“Why did it happen?” A review and conceptual framework for research on perpetrators' and victims' explanations for intimate partner violence

Abstract

Although there is an extensive research literature on individual and cultural risk factors for intimate partner violence (IPV), much less is known about the factors that victims and perpetrators of IPV perceive as playing a role in violent events. In part, lack of systematic research on perceived reasons for violence is due to the lack of a clear conceptual model and comprehensive measures of perceived reasons why partner violence occurs. In this paper, we provide a conceptual model for domains of factors influencing IPV and use this model to frame our review of existing research on victims' and perpetrators' explanations for IPV. We discuss differences in explanations for IPV in terms of gender and whether explanations refer to the respondents' own or their partners' use of violence. Our review findings suggest a need for more standardization of measurement and larger representative samples in order to identify more systematically reasons that are perceived by victims and perpetrators to be most the important contributors to IPV. Further research on perceived reasons for IPV also needs to address gender differences as well as differences related to self-partner attributions.

1. Introduction

Intimate partner violence (IPV) is a widespread health and social problem that cuts across socio-economic, cultural, and ethnic boundaries (Balci & Ayranci, 2005; Graham et al., 2008; Hegarty et al., 2000; Johnson et al., 2008; Klap et al., 2007). Although there is no universally agreed-upon definition of IPV (Hamberger, 2005), it can be broadly understood as the use or threat of physical, sexual, and/or psychological violence among current or former intimate partners (Arias & Corso, 2005). Research indicates that both men and women perpetrate violence against their partners, although the effects of this violence are overall more detrimental to women than men in terms of frequency and severity of injuries, time required off work, and use of medical, mental health and justice system services (Archer, 2000; Arias & Corso, 2005; Tjaden & Thoennes, 2000). Existing research suggests that as many as 30 to 54% of women in Western countries will experience at least one episode of physical IPV in their lifetime (Reisenhofer & Seibold, 2007).

Despite prevalence of IPV, few studies have examined the perceptions of those involved in partner violence in an effort to understand why, in their minds, the violence occurred – that is, the reason, cause or explanation people give for why they or their partner acted aggressively. Yet, individuals' perceptions are central to their own experiences of violence and the manner in which they behave in the face of violence from others. Therefore, perceptions of both female and male perpetrators and victims can provide important insight into intimate partner violence that may not be apparent from more objective measures of risk factors, such as demographic characteristics of victims and perpetrators or cultural factors associated with IPV (see WHO, 2009).

Causal attribution theory is useful for understanding the importance of perceptions surrounding IPV and for articulating how this knowledge can be used in the development of effective prevention programming. Attribution theory focuses on the ways in which people explain the events that happen to them and on how these perceptions are determined by their perspective in the event (Fiske & Taylor, 1991). Weiner's (1992) attribution theory of motivation is particularly helpful for understanding the relevance of perceived reasons for IPV for intervention and prevention efforts. First, his dimension of locus (i.e., whether the cause is internal or external to the individual) distinguishes whether a person attributes responsibility to him/herself, to someone else, or to other contextual factors. The second dimension, stability, refers to whether the perceived cause is malleable. For example, aggression attributed to someone's character (e.g., “he was aggressive to me because he is mean”) will likely be perceived as less malleable aggression attributed to someone's transient mood (e.g., “he was aggressive to me because he was so angry”). The third dimension is controllability. This dimension involves the extent to which the perpetrator is perceived as being able to control, for example, his/her mood and actions; again, the extent to which behaviour and emotions are perceived as controllable will affect how readily they can be modified in prevention and intervention efforts. These three dimensions of Weiner's (1992) attribution theory provide insight into the different levels of perceived influences on IPV and how these levels may be more or less amenable to change through intervention approaches.

Perceptions of why aggression occurred are especially relevant for understanding factors influencing the severity of aggression, especially the distinction between “intimate terrorism” and “situational couple violence.” Intimate terrorism involves one-sided violence constructed largely around systematic assertion of power and desire to control the victim, usually by a man toward a female partner. Situational couple violence (Collins, 2008; Johnson, 1995; Johnson & Leone, 2005), on the other hand, involves (usually less severe) aggression perpetrated by either partner, sometimes under circumstances of mutual aggression, without systematic control by one partner. Knowledge of perceived explanations for partner aggression may help to clarify the different forms of IPV (Johnson, 2006) and differentiate intimate partners who engage in minor situational aggression from those whose aggression is likely to lead to more severe emotional or physical harms (Finkel, 2007; O'Leary & Slep, 2006; Wilkinson & Hamerschlag, 2005). Perceived reasons for aggression may also distinguish the factors that determine whether IPV involves more severe versus less severe aggression on different occasions within the same couple (Collins, 2008).

Preventing escalation of a minor conflict to emotional and physical violence is fundamental to reducing the occurrence of intimate partner violence. Thus, prevention and treatment programs need to address attributional biases in order to increase awareness by perpetrators of how their perceptions regarding the reasons for aggression affect their behavior. Also, a better understanding of victims' and perpetrators' perceptions for IPV can be used to develop programs that will help couples in violent relationships to reinterpret their perceptions in ways that will avoid aggression and violence in the future. Analysis of perceptions may also help to identify perpetrators who are unwilling to change problematic attributions for their aggression, thus indicating a systemic abusive relationship that would not be suitable for couples counseling.

2. A Conceptual Model of Different Levels of Explanations for IPV

2.1. Why is a conceptual model necessary?

The research that has addressed perceptions of why IPV occurs has lacked an overall conceptual model or a common approach to facilitate synthesis of research findings. Related to lack of a comprehensive conceptual model, the existing research does not distinguish different levels of perceived influences on IPV in terms of their relationship to the perceiver and to the violent event. Violence attributed to a personality characteristic versus violence attributed to a situational reaction, for instance, are not only conceptually different, but are also likely to require different prevention and treatment approaches. Distinguishing between these different levels of perceived influences on IPV is thus an important first step in making better sense of the existing research and in directing future research and program development. In addition, because persons involved in IPV often attribute aggression to multiple causes or explanations (DeKeseredy et al., 1997; Olson & Lloyd, 2005), a common framework is needed to allow for examination of interrelationships among reasons for different perpetrator/victim groups.

Following from this point, it is important to assess whether there are differences in perceptions of why aggression occurs depending on the type of population under study and the form of violence being addressed. If perceived reasons for violence are found to vary across different types of violence and perpetrator groups, different prevention and intervention approaches may be required (depending on the relevant perceived causes). A better understanding of how perceptions differ by gender, role in violence (victim, perpetrator, or mutual aggressor), whether an individual is explaining their own violence or their partner's violence (“self” or “partner” attribution), and type of aggression can be used to develop programs that will help individuals and couples in violent relationships to understand their perceptions in ways that can help them avoid or minimize aggression and violence in the future.

In the present paper, we propose a three-level conceptual model for perceived reasons for IPV. Although previous research has included a variety of perceived reasons or “motives” offered by perpetrators and victims to explain their IPV, the different levels at which these influences operate have not been clearly distinguished. We identify a model of different levels of perceived reasons for IPV in terms of the proximity of the explanation to the violent event. We then use this model to organize and analyze existing research on perceived reasons for violence, focusing on differences by gender, and victim/perpetrator role. For studies where respondents reported on both perpetration and victimization, we also analyzed self versus partner attributions (i.e. whether the respondent is explaining their own aggression or their partner's aggression). The notion of self-partner differences in attributions overlaps to some degree with victim-perpetrator differences, in that individuals explaining their own violence can be regarded as “perpetrators,” while those explaining their partner's violence can be viewed as “victims.” However, the notion of self-partner differences recognizes that individuals in bidirectionally-violent relationships simultaneously occupy the roles of victim and perpetrator, and may use different explanations for violence depending on whether they are specifically addressing their own violence or their partner's violence.

This review sets the stage for future efforts to develop a comprehensive instrument for measuring explanations for partner violence and for tailoring intervention efforts to the reasons that victims and perpetrators of both genders view as most important. We conclude the paper by discussing the implications of our review for future research on perceptions of partner violence.

2.2. Existing measures of perceived reasons for IPV

Much of what is known about victims' and perpetrators' perceived reasons for IPV is scattered throughout a heterogeneous body of literature involving very different samples and very different research questions that address in some way why victims and/or perpetrators believe their IPV occurred. These questions have typically not been part of a systematic or validated instrument designed specifically for a comprehensive investigation of victims' and perpetrators' explanations for IPV. There are, however, a number of existing measures that partially tap into perceived reasons for partner violence. The PAVE (Proximal Antecedents to Violent Episodes) scale (Babcock et al., 2004), for example, includes items on motives or reasons for aggression. However, this questionnaire was designed to tap into instrumental and expressive types of violence with the goal of categorizing batterers by three specific types of motives (violence to control, violence out of jealousy, violence following verbal abuse), not to measure systematically perceived explanations for aggression. Therefore, the existing work with the PAVE does not provide a comprehensive picture of perceived causal factors among victims and perpetrators of IPV.

Three other measures have examined perceived reasons for IPV. The Relationship Abuse Questionnaire (RAQ) (Barnett et al., 1997) is a 28-item instrument that is based on the CTS but also includes questions on outcomes of abuse and attributions for abuse, which respondents rate on a scale of 1 (never) to 7 (more than once a week) in reference to each of twenty-eight forms of abuse. The nine possible attributions are: You were teasing your partner, just playing around; You were letting out your violent feelings; You were teaching your partner a lesson; You were trying to upset your partner emotionally; You were showing your partner who was the boss; You were protecting yourself from physical harm; You were trying to get your partner's attention; You were unaware of any particular intention; and You were trying to hurt your partner physically. Several pilot studies were undertaken by the authors to construct and validate the RAQ (see Barnett et al., 1997, for a description of these studies).

The Motivations and Effects Questionnaire (MEQ) was developed for a study of dating violence (Follingstad et al., 1991) in which victims and perpetrators of dating violence were asked whether thirteen possible motivations (to show anger; due to an inability to express self verbally; to feel more powerful; to get control over the other person; in retaliation for being hit first; to protect self; in retaliation for emotional hurt; anger displaced onto partner; to punish person for wrong behaviour; to prove love; because it was sexually arousing; to get attention; because of jealousy) were present in the person who used force against them or were part of their own motivations for using force. The 29-item Reasons for Violence Scale (RVS), developed by Stuart et al. (2006) asks the percent of time the respondent was physically aggressive toward the partner for each reason given. The items, which were selected by the authors on the basis of a literature review, include reasons such as self-defense, to get away from your partner, to feel more powerful, to hurt your partner's feelings, because you were jealous, etc.

These measures suffer from two major weaknesses. First, all of the measures include a variety of different “reasons” or perceived “motives” for IPV, without clear distinction between their very different levels of influence or a systematic attempt to address all possible domains at these levels. By and large, these measures are limited in types of explanations that they present to respondents, and typically do not include a variety of proximal and distal reasons for IPV. Second, all measures except the PAVE address general tendencies rather than explanations related to specific incidents of IPV. It is critically important that perceived reasons be measured for specific incidents of IPV because focusing on specific incidents allows for concrete referents for the attributions as opposed to using less reliable mental averaging approaches. In addition, focusing on specific episodes of violence provides a basis for identifying perceived reasons that differentiate between different types or severity of aggression within the same couple as well as between different couples.

2.3. A three-level model of perceived influences on IPV

As shown in Figure 1, our conceptualization of perceived reasons for IPV defines three distinct levels of influence on IPV based on their proximity to violence. Level 1 (the most distal) involves stable attributes of individuals such as an aggressive personality, attitudes or beliefs conducive to violence, and an upbringing or childhood experiences that may make a person generally more likely to be aggressive. We label this level background and personal attributes of the perpetrator or victim. This level of explanation has often been invoked in studies of risk factors for IPV (e.g., Johnson et al., 2008; Lloyd & Emery, 2000; Schumacher et al., 2001) but may also be perceived by perpetrators and victims as relevant explanations for why IPV occurred. For example, these attributes might be used to explain a person's propensity for using aggressive solutions or for being unable to restrain him or herself from aggression in certain situations. Problems with alcohol and drugs and long-term mental health problems can be categorized at this level when they are defined as personal attributes of the aggressor rather than as transitory problems.

Figure 1.

Three-level model of perceived reasons for aggression

Level II involves current life circumstances such as stress, depression, poor physical health, current alcohol and drug abuse, and other life circumstances that might make a person more likely to respond aggressively in intimate conflict because of the stress or other effects associated with these life situations or factors. These broader issues can be seen as producing heightened tension that might directly or indirectly lead to conflict and violence.

Level III refers to immediate precursors or precipitators that lead to physical violence. This level includes acts by the partner seen as provoking or aggressive, the emotional or mental state of the perpetrator (including intoxication), and other situational factors. Because most knowledge of precipitants is necessarily based on self-report (rather than, e.g., observation), there is little “objective” knowledge of the specific types of situations that comprise the greatest risks for IPV. Therefore, a comprehensive and systematic examination of perceived precipitators can be useful not only for identifying perceived high risk situations but also for developing more objective ways to identify key precipitators and situations that increase the risk of IPV.

As indicated by the arrows in the model, Level I factors could be perceived as sole explanations for why an aggressive act occurred (i.e., direct link to aggression) but may also be mediated by life circumstances and immediate precursors or precipitators. Similarly, Level II factors may be perceived as a sufficient explanation or be mediated by immediate precursors.

In the sections that follow, we use this model to frame our review of empirical research on perceived explanations for IPV. We summarize the items found in the existing research literature that, in our interpretation, fall into each of the three levels of influence in our model and indicate the relative importance of these explanations based on rates of endorsement provided in the literature. We also identify significant group differences when sufficient data were provided in the original study to allow for significance testing. Finally, we summarize the limitations of the existing research and identify important directions for future research.

3. Methods Used in Reviewing and Summarizing Existing Research

Relevant publications were obtained using a keyword search of three electronic databases: PsycInfo, SocIndex, and PubMed. The keyword search criteria were: (motiv* or reason* or trigger*) AND (violen* or abus*) AND (intimate partner or spous* or dating). Article title and abstracts were reviewed, and all articles that appeared relevant were retrieved in full-text format and evaluated for inclusion in the review. In addition, articles accessed electronically were hand-searched for other relevant studies. A total of sixteen empirical articles that identified specific perceived reasons for IPV form the basis of the present review. A description of these studies can be found in Appendix A.

Because of the large variability in study methods and item wording across the studies included in our review, a meta-analysis was not possible. Wherever possible, however, we conducted statistical analyses to determine whether gender and victim-perpetrator/self-partner differences on specific items within studies met the conventional criterion (p < .05) for statistical significance. If only one comparison was possible (e.g., male versus female perpetrators), statistical significance was estimated using the Chi-squared statistic. Where rates of endorsement for both female and male perpetrators and victims were available, we created a data file in SPSS (version 16) in order to conduct logistic regression analysis for each item, regressing endorsement of the item on gender of the respondent and on the victim-perpetrator status of the respondent. We also tested for significant interactions between gender of respondent and self-partner differences. In addition, we assessed the statistical significance of differences in endorsement rates by the gender of the aggressor (i.e., male aggression versus female aggression, regardless of whether reported by the perpetrator or victim) and severity (for the one study (Cascardi & Vivian, 1995) included in our review that specifically measured perceptions for violence defined as mild and severe).

Throughout the paper, quotation marks are used to indicate the exact wording of closed-ended items and italicized font is used to indicate category labels constructed by authors of the study from open-ended responses. Bold font is used to indicate the themes within our model that were used to organize items.

4. Results

4.1. Explanations or attributions relating to background and personal attributes of the perpetrator or victim

In the studies included in our review, only two items were found that measured perceived explanations for IPV relating to the behavior of parents and other childhood experiences: “My mother would, at times, be physically aggressive to my father or my stepfather” and “I learned when growing up that I could be physically aggressive toward my brother and he would not fight back,” both from a study of college women who initiated violence against male dating partners (Fiebert & Gonzalez, 1997). These explanations accounted for 10% or less of the total number of violent incidents being addressed in the study.

Rates of endorsement for explanations for IPV related to the character, personality or other stable attributes of the aggressor or the victim varied substantially depending on the population under study and the manner in which the dimension was operationalized. Personality of perpetrator was cited by less than 10% of men and women in marital counseling reporting on reasons for the worst incident of partner aggression in the past year (Cascardi & Vivian, 1995). However, a study of explanations for the arrest incident among men and women convicted of domestic abuse against a partner (Henning et al., 2005) found that up to 31% of perpetrators attributed their aggression to some aspect of their own personality and up to 61% attributed their aggression to a characteristic of the victim.

Overall, explanations related to personality, character or stable attributes were more likely to be used by individuals to describe their partner's aggression than their own aggression. For example, while 44% of female and 31% of male respondents in a general population sample of women and men who had been in violent intimate relationships (Carrado et al., 1996) endorsed “it is in [his/her] character, it is the way [he/she] is” as an explanation for their partner's aggression, significantly fewer (16% of female and 27% of male respondents) explained their own aggression as being due to their character.

In the only study that included characteristics of the victim as options (Henning, et al., 2005), perpetrators were significantly more likely to attribute their aggression to attributes of the victim rather than to their own attributes.¹ There were also significant gender differences, with aggression due to alcohol and drug problems more frequently attributed to male partners, and violence related to insecurity attributed more frequently to female partners.

Various attitudes and beliefs were included in a closed-ended study of immediate and “deeper” reasons for IPV perpetrated by college women (Fiebert & Gonzalez, 1997). The most frequently endorsed item of the six items included in the study was “I believe that men can readily protect themselves so I don't worry when I become physically aggressive” (24%), and the least endorsed was “I have seen and admired women in the movies, and on TV, who strike their partners” (3%). Although our literature search did not identify any quantitative studies of perceived attitudes and beliefs of male perpetrators as reasons for IPV, qualitative research has indicated that beliefs and attitudes about gender roles are central to the reasons male perpetrators give for their use of violence, including, for example, the importance of the man being the breadwinner in an intimate relationship and the belief that a real man is strong and dominant and therefore has the right to control his woman (Coleman, 1980; Wood, 2004).

4.2. Explanations or attributions related to current life circumstances (e.g., stressors)

Specific items involving stress as a perceived reason for IPV were included in four studies in our review. Kernsmith (2005) reported that being “under a lot of stress” was one of the most commonly endorsed explanations (28%) for IPV among men in batterer intervention counseling but was not a common explanation among female perpetrators (specific rate not reported in original study). However, female respondents in a study of women who had been court-referred to batterer intervention programs (Stuart et al., 2006) reported that 36.5% of the time their violence was “because of stress.”

Cascardi and Vivian's (1995) study of couples in marital treatment involved coding the presence of various stressors (specifically, financial, marital, family/child, health, or multiple stressors) cited by participants as existing around the time that the worst episode of violence in the last year occurred. As such, stress in this case was not explicitly a perceived cause of IPV, but, rather, a contextual contributor to aggression. Financial/work stress was reported more frequently overall (20% to 79% depending on severity level of aggression and gender of respondent) than other types of stress, especially for male perpetrators explaining their own aggression (79% for severe aggression, 48% for mild), by male victims explaining severe aggression by a female partner (72%) and by female victims explaining severe aggression by a male partner (69%). Male respondents were significantly more likely than female respondents to describe financial stress as a factor contributing both to their own and to their partner's aggression, and this type of stress was significantly more likely to be endorsed for severe than for mild aggression. Related themes of economic reason and he reflects distress in his work were reported by battered women in Turkey as perceived reasons for their husbands' violence against them (Balci & Ayranci, 2005), although open-ended responses associated with these themes were endorsed at relatively low rates (16.3% for economic reason and 7.4% for he reflects distress in his work). However, the interview question eliciting these responses was not clear from the article, hampering clear interpretation of the findings.

In Cascardi and Vivian's (1995) study, marital stress was the most frequently cited form of stress by male victims of mild aggression (45%), and the second most frequently cited by female and male perpetrators of mild aggression (31% and 30%, respectively), with no significant differences by gender, self versus partner attributions, or aggression severity. Family/child stress was the most common type of stress contributing to IPV cited by women explaining their own mild and severe aggression (40% for both), and was tied for first with financial/work stress for female victims describing their partner's mild aggression (28%). Family/child stress was cited significantly more often by female than by male respondents, and was also significantly related to the gender of the aggressor (more likely to be attributed to female than male perpetrators). Other types of stress were cited by 15% or less of respondents in each group, except for multiple stressors, cited by 20% of female perpetrators of mild aggression and health stress, reported by 22% of female victims of mild aggression. Stress about health was described as a contextual factor significantly more often by female than by male respondents.

Stress was also coded in the Cascardi and Vivian study as a specific reason for violence, and was cited at very low rates (ranging from 0 to 10% across groups). This suggests that although current stressors may play a perceived role in the broader context of aggression, stress is not necessarily viewed by individuals as a proximal cause of aggression.

4.3. Explanations related to immediate precursors or precipitators

Level III relates to immediate precursors or precipitators of IPV and includes the following categories of perceived reasons: (1) perceived aggression, provocation or threat to self or relationship by partner; (2) the emotional, mental or physical state of the perpetrator; (3) aggression used as a way of communicating, getting attention or showing feelings; (4) aggression used to control, coerce or exert power over partner; and (5) “hot button” issues.

4.3.1. Perceived aggression, provocation or threat to self or relationship by partner

This category includes a wide range of explanations that has in common the notion that something one partner did prompted the other partner to respond violently, involving, for example, self-defense in response to the partner's aggression, retaliation for something the partner did, aggression in response to the partner's infidelity, and so on.

Self-defense as an explanation for IPV was measured in a number of studies with rates of endorsement varying substantially across studies. “Self–defense” was endorsed by 56% of male and 42% of female university students in aggressive dating relationships (Harned, 2001), 38.7% of women court-referred to batterer intervention programs (Stuart et al., 2006), 35.6% of female college students who had been aggressive toward a dating partner (Makepeace, 1986), and accounted for 46.2% of aggressive episodes reported by women arrested for domestic violence (Hamberger, 1997). In contrast, endorsement rates were considerably lower in a general population study of IPV (21% or less of male and female perpetrators and victims, Carrado et al., 1996), in a study of dating violence (17.7% of male and 18.6% of female respondents explaining their own aggression and 4.1% of male and 4.8% of female respondents explaining their partner's aggression, Follingstad et al., 1991), and among male college students who had been aggressive toward a dating partner (18.1%, Makepeace, 1986). The only significant gender difference in this subcategory was found in the Makepeace (1986) study of dating violence, with females endorsing self-defense more often than males. Significant self-partner differences did emerge, however, in both the Carrado et al. (1996) and Follingstad et al. (1991) studies, with respondents more likely to cite self-defense as a reason for their own aggression than for their partners' aggression.

The notion of retaliation implies that a violent act extends beyond the motive of self-defense to include an element of retribution for something said or done by the partner. This explanation can include retaliation for physical aggression, emotional hurt, or other acts by the partner. Rates of endorsing retaliation as an explanation for physical aggression varied across studies and by type of retaliation. For example, 0% of male victims of dating violence reported that the female partner was aggressive in retaliation for being hit first while 63.3% said the female partner was aggressive in retaliation for emotional hurt (Follingstad et al., 1991). A similar but not as large of an effect was found for female respondents in this study, with 21.7% reporting that their partner was aggressive in retaliation for being hit first but 40.3% attributing their partner's violence to retaliation for emotional hurt.

Our analysis found differences in endorsement of retaliation by gender of the aggressor. In the Follingstad et al. (1991) study of dating violence, retaliation for being hit first was significantly more likely to be used to explain male than female-perpetrated aggression (regardless of whether the respondent was reporting his/her own aggression or that of the partner) while retaliation for emotional hurt was more likely to be used to explain female than male-perpetrated aggression.² Similarly, Kernsmith (2005) found that a significantly larger proportion of female versus male perpetrators in batterer intervention counseling endorsed “getting back at partner for emotional hurt” as an explanation for their own violence (42% versus 22%, respectively).

A significant self-partner difference in retaliation was found in Carrado et al.'s (1996) general population sample for both “getting back for some physical action” (endorsed by 27% of males and 21% of females explaining their own aggression, compared to 8% of males and 12% of females explaining their partner's aggression) and “getting back for something nasty said or threatened” (53% of male and 52% of female respondents explaining their own aggression, versus 44% of males and 32% of females explaining their partner's aggression).

Perceived threats to the relationship as explanations for aggression have been measured as follows: “because your partner cheated on you” (Stuart et al., 2006); “[I/victim] was unfaithful” (Henning et al., 2005); woman [victim] suspected of infidelity (Karamagi et al., 2006); “because you didn't believe that your partner cared about you” (Stuart et al., 2006); “because you were afraid your partner was going to leave you” (Stuart et al., 2006); woman's [victim's] attempts to leave (Dobash & Dobash, 1984); “[I/victim] was not fully committed to [victim/me]” (Henning et al., 2005); and [victim] objecting to mistress (Balci & Ayranci, 2005). Across several studies, infidelity was used to explain violence by between 7.1% and 46.2% of respondents. The low end of this range (7.1%) reflected the percent of a sample of battered women in Uganda who reported that their husbands beat them because of the suspicion that they (the wives) were being unfaithful (Karamagi et al., 2006). A higher rate was found in a study of women arrested for domestic violence, 24.6% of whom said they were violent because “[their] partner cheated on [them]” (Stuart et al., 2006). The highest rates of violence attributed to infidelity were reported in a study of men and women convicted of domestic abuse (Henning et al., 2005), with 46.2% of women and 31.5% of men reporting that they were violent because their partners were “unfaithful.”

Endorsement of explanations relating to commitment issues also varied across studies. Over a third (37.5%) of male perpetrators and half (50.0%) of female perpetrators convicted of domestic abuse (Henning et al., 2005) attributed their violence to their partner's lack of commitment. In the study of women arrested for domestic violence (Stuart et al., 2006), 25.1% said they were violent because they did not believe their partner cared about them and 18.2% attributed their violence to the belief that their partner was going to leave them.

Other types of precipitating actions by the partner measured in previous research on perceived explanations for IPV include: general provocation, starting an argument, lying, disrespect and insensitivity, and sexual refusal. Rates ranged from 0% for the general category provocation coded in an open-ended study of couples in marital treatment (Cascardi & Vivian, 1995) to 48% for “did not receive the respect they deserved” in a study of women in batterer intervention programs (Kernsmith, 2005).

In Stuart et al.'s (2006) study of women arrested for domestic violence, “because your partner was going to walk away or leave a conflict before it was solved” was seen as an explanation for 25.2% of violence, while “because your partner provoked you or pushed you over the edge” was endorsed as a reason for 38.9% of violence. “Partner lied” was endorsed by 10.2% of female university students explaining why they were aggressive toward a dating partner (Hettrich & O'Leary, 2001); “partner started an argument/started yelling” was endorsed by 32% of men in batterer intervention programs (Kernsmith, 2005); and “partner wasn't sensitive to their needs” was endorsed for 46% of violence reported by college women who were aggressive toward male dating partners (Fiebert & Gonzalez, 1997).

Sexual refusal may in some cases involve a perceived threat to the relationship, or, alternatively, may relate to one partner's desire to dominate or exert power over the other partner. We chose to categorize items about sexual refusal in the present category because the manner in which they were defined in the research studies did not provide sufficient detail to identify control or coercion motives. As such, sexual refusal is interpreted here as an action taken by one partner that precipitated the other partner's violent behavior. Rates of endorsement for violence precipitated by sexual refusal were very low (2% to 5.9%) across the three studies that included this item (Dobash & Dobash, 1984; Karamagi et al., 2006; Stuart et al., 2006). These findings suggest that neither perpetrators nor victims view sexual refusal as a salient reason for IPV.

4.3.2. Emotional, mental or physical state of the perpetrator

The second major category of perceived precursors to IPV concerns the perpetrator's state of mind and/or body. The most commonly addressed emotional state that has been measured as a perceived reason for IPV is anger. Six different studies addressed anger in some form as an explanation for partner violence with all studies finding relatively high rates of endorsement. Over a third (39.4%) of respondents endorsed anger as a reason for their aggression in a study of women arrested for domestic violence (Stuart et al., 2006). Similarly, high proportions of men and women endorsed anger as a perceived explanation for IPV in Cascardi & Vivian's (1995) study of mild and severe violence between couples in marital treatment (28% to 52% depending on gender and severity of aggression). For dating violence, 28.3% of men and 24.2% of women endorsed uncontrollable anger as an explanation for their violence in one study (Makepeace, 1986) and 23.6% of female perpetrators endorsed anger as an explanation in another study (Hettrich & O'Leary, 2001). Displaced anger, however, measured in two studies (Follingstad et al., 1991; Stuart et al., 2006), was much less commonly endorsed as an explanation for IPV, with rates varying from 3.4% to 18.4%.

Two studies found that anger was used more often to explain violence perpetrated by women than by men, regardless of whether it was being reported by the perpetrator or victim. For instance, in one study of dating violence (Follingstad et al., 1991), “to show anger” was endorsed by 57.6% of female respondents explaining their own aggression and 59.2% of males explaining their female partner's aggression, compared with 37.5% of male respondents and 40.3% of females explaining the male partners' aggression. Similarly, 29% of females versus 10% of males in batterer intervention counseling endorsed “expressing anger” as an explanation for their own violence (Kernsmith, 2005).

Overall, these results suggest that anger is a common explanation for aggression; however, this explanation is not particularly informative about the underlying conflict or reasons that led to the emotion. As such, meaningful interpretation of findings relating to anger is hampered by the non-specific nature of the measures used to date.

Other aspects of the perpetrator's mental/physical state such as alcohol intoxication, have been measured as perceived immediate precipitators of IPV. Three studies (Carrado et al., 1996; Karamagi et al., 2006; Stuart et al., 2006) addressed substance use/intoxication at the time that violence occurred as an explanation for IPV. Endorsement rates varied by gender of the perpetrator, whether the respondent was the victim or perpetrator and by type of substance.

In the Carrado et al. (1996) general population study, substances were more likely to be used to explain male than female aggression, and were also more likely to be used to explain the partners' rather than the respondents' aggression. Specifically, 13% of female and 35% of male respondents attributed their violence to “being ‘under the influence’ of, for instance, alcohol at the time,” while 31% of male and 45% of female respondents stated that their partner was violent because of being under the influence. An interview study of battered women in Uganda (Karamagi et al., 2006) revealed a much lower rate of violence being attributed to alcohol use at the time of violent episodes, with only 5.3% of the sample attributing the husband's violence to his being drunk. With regard to type of substance, Stuart et al. (2006) found that women arrested for domestic violence were more likely to attribute their aggression to being under the influence of alcohol (17.8%) than drugs (8.4%).

4.3.3. Aggression used as a way of communicating, getting attention or showing feelings

Rates of endorsement for being aggressive to get a partner's attention were generally high, although rates varied by study, perpetrator-victim role and how the construct was measured. For example, 9.6% of women arrested for domestic violence cited reasons for their aggression that were categorized as get partner to talk, attend, listen/do something (Hamberger, 1997) while 64% of male perpetrators of IPV participating in a UK general population survey endorsed “thought it was the only way to get through to her” as a reason for their aggression (Carrado et al., 1996). In between these extremes, about 25% of females arrested for domestic violence (Stuart et al., 2006) reported having been violent in order “to get your partner's attention,” while female college students endorsed “I wished to gain my partner's attention” as a reason for their violence against a dating partner in 44% of violent episodes (Fiebert & Gonzalez, 1997). The similar construct of being aggressive because “my partner was not listening to me” was endorsed by female college students in 43% of incidents in which they were violent toward male dating partners (Fiebert & Gonzalez, 1997), and “partner wasn't listening” was endorsed by 30% of women in batterer intervention counseling (Kernsmith, 2005).

There was a significant self-partner difference in the UK general population study by Carrado et al. (1996), with 53% of female and 64% of male respondents endorsing that violence was “the only way to get through to” their partner as an explanation for their own aggression, compared with 32% of females and 43% of males endorsing this reason to explain their partner's aggression. One study of dating violence (Follingstad et al., 1991) that included the explanation “to get attention” highlighted the impact of how closed-ended items are worded on endorsement rates. In particular, the omission of the word “partner's” before the word “attention” seemed to produce a different pattern of results in terms of self-partner differences compared to those found in Carrado et al.'s (1996) study. Specifically, “to get attention” was more likely to be endorsed for a partner's aggression than for the respondent's own aggression, a pattern similar to that found for some characterological explanations included in Level I and opposite of that found for “to get partner's attention.” These findings suggest that the explanation “to get attention” may be more about the characteristics of the perpetrator (i.e., attention-seeking) than about using aggression to get a partner to pay attention at any given moment.

Communication problems as perceived reasons for aggression were endorsed by moderate to high percentages of respondents, depending on the study and the way the construct was worded. “Inability to express self verbally” was endorsed as an explanation for IPV by 27.1% of female perpetrators, 20.8% of male perpetrators, 28.2% of female victims and 32.7% of male victims of dating violence in one study (with no significant differences) (Follingstad et al., 1991) and by 50% of female perpetrators and 41% of male perpetrators of dating violence in a different study (also not a significant gender difference) (Harned, 2001). Similar constructs, “to show feelings that you couldn't explain in words” and “because you didn't know what to do with your feelings,” were endorsed by 38.0% and 35.2% of women arrested for domestic violence, respectively (Stuart et al., 2006). Qualitative interviews identified express feeling/tension as a motivation for IPV among 19.2% of women arrested for domestic violence (Hamberger, 1997).

Miscommunication was endorsed as a reason for IPV by 24% of perpetrators and 23% of victims (males and females combined) in a general population survey in which respondents were asked about the cause of their most recent conflict (Stets & Henderson, 1991). However, only 4.7% of female university students who had perpetrated dating violence endorsed “poor communication” as a reason for their aggression (Hettrich & O'Leary, 2001). “Communication” as an explanation for aggression was endorsed by 50% of men explaining their own violence and 21% of women explaining their husband's violence (significant difference between male perpetrators and female victims) in a study of men in treatment for alcoholism and wives of men in alcoholism treatment (Murphy et al., 2005).

Proving love (measured by Follingstad et al., 1991; Harned, 2001; Stuart et al., 2006) and showing affection (Murphy et al., 2005) were also identified in some research as perceived reasons for IPV, with generally low endorsement rates (ranging from 0% to 15%), except for the findings from one study of females arrested for domestic violence (Stuart et al., 2006), in which 27.1% of respondents endorsed using violence “to prove you love your partner.” One study of physical aggression in dating relationships (Harned, 2001), found that men were significantly more likely than women to endorse being aggressive to prove love (15% compared to 4%).

4.3.4. Aggression used to exert power or intimidate, to control or coerce partner, and to punish partner

Several studies measured exerting power over or intimidating partner as reasons for IPV. In a study of women arrested for domestic violence (Stuart et al., 2006), 26.1% endorsed “to feel more powerful” as a reason for their own aggression. In a study of dating violence (Fiebert & Gonzalez, 1997), female perpetrators endorsed “I feel personally empowered when I behave aggressively against my partner” for 12% of their violence. In another study of dating violence (Follingstad et al., 1991), endorsement of “to feel more powerful” as a motive for aggression depended on whether the respondent was the victim or perpetrator, with 20.4% of male and 31.5% of female respondents endorsing this explanation for their partner's aggression, versus 0% of male and 3.4% of female respondents who endorsed this as a reason for their own aggression.

Intimidation as a reason for IPV was measured in two samples of perpetrators. In a study of females arrested for IPV (Stuart et al., 2006), 11% of respondents said they used violence “to make your partner scared or afraid,” while in a study of dating violence (Makepeace, 1986), “intimidate” was endorsed by 21.3% of male and 6.8% of female respondents (significant gender difference) to explain their own aggression.

Several studies included items relating to aggression to coerce or control partner such as making the partner do something or stop doing something. High endorsement (with a significant victim-perpetrator difference) was found for the item “to make [victim] do what [perpetrator] wanted” in a general population study of reasons for conflict, with 26% of both male and female perpetrators endorsing this reason for their own aggression versus 51% of male victims and 43% of female victims attributing their partners' aggression to this reasons (Carrado et al., 1996). In a sample of women arrested for domestic violence (Stuart et al., 2006), 22.4% endorsed being aggressive “to get your partner to do something or stop doing something” and 16.9% endorsed using violence “to make your partner agree with you” as an explanation for their aggression

Moderately high rates were found for aggression done for the purpose of preventing a partner from doing something, as measured in the following studies: “to stop [victim] from doing something,” endorsed by 29% of female victims, 37% of male victims, 33% of female perpetrators and 43% of male perpetrators in the UK general population study (Carrado et al., 1996); “stopping partner from doing something,” endorsed by 21% of women and 15% of men in a batterer intervention counseling program (Kernsmith, 2005); “to shut your partner up or to get your partner to leave you alone,” endorsed by 23% of women arrested for domestic violence (Stuart et al., 2006); “when partner was nagging them,” endorsed by 40% of men in batterer intervention counseling (Kernsmith, 2005); and stop nagging/get other to shut up, coded as a motive for 11.5% of women arrested for domestic violence (Hamberger, 1997).

Endorsement rates for the motive “to get control over the other person” were generally high, although one study found large differences by gender and victim-perpetrator role. Specifically, 55.6% of female victims, 26.5% of male victims, 22.0% of female perpetrators and 8.3% of male perpetrators endorsed this as a reason for IPV in a study of dating violence (Follingstad et al., 1991). This motive was also endorsed by 27% of female and 44% of male perpetrators in an internet survey of motives for physical IPV among students (Harned, 2001) and by 21.7% of women arrested for domestic violence (Stuart et al., 2006).

Punishment as a motive was measured directly in two studies and implied as a motive in a third. In a study of female perpetrators arrested for domestic violence (Stuart et al., 2006), respondents endorsed “to punish your partner” as a reason for their aggression in 24.7% of incidents. In another study of university students who were victims or perpetrators of dating violence (Follingstad et al., 1991), there was a (nonsignificant) trend for victims to be more likely to attribute aggression by the partner to “punishment for wrong behavior” (26.6% of female victims and 28.6% of male victims) than for perpetrators to attribute their own aggression to this motive (16.9% of female and 12.5% of male perpetrators). In interviews with battered women in Uganda (Karamagi et al., 2006), punishment for perceived wrongdoing by the women was reflected in the following reported reasons for IPV by their husbands: neglect of housework (reported by 13.5% of respondents), being away without husband's permission (12.9%) and disobedience/disrespect of husband/in-laws (8.2%).

4.3.5. “Hot button” issues

Certain topics or “hot button” issues have been perceived as playing a contributing role to IPV. In one study (Murphy et al., 2005), men in treatment for alcoholism and wives of men in alcoholism treatment were asked about topics leading to aggression by the male partner. Two other studies identified topics leading to IPV from open-ended interviews with battered women in refuges in Scotland (Dobash & Dobash, 1984) and with female victims of IPV in Uganda (Karamagi et al., 2006). “Money” was a frequent topic of conflict leading to IPV in the alcoholism treatment sample, as reported both by men in alcoholism treatment (40%) and the wives of men in alcoholism treatment (35%) (Murphy et al., 2005), but was identified as the topic leading to aggression by less than 20% of battered women in Scotland (Dobash & Dobash, 1984), and by only 8.8% of Ugandan women who were asked why their husband beat them (Karamagi et al., 2006). IPV was attributed to conflict over children by 22% of male alcoholics who had perpetrated IPV and 16% of the wives of male alcoholics (victims of IPV) (Murphy et al., 2005) but by less than 6% of respondents in the study of battered women in Scotland (Dobash & Dobash, 1984).

Expectations about domestic work was identified by battered women as a frequent topic of conflict associated with IPV (Dobash & Dobash, 1984) (37%, 34% and 32% for first, worst and last episode of violence, respectively). The similar item “household tasks,” however, was less frequently endorsed by men in alcoholism treatment (20%) and wives of male alcoholics (12%) (Murphy et al., 2005). This difference in rates of endorsement between these two studies may reflect the fact that expectations about domestic work for the battered women sample is more about power and control relating to the patriarchal dynamics involved in wife battering than it is about being a “hot button” topic.

“Husband's alcohol use” was endorsed as a topic of conflict precipitating violence by 68% of female victims and 55% of male perpetrators, while “husband's drug use” was endorsed by 29% of females and 25% of males in the study of men in alcoholism treatment and wives of men in treatment (Murphy et al., 2005). Other items in this study included: conflict over “living situation” (endorsed by 18% of husbands and 21% of wives); conflict over “making decisions” (12% of both husbands and wives); and conflict over “parents or in-laws” (12% of husbands and 10% of wives). There were no significant differences between male perpetrators and female victims on any of these items.

Less frequently endorsed “hot button” issues were related to relationship issues; specifically, “amount of time spent together” was endorsed by 18% of male alcoholics who perpetrated IPV against their wives and 16% of wives of alcoholic males (Murphy et al., 2005), and “how to spend time together” was endorsed by 15% of male perpetrators and 12% of female victims in Murphy et al.'s (2005) study.

4.4. Other perceived reasons

The literature we reviewed included some perceived reasons that could not be classified, either because their meanings were ambiguous or because they did not represent a clearly defined category of reasons for IPV. Reasons that could not be categorized that were endorsed relatively frequently included: “I/victim was not willing to make compromises,” cited by 22.6% of females and 21.2% of males convicted of domestic assault in reference to their own unwillingness to compromise and significantly more (55.7% of females and 45.4% of males) in reference to their partner's (the victim's) unwillingness to compromise; “I did not believe my actions would hurt my partner,” endorsed for 38% of violence reported by female perpetrators of dating violence (Fiebert & Gonzalez, 1997); “[perpetrator] felt partner [victim] was trying to control them,” endorsed by 37% of women in batterer intervention counseling (Kernsmith, 2005); husband [perpetrator] has another partner, cited by 24.1% of female victims of IPV in Uganda; “to hurt your partner's feelings,” cited for 20.4% of incidents of violence by women arrested for domestic violence (Stuart et al., 2006).

Less frequently endorsed reasons in our “other” category included: “to harm,” endorsed by 8.3% of female and 2.4% of male perpetrators of dating violence (Makepeace, 1986); lack of love and respect, cited by 4.4% of female victims of IPV in Turkey (Balci & Ayranci, 2005); “embarrassed,” endorsed by 3.9% of female perpetrators of dating violence (Hettrich & O'Leary, 2001); “because it was sexually arousing” (endorsement rates of 0% to 18% across different studies of dating violence (Follingstad et al., 1991; Harned, 2001) and among women arrested for domestic violence (Stuart et al., 2006)); “I sometimes find when I express my anger physically I become turned on sexually” (accounting for 8% of violent incidents in Fiebert & Gonzalez's (1997) study); and husband influenced by friends or relatives, cultural differences and not being sexually satisfied (cited by 14.1%, 3.7% and 2.2% of Turkish battered women, respectively, Balci & Ayranci, 2005).

Several studies included reasons related to substance use that could not be categorized because it was unclear if the item referred to substance use at the time of violence or substance abuse problems more generally. These included items infrequently endorsed (less than 7% of battered women who cited husband's drinking behavior, Dobash & Dobash, 1984), frequently endorsed (alcohol or gambling cited by 37.0% Turkish battered women, Balci & Ayranci, 2005), and variable endorsement depending on gender and victim-perpetrator role of respondent (substance use cited by 0% of female perpetrators and male victims, 17% of male perpetrators and 28% of female victims among men and women in marital counseling, Cascardi & Vivian, 1995).

5. Discussion and Conclusion

In the present paper, we proposed a three-level conceptual model for perceived reasons for IPV that includes the following domains of factors: I. Background and personal attributes of the perpetrator or victim; II. Current life circumstances; and III. Immediate precursors or precipitators. We used this model to categorize and analyze the existing research literature that addresses victims' and perpetrators' explanations for IPV, reviewing a total of 16 empirical studies. We conclude this paper by summarizing our review findings and the implications of our review for the study of perceived reasons for IPV.

5.1. Summary and discussion of review findings

For the first level of explanation (background and personal attributes), rates were generally low for factors relating both to attitudes and childhood experiences of violence. Only one study was found that explicitly measured the perceived link between attitudes and IPV, and this study addressed dating violence committed by a sample of female undergraduates evaluating their own aggression (Fiebert & Gonzalez, 1997). However, links between attitudes and male violence have been reported in the qualitative research literature (Coleman, 1980; Wood, 2004) and are evident in some of the perceived explanations we reviewed as precursors and precipitators. Thus, the present review underscores the need for more research that examines systematically perceptions related to this level of influence on IPV in diverse and representative samples, especially pertaining to the role of attitudes of the perpetrator.

Findings relating to character/personality factors varied across studies, with endorsement rates ranging from 0% to 61%. For the majority of items, significance testing revealed no gender differences. However, consistent with the broader psychological research on attributions (Fiske & Taylor, 1991; Jones & Nisbett, 1972; Miller & Ross, 1975; Ross & Nisbett, 1991), respondents were more likely to use characterological explanations in reference to their partner's violence than their own. The large variability in responses relating to the extent to which personality or character are perceived as contributing to IPV suggests a need both for greater standardization of measurement and larger more representative samples. Such research could help to tease apart gender and self-partner differences and allow for the identification of attributes that are perceived to be the most frequent contributors to partner violence.

Our second level of explanations involves current life circumstances that are perceived as contributing to IPV. In one study (Cascardi & Vivian, 1995), importance of different types of stress varied by gender, with financial and work stress topping the list of male stressors and family/child and health stressors as the top stressors associated with violence reported by women. In general, it appears that stress may be viewed as an important contextual contributor to violence, although perhaps not as a specific reason for violence. However, because the perceived role of stress as a contributing factor was examined in only a few studies with small select samples, it is difficult to draw conclusions about the extent to which victims and perpetrators view stress and other factors at this level as reasons for IPV. More definitive conclusions about gender differences and the perceived importance of various stressors require further investigation.

Research at this level should also go beyond the general concept of stress to examine other contextual factors (such as depression and health problems). For instance, in a study of opinions about the importance of a variety of potential causes and triggers of IPV among 460 married women in Iran (Hamzeh, 2008), there was high endorsement of items related to current life circumstances including: “addiction of a partner” (endorsed by 92.2% of the sample), “mental disorder of a partner” (88.7%), “husband's long-term unemployment” (83%), “difficulties at husband's work” (62.1%), and “economic problems in the household” (45.7%). Because this study examined opinions about reasons for IPV among a community sample, the above rates may or may not be reflective of perceived reasons for violence personally experienced by the sample. Nevertheless, these results suggest a need to further explore how life circumstances are perceived as contributing to IPV, including whether issues such as addiction and mental health problems are perceived as character or personality problems of the perpetrator (i.e., Level I contributors), current life circumstances leading to IPV (i.e., Level II contributors), or current mental state of the perpetrator (e.g., alcohol intoxication) or “hot button” issues that trigger aggression (i.e., Level III contributors).

Both the first and second levels of explanations in our model were addressed in relatively few of the studies included in our review. Some of the items associated with these levels, particularly the background factors in Level I, are more commonly seen in the IPV literature as “risk factors” – that is, characteristics of the person, family or culture that increase risk of IPV (WHO, 2009). Future efforts to develop a comprehensive measure of perceived influences on IPV should evaluate more fully the extent to which victims and perpetrators themselves view attitudes, personality and background factors as playing a role in IPV. The relevance of understanding victims' and perpetrators' perceptions pertains at least in part to the extent that these influences are seen as malleable factors that can be addressed using approaches that focus on the conflict situation versus stable personality and attitudes of the perpetrator for which personal counseling and other interventions directed toward the perpetrator are needed. As we discuss below, self-partner differences in perceptions relating to these factors may also play an important role in prevention of IPV.

The third level of explanations, involving perceptions regarding immediate precursors or precipitators relating to IPV, contained the most numerous and diverse set of items of all three levels of explanations in the review. Perceived provocation by a partner was associated with the motives of self-defense and retaliation, both of which were commonly reported reasons for IPV among some samples. Threats to the relationship, especially infidelity and lack of commitment, were commonly endorsed as perceived reasons for aggression. Anger (but not displaced anger) was also a frequently endorsed explanation for IPV. Using violence to show feelings that could not be explained in words and as a way to get through to a partner were important motives across various types of samples. Certain forms of controlling a partner and influencing his/her behavior were also frequent explanations for IPV, with explanations relating to coercion and control generally perceived as more important than explanations related to feeling more powerful, intimidation, and punishment.

5.2. Gender differences in explanations for IPV

The issue of gender differences in IPV perpetration represents a source of ongoing debate in the partner violence literature. According to Holtzworth-Munroe (2005), the question of whether male and female aggression is similarly motivated remains unresolved, with existing research showing some overlap in women and men's reported motivations as well as some differences (with women more frequently claiming reactive or expressive aggression and males more often endorsing the use of violence for instrumental reasons (i.e., to achieve a desired outcome) (Cercone, Beach & Arias, 2005)). In contrast, Straus (2009) views the empirical evidence as more symmetrical, arguing that both women and men are most commonly motivated by anger and desires to coerce and punish a partner's misbehavior.

Our review suggests that violence perpetrated by women is seen in reactive or expressive terms somewhat more than is the violence of men. Specifically, we found that reasons for violence associated with retaliation for emotional hurt and out of anger were more frequently endorsed by female than by male perpetrators. Greater use of aggression for instrumental reasons by men, however, was somewhat less evident from our review, with coercion and control emerging as important for both sexes. However, gender and victim-perpetrator differences in the perceived role of coercion suggest that this issue needs further investigation. In particular, one study (Follingstad et al., 1991) found that although a large proportion (55.6%) of female victims endorsed “to get control over other person” as a reason for their male partner's aggression, corresponding endorsement by male perpetrators was rare (8.3%), suggesting important differences in victim-perpetrator perspective on this issue. The discrepancy was less evident for male victims (26.5%) and female perpetrators (22.0%), suggesting that victim-perpetrator differences may vary by gender. The same pattern emerged from this study for “to feel more powerful” as a reason for aggression.

Gender differences in the extent that control and power are perceived as explanations for aggression may also depend on how these constructs are defined. For example, in one study of dating/courtship violence, female perpetrators were more likely than male perpetrators to endorse “to feel more powerful” as a reason for their aggression (Harned, 2001); however, in another study of dating violence, female perpetrators were less likely than male perpetrators to endorse “to intimidate” (Makepeace, 1986).

Some gender differences were found that reflect, at least in part, gender roles, such as the differences in the type of stress that were perceived as contributing to IPV (Cascardi & Vivian, 1995), with financial and work stress more likely to be identified for men and family stress more relevant for women. Consistent with heavier alcohol and drug use by men than by women in all societies (Wilsnack et al., 2009), substance use tended to be identified as more of a contributor to male-perpetrated than female-perpetrated IPV.

Other types of gender differences were apparent in a study of men and women convicted of domestic abuse (Henning et al., 2005), with female perpetrators being more likely than male perpetrators to attribute their aggression to their partners' infidelity and lack of commitment. Self-defense as an explanation for IPV was endorsed significantly more often by female than by male perpetrators in one study of dating violence (Makepeace, 1986), but no significant gender difference in endorsement of self-defense was found in a different dating violence study (Follingstad et al., 1991). Finally, Follingstad et al. (1991) found that male perpetrators were more likely than female perpetrators to endorse “proving love” as the reason for their aggression.

It is clear that gender differences exist for at least some perceived reasons for IPV, although few consistent patterns were found across studies, even for studies involving similar populations. Therefore, the present review underscores the need for systematic research with large representative samples in order to identify consistent patterns of gender differences and develop a clearer understanding of how gender differences in perceptions of reasons for IPV may contribute to gender differences in experiences of IPV.

5.3. Self-partner differences in explanations for IPV

Compared to the issue of gender differences, self-partner differences in perceived reasons for IPV are less commonly addressed in the existing literature. Nevertheless, such differences emerged more prominently in our review than was the case for gender differences. Significant self-partner differences in perceptions were found for explanations pertaining to personality/character, with individuals being more likely to blame violence on their partner's character ‘deficiencies’ than their own. This result is consistent with findings from attribution theory research that people tend to associate dispositional influences (e.g., character) to the behavior of others while being more likely to attribute their own behavior to situational factors (Jones & Nisbett, 1972; Ross & Nisbett, 1991). Knowledge of basic patterns in self-partner attributions may be especially important for improving prevention programming directed toward situational couple violence. This knowledge may also be useful in addressing perceptions and misperceptions in treatment programs for those who engage in sustained abusive or aggressive acts by acknowledging the situational factors that trigger their aggression while identifying their own role in choosing violence over alternative responses.

Self-partner differences were also found for self-defense, with both men and women being more likely to explain their own aggression as having occurred in self-defense than they were to endorse this as an explanation for their partner's aggression. It is perhaps not surprising that respondents would tend to attribute their own behavior, but not their partner's, to self-defense (even if it may not have been). Both legally and morally, self-defense represents a legitimate ‘excuse’ for aggression, and there are perceptual biases that make it more likely that a person would see their own motives as self-defensive while being less likely to attribute their partner's violence to self-defense. For instance, given that threat is often a subjective experience, a person may perceive their partner's behavior as threatening at times when the partner does not intend or perceive his/her own behavior that way. In some contexts, the result of this disconnect in experience might be aggression that one partner perpetrates with the goal of self-protection, while the other partner, not perceiving or intending their own behavior to be threatening, may view the same action as an unprovoked initiation of violence against them. Overall, these findings suggest that greater research attention needs to be paid to self-partner and victim-perpetrator differences in perceived reasons for IPV. Understanding these differences may also represent an important direction for prevention and intervention approaches.

5.4. The need for standardization of construct conceptualization

Our review found evidence to suggest that variations in how constructs are specified can have an important impact on rates of endorsement. Take, for example, the concept of retaliation. In Makepeace's (1986) study of dating violence among university students, “retaliation” did not receive particularly high rates of endorsement (18.9% of females and 16.5% of males reporting on their own aggression). However, when retaliation was measured separately for physical (“being hit first”) versus emotional (“emotional hurt”) harm (Follingstad et al., 1991), 55.9% of female and 25.0% of male respondents endorsed retaliation for emotional hurt to explain their own aggression, but only 13.6% of females and 29.2% of males endorsed retaliation for being hit first. Such variations in rates of endorsement suggest that retaliation may not be a unidimensional construct, and that the manner in which it is conceptualized and specified may have an impact on the importance it is assigned by victims and perpetrators. A similar conclusion can be applied to the findings related to use of violence to “get partner's attention” (which generally received high endorsement by perpetrators) versus “to get attention” (which received low endorsement and was endorsed more by victims than by perpetrators).

In sum, development of a comprehensive measure of perceived reasons for IPV requires analysis of the effect of wording on endorsement of complex emotional and behavioral constructs. In the case of “anger” as an explanation for aggression, for instance, more specific definitions of this construct are needed to clarify the perceived role played by anger – for example, anger (or poor anger management) as a general temperament, anger provoked by hot button issues or some other aspect of the conflict, being angry at something else and taking it out on a partner, and so on.

Imprecise specification of perceived reasons generally makes it difficult to review and draw conclusions from the existing literature. This limitation was particularly evident in the case of explanations involving jealousy, which was included as a perceived reason for IPV in a number of studies (Cascardi & Vivian, 1995; Dobash & Dobash, 1984; Follingstad et al., 1991; Hettrich & O'Leary, 2001; Murphy et al., 2005; Stets & Henderson, 1991; Stuart et al., 2006), but not reported in our results because of the inability to identify the level of influence being measured. Endorsement rates for explanations involving jealously varied from 0% to 42%, partly reflecting the diversity of samples, but also likely reflecting imprecision of measurement. Interpreting the meaning of these results relative to other types of explanations in the research literature is hampered by the lack of conceptual clarity in the manner that jealousy was being addressed. Conceptually, jealousy can be a stable attribute of an individual's character (i.e., the aggression occurred because the perpetrator is a jealous person), a transient emotional state, or a response to a perceived relationship threat. Use of a conceptual model, such as the one that we have proposed, can thus lead to better conceptualization and operationalization of constructs, which in turn can lead to better elaboration of a theoretical model for perceived influences on IPV and an increase in knowledge that has practical applications.

5.5. Future directions

Lack of a comprehensive and psychometrically-sound instrument for measuring motivations of partner violence has resulted in a body of literature characterized by inconsistencies and wide gaps in knowledge. Our review has indicated major limitations in conceptual clarity and specification of the proximity of perceived influences to incidents of IPV. Our findings of large differences in endorsement depending on how an item is worded or measured suggest that greater attention to content and construct validity and construct coverage would contribute to a better understanding of the different levels of perceived influences on IPV. In addition, there is a need both for more standardization of measurement and larger representative samples in order to identify the reasons that are perceived to be most frequent contributors to IPV in given situations, as well as to produce more systematic knowledge of differences by gender, self versus partner attributions, and severity of violence.

Appendix A: Methodological details of studies reporting rates of perceived reasons/explanations for IPV

1. Balci & Ayranci (2005): Face-to-face interviews and questionnaires were used to collect data from 135 battered women in Turkey, including a question on women's views of the reasons for their partner's violence. The items reported in our paper represent the authors' coding of women's open-ended responses about their general experience of violence.

2. Carrado et al. (1996): From a representative sample of 1,978 heterosexual men and women aged 15 and over from the general UK population, data were collected from a subsample that had ever sustained (155 men, 130 women) or inflicted (85 men, 106 women) IPV. On a survey with face-to-face and self-report components, respondents could endorse any of 8 items about the reasons/context they associated with their previous experiences of IPV.

3. Cascardi and Vivian (1995): The open-ended Semi-Structured Marital Interview was used to obtain information on the worst episode of physical IPV perpetration and/or victimization in the past year among men and women in marital counseling. Relevant questions addressed stressors at the time of the violence and perceived function of the violence. Respondents reporting mild aggression included 20 female perpetrators and 18 female victims, 29 male perpetrators and 13 male victims. Respondents reporting severe aggression included 25 female perpetrators and 23 female victims, 14 male perpetrators and 32 male victims.

4. Dobash & Dobash (1984): In-depth interviews were conducted with 109 battered women connected with Women's Aid refuges in Scotland. Primarily open-ended questions were asked about the sources of conflict during the first, worst, last, and typical incidents of IPV perpetrated by their husbands.

5. Fiebert and Gonzalez (1997): Initial data were gathered on 978 women enrolled in undergraduate classes in community colleges and state universities. Potential participants completed the Domestic Behavior and Analysis Form, on which 285 reported initiating violence against a male partner in the past five years. This subsample answered closed-ended questions on their reasons, both ‘immediate’ (5 possible items to endorse) and ‘deeper’ (10 possible items), for their use of violence in dating relationships.

6. Follingstad et al. (1991): A sample of 207 male and 288 female university students, recruited from undergraduate university classes, completed the Motivations and Effects Questionnaire (MEQ) developed for this study, in which victims and perpetrators of dating violence could endorse any of 13 possible motives. We report rates for 33 males and 82 females who experienced physical violence victimization and 24 males and 59 females who reported physical violence perpetration.

7. Hamberger (1997): Descriptive information about the context of partner violence was collected through open-ended interviews with 52 women arrested for domestic violence. Participants were asked about their typical motivation for IPV.

8. Harned (2001): Data on dating violence were collected through an Internet-based survey, which included items on motives for the use of physical violence. Of a stratified random sample of 1,150 students completing the initial survey, 83 women and 39 men who reported perpetrating dating violence completed a 12-item version of the MEQ developed by Follingstad et al. (1991). A 6-point scale (ranging from 0—never to 6—always) measured how often their violence against dating partners was motivated by each item.

9. Henning et al. (2005): Data from the Domestic Violence Assessment Center database were used to examine motivations for IPV perpetrated by 1,267 men and 159 women convicted of domestic assault. In reference to the arrest incident, sixteen items (8 reasons asked once for self-blame and again for victim-blame) were prefaced with the statement “This incident happened because…” to which respondents indicated “not true,” “somewhat true,” or “very true.” Rates reported in the study are for respondents who indicated either “somewhat true” or “very true.”

10. Hettrich and O'Leary (2001): A convenience sample of 483 female university students completed screening questionnaires for participation in a study of dating violence. Those reporting use of physical aggression toward a dating partner (n=127) completed the Reasons for Aggression Scale, a two-part questionnaire. In the open-ended section, women described the two most recent physically aggressive incidents with their partner and explained what led to the aggression. In the closed-ended component, participants could indicate the extent to which each of 12 reasons was a factor in these acts of aggression. Items were rated on a 5-point scale (ranging from 1—not a cause to 5—main cause). The reasons most commonly endorsed as the “main cause” of the two most recent incidents of IPV were reported by the authors.

11. Karamagi et al. (2006): This study involved a cross-sectional household survey of women aged 18 and above with infants in Eastern Uganda. Data were collected using an interviewer-administered questionnaire with open-ended and closed-ended questions. We report on data collected from 170 female victims of IPV who answered an open-ended question about their perceptions of why their husbands beat them.

12. Kernsmith (2005): A cross-sectional survey of 125 participants (66 men and 59 women) in batterer intervention counseling programs asked participants to identify the typical reasons they use violence in intimate relationships. A 16-item version of the MEQ (Follingstad et al. 1991) was used, with each item measured on a 5-point Likert-type scale ranging from “very much [a reason]” to “not at all [a reason].” Selected results of most frequently endorsed motives were provided in the study; therefore, we do not have data for items that received low endorsement rates, and thus some items could not be evaluated statistically.

13. Makepeace (1986): Data were gathered as part of the College Premarital Abuse Project, a survey of 2,238 students from 7 U.S. colleges and universities. Questionnaires addressed experiences of courtship violence, including motives for perpetrating violence in dating relationships. This article is based on the worst incident data from 127 men and 264 women who reported inflicting or sustaining physical aggression in a courtship context.

14. Murphy et al. (2005): This study involved a comparison of violent and non-violent conflicts involving alcoholic men and their female partners. 40 alcoholic men and 68 female partners of alcoholic men completed a questionnaire and situational conflict interview. Closed-ended questions included identification of whether each of 25 given topics was involved in the couple's violent and non-violent conflicts. Our analyses use only the rates for conflicts that involved physical aggression. Selected results of the most common topics for each sex were provided in the study, meaning that we do not have specific rates for items that received low endorsement, and thus some items could not be evaluated statistically.

15. Stets and Henderson (1991): A national U.S. sample of 272 individuals (54% male, 46% female) between the ages of 18 to 30 was obtained through random digit dialing. The telephone interview involved open- and closed-ended questions, including an open-ended question about the cause of the most recent conflict tactic used or received. Note that results were provided only for incidents (that is, data on motivations were not separated by gender).

16. Stuart et al. (2006): 87 women arrested for domestic violence and court-referred to batterer intervention programs completed the Reasons for Violence Scale (developed for this study). The scale consisted of 29 possible reasons for perpetrating partner violence; respondents rated each reason on a scale of 0% to 100%, indicating how much/how often each reason was a factor in their perpetration of violence.