXI, Spielberger, 1988) The factor-derived STAXI is a 44-item measure of the experience and expression of anger. The STAXI was designed to assess anger in accordance with statetrait personality theory and to measure the multi-dimensional components of anger that are distinct from the constructs of hostility and aggression. The STAXI is composed of the State Anger Scale (SAS), Trait Anger Scale (TAS), and Anger Expression Scales (AX; Spielberger et al., 1986) and have internal reliability coefficients that range from 0.74 to 0.93 (Fuqua et al., 1991; Musante, Treiber, Davis, & Waller, 1999; Spielberger, 1988; Spielberger, Reheiser, & Sydeman, 1995; Spielberger & Sydeman, 1994). The SAS was intended to measure the experience of current angry feelings and the TAS was designed to assess the frequency an individual experiences state anger over time. The TAS consists of two subscales, Angry Temperament and Angry Reaction, with alpha coefficients of 0.85 and 0.73, respectively (Fuqua et al., 1991). The Anger Expression Scales (AX; Spielberger et al., 1986) were developed to evaluate the expression and control of anger. The AX currently consists of the following three scales: (1) Anger In , which measures the tendency to suppress angry feelings; (2) Anger Out , which assesses the tendency to express anger outwardly toward individuals or objects through physically or verbally aggressive behavior; and (3) Anger Control , which assesses the tendency to successfully diminish the occurrence of angry feelings. Deffenbacher et al. (1996) provided support for the discriminant and convergent validity as well as the clinical utility of the STAXI in a series of studies with college student samples. The STAXI (as well as the more recently published STAXI-2; Spielberger, 1999) is based upon a solid conceptual model and possesses strong psychometric properties. 2.2.4. Single-item rating scales Perhaps the most basic method of assessing anger is to ask an individual to rate his or her perceived level of anger on a single rating scale. While this strategy affords great simplicity, it is fraught with the expected measurement errors inherent whenever a one-item scale is used to assess a complex content domain such as anger. Given the multidimensional nature of anger, a one-item anger rating is likely to offer little information about the variations and complexities of respondents anger experience and expression. 2.2.5. BussDurkee Hostility Inventory (BDHI; Buss & Durkee, 1957) The BDHI is a frequently used measure of general hostility (Buss & Perry, 1992) and in research investigating partner assaultive men (six studies). The BDHI is a 75-item logically derived, truefalse scale comprised of eight subscales: Assault, Indirect Hostility, Verbal Hostility, Irritability, Negativism, Resentment, Suspicion, and Guilt. Factor analyses have suggested the BDHI is best conceptualized as a measure of anger expression style (Biaggio, 1980) rather than a measure of internal, subjective feelings. A revision of the BDHI, the Aggression Questionnaire (Buss & Perry, 1992), has successfully addressed some of the BDHIs serious limitations (for more information, see Eckhardt et al., 2004). Of interest to the present review, OLeary and Curley (1986) utilized nine questions from the BDHI in the SpouseSpecific Aggression Sub-Scale (SS-agg) of the Spouse-Specific Assertiveness Scale (SSAS; OLeary & Curley, 1986).
133
2.2.6. Hostility and Direction of Hostility Questionnaire (HDHQ; Caine, Foulds, & Hope, 1967) The HDHQ is a 51-item MMPI-based measure of Extrapunitive (other-directed) and Intrapunitive (self-directed) hostility. It yields three Extrapunitive subscales (acting-out hostility, criticism of others, and delusional hostility) and two Intropunitive subscales (self-criticism and delusional guilt). While Moreno, Fuhriman, and Selby (1993) reported stability coefficients in the 0.840.93 range, Biaggio and Maiuro (1985) reported substantially lower stability coefficients (r s=0.230.70). Unfortunately, the empirical literature on the HDHQ is sparse and construct validity data is lacking. 2.2.7. Hostility Scale of the Brief Symptom Inventory (BSI; Derogatis & Melisiratos, 1983) This five-item scale derives from the BSI, a 53-item short form of the Symptom Checklist-90 (Derogatis, Rickels, & Rock, 1976). Of the five items comprising the BSI Hostility scale, four assess verbal and physical aggression and the remaining item assesses annoyance. While the BSI possesses excellent reliability and validity, and the Hostility subscale has adequate internal consistency and stability (Derogatis, 1992), the five items do not represent the construct of hostility in either breadth or definition, thus this scale is a questionable measure of hostility. 2.2.8. Specific Affect Coding System (SPAFF; Gottman & Krokoff, 1989) Researchers investigating how violent men interact with their spouses have used specific ratings systems for observations of angry behaviors exhibited during laboratory dyadic interactions. As compared to previously described self-report formats variables such as anger-related vocal tone, facial expressions, and verbal content are available for assessment during specific relationship situations. Trained raters code relevant SPAFF variables while viewing videotape of a couple discussing an area of conflict within their relationship. Variations of the SPAFF (Gottman & Krokoff, 1989) assess both positive (e.g., humor) and negative (e.g., anger, contempt, and belligerence) emotions and behaviors, which are collected at specific intervals to assess antecedents and consequences of couples behavior. 2.2.9. Articulated Thoughts in Simulated Situations (ATSS; Davison, Robins, & Johnson, 1983) The Articulated Thoughts in Simulated Situations (ATSS) paradigm was developed as an alternative to questionnaire assessment that allows for open-ended, bon-lineQ verbal reports of thoughts and affective labels that occur during emotional arousal. In this procedure, participants listen to audiotaped scenarios and are instructed to imagine that the situations are actually occurring and, when prompted by a tone, to articulate their thoughts and feelings during a 30-s pause. Another short (1020 s) audiotaped segment is then played, followed by participants articulations, and so on. These articulations are later transcribed and coded along relevant dimensions by trained raters. The validity of the ATSS procedure has been demonstrated in several populations, including IPV men (Barbour et al., 1998; Eckhardt et al., 1998; Eckhardt, Jamison, & Watts, 2002) (for a review of extant ATSS research, see Davison, Vogel, & Coffman, 1997). 2.2.10. Biglan rating scales (Biglan, Rothlind, Hops, & Sherman, 1989) Holtzworth-Munroe and Smutzler (1996) used a subset of rating scales adapted from Biglan et al. (1989) with a population of IPV men. The scales are designed to assess how a husband would feel and respond to scenarios presented in written or videotaped stimuli. Ratings of seven feeling states (three positive and four negative) were made on seven-point Likert scales. Respondents also predicted the likelihood they would behave in particular ways during each situation by making six
134
ratings on a seven-point scale. Three of the ratings concerned the likelihood of making positive responses (e.g., try to comfort/support her, say something supportive, and discuss the subject with her), and three other ratings concerned the likelihood of making negative responses (e.g., make a critical remark).
3. Results 3.1. Population/study level of analysis Effect size comparisons between IPV perpetrators and their nonviolent counterparts are based on results obtained from 23 study samples reported in 28 separate publications (see Table 1). The combination of effect sizes from each participant sample comparing partner assaultive men and nonpartner assaultive men suggested that as a whole there was a moderate relationship between anger, hostility, and IPV (d +=0.51, k =23; 95% CI=0.440.57; p b0.001). Tests of heterogeneity ( Q w(22)= 53.44; p b0.001) indicated that the composite effect was not representative of the individual effects included in the sample. Removal of the largest outlier (Hastings & Hamberger, 1988; NAS, d =0.45) reduced heterogeneity ( Q w(21)=23.53, p =0.32) and raised the overall effect size (d +=0.54, 95% CI=0.480.60, p b0.001). This study was unique compared to the rest of the sample in that a group of male IPV perpetrators were recruited from a drug and alcohol treatment center and were included as a participant group with other IPV perpetrators without drug or alcohol problems. Despite the unique aspects of this study it will be retained for further analyses as the purpose of this study was to evaluate the entire research literature to estimate the overall relationship among anger, hostility, and IPV perpetration using each relevant participant population regardless of the direction or size of effect. Due to the potential for bias created by the exclusion of unpublished data, it was necessary to examine the influence of the bfile drawerQ problem (Cooper & Hedges, 1994; Rosenthal, 1991), which assumes that an unknown number of studies with effect sizes of zero remain unpublished because they have either not been submitted for publication or have been rejected. To evaluate this, the overall effect size for each assessment of anger and hostility was plotted in a stem and leaf graph (see Fig. 1). The relatively normal distribution of effect sizes suggested that the data included in this study were not influenced by publication bias (Light & Pillemer, 1984). Additionally, a bfail safe N Q of 77 unpublished studies, with an effect size of zero, would be required to reduce the mean effect size (d +=0.51) down to the critical value of a medium effect size and over 6000 studies would be required to reduce the mean effect size to the critical value of a small effect size [Orwin, 1983; N fs=(N total(meanD D crit))/D crit)]. Consequently, further analysis of this population of effect sizes was considered appropriate. 3.2. Measurement level of analysis The effect size calculated from each reported anger/hostility measurement obtained from the 28 studies suggested that across measures, anger and hostility had a moderate relationship with IPV (d +=0.51, k =43, p b0.001; 95% CI=0.450.55; see Table 2). Tests of heterogeneity ( Q w(42)=99.84, p b0.001) suggested that individual effect sizes were not accurately represented by the grand effect. Removal of the largest outlier (Hastings & Hamberger, 1988; NAS, d =0.45) reduced heterogeneity
135
Fig. 1. Plot of measure effect sizes.
( Q w(41)=70.93, p b0.001) and raised the overall effect size (d +=0.54, 95% CI=0.470.57). Removal of a second outlier (Holtzworth-Monroe, Rehman, & Herron, 2000; SPAFF Anger, d =0.07, p =0.66) reduced heterogeneity ( Q w(40)=57.32, p b0.07) and raised the overall effect size (d +=0.55, 95% CI=0.490.59). Removal of the third largest outlier (Eckhardt et al., 1998; ATSS SpouseSpecific Hostility, d =1.30, p b0.001) reduced heterogeneity ( Q w(39)=47.41, p =0.34) and reduced the overall effect size (d +=0.53, 95% CI=0.480.58). The first two boutliersQ had smaller effect sizes than the measures as a whole and their removal resulted in the increase of the grand effect size, while the third largest outlier resulted in a reduction of the overall effect size. As the current study suggested a positive relationship between anger, hostility, and IPV, the removal of smaller effect sizes could be perceived as biased and inclusion of the larger outlying effect size could also be perceived as biased. The purpose of this study was to estimate the overall relationship among anger, hostility, and IPV using each instance of measurement, regardless of the direction or size of individual effect sizes. Consequently, the effect sizes were retained for further analyses. Given the diversity of participant samples and methods of measurement, significant variations in effect sizes should be expected and are not inherently attributable to the constructs being measured or a specific research methodology. 3.3. Construct level of analysis In order to identify independent relationships between anger, hostility, and IPV, separate effect sizes for the constructs of anger and hostility were calculated. The combination of effect sizes from each
136
Table 2 Effect sizes for each assessment of anger and hostility Study Barbour et al. (1998) Barnett et al. (1991) Beasley and Stoltenberg (1992) Boyle and Vivian (1996) Measure STAXI ATSS: Anger expression BDHI: Hostility STAS Modified BDHI STAS: Trait SS Anger Rating MAI-Hostility SPAFF: Anger TAS MAI Self-Ratings: Anger MAI MAI MAI ATSS: Anger control ATSS: Hostility SS ATSS: Anger control SS ATSS: Hostility ATSS: Anger Control HDHQ: Hostility BSI: Hostility NAS Biglan: Hostility Biglan: Anger SS anger and hostility STAXI: SS BPAQ: SS BPAQ STAXI General anger and hostility SPAFF: Hostility SPAFF: Anger TAS/BDHI: Anger and hostility TAS/BDHI: Anger and hostility BDHI BAAQ BDHI NAS 1-Item rating BDHI MAI SPAFF: Anger NV 57 93 35 143 IPV 31 93 49 169 N 88 186 84 312 d 0.58 0.26 0.37 0.77 0.69 0.43 0.50 0.22 0.48 1.47 0.36 0.59 0.16 0.31 0.41 0.75 1.30 0.68 0.34 0.32 0.11 0.67 0.45 0.41 0.34 0.75 0.69 0.61 0.56 0.52 0.41 0.40 0.07 0.56 0.60 0.92 0.66 1.02 0.03 0.82 0.45 0.44 0.29 0.51c 95% CI 0.131.02 0.17 to 0.70 0.080.66 0.331.23 0.450.93 0.190.66 0.270.74 0.01 to 0.45 0.09 to 1.06 0.772.17 0.020.71 0.061.11 0.39 to 0.70 0.05 to 0.67 0.070.75 0.111.40 0.821.78 0.241.13 0.10 to 0.78 0.11 to 0.77 0.49 to 0.72 0.331.01 0.80 to 0.10 0.010.81 0.06 to 0.73 0.421.08 0.361.02 0.290.93 0.230.88 0.200.84 0.090.73 0.080.73 0.290.93 0.400.72 0.490.72 0.481.37 0.201.13 0.571.47 0.67 to 0.73 0.281.36 0.080.84 0.060.82 0.13 to 0.72 0.450.55 p 0.009 0.224 0.012 0.001 0.001 0.001 0.001 0.060 0.056 0.001 0.017 0.026 0.504 0.050 0.006 0.023 0.002 0.002 0.120 0.131 0.720 0.001 0.004 0.051 0.100 0.001 0.001 0.001 0.005 0.001 0.012 0.011 0.001 0.001 0.001 0.001 0.002 0.001 0.077 0.001 0.004 0.021 0.142 0.001
Burman et al. (1993) Date and Ronan (2000) Dutton (1995) Dutton and Browning (1988) Dutton and Starzomski (1993) Dutton et al. (1994) Dutton et al. (1996) Eckhardt and Kassinove (1998) Eckhardt et al. (1998)
15 20 44 36 17 40 45 20 57
50 20 132 24 58 120 140 20 31
65 40 176 60 75 160 185 40 88
Else et al. (1993) Gavazzi et al. (1996) Hastings and Hamberger (1988) Holtzworth-Munroe and Smutzler (1996) Holtzworth-Munroe et al (2000)
21 100 43 51 61
21 52 125 46 102
42 152 168 97 163
Leonard and Blane (1992) Leonard and Senchak (1993) Maiuro et al. (1986) Maiuro et al. (1987) Maiuro et al. (1988) Margolin (1988) Margolin et al. (1988) Margolin et al. (1998) Tweed and Dutton (1998) Waltz et al. (2000) Overall
29 26 29 25 18 140 44 31
78 67 77 20 59 31 70 71
314a 607a 107 93 106 45 77 171 114 102 6489b
SS=Spouse Specific, NV=Nonviolent Males, IPV=Intimate Partner Violence Perpetrators. a Sample sizes reported did not identify violent or nonviolent. b This represents sample size based on each effect size. The actual sample size of studies is N =3897. c Q w(42)=99.84: p b0.001.
137
assessment of anger (d +=0.47; k =25; 95% CI=0.340.49; Q w(24)=56.421, p b0.001) and hostility (d +=0.58; k =14; 95% CI=0.450.63; Q w(13)=31.61, p b0.003) resulted in moderate effects. Effect sizes obtained from measures of hostility were significantly higher than those from anger ( Q B(1)=4.14, p b0.05). Effects calculated from measures purported to assess both anger and hostility (HoltzworthMunroe et al., 2000; Leonard & Blane, 1992; Leonard & Senchak, 1993) also produced a moderate effect (d +=0.59, k =4; 95% CI=0.500.67; Q w (3)=2.38, p =0.67).
Table 3 Differences in anger and hostility between relationship-distressed nonviolent males and IPV perpetrators Study Anger Barbour et al. (1998) Beasley and Stoltenberg (1992) Boyle and Vivian (1996) Eckhardt et al. (1998) Hastings and Hamberger (1988) Holtzworth-Munroe and Smutzler (1996) Holtzworth-Munroe et al. (2000) Measure n DNV STAXI STAS STAXI Self-rating ATSS NAS ATSS Biglan STAXI: AX SS STAXI: AX STAXI: Trait SS STAXI: Trait SPAFF SPAFF 23 35 94 23 43 23 23 IPV 31 49 69 31 105 21 102 54 84 163 54 148 44 125 0.54 0.77 0.42 0.01 0.19 0.45 0.30 0.06 1.18 1.14 1.03 1.01 0.34 0.20 0.38 0.01 to 1.09 0.321.22 0.110.74 0.31 to 0.31 0.35 to 0.73 0.81 to 0.09 0.250.83 0.65 to 0.53 0.701.66 0.671.62 0.601.50 0.541.48 0.11 to 0.80 0.22 to 0.63 0.260.50 0.054 0.001 0.007 0.999 0.500 0.007 0.296 0.853 0.001 0.001 0.001 0.001 0.059 0.309 0.001 N d 95% CI p
Waltz et al. (2000) Total Hostility Barnett et al. (1991) Boyle and Vivian (1996) Eckhardt et al. (1998) Holtzworth-Munroe and Smutzler (1996) Holtzworth-Munroe et al. (2000) Total Anger/hostility composite Holtzworth-Munroe et al. (2000) Total Overall
31
71
102 1481a
BDHI Modified BDHI MAI ATSS Biglan BPAQ: SS BPAQ
43 94 23 23 23
93 69 31 21 102
136 163 54 44 125 810b
0.39 0.84 0.21 1.04 0.72 1.30 1.08 0.70
0.03 to 0.81 0.521.17 0.10 to 0.52 0.471.61 0.111.33 0.821.78 0.601.55 0.540.86
0.068 0.001 0.187 0.001 0.022 0.001 0.001 0.001
SS General
23
102
125 250c 2504d
1.48 0.69 1.06 0.60
1.001.97 0.221.15 0.731.40 0.440.62
0.001 0.001 0.001 0.001
SS=Spouse Specific, DNV=Relationship-Distressed Nonviolent Males, IPV=Intimate Partner Violence Perpetrators. a This is the total combined from each effect size. The actual N from studies is N =774. b This is the total combined from each effect size. The actual N from studies is N =522. c This is the total combined from each effect size. The actual N from studies is N =125. d This is the total combined from each effect size. The actual N from studies is N =910.
138
3.4. Anger, hostility, relationship satisfaction, and IPV Given the strong and consistent linear relationship between relationship distress and IPV (e.g., Schumacher et al., 2001), we evaluated the moderating influence of this factor by examining differences among IPV perpetrator samples and maritally distressed nonviolent samples on measures of anger and
Table 4 Differences in anger and hostility between low-moderate and moderate-severe IPV perpetrators Study Anger Burman et al. (1993) Boyle and Vivian (1996) Eckhardt et al. (in press) Hamberger et al. (1996) Holtzworth-Munroe and Smutzler (1996) Holtzworth-Munroe et al. (2000) Measure n Low SPAFF Self-Ratings STAS: Trait STAXI NAS Biglan STAS: Trait STAS: Trait SS STAXI: AX STAXI: AX SS SPAFF BAAQ NAS AQ SPAFF 33 69 61 335 21 37 Severe 17 100 138 180 28 65 50 169 199 515 49 102 0.74 0.51 0.24 1.10 0.64 0.68 0.75 0.69 0.66 0.58 0.19 0.07 0.71 0.28 0.11 0.54 0.141.34 0.210.83 0.06 to 0.55 0.781.42 0.460.83 0.091.26 0.341.17 0.271.10 0.251.08 0.170.99 0.21 to 0.59 0.41 to 0.56 0.331.10 0.29 to 0.87 0.45 to 0.68 0.490.68 0.020 0.001 0.115 0.001 0.001 0.023 0.001 0.001 0.001 0.004 0.342 0.759 0.001 0.298 0.663 0.001 N d 95% CI p
Maiuro et al. (1987) Saunders (1992) Waltz et al. (2000) Total Hostility Boyle and Vivian (1996) Cadsky and Crawford (1988) Hershorn and Rosenbaum (1991) Holtzworth-Munroe and Smutzler (1996) Holtzworth-Munroe et al. (2000) Maiuro et al. (1988) Total Anger/hostility composite Holtzworth-Munroe et al. (2000) Total Overall
30 86 40
37 40 18
67 126 54 1966a
Modified BDHI MAI BDHI BDHI Biglan BPAQ BPAQ: SS BDHI
69 106 24 21 37 38
100 66 17 28 65 39
169 172 41 49 92 77 881b
0.16 0.12 0.40 0.85 0.36 0.63 0.57 0.07 0.30
0.15 to 0.47 0.19 to 0.43 0.090.71 0.201.50 0.94 to 0.20 0.221.04 0.160.98 0.37 to 0.52 0.150.42
0.302 0.439 0.009 0.011 0.212 0.001 0.005 0.748 0.001
SS Ratings General Ratings
102 204c 3051d
0.84 0.66 0.75 0.48
0.421.26 0.251.08 0.460.35 0.430.58
0.001 0.001 0.001 0.001
SS=Spouse Specific, Low=Low-Moderate Subtype Batterers, Severe=Moderate-Severe Subtype Batterers. a This is the total combined from each effect size. The actual N from studies is N =1277. b This is the total combined from each effect size. The actual N from studies is N =610. c This is the total combined from each effect size. The actual N from studies is N =102. d This is the total combined from each effect size. The actual N from studies is N =1993.
139
hostility. The nine studies included in these analyses (see Table 3) were selected based on the inclusion of a comparison group of nonviolent relationship-distressed men as defined by screening measures, or if a study reported that nonviolent subjects were recruited from clinical settings (e.g., relationship counseling). IPV perpetrators consistently reported moderately higher levels of anger and hostility as compared to relationship discordant nonviolent men (k =23, d +=0.60, p b0.001; 95% CI=0.440.62). Composite measures of anger and hostility produced a large effect (k =2, d +=1.06, 95% CI=0.731.40), which was significantly higher than individual measures of hostility (k =7, d +=0.70; 95% CI=0.540.86) and anger (k =8, d +=0.28, 95% CI=0.150.40; Q B(2)=20.55, p b0.001). Measures of hostility produced significantly larger effects than measures of anger ( Q B(1)=10.30, p b0.002). 3.5. Anger, hostility, and severity of intimate partner violence Researchers have proposed relatively distinct clusters or subtypes of IPV perpetrators, with specific hypotheses concerning the experience and expression of anger for each subtype. A variety of batterer typologies have been proposed conceptualized according to (among other things) the severity and generality of physical abuse (see Holtzworth-Munroe & Stuart, 1994; Waltz, Babcock, Jacobson, &
Table 5 Group differences in hostility according to assessment method Study Self-report measures Barnett et al. (1991) Boyle and Vivian (1996) Gavazzi et al. (1996) Else et al. (1993) Holtzworth-Munroe et al. (2000) Maiuro et al. (1986) Maiuro et al. (1988) Margolin et al. (1998) Measure n NV BDHI: Hostility Hostility/Agg.: SS MAI-Hostility BSI: Hostility HDHQ: Hostility BPAQ: SS BPAQ BDHI BDHI BDHI 93 123 100 21 61 29 29 140 IPV 93 169 52 21 102 78 77 31 186 292 152 42 163 107 106 171 16 741a 0.37 0.69 0.22 0.67 0.11 0.61 0.56 0.92 1.02 0.45 0.56* 0.080.66 0.450.93 0.01 to 0.45 0.331.01 0.49 to 0.72 0.290.93 0.230.88 0.481.37 0.571.47 0.080.84 0.430.64 0.012 0.001 0.060 0.001 0.720 0.001 0.005 0.001 0.001 0.004 0.001 N d 95% CI p
Observational measures Eckhardt et al. (1998) Holtzworth-Munroe and Smutzler (1996) Holtzworth-Munroe et al. (2000) Overall
ATSS: Hostility SS ATSS: Hostility Biglan: Hostility SPAFF: Hostility
57 51 61
31 46 102
88 97 163 4362b 21103c
1.30 0.34 0.41 0.40 0.61* 0.58
0.821.78 0.10 to 0.8 0.060.81 0.080.73 0.350.75 0.450.63
0.002 0.120 0.051 0.011 0.001 0.001
SS=Spouse Specific, NV=Nonviolent Males, IPV=Intimate Partner Violence Perpetrators. a This is the total combined from each effect size. The actual N from studies is N =1219. b This is the total combined from each effect size. The actual N from studies is N =348. c This is the total combined from each effect size. The actual N from studies is N =1492. * Q B(1)=0.0124, p =0.911.
140
Gottman, 2000). The present set of analyses evaluated the ability of measures of anger and hostility to differentiate between men classified into low-moderate severity IPV subtypes and moderate-high severity IPV subtypes. Effect sizes were generated from 11 studies that yielded 25 effect sizes that involved a comparison of at least two IPV subtypes based on the severity of IPV (see Table 4). Men in moderate-high severity IPV subtypes consistently reported moderately higher levels of anger and hostility as compared to men in low-moderate IPV subtypes (d +=0.48, p b0.001; 95% CI=0.420.58). Composite measures of anger and hostility produced a high-moderate effect (k =2, d +=0.75, 95% CI=0.461.04), which was significantly
Table 6 Group differences in anger according to assessment method Study Self-report measures Barbour et al. (1998) Beasley and Stoltenberg (1992) Boyle and Vivian (1996) Date and Ronan (2000) Dutton and Browning (1988) Dutton and Starzomski (1993) Dutton (1995) Dutton et al. (1994) Dutton et al. (1996) Hastings and Hamberger (1988 ) Holtzworth-Munroe et al. (2000) Maiuro et al. (1987) Margolin et al. (1988) Margolin (1988) Tweed and Dutton (1998) Total Observational measures Barbour et al. (1998) Burman et al. (1993) Eckhardt et al. (1998) Eckhardt and Kassinove (1998) Holtzworth-Munroe and Smutzler (1996) Holtzworth-Munroe et al. (2000) Waltz et al. (2000) Total Overall Measure n NV STAXI STAS Anger Rating STAS: Trait TAS Self-Ratings: Anger MAI MAI MAI MAI NAS STAXI: SS STAXI BAAQ 1-Item NAS MAI 57 35 143 20 36 17 44 40 45 43 61 26 18 31 44 IPV 31 49 169 20 24 58 132 120 140 125 102 67 59 140 70 88 84 312 40 60 75 176 160 185 168 163 93 78 171 114 2275a 0.58 0.77 0.50 0.43 1.47 0.59 0.16 0.36 0.31 0.41 0.45 0.69 0.52 0.66 0.82 0.45 0.44 0.51* 0.131.02 0.331.23 0.270.74 0.190.66 0.772.17 0.061.11 0.39 to 0.70 0.020.71 0.05 to 0.67 0.070.75 0.80 to 0.10 0.361.02 0.200.84 0.201.13 0.281.36 0.050.84 0.060.82 0.360.54 0.009 0.001 0.001 0.001 0.001 0.026 0.504 0.017 0.050 0.006 0.004 0.001 0.001 0.002 0.001 0.004 0.021 0.001 N d 95% CI p
ATSS: Anger expression SPAFF: Anger ATSS: Anger Control SS ATSS: Anger Control ATSS: Anger control Biglan: Anger SPAFF: Anger SPAFF: Anger
57 15 57 20 51 61 31
31 50 31 20 46 102 71
88 65 88 40 97 163 102 731b 3132c
0.32 0.48 0.68 0.33 0.75 0.34 0.07 0.29 0.38* 0.47
0.11 to 0.70 0.09 to 1.06 0.241.13 0.11 to 0.77 0.111.40 0.06 to 0.73 0.390.25 0.130.72 0.140.45 0.340.49
0.224 0.005 0.002 0.131 0.023 0.100 0.001 0.142 0.001 0.001
SS=Spouse Specific, NV=Nonviolent Males, IPV=Intimate Partner Violence Perpetrators. a This is the total combined from each effect size. The actual N from studies is N =2028. b This is the total combined from each effect size. The actual N from studies is N =555. c This is the total combined from each effect size. The actual N from studies is N =2333. * Q B(1)=4.14, p b0.05.
141
higher than measures of anger (k =15, d +=0.62, 95% CI=0.520.72) and hostility (k =8, d +=0.28, 95% CI=0.150.40; Q B(2)=15.05, p b0.001). Interestingly, measures of hostility produced smaller effect sizes than those of anger ( Q B(1)=12.193, p b0.001), which stands in contrast to comparisons between IPV and NV males in general, as well as between IPV and DNV males. 3.6. Method of assessment In order to evaluate the ability of different anger/hostility assessment methods to differentiate IPV perpetrators from their nonviolent counterparts, a grand effect was size was calculated from each instance of self-report and observational measurement. Comparisons of effects derived from self-report measures (d +=0.53, k =29; 95% CI=0.470.58, p b0.001) and observational assessments (d +=0.47, k =14; 95% CI=0.330.54) identified no significant differences at the study, construct, or measurement levels of analysis (see Tables 57). While significant differences were not found, comparisons of effect sizes from observational (d +=0.38, k =8; 95% CI=0.140.45, p b0.001) and self-report (d +=0.51, k =17; 95% CI=0.360.54) measures of anger showed a trend towards significance ( Q B(1)=2.73, p =0.09), as did a comparison of effect sizes from measures of self-reported anger (d +=0.51, k =17; 95% CI=0.360.54), self-reported hostility (d +=0.56, k =10; 95% CI=0.430.64), and self-reported composite anger and hostility (d +=0.59, k =2; 95% CI=0.500.68; Q B(2)=4.57, p =0.11). Similarly, a comparison of observational assessments of anger (d +=.38, k =8; 95% CI=0.150.46), hostility (d +=0.61, k =4; 95% CI=0.35 0.75), and composite anger/hostility measures (d +=0.57, k =2; 95% CI=0.340.80) approached significance ( Q B(2)=5.54, p b0.07). 3.7. Anger and hostility assessment context Traditionally, the constructs of anger and hostility have been assessed as global personality traits. Recently, researchers have modified existing measures or developed new assessment tools to evaluate
Table 7 Group differences across composite measures of anger and hostility Study Observational measures Holtzworth-Munroe et al. (2000) Measure n NV SS anger and hostility General anger and hostility 61 IPV 102 163 326a Self-report/general anger Leonard and Blane (1992) Leonard and Senchak (1993) Overall 0.75 0.41 0.57 0.421.08 0.090.73 0.340.80 0.001 0.012 0.001 N d 95% CI p
TAS/BDHI TAS/BDHI
314 607 921 1247b
0.56 0.60 0.59 0.59
0.400.72 0.490.72 0.500.68 0.500.67
0.001 0.001 0.001 0.001
SS=Spouse Specific, NV=Nonviolent Males, IPV=Intimate Partner Violence Perpetrators. TAS=Trait Anger Scale; BDHI=Buss Durkee Hostility Inventory. a This is the total combined from each effect size. The actual N from studies is N =163. b This is the total combined from each effect size. The actual N from studies is N =1084.
142
Table 8 Group differences in hostility according to assessment target Study General hostility Barnett et al. (1991) Boyle and Vivian (1996) Gavazzi et al. (1996) Eckhardt et al. (1998) Else et al. (1993) Holtzworth-Munroe et al. (2000) Maiuro et al. (1986) Maiuro et al. (1988) Margolin et al. (1998) Measure n NV BDHI: Hostility MAI-Hostility BSI: Hostility ATSS: Hostility HDHQ: Hostility BPAQ BDHI BDHI BDHI 93 143 100 57 21 61 29 29 140 IPV 93 169 52 31 21 102 78 77 31 186 312 152 88 42 163 107 106 171 1307a 0.37 0.22 0.67 0.34 0.11 0.56 0.92 1.02 0.45 0.52* 0.080.66 0.01 to 0.45 0.331.01 0.10 to 0.78 0.49 to 0.72 0.230.88 0.481.37 0.571.47 0.080.84 0.360.59 0.012 0.060 0.001 0.120 0.720 0.005 0.001 0.001 0.004 0.001 N d 95% CI p
Spouse-specific hostility Boyle and Vivian (1996) Eckhardt et al. (1998) Holtzworth-Munroe and Smutzler (1996) Holtzworth-Munroe et al. (2000) Holtzworth-Munroe et al. (2000) Overall
Modified BDHI ATSS: Hostility Biglan: Hostility BPAQ SPAFF: Hostility
143 57 51 61
169 31 46 102
312 88 97 163 803b 2110c
0.69 0.30 0.41 0.61 0.40 0.68* 0.58
0.450.93 0.821.78 0.000.81 0.290.93 0.080.73 0.490.78 0.450.63
0.001 0.002 0.051 0.001 0.011 0.001 0.001
these constructs in relationship-specific contexts (i.e., spouse/partner anger). A comparison of measures of spouse-specific anger and hostility (k =16, d +=0.53, 95% CI=0.430.60) and general measures of anger and hostility (k =27, d +=0.51, 95% CI=0.440.55) revealed no significant differences between the assessment of the two contexts ( Q B(1)=0.14, p =0.71; see Tables 79). However, a comparison of spouse-specific hostility (k =5, d +=0.68, 95% CI=0.490.78) and general hostility (k =9, d +=0.52, 95% CI=0.360.59) approached significance ( Q B(1)=2.908, p =0.08).
4. Discussion The purpose of the present review was to evaluate whether male perpetrators of IPV differed from relevant nonviolent comparison males in their experience and expression of anger and hostility. These differences were evaluated not only according to the construct under investigation (anger vs. hostility), but the method of measurement used to assess these constructs, which included traditional endorsement-type self-report measures, observations of dyadic communication, and individuals reactions to imagined conflicts. We also evaluated whether levels of relationship distress and the presence of proposed batterer subtype affected effect size differences. In discussing the findings of this
B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152 Table 9 Group differences in anger according to assessment target Study General anger Barbour et al. (1998) Beasley and Stoltenberg (1992) Boyle and Vivian (1996) Date and Ronan (2000) Dutton (1995) Dutton and Browning (1988) Dutton and Starzomski (1993) Dutton et al. (1994) Dutton et al. (1996) Hastings and Hamberger (1988) Holtzworth-Munroe et al. (2000) Maiuro et al. (1987) Margolin (1988) Margolin et al. (1988) Tweed and Dutton (1998) Measure n NV STAXI ATSS: Anger Control STAS STAS: Trait TAS MAI Self-Ratings: Anger MAI MAI MAI NAS STAXI BAAQ NAS Affect Coding MAI 57 35 143 20 44 36 17 40 45 43 61 26 31 18 44 IPV 31 49 169 20 132 24 58 120 140 125 102 67 140 59 70 88 84 312 40 176 60 75 160 185 168 163 93 171 78 114 2034a 0.58 0.32 0.77 0.43 1.47 0.36 0.59 0.16 0.31 0.41 0.45 0.52 0.66 0.45 0.82 0.44 0.49* 0.131.02 0.11 to 0.77 0.331.23 0.190.66 0.772.17 0.020.71 0.061.11 0.39 to 0.70 0.05 to 0.67 0.070.75 0.80 to 0.10 0.200.84 0.201.13 0.050.84 0.281.36 0.060.82 0.320.51 N d 95% CI
143
0.009 0.131 0.001 0.001 0.001 0.017 0.026 0.504 0.050 0.006 0.004 0.001 0.002 0.004 0.001 0.021 0.001
Spouse-specific anger Barbour et al. (1998) Boyle and Vivian (1996) Burman et al. (1993) Eckhardt and Kassinove (1998) Holtzworth-Munroe and Smutzler (1996) Holtzworth-Munroe et al. (2000) Waltz et al. (2000) Total
ATSS: Anger control ATSS: Anger expression Self-Rating SPAFF: Anger ATSS: Anger control Biglan: Anger STAXI SPAFF: Anger SPAFF: Anger
57 143 15 20 51 61 31
31 169 50 20 46 102 71
88 312 65 40 97 163 102 1118b 3132c
0.68 0.26 0.50 0.48 0.75 0.34 0.69 0.07 0.29 0.45* 0.47
0.241.13 0.17 to 0.70 0.270.74 0.09 to 1.06 0.111.40 0.06 to 0.73 0.361.02 0.290.93 0.13 to 0.72 0.290.54 0.340.49
0.002 0.224 0.001 0.056 0.023 0.100 0.001 0.001 0.142 0.001 0.001
review, we first present a brief overview of the results, followed by the theoretical and practical implications of these data. The meta-analytic results suggested that relative to nonviolent males, IPV perpetrators consistently indicated higher levels of anger and hostility across various types of measurement approaches. Since it is unknown if such a difference is due to the presence of violence per se, or to other third variables that potentially relate to both anger and IPV, we evaluated the moderating role played by relationship distress. These analyses indicated that the violentnonviolent difference remained significant even after accounting for relationship distress, as IPV perpetrators showed moderately higher levels of anger and
144
hostility than relationship discordant-nonviolent males. In addition, higher levels of anger and hostility differentiated between moderate-severe IPV men and low-moderate IPV perpetrators, suggesting a potential linear relationship between levels of anger and hostility and the severity of IPV. From a measurement standpoint, composite measures of anger and hostility produced the highest effect sizes, followed by measures of hostility and then measures of anger. This pattern of results was consistent across comparisons of self-reported and observational measures as well as comparisons of general and relationship-specific measures. 4.1. Implications We began this review by noting the rather shaky theoretical, empirical, and treatment grounds upon which the angerIPV relationship currently rests. Despite the seemingly obvious connection between feeling angry and acting aggressively, serious gaps exist regarding our ability to make a confident statement about whether anger bmattersQ with regard to IPV. This gap is due, at least in part, to the fields limited knowledge about the anger construct in general, problems relating to construct definition and measurement, and resistance from segments of the IPV research and treatment communities about the very notion of anger being related to partner violence. But the resolution of this dilemma also depends, in part, upon the nature of the research question being posed. If the question is bAre men with a history of intimate partner violence angrier than relationally nonviolent men? Q then the answer is a firm byes.Q In the present review, high levels of anger and hostility, as measured by a variety of different instruments, reliably differentiated between men who physically assault their intimate partners from men who report no such violent history, a differentiation that persisted even after taking into account the powerfully predictive role played by relationship distress in partner violence as well as considering the variety of anger/hostility assessment methods used to assess these constructs. The overall effect is a moderate one, suggesting that while anger and hostility may indeed differentiate IPV perpetrators from nonviolent males, there are indeed a variety of other factors both internal and external to the perpetrator that collaborate to produce IPV. Thus, anger problems should be regarded as one among many known correlates of IPV (for a detailed review of these factors, see Schumacher et al., 2001). However, if the question is modified slightly to bDo anger and hostility problems differentiate violent from nonviolent men during relationship conflicts? Q the conclusions become more equivocal. The present review suggests that if the assessment target is related to anger expression (insulting, belligerent, contemptuous, and angry facial expressions and verbal behaviors), with researchers using methodologies involving observations of couples discussing a relationship conflict in the lab or recording males reactions while imagining such a conflict, then one can conclude that IPV perpetrators indeed exhibit more responses characteristic of these more destructive forms of anger and hostility relative to nonviolent males. However, if the assessment target concerns the subjective experience of anger during such conflicts, it is not clear whether IPV males differ from nonviolent males. For example, Barbour et al. (1998) and Eckhardt et al. (2002) found that nonviolent individuals were more likely than IPV perpetrators to articulate statements of annoyance and anger (specific emotion words such as bmadQ or bfuriousQ) during laboratory anger arousal, perhaps suggesting that IPV perpetrators have problems relating to affective labeling. Nevertheless, researchers using couple interaction methodologies (e.g., Jacobson, Gottman, Waltz, Rushe, Babcock & Holtzworth-Munroe, 1994) have
145
not typically assessed the experience of anger or hostility during conflict discussions, confining their outcome measures to facial expressions and dyadic interactions. Thus, while IPV perpetrators reliably differ from nonviolent men in the obvious and predicted direction in terms of the outward expression of anger, precious little data are available to address whether such individuals are indeed experiencing the more subjective aspects of anger and hostility. The dilemma presented above proves similarly vexing if the question is altered to read: bIs there a functional relationship between anger/hostility and IPV? Q In other words, if we accept the conclusion that anger and hostility differentiate violent from nonviolent males, and we tentatively agree with the contention that IPV perpetrators also show anger-related disturbances when bprovokedQ in the laboratory, does this also suggest that problematic anger and hostility arousal serve as risk factors for the perpetration of IPV? General models of aggressive behavior (e.g., Anderson & Bushman, 2002) and IPV perpetration (e.g., Holtzworth-Munroe & Stuart, 1994) would suggest that problematic anger arousal should serve as a risk factor for IPV, in that individuals experiencing intense anger arousal are more focused on anger and aggression-related informational cues, more likely to recall prior usage of aggressive problems solving tactics, more likely to see an aggressive response as justified, and thus may experience fewer prohibitions against aggressive responding. Empirically, however, the picture is less clear. While prospective data are available to suggest that high anger is a moderate risk factor for violent behavior in general (Monahan et al., 2001) and IPV in particular (Leonard & Senchak, 1996), it is not clear if this relative risk relationship can indeed be translated into conclusions about the acute effects of anger arousal preceding discrete episodes of male-to-female violence. The types of methods used to assess anger and hostility have been fairly limited in breadth and scope, and have often possessed questionable psychometric properties. Data derived from such measures, while suggestive of the relationships they purport to represent, provide a rather tenuous conceptual foundation. Similarly, most research strategies implemented in this area have been fairly simple: administer measures of anger to samples of violent and nonviolent males, and make inferences about etiology and risk of IPV enactment from obtained group differences. However, from a methodological standpoint, serious questions concerning construct validity and external validity are raised if such relatively static data are used to make more dynamic conclusions about IPV perpetration. That is, in a given situation, and in the context of other potentially important factors related to IPV enactment, to what extent does the intense and problematic experience of anger portend attributable risk to subsequent instances of IPV? There are no available data to adequately answer this question. A related dilemma within the IPV literature concerns the role of alcohol usage and IPV perpetration, and a recent resolution of this dilemma may shed some light on how researchers can address issues relating to the relationship between anger and IPV. It has long been known that a relationship exists between problematic alcohol usage and IPV (for a review, see Wekerle & Wall, 2001), but the precise nature of this relationship has been unclear. Thus, this relationship could be due to the direct pharmacological effects of alcohol, a moderation effect wherein alcohol interacts with other risk factors to produce IPV, or this could be a spurious relationship with IPV and alcohol problems perhaps caused by some third variable. Regardless, little data addressed whether the acute effects of alcohol usage were meaningfully related to temporally proximal incidents of IPV. Recently, Fals-Stewart (2003) addressed this dilemma using the Time-Line Follow-Back (TLFB) interview, a calendar method that assesses daily patterns and frequency of alcohol usage, among samples of IPV perpetrators entering an alcoholism treatment program and batterer intervention program. His results indicated that, controlling for other
146
factors, IPV was 8 times more likely, and severe IPV 11 times more likely, on drinking days versus nondrinking days. Fals-Stewart et al. (Fals-Stewart, Birchler, & Kelley, 2003) have developed an IPVspecific version of this measure (the Time-Line Follow-Back Spousal Violence Interview), which offers much promise as a methodological improvement to better determine the attributable risk of IPV associated with discrete episodes of anger arousal. In light of the limitations noted above, the results of the present review do not specifically address the viability of anger based interventions for violent individuals. Given the suggestive, but indefinite, relationship between anger disturbances and IPV perpetration, it is premature to assume that angerrelated factors should be incorporated into intervention programs for men who assault intimate partners. While IPV perpetrators consistently report higher levels of anger and hostility than nonviolent males, both on questionnaire measures and on observational measures of more expressive-motor components of anger, we still know little about whether acute experiences of anger cause perpetrators to act aggressively. Nevertheless, it would be similarly ill-advised to assume that anger factors are an inappropriate target for interventions with men who batter. Indeed the consistency of the moderate relationship found between anger disturbances and IPV reported in this review suggest a door that rests partly open and not one that should be firmly shut. This dilemma may be resolved through continued collaborations between laboratory researchers studying anger arousal patterns of abusive males and clinicians willing to integrate these basic findings into existing IPV intervention models.
5. Conclusions The confusion surrounding the etiological role of anger in IPV has crossed the typical boundaries of academic inquiry, and has become a hotly contested matter within practice and policy circles regarding intervention programs for abusive men. As noted earlier, about half of states with guidelines for batterer intervention programs explicitly outlaw anger-focused interventions (Healy, Smith, & OSullivan, 1998), a policy which has also downgraded anger as an etiological factor in IPV perpetration. As an example to illustrate the fervency of these prohibitions, the Illinois Coalition Against Domestic Violence (2004) unequivocally states that batterers intervention programs b. . .are NOT and should not be Anger Management programs. An Abuser does not have a dproblem with angerT; the Abuser has a problem with the use of Power and Control over the Victim.Q In this and many similar guidelines in other states, very strong conclusions have been presented without empirical support suggesting that anger-related factors are not related to IPV occurrence and thus have no place in batterers intervention. This contention perhaps reflects a more general concern that invoking internal mechanisms such as disturbed anger will lead to a bmedical modelQ approach to IPV that might neglect roles played by the perpetrators choice to act violently or other societal/peer supports. But it would be equally counterproductive to ignore internal factors, such as anger disturbances, in the face of compelling evidence that such variables are indeed risk factors for IPV. The present quantitative review makes it clear that anger and hostility problems are, at the very least, factors that characterize IPV perpetrators. However, the available data also clearly point out how little we know about the causes of IPV in general as well as the specific, functional role played by anger in particular. It is hoped that this notable gap in our knowledge fosters continued collaborative efforts by IPV researchers, practitioners, and policy makers to better understand how and whether anger-related factors relate to the etiology, treatment, and prevention of IPV.
147
Acknowledgements Portions of this research were presented at the 37th Annual Conference of the Association for Advancement of Behavior Therapy, Boston, MA. The authors wish to thank the extraordinary efforts of Gavin Brown.
References3
Abelson, R. P. (1981). The psychological status of the script concept. American Psychologist, 36, 715 729. Anderson, C. I., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53, 27 51. Anglin, K., & Holtzworth-Munroe, A. (1997). Comparing the responses of maritally violent and nonviolent spouses to problematic marital and nonmarital situations: Are the skill deficits of physically aggressive husbands and wives global? Journal of Family Psychology, 11, 301 313. Averill, J. R. (1982). Anger and aggression: An essay on emotion. New York7 Springer. Ax, A. F. (1953). The physiological differentiation between fear and anger in humans. Psychosomatic Medicine, 15, 433 442. Bandura, A. (1973). Aggression: A social-learning analysis. New York7 Prentice-Hall. *Barbour, K. A., Eckhardt, C. I., Davison, G. C., & Kassinove, H. (1998). The experience and expression of anger in maritally violent and maritally discordant-nonviolent men. Behavior Therapy, 29, 173 191. Barefoot, J. C. (1992). Developments in the measurement of hostility. In H. S. Friedman (Ed.), Hostility, coping, and health (pp. 13 31). Washington, DC, US7 American Psychological Association. Barefoot, J. C., & Lipkus, I. M. (1994). The assessment of anger and hostility. In A. W. Siegman, & T. W. Smith (Eds.), Anger, hostility, and the heart (pp. 43 66). Hillsdale, NJ7 Lawrence Erlbaum Associates. Bargh, J. A. (1994). The four horsemen of automaticity: Awareness, intention, efficiency, and control in social cognition. In R. S. Wyer, Jr., & T. K. Srull (Eds.), Handbook of social cognition, vol. 1, Basic processes; vol. 2, Applications, (2nd ed.). Hillsdale, NJ7 Lawrence Erlbaum Associates. *Barnett, O., Fagan, R., & Booker, J. (1991). Hostility and stress as mediators of aggression in violent men. Journal of Family Violence, 6, 217 241. *Beasley, R., & Stoltenberg, C. D. (1992). Personality characteristics of male spouse abusers. Professional Psychology, Research and Practice, 23, 310 317. Beck, A. T. (1999). Prisoners of hate. New York7 Harper Collins. Berkowitz, L. (1989). The frustrationaggression hypothesis: An examination and reformulation. Psychological Bulletin, 106, 59 73. Berkowitz, L. (1993). Towards a general theory of anger and emotional aggression: Implications of the cognitive neoassociationistic perspective for the analysis of anger and other emotions. In R. S. Wyer, & T. K. Srull (Eds.), Perspectives on anger and emotion. Advances in social cognition, vol. 6. (pp. 1 46). Hillsdale, NJ7 Lawrence Erlbaum Associates. Berkowitz, L. (1994). Is something missing? Some observations prompted by the cognitiveneoassociationist view of anger and emotional aggression. In L. R. Huesmann (Ed.), Aggressive behavior: Current perspectives. Plenum series in social/clinical psychology (pp. 35 57). New York7 Plenum Press. Berkowitz, L. (2001). On the formation and regulation of anger and aggression: A cognitiveneoassociationistic analysis. In W. G. Parrott (Ed.), Emotions in social psychology: Essential readings (pp. 325 336). Philadelphia7 Psychology Press. Biaggio, M. K. (1980). Assessment of anger arousal. Journal of Personality Assessment, 44, 289 298. Biaggio, M. K., & Maiuro, R. D. (1985). Recent advances in anger assessment. In C. D. Spielberger, & J. N. Butcher (Eds.), Advances in Personality Assessment, vol. 5. (pp. 71 111). Hillsdale, NJ7 Erlbaum. Biaggio, M. K., Supplee, K., & Curtis, N. (1981). Reliability and validity of four anger scales. Journal of Personality Assessment, 45, 639 648.
3
References marked with an asterisk indicate studies included in the meta-analysis.
148
Biglan, A., Rothlind, J., Hops, H., & Sherman, L. (1989). Impact of distressed and aggressive behavior. Journal of Abnormal Psychology, 98, 218 228. Blaney, P. H. (1986). Affect and memory: A review. Psychological Bulletin, 99, 229 246. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129 148. *Boyle, D. J., & Vivian, D. (1996). Generalized versus spouse-specific anger/hostility and mens violence against intimates. Violence and Victims, 11, 293 317. *Burman, B., Margolin, G., & John, R. S. (1993). Americas angriest home videos: Behavioral contingencies observed in home reenactments of marital conflict. Journal of Consulting and Clinical Psychology, 61, 28 39. Buss, A. H. (1961). The psychology of aggression. New York7 Wiley. Buss, A. H., & Durkee, A. (1957). An inventory for assessing different kinds of hostility. Journal of Consulting Psychology, 21, 343 349. Buss, A. H., & Perry, M. (1992). The aggression questionnaire. Journal of Personality and Social Psychology, 59, 73 81. *Cadsky, O., & Crawford, M. (1988). Establishing batterer typologies is a clinical sample of men who assault their female partners. Canadian Journal of Community Health, 7, 119 127. Caine, T. M., Foulds, G. A., & Hope, K. (1967). Manual of the Hostility and Direction of Hostility Questionnaire (HDHQ). London7 University of London Press. Cano, A., & Vivian, D. (2001). Life stressors and husband-to-wife violence. Aggression and Violent Behaviour, 6, 459 480. Clarke, J., Stein, M. D., Sobota, M., Marisi, M., & Lucy, H. (1999). Victims as victimizers: Physical aggression by persons with a history of childhood abuse. Archives of Internal Medicine, 159, 1920 1924. Cohen, J. (1977). Statistical power analysis for the behavioral sciences. (rev. ed.). Hillsdale, NJ, US7 Lawrence Erlbaum Associates 1977, reprinted 1987. Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological Review, 82, 407 428. Cooper, H., & Hedges, L. V. (1994). Research synthesis as a scientific enterprise. In H. Cooper, & L. V. Hedges (Eds.), The handbook of research synthesis. New York7 Russell Sage Foundation. Cooper, H. M. (1989). Integrating research: A guide for literature reviews. (2nd ed.). Thousand Oaks, CA7 Sage Publications. *Date, A. L., & Ronan, G. F. (2000). An examination of attitudes and behaviors presumed to mediate partner abuse: A rural incarcerated sample. Journal of Interpersonal Violence, 15, 1140 1155. Davison, G. C., Robins, C., & Johnson, M. (1983). Articulated thoughts during simulated situations: A paradigm for studying cognition in emotion and behavior. Cognitive Therapy and Research, 7, 17 40. Davison, G. C., Vogel, R. S., & Coffman, S. G. (1997). Think-aloud approaches to cognitive assessment and the articulated thoughts in simulated situations paradigm. Journal of Consulting and Clinical Psychology, 65, 950 958. Deffenbacher, J. L. (1994). Anger reduction: Issues, assessment, and intervention strategies. In A. W. Siegman, & T. W. Smith (Eds.), Anger, hostility, and the heart (pp. 239 269). Hillsdale, NJ7 Erlbaum. Deffenbacher, J. L., Oetting, E. R., Thwaites, G. A., Lynch, R. S., Baker, D. A., Stark, R. S., et al. (1996). Statetrait anger theory and the utility of the trait anger scale. Journal of Counseling Psychology, 43, 131 148. Derogatis, L. R. (1992). The brief symptom inventory: Administration, scoring, and procedures manualII . (2nd ed.). Baltimore, MD7 Clinical Psychometric Research. Derogatis, L. R., & Melisiratos, N. (1983). The brief symptom inventory: An introductory report. Psychological Medicine, 13, 595 605. Derogatis, L. R., Rickels, K., & Rock, A. F. (1976). The SCL-90 and the MMPI: A step in the validation of a new self-report scale. British Journal of Psychiatry, 128, 280 289. Deschner, J., McNeil, J. S., & Moore, M. G. (1986). A treatment model for batterers. Social Casework, 67, 55 60. DiGiuseppe, R., Tafrate, R. C., & Eckhardt, C. I. (1994). Critical issues in the treatment of anger. Cognitive and Behavioral Practice, 1, 111 132. *Dutton, D. G. (1995). Intimate abusiveness. Clinical Psychology: Science and Practice, 2, 207 224. *Dutton, D. G., & Browning, J. J. (1988). Concern for power, fear of intimacy and wife abuse. In G. T. Hotaling, D. Finkelhor, J. T. Kirkpatrick, & M. Straus (Eds.), New directions in family violence research (pp. 163 175). Beverly Hills, CA7 Sage. Dutton, D. G., Saunders, K., Starzomski, A., & Bartholomew, K. (1994). Intimacyanger and insecure attachment as precursors of abuse in intimate relationships. Journal of Applied Social Psychology, 24, 1367 1386.
149
*Dutton, D. G., Starzomski, A., & Ryan, L. (1996). Antecedents of abusive personality and abusive behavior in wife assaulters. Journal of Family Violence, 11, 113 132. *Dutton, D. G., & Starzomski, A. J. (1993). Borderline personality in perpetrators of psychological and physical abuse. Violence and Victims, 8, 327 337. Dutton, D. G., & Starzomski, A. J. (1994). Psychological differences between court-referred and self-referred wife assaulters. Criminal Justice and Behavior, 21, 203 222. Eckhardt, C., Jamison, T. R., & Watts, K. (2002). Anger experience and expression among male dating violence perpetrators during anger arousal. Journal of Interpersonal Violence, 17, 1102 1114. *Eckhardt, C. I., Barbour, K. A., & Davison, G. C. (1998). Articulated thoughts of maritally violent and nonviolent men during anger arousal. Journal of Consulting and Clinical Psychology, 66, 259 269. Eckhardt, C. I., Barbour, K. A., & Stuart, G. L. (1997). Anger and hostility in maritally violent men: Conceptual distinctions, measurement issues, and literature review. Clinical Psychology Review, 17, 333 358. Eckhardt, C. I., & Deffenbacher, J. L. (1995). Diagnosis of anger disorders. In H. Kassinove (Ed.), Anger disorders: Definition, diagnosis, and treatment (pp. 27 48). Washington, DC7 Taylor & Francis. Eckhardt, C. I., & Dye, M. L. (2000). The cognitive characteristics of maritally violent men. Cognitive Therapy and Research, 24, 139 158. *Eckhardt, C. I., Holtzworth-Munroe, A., Norlander, B., Sibley, A., & Cahill, M. (in press). Readiness to change, partner violence subtypes, and treatment outcomes among men in treatment for partner assault. Violence and Victims . Eckhardt, C. I., & Kassinove, H. (1998). Articulated cognitive distortions and cognitive deficiencies in maritally violent men. Journal of Cognitive Psychotherapy, 12, 231 250. Eckhardt, C. I., Norlander, B. J., & Deffenbacher, J. L. (2004). The assessment of anger and hostility: A critical review. Aggression and Violent Behaviour, 9, 17 43. Edleson, J. L., Eiskovits, Z., & Guttmann, E. (1985). Men who batter women. Journal of Family Issues, 6, 229 247. *Else, L., Wonderlich, S. A., Beatty, W. W., Christie, D. W., & Stanton, R. D. (1993). Personality characteristics of men who physically abuse women. Hospital and Community Psychiatry, 44, 54 58. Fals-Stewart, W. (2003). The occurrence of partner physical aggression on days of alcohol consumption: A longitudinal diary study. Journal of Consulting and Clinical Psychology, 71, 41 52. Fals-Stewart, W., Birchler, G. R., & Kelley, M. L. (2003). The timeline followback spousal violence interview to assess physical aggression between intimate partners: Reliability and validity. Journal of Family Violence, 18, 131 142. Fredrickson, B. L., Maynard, K. E., Helms, M. J., Haney, T. L., Siegler, I. C., & Barefoot, J. C. (2000). Hostility predicts magnitude and duration of blood pressure response to anger. Journal of Behavioral Medicine, 23, 229 243. Fuqua, D. R., Leonard, E., Masters, M. A., Smith, R. J., Campbell, J. L., & Fischer, P. C. (1991). A structural analysis of the StateTrait Anger Expression Inventory. Educational and Psychological Measurement, 51, 439 446. *Gavazzi, S. M., Julian, T. W., & McKenry, P. C. (1996). Utilization of the Brief Symptom Inventory to discriminate between violent and nonviolent male relationship partners. Psychological Reports, 79, 1047 1056. Glass, G. V. (1976). Primary, secondary, and meta-analysis of research. Educational Researcher, 5, 3 8. Gondolf, E. W., & Russell, D. (1986). The case against anger control treatment programs for batterers. Response, 9, 2 5. Gottman, J. M., & Krokoff, L. J. (1989). Marital interaction and satisfaction: A longitudinal view. Journal of Consulting and Clinical Psychology, 57, 47 52. Hamberger, L. K. (1997). Cognitive behavioral treatment of men who batter their partners. Cognitive and Behavioral Practice, 4, 147 169. *Hamberger, L. K., Lohr, J. M., Bonge, D., & Tolin, D. F. (1996). A large sample empirical typology of male spouse abusers and its relationship to dimensions of abuse. Violence and Victims, 11, 277 292. *Hastings, J. E., & Hamberger, L. K. (1988). Personality characteristics of spouse abusers: A controlled comparison. Violence and Victims, 3, 31 48. Healey, K., Smith, C., & OSullivan, C. (1998). Batterer intervention: Program approaches and criminal justice strategies. Washington, DC7 National Institute of Justice. *Hershorn, M., & Rosenbaum, A. (1991). Over vs. undercontrolled hostility: Application of the construct to the classification of maritally violent men. Violence and Victims, 6, 151 158. Holtzworth-Munroe, A., Bates, L., Smutzler, N., & Sandin, E. (1997). A brief review of the research on husband violence: Part I. Maritally violent versus nonviolent men. Aggression and Violent Behaviour, 2, 65 99.
150
Holtzworth-Munroe, A., & Hutchinson, G. (1993). Attributing negative intent to wife behavior: The attributions of maritally violent versus nonviolent men. Journal of Abnormal Psychology, 102, 206 211. Holtzworth-Munroe, A., Jacobson, N. S., Fehrenbach, P. A., & Fruzetti, A. (1992). Violent married couples attributions for violent and nonviolent self and partner behaviors. Behavioral Assessment, 14, 53 64. Holtzworth-Munroe, A., Rehman, U., & Herron, K. (2000). General and spouse specific anger and hostility in subtypes of maritally violent and nonviolent men. Behavior Therapy, 31, 603 630. *Holtzworth-Munroe, A., & Smutzler, N. (1996). Comparing the emotional reactions and behavioral intentions of violent and nonviolent husbands to aggressive, distressed, and other wife behaviors. Violence and Victims, 11, 319 339. Holtzworth-Munroe, A., Smutzler, N., & Bates, L. (1997). A brief review of the research on husband violence: Part III. Sociodemographic factors, relationship factors, and differing consequences of husband and wife violence. Aggression and Violent Behaviour, 2, 285 307. Holtzworth-Munroe, A., & Stuart, G. L. (1994). Typologies of male batterers: Three subtypes and the differences among them. Psychological Bulletin, 116, 476 497. Huesmann, L. R. (1988). An information processing model for the development of aggression. Aggressive Behavior, 14, 13 24. Hotaling, G. T., & Sugarman, D. B. (1986). An analysis of risk markers in husband to wife violence: The current state of knowledge. Violence and Victims, 1, 101 124. Illinois Coalition Against Domestic Violence, 2004. Batterer intervention. Retrieved March 30, 2004 from http:// www.ilcadv.org/legal/bat_interven.htm Jacobson, N. S., Gottman, J. M., Waltz, J., Rushe, R., Babcock, J., & Holtzworth-Munroe, A. (1994). Affect, verbal content, and psychophysiology in the arguments of couples with a violent husband. Journal of Consulting and Clinical Psychology, 62, 982 988. Johnson, B. T. (1989). DSTAT: Software for the meta-analytic review of research literatures. Hillsdale, NJ7 Erlbaum. Kassinove, H., & Sukhodolsky, D. G. (1995). Anger disorders: Basic science and practice issues. In H. Kassinove (Ed.), Anger disorders: Definition, diagnosis, and treatment (pp. 1 27). Washington, DC7 Taylor & Francis. Kessler, R. C., Molnar, B. E., Feurer, I. D., & Appelbaum, M. (2001). Patterns and mental health predictors of domestic violence in the United States: Results from the National Comorbidity Survey. International Journal of Law and Psychiatry, 24, 487 508. Kneip, R. C., Delamater, A. M., Ismond, T., Milford, C., Salvia, L., & Schwartz, D. (1993). Self-and spouse ratings of anger and hostility as predictors of coronary heart disease. Health Psychology, 12, 301 307. Kroner, D. G., Reddon, J. R., & Serin, R. C. (1992). The Multidimensional Anger Inventory: Reliability and factor structure in an inmate sample. Educational and Psychological Measurement, 687 693. *Leonard, K. E., & Blane, H. T. (1992). Alcohol and marital aggression in a national sample of young men. Journal of Interpersonal Violence, 7, 19 30. *Leonard, K. E., & Senchak, M. (1993). Alcohol and premarital aggression among newlywed couples. Journal of Studies on Alcohol, 11, 96 108. Leonard, K. E., & Senchak, M. (1996). Prospective prediction of husband marital aggression within newlywed couples. Journal of Abnormal Psychology, 105(3), 369 380. Light, R. J., & Pillemer, D. B. (1984). Summing up: The science of reviewing research. Cambridge, MA7 Harvard University. Loevinger, J. (1957). Objective tests as instruments of psychological theory. Psychological Reports, 3, 635 694. *Maiuro, R. D., Cahn, T. S., & Vitaliano, P. P. (1986). Assertiveness deficits and hostility in domestically violent men. Violence and Victims, 1, 279 289. *Maiuro, R. D., Cahn, T. S., Vitaliano, P. P., Wagner, B. C., & Zegree, J. B. (1988). Anger, hostility, and depression in domestically violent versus generally assaultive men and nonviolent control subjects. Journal of Consulting and Clinical Psychology, 56, 17 23. *Maiuro, R. D., Vitaliano, P. P., & Cahn, T. S. (1987). A brief measure for the assessment of anger and aggression. Journal of Interpersonal Violence, 2, 166 178. *Margolin, G. (1988). Interpersonal and intrapersonal factors associated with marital violence. In G. T. Hotaling, D. Finkelhor, J. T. Kirkpatrick, & M. A. Straus (Eds.), Family abuse and its consequences (pp. 203 217). Beverly Hills, CA7 Sage Publications. *Margolin, G., John, R. S., & Foo, L. (1998). Interactive and unique risk factors for husbands emotional and physical abuse of their wives. Journal of Family Violence, 13, 315 344.
151
*Margolin, G., John, R. S., & Gleberman, L. (1988). Affective responses to conflictual discussions in violent and nonviolent couples. Journal of Consulting and Clinical Psychology, 56, 24 33. McKenry, P. C., Julian, T. W., & Gavazzi, S. M. (1995). Toward a biopsychosocial model of domestic violence. Journal of Marriage and the Family, 57, 307 320. Miller, T. Q., Smith, T. W., Turner, C. W., Guijarro, M. L., & Hallet, A. J. (1996). A meta-analytic review of research on hostility and physical health. Psychological Bulletin, 119, 322 348. Mills, J. F., Kroner, D. G., & Forth, A. E. (1998). Novaco Anger Scale: Reliability and validity within an adult criminal sample. Assessment, 5, 237 248. Monahan, J., Steadman, H. J., Silver, E., Appelbaum, P. S., Robbins, P. C., Mulvey, E. P., et al. (2001). Rethinking risk assessment: The MacArthur study of mental disorder and violence. New York7 Oxford University Press. Moreno, J. K., Fuhriman, A., & Selby, M. J. (1993). Measurement of hostility, anger, and depression in depressed and nondepressed subjects. Journal of Personality Assessment, 61, 511 523. Musante, L., Treiber, F. A., Davis, H. C., & Waller, J. L. (1999). Assessment of self-reported anger expression in youth. Assessment, 6, 225 233. Novaco, R. W. (1975). Anger control: The development and evaluation of an experimental treatment. Lexington, MA7 Lexington. Novaco, R. W. (1977). A stress inoculation approach to anger management in the training of law enforcement. American Journal of Community Psychology, 5, 327 346. OLeary, K. D. (1993). Through a psychological lens: Personality traits, personality disorders, and levels of violence. In R. J. Gelles, & D. R. Loseke (Eds.), Current controversies on family violence (pp. 7 30). Newbury Park, CA7 Sage. OLeary, K. D., Barling, J., Arias, I., Rosenbaum, A., Malone, J., & Tyree, A. (1989). Prevalence and stability of physical aggression between spouses: A longitudinal analysis. Journal of Consulting and Clinical Psychology, 57, 263 268. OLeary, K. D., & Curley, A. D. (1986). Assertion and family violence. Correlates of spouse abuse. Journal of Marital and Family Therapy, 12, 281 290. Orwin, R. G. (1983). A fail safe N for effect size in meta-analysis. Journal of Educational Statistics, 8, 157 159. Pence, E., & Paymar, M. (1993). Education groups for men who batter: The Duluth model. New York7 Springer. Purdy, F., & Nickle, N. (1981). Practice principles for working with groups of men who batter. Social Work with Groups, 4, 111 122. Riley, W. T., & Treiber, F. A. (1989). The validity of multidimensional self-report anger and hostility measures. Journal of Clinical Psychology, 45, 397 404. Rosenbaum, A., & OLeary, K. D. (1981). Marital violence: Characteristics of abusive couples. Journal of Consulting and Clinical Psychology, 49, 63 71. Rosenthal, R. (1991). Meta-analytic procedures for social research. (rev. ed.). Thousand Oaks, CA7 Sage. *Saunders, D. G. (1992). A typology of men who batter: Three types derived from cluster analysis. American Journal of Orthopsychiatry, 62, 264 275. Schumacher, J. A., Feldbau-Kohn, S., Slep, A. M. S., & Heyman, R. E. (2001). Risk factors for male-to-female partner physical abuse. Aggression and Violent Behaviour, 6, 281 352. Siegel, J. M. (1986). The Multidimensional Anger Inventory. Journal of Personality and Social Psychology, 51, 191 200. Siegel, J. S. (1985). The measurement of anger as a multidimensional construct. In M. A. Chesney, & R. H. Rosenman (Eds.), Anger and hostility in cardiovascular and behavioral disorders. New York7 Hemisphere. Smith, T. W. (1994). Concepts and methods in the study of anger, hostility, and health. In A. W. Siegman, & T. W. Smith (Eds.), Anger, hostility, and the heart (pp. 23 42). Hillsdale, NJ7 Lawrence Erlbaum Associates. Sonkin, D. J., Martin, D., & Walker, L. E. A. (1985). The male batterer: A treatment approach. New York7 Springer. Spielberger, C. D. (1988). Manual for the StateTrait Anger Expression Inventory. Odessa, FL7 Psychological Assessment Resources. Spielberger, C. D. (1999). STAXI-2: The State Trait Anger Expression Inventory professional manual. Odessa, FL7 Psychological Assessment Resources. Spielberger, C. D., Jacobs, G., Russell, J. S., & Crane, R. S. (1983). Assessment of anger: The StateTrait Anger Scale. In J. N. Butcher, & C. D. Spielberger (Eds.), Advances in Personality Assessment, vol. 2. Hillside, NJ7 Erlbaum. Spielberger, C. D., Johnson, E. H., Jacobs, G. A., Krasner, S. E., Oesterle, S. E., & Worden, T. J., (1986). The Anger Expression (AX) scale . Unpublished manuscript, University of South Florida, Center for Research in Community Psychology, Tampa.
152
Spielberger, C. D., Reheiser, E. C., & Sydeman, S. J. (1995). Measuring the experience, expression, and control of anger. In H. Kassinove (Ed.), Anger disorders: Definition, diagnosis, and treatment (pp. 49 67). Washington, DC7 Taylor & Francis. Spielberger, C. D., & Sydeman, S. J. (1994). StateTrait Anxiety Inventory and StateTrait Anger Expression Inventory. In M. E. Maruish (Ed.), The use of psychological tests for treatment planning and outcome assessment (pp. 292 321). Hillsdale, NJ7 Erlbaum. Suarez, E. C., & Williams, R. B. (1989). Situational determinants of cardiovascular and emotional reactivity in high and low hostile men. Psychosomatic Medicine, 51, 404 418. Tafrate, R. C., & Kassinove, H. (2002). Anger control in men: Barb exposure with rational, irrational, and irrelevant selfstatements. Journal of Cognitive Psychotherapy, 12, 187 211. Tavris, C. (1989). Anger: The misunderstood emotion. New York7 Simon & Schuster. Tolman, R. M., & Bennett, L. W. (1990). A review of quantitative research on men who batter. Journal of Interpersonal Violence, 5, 87 118. Tolman, R. M., & Saunders, D. G. (1988). The case for the cautious use of anger control with men who batter: A response to Gondolf and Russell. Response to the Victimization of Women and Children, 11, 15 20. *Tweed, R. G., & Dutton, D. G. (1998). A comparison of impulsive and instrumental subgroups of batterers. Violence and Victims, 13, 217 230. Walker, L. D. (1979). The battered woman. New York7 Harper and Row. *Waltz, J., Babcock, J. C., Jacobson, N. S., & Gottman, J. M. (2000). Testing a typology of batterers. Journal of Consulting and Clinical Psychology, 68, 658 669. Wekerle, C., & Wall, A. M. (2001). Introduction: The overlap between relationship violence and substance abuse. In C. Wekerle, & A. M. Wall (Eds.), The violence and addition equation (pp. 1 24). New York7 Brunner-Routledge. Zillman, D. (1979). Hostility and aggression. New Jersey7 Erlbaum.

Ostil

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ostil

Uploaded by

Copyright:

Available Formats

Clinical Psychology Review 25 (2005) 119 152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

Barnett et al. (1991)

Beasley and Stoltenberg (1992) Boyle and Vivian (1996)

Burman, Margolin, and John (1993)

Community; N =65; 33 LMIPV, 17 MSIPV IPV, 15 NV

Cadsky and Crawford (1988) Date and Ronan (2000)

Dutton et al. (1994) Dutton, Starzomski, and Ryan (1996)

130 Table 1 (continued ) Author Margolin, John, and Gleberman (1988)

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

Margolin, John, and Foo (1998)

Community, N =171 31; IPV, 140 NV

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

Fig. 1. Plot of measure effect sizes.

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

50 20 132 24 58 120 140 20 31

65 40 176 60 75 160 185 40 88

42 152 168 97 163

314a 607a 107 93 106 45 77 171 114 102 6489b

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

BDHI Modified BDHI MAI ATSS Biglan BPAQ: SS BPAQ

136 163 54 44 125 810b

0.39 0.84 0.21 1.04 0.72 1.30 1.08 0.70

0.068 0.001 0.187 0.001 0.022 0.001 0.001 0.001

125 250c 2504d

1.48 0.69 1.06 0.60

1.001.97 0.221.15 0.731.40 0.440.62

0.001 0.001 0.001 0.001

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

Modified BDHI MAI BDHI BDHI Biglan BPAQ BPAQ: SS BDHI

169 172 41 49 92 77 881b

0.16 0.12 0.40 0.85 0.36 0.63 0.57 0.07 0.30

0.302 0.439 0.009 0.011 0.212 0.001 0.005 0.748 0.001

SS Ratings General Ratings

102 204c 3051d

0.84 0.66 0.75 0.48

0.421.26 0.251.08 0.460.35 0.430.58

0.001 0.001 0.001 0.001

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

ATSS: Hostility SS ATSS: Hostility Biglan: Hostility SPAFF: Hostility

88 97 163 4362b 21103c

1.30 0.34 0.41 0.40 0.61* 0.58

0.821.78 0.10 to 0.8 0.060.81 0.080.73 0.350.75 0.450.63

0.002 0.120 0.051 0.011 0.001 0.001

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

88 65 88 40 97 163 102 731b 3132c

B. Norlander, C. Eckhardt / Clinical Psychology Review 25 (2005) 119152

314 607 921 1247b

0.56 0.60 0.59 0.59

0.400.72 0.490.72 0.500.68 0.500.67