Introduction

This commentary is a response to “Controversies involving Gender and Intimate Partner Violence in the United States” by Langhinrichsen-Rohling (2010) I argue that her emphasis on gender parity in violence as a basis for intervention confounds two distinct realities, the normative use of violence in families and relationships and the types of partner assault and coercive control targeted by the advocacy movement and by public policy and that lead their primarily female victims to seek outside assistance. I trace this confusion to the domestic violence paradigm that abstracts violent acts from the history, contextual dynamics, experience and consequences of abuse in relationships. After reviewing the evidence for gender parity in violence and asymmetry in abuse, I outline an alternative paradigm that distinguishes partner assaults from coercive control; sets violent acts in an historical, experiential and political context; includes multiple tactics in addition to violence; and emphasizes harms to autonomy, personhood and basic rights as well as to physical security. I close by describing why I believe feminist researchers should focus on domestic violence that reinforces gender inequality or male domination, what Michelle Dempsey (2009) terms “domestic violence in its strong sense,” rather than domestic violence that does not, “domestic violence in the weak sense.”

Overview

Relying largely on evidence from population surveys, Langhinrichsen-Rohling (2010) argued that women are as prone to use violence as men in relationships and that the modal dynamic in these situations involves “bi-directional” violence. She reviewed the challenges these data pose to measuring abuse, showed why it is critical to address women’s violence even among victims who use shelters, proposed a typology of bi-directional violence and highlighted the role of alcohol and psychopathology in abuse, two important problems that deserve more attention in domestic violence research. She acknowledged important differences in the dynamics and outcomes of male and female partner violence, but argued that these reflect cultural and individual factors that are best managed when clinicians attend to the bi-valent patterns she highlighted.

With the qualifications outlined below, I concur with Langhinrichsen-Rohling’s general empirical claim, that female as well as male partners resort to violence in large numbers, that their motives for doing so are often similar, and that both parties are often violent. I also agree that advocates as well as clinicians would benefit from a greater appreciation of the extent, dynamics and roots of women’s use of force in relationships, including the role of behavioral and psychological problems. The typology Langhinrichsen-Rohling (2010) proposed might be useful in clinical work with couples or families where violence is an issue and may have specific applicability to designing programs to treat relationship violence by women.

Where I depart from Langhinrichsen-Rohling’s analysis is in whether and how the evidence of gender parity in violent acts bears on our understanding of or response to “abuse” or battering, the problem which the domestic violence revolution set out to address (Stark and Flitcraft 1996; Stark 2007; Schechter 1982). I use the term “abuse” cautiously, recognizing that where dependent persons are involved, it refers to the illegitimate use of power by someone who has an otherwise legitimate or formal responsibility for their care and protection such as a parent or guardian. Since formal equality is the norm among adult partners, however, at least in highly industrialized societies like the U.S., I use partner abuse to refer to the non-voluntary establishment of unreciprocated authority by one party over the other and the corresponding reallocation of resources and opportunities in ways that benefit the dominant party (Young 1990). To clarify our differences, I return to the puzzle that underlies the challenges Langhinrichsen-Rohling (2010) identifies, how to reconcile evidence of gender parity in violent acts by partners with the equally well documented gender asymmetry in the dynamics and effects of partner abuse. I focus on the distinction between the patterned subjugation of one partner by the other that characterizes the dynamic the advocacy movement has identified as abuse with the widespread propensity for individuals or couples to use violence when they fight, to express jealousy, frustration or anger, settle conflicts, or to negotiate power differences. While I recognize that clients and treaters may contract to address violence used in these latter contexts through individual or couples’ work, this behavior rarely involves abuse as I use the term and does not merit non-voluntary consignment to institutional or therapeutic management.

The current domestic violence paradigm takes violent acts as its unit of analysis so long as their source is a partner and disregards their history, motive, dynamic, context, consequence or meaning to those involved. From this vantage, the only way to distinguish normative from abusive violence is in the physical valence and proximate outcome of these acts. The same confusion attends the distinction between so-called “psychological abuse” (Tolman 1989) and the insults, jealousy, demands for compliance and types of controlling behavior that are commonplace in relationships. Application of this paradigm, I believe, has led to mistaken generalizations about the nature of abuse based on survey evidence of violent acts.

Except in the relatively rare instances involving extreme violence, to understand whether and how the use of force in relationships contributes to abuse, we have to set violent acts in their historical context, consider the co-occurrence of other coercive and controlling tactics, take the subjective experience of those who are targeted into account, whether they experience themselves as victimized and seek outside assistance, for instance, and assess individual and societal harms that extend beyond physical injury, fear or psychological trauma. The model of coercive control has the distinct advantage of allowing us to make these connections and so eliminates the confusion occasioned when evidence of gender parity in violent acts is juxtaposed to gender asymmetries in help-seeking by domestic violence victims.

I remain agnostic about whether force is always unacceptable in relationships or families. I exclude what I call “fights” involving physical force from the category of abuse not because I believe they are inconsequential or acceptable, but because I believe that special societal attention is merited by coercion that occurs in the context of and reinforces inequality, in this case gender inequality. This is both an ethical position that derives from the special vulnerability of persons who are unequal because of their age, race, sex or societal status, and reflects the fact that coercion in the context of structural inequality has different dynamics and individual and societal consequence even when it is bi-directional than the use of force among relative equals. Because male–female abuse (but not female–male abuse) demonstrably harms women’s chances for equal personhood within and outside of relationships, I believe it should be the legitimate object of feminist research and activism. Conversely, I do not believe there is compelling evidence that any substantial proportion of men assaulted by female partners want or require more protections, assistance and support than are currently available.

The Domestic Violence Paradigm

Summarizing the dominant view in the domestic violence field, Gelles (1997) defined violence as an “act carried out with the intention or perceived intention of causing physical pain or injury to another person” (p. 14). From here, identifying domestic violence should have been a simple matter of determining whether partners or former partners are responsible for assaults. Instead, applying the definition has proved extremely problematic. Despite the publication of thousands of research monographs on domestic violence since the mid-70’s, there is still no consensus on the prevalence of abuse, who commits it, the principal causes and dynamics involved, and what types of assistance are required or effective. To cite just one example, surveys using the incident-specific definition have generated vastly divergent estimates of the annual prevalence of woman abuse based on whether respondents were asked about “conflicts” (136/1000), “safety” concerns (15/1000) or “crime”(7/1000) (Straus 1995; Tjaden and Thoennes 2000; Schafer et al. 1998). Without a consensus on the baseline we’re working from, service planning and evaluation will remain scattershot.

The definition contains three core assumptions that have guided research, policy and intervention in the field. The first is that partner abuse can be equated with violence. Despite attempts to broaden the definition of domestic violence to include “psychological abuse” or “control” (Saltzman 2000; DeKeseredy 2000; Dutton and Goodman 2005), criminal and family law, medical identification protocols, and other major intervention strategies or policies continue to focus on violence almost exclusively.

The second assumption is that violence can be usefully disaggregated into discrete acts or episodes of coercion. Like the criminal law, the population surveys from which Langhinrichsen-Rohling (2010) draws her evidence of gender parity ask respondents about whether they have engaged in or been the target of specific violent acts during the study year. Reflecting the same approach, safety planning, protection orders, counseling for batterers, and other intervention strategies are predicated on the assumption that perpetrators and victims exercise sufficient decisional autonomy ‘between’ episodes to end the abuse, what is called “time to violence” in the treatment literature.

The third assumption is that the severity of abuse can be assessed by applying a calculus of physical injury and psychological trauma to violent episodes. The presence of injury is typically not a formal prerequisite to access criminal justice, shelter, medical or other services. As a practical matter, however, decisions regarding intervention are often based on a calculus of relative physical and/or psychological harm (Buzawa and Buzawa 2003; Berk and Loseke 1980–1981; Loeseke 1992). In the National Family Violence Surveys (NFVS) for instance, acts with a low probability of causing injury were classified as “not abuse” or as a “normal” part of interacting with children or a spouse (Gelles 1997, p.14).

The Failure of the Paradigm

There is considerable evidence that the actual experience of abuse victims contradicts each of the major assumptions at the heart of the domestic violence paradigm. Evidence on the context, dynamic, meaning and consequence of partner violence comes from studies conducted at points of service, crime surveys, such as the National Crime Victimization Survey (NCVS) (Klein 2009), and from the National Violence Against Women Survey (NVAWS) conducted in 1995–1996 (Tjaden and Thoennes 1998, 2000), which identified coercive acts that caused safety concerns. These data consistently show that the hallmark of women’s physical abuse is frequent, but predominantly minor assault extending over a considerable period and with a cumulative impact on women’s health that is unparalleled among other classes of assault victims.

Partner assault is limited to isolated incidents in only a tiny proportion of cases (Stark and Flitcraft 1996). The average victim suffers between 3.5 and 8 assaults annually with well over a third report experiencing “serial” abuse (once a week or more) (Teske and Parker 1983; Klaus and Rand 1984; Mooney 1993; Stark and Flitcraft 1996). Meanwhile, the average duration of an abusive relationship is between 5.5 and 7 years (Stark and Flitcraft 1996; Campbell et al. 1998). Importantly, well over 95% of partner assaults are minor from a medical or criminal justice standpoint, even among the populations where we would expect to find the highest rates of serious injury—in the Emergency Room (96%), among women who call police (97%), and in the military (93%) (Stark and Flitcraft 1996; Connecticut Department of Public Safety, Family Violence Arrests, Annual Report 2008; Caliber Associates 2002). This does not mean the consequences are minor, of course. Battered women who use the health system have been distinguished from non-battered women by a profile of medical, psychosocial and behavioral problems that includes disproportionate rates of alcohol and drug abuse, depression, attempted suicide, homelessness, an elevated risk of HIV and other SDTs, unwanted high risk pregnancies and fear of child abuse (Stark and Flitcraft 1996; Frieze 2005). The emergence of this profile post-dates the onset of abuse, pointing to ongoing abuse as its cause (Stark and Flitcraft 1996). Nor is violence typically the only or even the most salient issue in these cases. Depending on whether the data are drawn from arrests, shelter populations or batterer intervention programs, in 60% to 80% of the cases where female victims are involved coercion is accompanied by a range of tactics designed to isolate, intimidate, exploit, degrade and/or control a partner in ways that violate a victim’s dignity, autonomy and liberty as much as their physical integrity or security. (Buzawa and Hotaling 2001; Tolman 1989; Rees et al. 2006; Stark 2007). In one well-designed study, for instance, 6 out of 10 of the men arrested for domestic violence reported they had taken their partner’s money as well as assaulted them, and had restricted their partners in three or more additional ways (Buzawa et al. 1999).

Given these findings, it is hardly surprising that the female population who seek assistance reports experiencing abuse as continuous rather than as comprised of discrete episodes (Hall-Smith et al. 1990).

When shelters, police, courts, health or mental health providers interpolate male partner abuse through the prism of the violent-incident frame, the oppression battered women experience is disaggregated, trivialized, normalized or rendered invisible, with interventions actually becoming more perfunctory as abuse becomes more comprehensive over time (Stark and Flitcraft 1996; Stark 2007). This has tragic consequences for victims. Since the vast majority of violent episodes are minor, for instance, when criminal justice bases its assessment of the severity of abuse on these incidents, domestic violence is reduced to a second-class misdemeanor for which few are sanctioned (Stark 2007).

Langhinrichsen-Rohling (2010) suggested that the greater fear reported by female than male victims of similar violent acts may reflect the differential socialization of men to minimize pain and fear. But a more parsimonious explanation is that the fear presented by female victims is the cumulative byproduct of the history of coercion and control in their relationships, an experience which may have no parallel in the experience of male victims of female partner abuse. Approaching abuse as an ongoing or chronic problem rather than as episodic changes everything. A doctor who views each complaint of chest pain as separate may become frustrated by multiple visits with identical complaints, as many police, physicians, mental health practitioners, judges and advocates are when abused women return repeatedly for help or “remain” in abusive relationships. The result is that victims are viewed as exaggerating their situation, particularly when the proximate incident is minor, being “ambivalent” about separation, or even fabricating their reports of abuse, an allegation that is particularly common in family court. However, when physicians recognize the particular complaints as symptoms of heart disease, a chronic problem, they become proactive, view repeated use of their services as appropriate and even desirable and take steps to ensure long-term risk-reduction. Even when proponents of the domestic violence paradigm recognize the frequency of assaults against women, instead of broadening the definition to include this reality, they attribute this to a sub- type of batterer who is prone to “recidivism”(Jacobson and Gottman 1998; Gondolf 1988).

The Gender Parity Challenge

Starting with the National Family Violence Surveys (NFVS) conducted in the late 1970’s and 80’s (Straus and Gelles 1990; Straus 1995), women consistently reported that they used violence against partners as often as men and, in some samples of younger women, even more often (Archer 2000; Swan et al. 2009). Dozens of similar surveys have shown that many of these women initiate the use of force, employ injurious levels of violence, stalk and/or sexually coercive their partners, and insult or humiliate them (“psychological aggression”) (Archer 2000; Tjaden and Thoennes 2000; Swan et al. 2009). While a larger proportion of women than men identified their violent acts as retaliatory (though not necessarily defensive), like men, they often reported being motivated by jealousy or a desire to punish or ‘control’ their partners. (Swan et al. 2009; Archer 2000; Felson 1996; Tjaden and Thoennes 2000). Recent surveys have confirmed earlier findings regarding the prevalence, scope and consequences of women’s violence against partners (Archer 2000; Graham-Kevan and Archer 2008).

Langhinrichsen-Rohling (2010) is one of a number of psychologists who have raised concerns that evidence of women’s use of force has had little impact on public policy or intervention, particularly compared to initiatives to stem male violence against female partners (Dutton 2005a, b). Advocates who initially publicized survey evidence on violence against women were largely silent about women’s reports of their own violence because they feared it could undermine support for female victims. But the principal reason there has been no public outcry for state intervention to protect “battered men” is that, apart from cross-sectional survey data, there has been little other evidence that female abuse constitutes a significant problem in public health or safety or that male victims have needs for protection, treatment or support that require new funding streams or services.

Calls to police from male victims are a marked exception to the absence of male abuse victims from service sites. By 2000, the U.S. National Incident Based Reporting System showed that women comprised 23.4% of individuals arrested for domestic violence nationwide, approximately the same call to arrest ratio found among female complainants (Hirschel et al. 2007). This suggests that the criminal justice system is responding similarly and ostensibly without bias to male and female victims of partner assault . There are also indications that some female perpetrators use control tactics as well as violence. Women arrested for domestic violence reported they threatened or used violence at least sometimes to make their partner do things they wanted him to do (38%) and these threats were sometimes effective (53%), to “get control” of their partner (22%) or to make their partner “agree” with them (17%). (Swan et al. 2009). However, although the motives and some behaviors reported by female perpetrators closely resemble reported dynamics identified among male perpetrators of assault (Klein 2009), there may be significant differences. Case studies of violent women (Dasgupta 2002) and voluntary samples of women arrested for partner violence (Swan and Snow 2002) consistently report that the vast majority had been victimized by the male partner they assaulted, with the proportion reporting victimization ranging from a low of 64% (the NFVS) to a high of 92% (Swan et al. 2009). The profiles of female offenders reveal higher levels of child sexual abuse, depression and substance use than among male offenders (Swan et al. 2009) Nor is there evidence that female partner assault evolves into the patterned subjugation that typifies women who use shelters, emergency rooms or other services. These data support Langhinrichsen-Rohling’s (2010) suggestion that a specialized treatment modality that targets the unique problems female perpetrators present might be more useful than the psycho-educational model used by most programs for male offenders. However, while such a protocol would encompass women’s role in bi-directional violence, the proportion of separated couples where violence occurs and women’s reported levels of childhood and adult abuse would appear to be incompatible with the forms of couple or family work she favors.

During the same period that arrests of female perpetrators of domestic violence have increased sharply, the most serious forms of female partner violence have declined. Women tend to seriously hurt or kill male partners when they feel themselves or their children are being threatened, though not necessarily in self-defense (Dasgupta 2002). In the mid-1970’s, male and female partners were equally likely to be killed in a violent confrontation. Since 1976, however, the number of men killed by women in intimate relationships has dropped 71%, much more than the overall drop in homicides, and by 80% among African–Americans. By contrast, the number of women killed by male intimates has dropped only 26% since 1976, far less than the overall drop in homicide. According to the FBI, in 2004 1, 596 females but only 385 males were killed by partners. (Uniform Crime Report 2006). Since men typically kill female partners when women threaten to leave or actually do so, these differential rates of decline suggest that available options have afforded women only temporary respite, another reflection of the episodic approach and one reason why women’s retaliatory violence has remained widespread.

Does Gender Parity in Violence Mean Gender Parity in Abuse?

The evidence Langhinrichsen-Rohling (2010) summarizes on gender parity in violent acts comes almost exclusively from general population surveys that utilized the Conflict Tactics Scale (CTS) to determine how often partners use force to address their conflicts and which tactics they have used. The researchers sought to map the use of force in families because they were morally opposed to any and all violence as a means to address disagreement and believed violence would be reproduced in subsequent generations (Gelles 1995; Gelles and Straus 1988). Because they also believed that millions of Americans were acculturated to use violence as a normal part of their problem solving, they were unconcerned about whether respondents found the violence in their relationships problematic. Even so, they distinguished the commonplace violent tactics they identified from the forms of woman abuse or “battering” with which the advocacy movement was concerned. Thus, Gelles (1995) termed the belief that men and women are equal perpetrators of ‘domestic violence’ a “myth” and Straus (as cited by Ellis and Stuckless 1996) termed the equation of the violent acts they identified with battering “ridiculous” and “unethical.” These early surveys concluded that 75% of women’s reported violent acts involved self-defense (Straus and Gelles 1990). Only a tiny proportion of respondents to the NFVS had called police or used other resources and almost all were women. This suggested that these respondents might be a different population than the abuse victims regularly use police or other services, an argument that has been fully spelled out by Johnson (2008).

Gender Asymmetry in Abuse

Not all survey evidence supported the gender parity hypothesis. Langhinrichsen-Rohling (2010) relied heavily on Archer’s meta-analysis of population surveys. Archer (2000) concluded that women perpetrated slightly more than 50% of reported partner violence and inflicted 35% of domestic violence injuries on men. By contrast, the NFVS reported that both the proportion of injurious assaults by men and their frequency were roughly 6 to 8 times greater than those committed by women (Straus and Gelles 1990), though the absolute numbers in both cases were small. A major limit of Archer’s meta-analysis was his decision to omit the National Violence Against Women Survey (NVAWS) from his overview, the largest population survey to date, presumably because its size (8,000 male and 8,000 female respondents) and focus on safety would have skewed his results. Although the NVAWS was conducted more than a decade ago, it remains one of the few population surveys to look at abuse over the adult life course and to compare male and female rates of being stalked and sexually assaulted alongside rates of domestic violence.

According to the NVAWS, with the marked exception of knives, which women and men use equally, men used every other means of serious assault much more often than women, including kicking, biting, choking, trying to drown, hitting with an object, “beating up,” and threatening with a knife and a gun, with ratios extending from 2:1 (for kicking and biting) to more than 14:1 (for beating up) (Tjaden and Thoennes 1998, exhibit 11, p.28). But the finding from the NVAWS that is most relevant here were the large gender differences in victimization that emerged when attention shifted from reported violent acts during the study year where the male:female ratio was only 1.4: 1, to partner assault over the life-course, where the ratio was more than 3:1 (22.1% vs. 7.4%) (Tjaden and Thoennes 2000), the same ratio reflected in domestic violence arrests. Gender differences in sexual abuse and stalking were considerably larger.

Proponents of gender parity dismissed the findings from the NVAWS because its focus on safety concerns excluded the less serious violent acts picked up by the NFVS and other surveys focused on conflict tactics (Straus 1998; Archer 2000). However, if we understand abuse to include the types of coercion that lead victims to apply a calculus of fear and potential suffering to their decision-making, a subjective safety concern would seem far more relevant than reports that force was used.

Relatively few studies have compared the abuse experiences of males and females in clinical settings. As Langhinrichsen-Rohling (2010) pointed out, these studies generally found that the modal pattern of violence is bidirectional, with a substantial proportion of women reporting having used violence, including some groups of women in shelters. However, these same studies also documented significant differences in dynamics and outcomes by gender. For instance, Phelan et al. (2000) compared men and women presenting complaints of injury at a Level 1 trauma center for emergency medical services and who reported being in a currently violent or abusive relationship. Men reported significantly higher rates of violence initiation than did women, with 100% of the men reporting initiating violence between 50% and 100% of the time. In contrast, 91% of the women reported initiating violence between zero and 20% of the time. These results are also important because they suggest that, even when women initiate violence in an encounter, this is not the typical dynamic in bi-directional violence.

Langhinrichsen-Rohling (2010) cited evidence from couples in counseling to support her emphasis on bi-directional violence, including work based on the couples’ clinic at SUNY- Stony-Brook. But were these couples engaged in “fights” or abusive assaults? For one thing, O’Leary and his colleagues excluded couples from treatment whose violence they deemed severe or abusive (O’Leary et al. 1989). For another thing, couples who reported violence had higher levels of marital satisfaction than the nonviolent couples and their relationships were more stable over time. Violence in these relationships was largely bi-directional, but probably not abusive.

The vast majority of point-of-service studies assessed the experiences of female victims only. So it is possible that a significant population of battered men remains hidden in medical, child welfare or mental health records. But the limited evidence we have shows that female/male ratios in help-seeking prompted by domestic violence range from 4:1 (for police calls and arrests) to between 15:1 and 20:1 for calls to hot-lines, requests for protection orders, ER visits, or use of child welfare or sexual assault services (Stark 2007).

Straus (1995) argued that men’s greater physical strength, hence the greater likelihood they will inflict injury on partners, explained why there is asymmetry in reporting despite parity in violent acts. There is some evidence to support this claim. Because it asks respondents about partner violence they consider a crime, the NCVS picks up the most serious acts of partner violence. According to the NCVS for 2004, between 1993 and 2004, the annual rate of male partner violence against women was roughly four times the rate for female partner violence (8.6 vs. 2.5) and the rate of injury was about twice as high for female than male victims, though the rate of “serious” injury to men was just slightly lower (Klein 2009). Gender discrepancies in injury can only explain a small proportion of asymmetrical help-seeking, however, since, as we’ve seen, very few of women’s calls to police or visits to the Emergency Room are prompted by injury. Instead, the disproportionate help-seeking by female victims is almost certainly a function of their far greater risk than men of suffering “systematic abuse,” the pattern of frequent but minor assaults combined with other tactics I subsume under coercive control (Macmillan and Kruttdschnitt 2005).

A New Paradigm: Domestic Violence as Coercive Control

The gender parity/asymmetry dilemma cannot be satisfactorily resolved so long as we embrace the violent incident-definition of abuse. When violent acts are abstracted from their historical and experiential context and assigned meaning based on their physical valence, the parity in violence appears to reflect a parity and relative symmetry in abuse, making it appear that the gender asymmetry in outcomes and help-seeking is a function of individual differences or gender socialization, the argument that grounded Langhinrichsen-Rohling (2010) article.

Any alternative model that hoped to make the asymmetry in help-seeking by female victims intelligible would begin by incorporating the elements of abuse research shows to be typical. Abuse would be defined as a course of conduct, i.e. as ongoing rather than incident-specific, its historical dimension; as typically involving frequent, even routine, but generally low-level assault; and as including a range of tactics in addition to threats or physical force. Because the violence it anticipates is generally low-level and frequent rather than severe and accompanied by multiple tactics, the alternative model would also highlight the cumulative nature of the harms inflicted and a range of harms in addition to physical injury and/or psychological trauma. The model of coercive control offers such an alternative. Although this alternative model has been available at least since the mid-1970’s and is widely used by advocates in one variant or another, it has remained marginal to research in the field.

Shortly after the first shelters opened in the U.S., advocates advanced a definition of abuse predicated on the use of violence to exercise “power and control” (Schechter 1982; Adams 1988; Jones and Schechter 1992) rather than simply to hurt or injure a partner. This work built on a picture of what psychologists had called “coercive control” or “conjugal terrorism,” terms they borrowed from the brain-washing and hostage literature to apply to the range of tactics abusers used over time to break down a victim’s will, constrain their decision-making power, exploit their resources, and entrap them (Singer 1979; Serum 1979; Morgan 1982). In a definitive chapter on coercive control, Okun (1986) drew an extended analogy between the techniques of “coerced persuasion” the Chinese had used on American POWs, the experience of women being conditioned to prostitution by pimps, and the experiences recounted to him in his counseling work with abusive men and battered women. A similar model of domestic violence as “control” provided the foundation for Emerge in Boston, one of the first programs for abusive men (Adams 1988), as well as for the “Power and Control” wheel developed by the Domestic Abuse Intervention Project (DAIP) in Duluth, Minnesota. With “power and control” as its hub and surrounded by a rim of physical and sexual violence, the spokes of the wheel are subdivided into economic abuse; coercion and threats; intimidation; emotional abuse; isolation; minimizing, denying and blaming; using children; and abusing male privilege. Jones and Schechter (1992) added “control through decision-making” to their list of abusive tactics and focused particular attention on the abuser’s micro-management of a victim’s everyday life, including “picking out your clothes,” “telling you what to wear” and “forbidding you to shop.”

In Coercive Control (Stark 2007), I classified all of the dimensions of abuse identified earlier under four tactical headings: violence, intimidation, isolation and control, including under intimidation the varied forms of sexual degradation often associated with domestic violence and under control, constraints over everything from basic material resources (money, food, sleep, etc.) to imposed ‘rules’ about everyday living. In contrast to models of “psychological abuse,” I called these control tactics “structural” because they affected forms of material deprivation and dependence that were independent of their psychological effect on victims. Summarizing the limited empirical data available on intimidation, isolation and control tactics, I argued that 60% to 80% of the cases seen at points-of-service in the U.S. involved coercive control. The remainder involved partner assaults primarily, where physical and psychological abuse were combined. These categories were roughly equivalent to those described by Johnson (2006, 2007, 2008) as “intimate terrorism” and “situational partner violence.” However, I distinguished assaults, where one party was victimized by violence, from “fights,” where relatively equal partners used force to address situationally specific issues, typically bi-directionally and neither considered themself victimized. In contrast to Johnson, I argued that the majority of violent acts identified by population surveys fell into the category of “fights” and should not be considered “abuse.”

Three final elements of the model are relevant to the current discussion, that it linked men’s greater capacity than women to deploy coercive control to their ability to exploit persistent gender inequalities; identified women’s enactment of stereotypic gender roles as the principal target of control tactics, particularity their roles as housekeepers, wives and mothers; and depicted coercive control as a “liberty” crime that caused a range of harms to women’s autonomy, dignity, personhood, and capacity to fulfill their responsibilities as citizens as well as to their physical security. Because the domestic roles targeted in this form of abuse are already devalued by their default consignment to women, the micro-management of daily activities that often accompanied coercive control was typically “invisible in plain sight.”

Coercive control may be the primary dynamic in the lives of between 8 and 10 million of the 14.5 million women in the U.S. I estimated are currently being abused by their partners in the U.S. (Stark 2007). Whatever the proximate motive or dynamics of a man’s abuse in a given instance, when so large a population of persons adapt a similar strategy to establish nonreciprocal relations in personal life, we should assume that a broad social logic is at work, constraining the selection of means, including violence, according to their relative efficacy. In this analysis, as in much of the early feminist work, I conceived of abuse as a rational and instrumental strategy which affords benefits that can be measured in increments of resources garnered, personal service, sexual exclusivity, and subjective reinforcement of gender identity. In a given case, childhood exposure to violence, family dysfunction or medical, psychological or behavioral problems can play an important role. Male socialization and other individual factors identified by Langhinrichsen-Rohling (2010) may also play a role. But coercive control is largely a male phenomenon not because men are empirically more likely to deploy the requisite tactics or are socialized to expect ‘control,’ but because inequality constrains women’s options in relationships independently of their personality, mental acuity, socialization, or physical prowess. This reality is not changed because women check many of the same boxes on surveys as men, including the box in which they acknowledge acts of jealousy, possessiveness or control, or because, in any given relationship, women may hurt, control or even dominate male or female partners in many of the same ways they are controlled.

Rather than portray the primarily female victims of abuse one-dimensionally, the coercive control model anticipates high levels of aggression by women in relationships that correspond to their newly won rights and resources, including their widespread use of violence. If men’s resort to coercive control exploits persistent gender inequalities in income or opportunity for instance, so too do women increasingly demand a status in families and relationships comparable to their newly one rights in the public arena, challenging and attempting to displace the traditional privileges men have enjoyed simply because they are male. The ways in which this tension is played out in millions of homes and relationships includes the increasing willingness of women to use violence to create an environment in which they enjoy the same autonomy, liberty and dignity they could increasingly command in school, work and in the political world. Many of the violent women with whom I work in my forensic practice view deference to male partners as a choice rather than as their fate and expect a quid pro quo in exchange that includes sexual fidelity and a certain financial commitment. These women feel entitled to punish male partners who fail in these roles and are no less possessive or prone to jealous rages than men. Even so, as structural inequalities ‘entered’ the relationship over time, even in many of the relationships in which violence was initially bi-directional or female initiated, the pattern changed shape and similarities in motive, behavior and even in the occasions for male or female violence gave way to the differential capacity of men to construct a regime of domination based on the greater shares of power and control they inherit from persistent inequalities simply because they are male. At this point, there is a convergence between societal inequalities (in income, e.g.) and structural inequities imposed in the relationship. The result is that relationships where violence seemed symmetrical to start evolve into the type Johnson (2008) called “violent resistance,” where women’s violence is met by the pattern of coercive control. In sum, women’s greater equality instills a female capacity for autonomy that can turn male–female relationships into a “battleground,” where bi-directional violence may be the norm as Langhinrichsen-Rohling (2010) claimed it is. Differences in race, social class or in the cultural, familial, behavioral or psychological factors she emphasized may influence how an individual man or woman chooses to relate in this battleground, whether they will exploit, assert, ignore, resist or submit to the assertion of female equality and male privileges and even who makes what choice to some extent. But they cannot change the basic fact that structural inequality endows men with an incremental advantage in relationships that is reflected in the peculiar cast of coercive control. Asymmetry in the social position of men and women, the fact that men cannot be unequal to women in the same way and at the same time that women are unequal to men, explains why male and female partner abuse have different dynamics and long-term consequences even when the motives, tactics and proximate consequences are the same. Coercive control does not merely play off sexual inequality. Because its deployment disables the self-interested decision-making of millions of women, coercive control also exacerbates gender inequality.

Feminist Therapy

The challenge to feminist therapy follows from the political context that differentiates female from male abuse. The abusive element in classic domestic violence is its use to elicit obedience, submission or dependence by violating a partner’s physical integrity or threatening to do so. The controlling effects of violence are mediated by fear of what will happen if the victim fails to accommodate the assailant, what can be termed its “or else” proviso. Some victims respond with violence to threats, assaults or fear, and this retaliatory violence may limit their partner’s abuse, even if they get the worst of a given violent encounter (Swan et al. 2009), the leveling effect noted earlier. The classic therapeutic goal in these cases is to combine harms reduction with safety planning. But in coercive control, the asymmetry is structural requiring a reallocation of power (not merely of resources or in the allocation of responsibilities) and the primary harms victims suffer are to realms of personhood, identity, and to their economic, social and political activity. In these cases, therapeutic intervention focuses less on what the abusive partner has done to the victimized woman than on what he has kept her from doing for herself, recovering and then supporting her capacity for self-direction. Since the disproportionate risk of psychological or behavioral problems among victimized women typically develop to accommodate and manage the stress of ongoing abuse, they cannot be ameliorated unless the objective situation is also changed. What distinguishes feminist therapy from the generic emphasis on personal empowerment that characterizes most therapeutic work is the recognition that a woman’s long-term vulnerability (and her capacity for self-direction) is inextricably bound to her disadvantaged social status and its reiteration in a specific abusive relationship. The alliance between “victim” and “witness” that Judy Herman (1992) sees as the antidote to the trauma of interpersonal violence is rooted in our social alignment as an advocate with women’s liberation, not merely in our clinical alignment with a particular client.

A Final Word

Feminism works backwards from the fact of gender inequality to identify its means and remove them through activism. Advocates selected partner violence as a point of intervention because it appeared to be a critical link in the chain of gender inequality (Schechter 1982; Stark 2007). Over time, as the advocacy movement evolved and scholars with a range of orientations entered the field alongside the feminists, a consensus formed around “violence” as the common harm to study and prevent, replacing the earlier focus on coercion as merely one means used to subordinate women in personal life. When violent acts moved center-stage, the only plausible rationale for focusing on violence against women rather than other common forms of relationship violence was empirical, the claim that women merited special attention because they were more likely than men to be victims of violence. Evidence of gender equity in violent acts is incompatible with this rationale.

From a feminist standpoint, the singular focus on violent acts is a conceptual dead-end, forcing us to account for the well-documented asymmetry in the experience of abuse by highlighting individual differences rather than the interplay of structural inequalities, the enactment of gender roles, the formation of personhood and the relational dynamics that mediate these levels of experience in abusive relationships.

The coercive control paradigm I propose restores the original emphasis on gender inequality as the principal wrong we hope to counter and looks beyond violence to encompass the multiple tactics men deploy in personal life to preserve what is left of gender inequality against women’s growing capacity for full personhood. Coercion continues to be one key to this process. But rather than ask ‘who uses violence,’ I believe our responsibility as feminist scholars is to identify how violence functions in relationships to preserve and extend gender inequalities.

Feminist scholars need not deny the realities of women’s violence or posit essentialist qualities in women that make their assaults somehow less significant or harmful than men’s assaults. To the contrary, as Langhinrichsen-Rohling (2010) argued, incorporating a full understanding of women’s violence into our picture of their oppression will facilitate a shift away from the patronizing and victim-blaming practices that are commonplace at the sites where battered women seek help, including shelters. A full appreciation of women’s violence entails embracing a broader view of what is at stake in abusive relationships, understanding that it is liberty and personhood and the larger rights of women as fully entitled citizens that require defense and our support as clinicians, not merely their physical integrity. This does not mean that we should minimize the suffering of men abused by female partners or the challenges to an equitable partnership posed by the use of force. What it suggests, however, is that we make clearer than we have not only how and why male partner abuse is different than female partner abuse and therefore merits different forms of policy and intervention, but why our political commitment to social justice compels making a priority of forms of abuse—“domestic violence in the strong sense”—that perpetuate structural inequality and subordination compared to other forms of abuse that leave the status quo largely in tact, though are no less painful for this fact. This does not change the pain experienced by males victimized by female partners, negate the possibility that women will assault, abuse or dominate individual men or that men’s violence can also have a leveling effect in relationships where women’s status is higher than theirs.