Summary of Workshop Discussion: Gender Symmetry
November 20, 2000, Arlington, Virginia
Among Racial, Ethnic, and Cultural Minorities
Table of Contents
- Welcome and Opening Remarks 
- Conflict and Control
- Open Discussion
- Summation: A Typology of Violence
- Future Directions in the Field
- Open Discussion
- Women's Life Experiences Project
- Open Discussion
- Summation: Measurement Issues
- Women's Use of Violence
- Open Discussion
- Summation: Women's Use of Violence
- Future Directions
- Closing Remarks
Welcome and Opening Remarks
Bernard Auchter, Acting Director, Violence and Victimization Division, Office of Research and Evaluation, National Institute of Justice (NIJ), U.S. Department of Justice, welcomed the group and set the context for this workshop. Following recommendations in the report from the National Academy of Sciences, NIJ is particularly interested in the development of clear definitions, connections of specific instruments to particular sub-populations (ethnic group, etc.) with whom they have been used, and validation of violence against women measures. We would like to take critical stock of where we are, he said, gathering pros and cons of different measurement strategies in order to improve the underlying science. He thanked each of the authors who provided papers as background to this workshop.
Leora Rosen, Social Science Program Manager, NIJ, led the extended mutual introductions among participants and gave a history of the selection process of persons participating in this focus group. Solicitations for research on gender symmetry had been drafted on two occasions in the last couple years, and there was disagreement about what to include for this topic. All agreed, however, to convene a workshop of those working in this field to collect some "best thinking" on the topic, to capture the emerging ideas from people with different vantage points in the field, and make these available to guide research.
In the course of the introductions, the variety of research interest/experience among the people covered a wide range, with study topics mentioned as:
- Types of violence in military families, using the conflict tactics scale (CTS)
- Forensic social work with children
- Examination of the two-stage measurement strategy for the National Crime Victimization Survey (NCVS)
- Police responses to violence and prevention measures that are culturally specific to minorities
- Poverty and joblessness in public housing in relation to acts of violence
- Women's victimization and connections to pathways to crime
- Methodological questions arising in connection with the National Violence Against Women Survey (NVAWS)
- Measurement issues relating to meaning and motive for violent actions
- Family violence in general, including ordinary corporal punishment of children
- Prosecution issues for domestic violence
- Rural types of victimization
- Effects of dual arrests under mandatory arrest laws for domestic violence
- Effect of violence definitions on delivery of services to victims
- Social stress context and violence against marginalized populations
- Coping strategies and help-seeking behavior of women in violent relationships
- Risk of homicide in intimate partner violence
- Control and commitment issues in violent intimate partnerships and analysis of the shelter movement as a step to neutralize this
Facilitator for the meeting, John Jeffries, Director and Senior Economist, National Associates Program, Vera Institute of Justice, encouraged an open discussion while reminding everyone that this NIJ forum was oriented towards issues of justice enforcement. As an economist and researcher, he also had strong interest in the measurement issues to be covered.
Conflict and Control
Presentation and Group Discussion
Michael Johnson presented the first paper, in which he spoke about critical distinctions in types of domestic violence. Too often research literature has made no distinction between kinds of violent relationships, which have different causes, developmental dynamics, and (probably) requirements for different kinds of interventions. Dr. Johnson has argued that the sampling frames of agencies (police, hospitals, shelters) and the sampling frames of "representative selections" are alike flawed, the former through biased perspective and the latter through refusals to disclose experiences. Johnson's study was based on secondary data sets and multiple sampling strategies, using factors collected in the late 1970s. Women who filed for protection in the courts were identified. In each case, a neighborhood person (somewhat analogous) was interviewed. He developed a typology based on perpetrator's motivation, particularly oriented to the general attempt to control. Through his meta-analysis of studies of partners' behavior in violent relationships, he has distinguished a four-category typology of partner violence:
- "Intimate terrorism," which had been earlier labeled "patriarchal terrorism" (one partner is violent and generally controlling)
- Mutual violent control (both partners are violent and controlling)
- Common couple violence (either partner is violent, or both partners are violent, but neither partner is generally controlling)
- Violent resistance (one partner is violent and generally controlling, the other partner - usually the woman - is violent in response)
The information on the effects of control motives of the violent individual was developed with cluster analysis and a controlled scale comparison. Dr. Johnson said he distinguished "common couple" violence from "patriarchal terrorism" by doing cluster analysis with a two-cluster solution. His purpose was to compare men who were high (on average) for control items and to men who were low on control items, keeping the violence itself out of the "control measure." Using a scale with a cut-off and cross-tabs, he could maximize the effect of control tactics. Outcomes in the examined sample (primarily poor urban women from Chicago, mostly African American and Hispanic), were measured in terms of frequency of injury and effects on psychological health. These differed based on the type of violence involved. For example, there was no difference in injury experience between the couples without violence and those that exhibited common couple violence; but there was a major jump in significant injuries when the kind of violence was "intimate/patriarchal terrorism."
Again using the Chicago data, experience of PTSD was seen with more than twice as many women in relationships that scored as "patriarchal terrorism" than among women in relationships scoring as "common couple violence." The violent controlling behavior ("intimate or patriarchal terrorism") was found to be 80 percent from the male partner across different samples; while mutual violent control and common couple violence were roughly gender symmetrical. Only an extremely small proportion of men had exhibited "violent resistance" with a partner who was violent and controlling, as defined in the study. Most importantly, there is significant difference between common couple violence and patriarchal terrorism in terms of long-term effect on the psychological health of the victim.
Dr. Rosen said it would be interesting to test a hypothesis about the relation of familial patriarchy to types of violence. Dr. Johnson said there is other work showing that individuals ascribing to patriarchal ideologies were more likely to be violent. A participant said Dr. Johnson's study does not link structures outside of the family to the family structure, and patriarchy refers to an authority structure rather than motives. There are also nonviolent patriarchal relationships. One person referred to research on patriarchy by Michael Smith. Self report research can be used to look at motives, but another presenter cautioned against taking research outcomes and then "determining causes." Dr. Johnson said motives for generally controlling behavior can vary: some men who exhibit controlling behavior do not score high on the "patriarchy scale"; also, there are women (a very small number) who act as "intimate terrorists." Complicated family dynamics involve more than individual motives, possibly historical "assumptions by men that they have rights by virtue of being male."
Other social science studies have related the macro level to the micro in order to demonstrate the effects of broad social norms on people's behavior. An NIJ staff member asked whether data on employment and divorce had been examined. "Paternalism" can relate to the career and material opportunities that women have. In Germany after World War II, as women were given the right to property ownership and more employment, the divorce rate rose to a large degree.
One participant asked about the intersection of race and class factors. Johnson said that women experiencing intimate terrorism (type of violence) were more likely to have interference with work and school activities. Do some relationships begin as "intimate terrorism" and become "violent resistance?" Dr. Johnson said he was trying to keep the control motive generally separate from characterization of the violence. His questions tried to determine the "basis for the control context," with questions such as: Does the partner always have to know where you are? Does he direct anger toward the children? Do you "get an allowance?" The manipulative style of life may cut across all kinds of relationships. There is a whole range of coercive activities in intimate partnerships, such as denying the woman the use of contraception or safe sex practices. Johnson's study has really only "scratched the surface."
Satya Krishnan said she felt uncomfortable with the categorization. Other distinctions might be more important to women. She asked how to consider the significance of race, class, and culture. When speaking, for example, to Asian women, certain motives may look different. Women have perhaps lived with lack of access to money as a normal situation and explain this to themselves in a different way (not perceiving this to be a controlling action). Motivations for violent and controlling behavior vary (Jacobsen and Gottman, etc.) Dr. Johnson said that qualitative research and rich interview data would be necessary to thoroughly understand the meaning and social context. Another participant noted that it is very difficult to untangle different cultural ideas about coercive control and the "male role," with corresponding "rights and resources." There are complicated cultural specificities. In Japan, for example, the word for "man" stems from a word meaning "work slave."
A participant asked whether "intimate terrorism" might stimulate (over time) a case of "violent resistance," in the same couple. Dr. Johnson said most research shows over time (even a two-year period) that the majority of women in a violent relationship do change the nature of it or escape from it. We are always influencing one another, said Ileana Arias, at what point does this pass into "power and control" manifestation for purposes of this kind of classification? She thought that women's tactics for control might not have been represented well in the study. The cluster analysis should identify a variety of patterns of control tactics.
Anthropological data fits this conceptualization better than many other types, said Jacquelyn Campbell. The Johnson framework could help explain the different viewpoints of people working in shelters compared to people, for example, in the military. The two groups are looking at different types of violence: the shelter staff more often see "intimate terrorism," while the military more often encounters "common couple violence."
Controlling behavior is a fairly new area of study. A good first step might be to document differences between male and female victimization in intimate partnerships (and see how these relate to occurrence of "intimate terrorism."). Violence patterns may show important differences aligned with marital status. Control and violence are closely connected in some couples who are not living together (and sometimes who have never even lived together). Can the measurement strategy for this area be made more "culturally generic?" One study (Journal of Marriage and the Family, R. Macmillan and R. Gartner, 1999) examined risk of spousal violence in connection with workforce participation, but did not look particularly at marital status in relation to "common couple violence."
Dr. Jeffries said he noted two issues deserving further focus: how current instruments should be constructed (with implications for current research practice) and defining policy questions that will favorably affect service delivery in a practical way. Many individuals "digest the nuances of our work," he said. Sometimes this can be "generalized in ways that are terrifying."
Dr. Rosen asked participants to think about severity of the violence. Does intimate terrorism necessarily mean more serious violence? What is the intersection of minor and severe violence in the context of this typology? Dr. Johnson said that only 11 percent of the violence appearing in the survey came from the "terrorism" type. Common couple violence drives most of the patterns. Among the shelters' data, however, a higher percentage of violence of the intimate terrorism type appeared. Perhaps one setting tends to miss other types of violence. The definition of "violence" itself needs clarification. How does one judge intent? Can you talk about behaviors having a numeric scale of "possible harm?" Patricia Tjaden said the crime of stalking exemplified this kind of definitional problems. A high prevalence of stalking has been found to be connected to murder of women.
One participant thought that Johnson's discussion did not successfully separate elements of control from the violence itself. Violent acts should be included under a broader umbrella of abusive behaviors, adding stalking. The implications for gay and lesbian couples also have not been much considered. What would be the policy implications if 80 percent of behaviors or family infrastructures that deserve most concern were not related to "assault" but to "abusive controlling behavior?"
The group spoke about keeping the significant political issues in mind when choosing research terminology. This "battleground" has serious implications for the services delivered to women, children, and others. The child maltreatment area has experienced similar problems with terminology. In Canada, fathers' rights groups have taken "crude counts" (not really gender symmetrical) to the court to prevent some shelter organizations from providing services to women and children. Disparities must be explained clearly to keep political support behind the research effort. Civil conflict on broad issues may continue, but the public should be coherently informed about the research.
Summation: A Typology of Violence
Evan Stark provided a summation of the discussion on Dr. Johnson's presentation. He said that the research highlighted the seriousness of coercive control and "intimate terrorism." He thought that the more tyrannical pattern, involving non-severe routine abuse, combined with coercive control tactics, has been increasing in society rather than decreasing. There are also many ways that violence is used in families that do not involve subordination of women. We can accept evidence of women's aggression and still recognize the fact that primarily women are battered, said Dr. Stark, if we change the legal emphasis to actions of coercive control.
Motive, said Dr. Stark, has little use as a concept for measurement purposes. Yet, a study focus on episodic violent situations, particularly on severity of injuries, may not perceive the ongoing process of control and coercion, in which the effect is largely a function of duration (time of access). Dr. Johnson suggested examining violence as "embedded in the context of general power and control." The system has failed to comprehend this problem, in part, because institutional data confuses different types of violence ("common couple" vs. "intimate terrorism"). Secondary data analysis sometimes makes large assumptions that group too much data into the category "common couple violence." Over time, this mixing of kinds of violence could inadvertently trivialize the circumstances of battered women. They have been recorded as saying that the "violence is not the worst part." Intimidation and isolation (like violence) are aspects of a "complex of domination."
Domestic violence is not like simple assault; perhaps physical force is not even the (sentinel) event that defines abuse in intimate relationships that are characterized by coercive control. In a recent article in the Georgetown Law Review, Ruth Jones argued that battered women experiencing coercive control should have court guardians appointed to make critical decisions about their safety and the safety of their children. While coercive control can reduce not only resistance but the capacity for autonomy and decision-making, it seems much more sensible to remove the source of the violation (the batterer) rather than further undermine a woman's independence. Because the field has attended to domestic violence almost exclusively, there is not a lot of data on coercive control. But there are hints. In recent research in Quincy, Buzawa and her colleagues found that 38 percent of the victims whose partners were arrested reported they were not free to come and go as they pleased, 45.8 percent were denied access to social support, 58.5 percent were denied access to money, and 46 percent experienced between three and 15 other restrictions in their daily routines. Can it be shown empirically that violence is the most salient feature of coercive control when it is used for domination? Perhaps using shelters, research could investigate this; unfortunately, it has become very hard to gain access to shelters, said Dr. Stark, unless you have experienced some injury.
We know (for comparison), he continued, that many people are prejudiced though they do not have the opportunity to discriminate. Discrimination as a practice rests not on personal prejudice in people, but in opportunity structures and enforced inequalities. The same may be said of coercive control. Women may be no less likely than men to want to control their partners. But because of sexual inequality and men's differential access to opportunity structures, coercive control is almost exclusively a male phenomenon. We need non-survey research (open-ended interviews, for instance) to tap the experiential base behind the terminology (coercive control). This may reveal that the moment of violence is not what is decisive for women's subjugation. Some groups oppose violence in all contexts, said Dr. Stark; but this was not his own view. Whether in families, communities, or among nations, "fighting" as well as "fighting back" (resistance) can be extremely productive in winning rights and resources, particularly in situations of unequal power.
From the criminal justice standpoint, violence against women is a "course of conduct" crime, different than a discrete event like simple assault, for which a police arrest is used. This view challenges some fundamental assumptions about intimate relationships for women. When does manipulation become coercive power and control? The higher risk of injury and death that women face in relationships of coercive control is more a function of women's inability to escape from abusive men due to structural dimensions of control than of the absolute level of force used against a partner. In fact, the Family Violence Survey has shown that there is a spectrum of ways in which violence is used in families, only some of which relate to subordination and domination of women. A new theory of harms and causation, based on what is known about coercive control (use of intimidation, isolation, etc.), should replace the focus on injuries. The emphasis should be on liberty, autonomy, and fundamental political rights.
Future Directions in the Field
Murray Straus gave the luncheon address on future directions in the field. He presented information on the controversy over domestic assaults by women and some methodological guidelines for measuring maltreatment within the family. His presentation contrasted findings from studies he characterized as family conflict studies with studies of crime. More than 100 family conflict studies found that rates of assault by women are about the same as the rates for men. However, in studies of crime victimization and police records, male perpetrators always outnumber female perpetrators by a large amount. This discrepancy occurs for several reasons documented in Straus's article on this issue. An example is that only about two percent of domestic assaults are reported to the police, and those that are reported tend to be assaults in which there is a serious injury or fear of a serious injury (more likely to be committed by a man).
National surveys conducted by Straus and his colleagues show that among young married and cohabiting couples (ages 18-20), about 35 percent of both males and females have perpetrated a physical assault against a partner in the previous 12 months. The rates declined with increasing age for both men and women. By the median age of American couples (40), it is down to about 10 percent for both men and women. By age 70, assaults on partners are still occurring, but at a much lower rate (four percent in the previous 12 months) but gender symmetry remains.
To understand the presence and persistence of domestic assaults by both men and women, even into old age, Dr. Straus suggested that it is necessary to study violence by both men and women. Moreover, female perpetrators are present at all levels of violence, including the most severe type. As an example, Dr. Straus pointed out that Dr. Johnson's morning presentation reported about 20 percent of persons who could be classified as "intimate terrorists" were women. In Straus's own and others' research, although injuries inflicted by men are more frequent than those inflicted by women, women do cause significant injuries to male partners. From one-seventh to one-third of intimate partner violence (IPV) injuries (depending on the study) are caused by women. Statistics on murder of partners by women between 1991 and 1996 show that women commit about one-third of intimate partner murders (Greenfeld, et al. 1998). Thus, from both criminological and public health perspectives, injuries and deaths resulting from female perpetrated assaults are a serious problem.
Dr. Straus also challenged the prevailing belief that sexual coercion is almost entirely a male offense. He showed a table summarizing six studies of college student relationships showing sexual coercion by women. Although the rates are even higher for men, the percent of women who reported having used physical force for sex ranged from two to 10 percent. Rates of "getting partner drunk to have sex" ranged from nine to 28 percent.
Dr. Straus argued that it is important to recognize that IPV by women as well as men, has important harmful effects. In the Canadian government national survey of 1999, of those assaulted by a partner 40 percent of women and 13 percent of men had been physically injured, and 38 percent of women and seven percent of men had feared for their lives. In general, although the rates are lower for men, the harmful effects of IPV are experienced by male as well as female victims, including fear, physical injury, economic loss, and psychological damage to the children. The injury rates for men are lower than for women, but substantial. Moreover, even seemingly "harmless" assaults by women have been shown to be harmful for children who witness them and are morally wrong and criminal acts.
These statistics indicate a need for criminal justice and public health policies that recognize and deal seriously with domestic assaults by women as well as men. Responses to maltreatment should be applicable to both at the individual level (including both psychological and criminal justice interventions), and at the societal level. Great progress in reducing attitudes that tolerate violence by men against their partners, but no progress in reducing tolerance of assaults by women. To a considerable extent informal social norms continue to make the marriage license a kind of hitting license for women, provided the speed limit (injuring a partner) is not exceeded. To change this, it is necessary to explicitly condemn violence by women as well as by men in public service announcements, school based programs, and training of police and health professionals. Even more fundamental would be programs to end use of corporal punishment by parents, which has been shown to be linked to IPV by women and men.
Some domestic assaults by women are in self-defense, but four different studies show that women are just as often as men the first to hit. So simplistic explanations such as self-defense are inadequate. Similarly, domestic assaults by women occur in egalitarian as well as male dominant families. So the theory that assaults by women are primarily a response to a patriarchal society and family structure is only a part of the explanation. Amelioration of the problem requires theory that explains maltreatment by both sexes. Studies of violence (an interactional event) need to address both parties to the incident.
Dr. Straus presented a chart listing criteria for developing or evaluating measures of partner maltreatment. The issues covered included the need (1) to separately measure each dimension, (2) to measure causes and context separately from acts of maltreatment, (3) to specify the time period (such as in the past 6 months, past year, or lifetime), and (4) to record the chronicity of acts of maltreatment. Chronicity is a key for understanding many deviant behaviors such as persistent drunkenness, child neglect, and intimate terrorism. Separate analysis of different kinds of maltreatment may show different antecedents and consequences. Hypotheses about the effects of maltreatment (such as injury) cannot be tested if context/cause is not separated from specific effects like injury.
Dr. Straus urged that (whenever possible) data needs to be obtained about maltreatment by both the respondent's and the partner's behavior. This is essential to understanding the dynamics of relationships involving maltreatment. It is also the best, and often the only, way to provide data on gender symmetry. In addition, studies such as the National Violence Against Women survey and the similar Canadian survey, which have asked only about victimization, have shown much lower rates of disclosure (prevalence).
Dr. Straus agreed that more use needs to be made of in-depth categorizations or typologies of offenders. He identified three different approaches to typologies of domestic violence offenders: (1) the Johnson approach (presented in the morning) which hinges on whether the offender's objective is general domination; (2) the Holtzworth-Munroe and Stewart approach, which hinges on psychopathology in the offender; and (3) Dr. Straus's own approach, using Conflict Tactics Scales data, which hinges on presence of asymmetrical severe assaults. If data from the Personal and Relationships Profile (PRP) is available, the combination of the CTS with that instrument enables both the Johnson and Holtzworth-Munroe and Stewart types to be identified. The single most important problem in the field is simply not gathering data. Dr.Straus recommended observation, cross-national and cross-cultural comparisons, and computer simulation as other measurement strategies.
One participant asked about the "representative sample fallacy," saying that any statistical manipulation based on the mean can often be misleading. Multinomial logistic regression takes more variation into account, to avoid "skewed" measurement. This analysis supports the distinction between different kinds of violence (as in Johnson's work).
Dr. Stark said the climb in minor violence compensates for a decline in severe violence that has occurred most dramatically in urban African American communities. Was there any race-specific data on changes in severe or minor violence? Geographic mapping of violence problems together with distribution of applied services over time, has shown more services to be targeted toward high poverty, African American communities. Possibly those areas (now receiving greater protection and control) are showing a greater change than the mainstream.
Women's Life Experiences Project
Sarah Cook, gave a presentation on context, meaning and method in the measurement of domestic violence, based on preliminary data from her study, the Women's Life Experience Project. The study used a form-based approach to measurement and sought to compare and contrast instruments and to evaluate computer based data collections that reflect women's experience of violence (either their own or perceptions of their partner's) in intimate relationships. To bring the focus on meaning and context of women's violence and to see the clear connection to "legal constructs," the survey instruments were adjusted with a range of additional questions and definitions of controlling behavior. The study tabulated and compared dimensions measured by the primary instruments currently most in use to look at violence against women, noting what aspects of context or event nature were missing (i.e., duration, character of victim-perpetrator relationship, psychological abuse, degree of non-consent, etc.)
For women, context usually carries the meaning of the event. Typical study questions might ask, "When the partner acted [this way] was it during a fight?" (conflict context); or, "Was he trying to keep you from doing something?" (coercive context). Researchers also wanted to find out how to externally validate findings concerning socio-cultural, interpersonal, or economic context in relation to behavioral and physical responses, and to see if effects are consistent over time. The instrument defined the contexts as:
1. Conflict, argument, or fight
2. Coercive (compelling or prohibiting behavior)
3. Control instigated ("provoked" in order to control or protect)
4. Unpredictable (out of the "blue")
5. Random (relating to a stranger)
Using a five-point Likert scale (from "none of the time" to "all of the time"), the project will seek, in the second step, to determine how varying context relates to different outcomes in terms of women's general health, risky sexual behavior, mental health, substance use, and injury. Data were collected on women entering prison at the Metro State Women's Prison (a maximum security prison near Atlanta) and the Grady Memorial Hospital. About half (30) of the population studied for this presentation were African American women.
The lifetime prevalence of violence among this group was very high. About 30 percent reported violence in connection with their most recent intimate partnership. Approximately 30 percent said that violence happened in the context of an argument (but few said that it was "always" this way). On category three ("control instigated"), a small number of women knew exactly what was meant by this context, though interviewers were not comfortable discussing this and some women interviewed were offended. Women said there are times when they perceived a partner's outburst as "inevitable," and they might try to "instigate" this in a time and place where they think they can control their batterer's actions or protect children. Another 30 percent pointed to the coercion motive, saying they were violent in response to being compelled to do or not do something. Different violence context items did not correlate clearly; but the research was able to link controlling behavior to physical violence, Dr. Cook said. It was unclear if the women understood "argue, prohibit, compel" in the same way. In the context of an argument, the woman typically assigns some responsibility for the conflict to herself and is less likely to go to a shelter in that case (also less likely to "do what the man wants," admitting defeat in the argument). Less than half of the group responded that violence was "always predictable." In relation to injuries experienced, there was a sizable correlation between violence occurring "in fights" and violence occurring "out of the blue" (random).
There were interesting correlations in help-seeking strategies. Dr. Cook showed 15 items out of a bevy of help-seeking strategies (actions such as, "go to shelter, pray, end the relationship, do what he wants, keep people around, take legal action, remove weapon, get counseling"). Total help-seeking action correlated strongly with the partner's "prohibit" behavioral context (coercion) and also with experience of injuries. Initial results have been promising, but are not to be considered conclusive, suggesting that controlling behavior could be seen as a subset of context.
A participant asked why "calling the police" was not included with the help-seeking behaviors. Dr. Cook said none of the criminal justice system actions correlated with particular contexts. It would be interesting to look at the relation of their (criminal) behavior to previous incidents of domestic violence. Dr. Cook said many of the women had been "in and out" of the criminal justice system. Another participant asked about the idea that violence is sometimes "scripted." Is this the idea behind the category "control instigated?" It points to the realization that these incidents are not necessarily uncontrolled violent outbursts. Dr. Rosen asked how "control instigated" would relate to "violent resistance." Dr. Cook said the latter topic deserved a paper of its own, perhaps research relating to how intimate conflict "leads" to criminal behavior.
Lorraine Malcoe expressed concern about violence (possibly more severe violence) that could be missed from the conflict tactics scale. She was very interested in issues of power in inter-gender conflict, especially as it relates to marginalized groups like the Native Americans. Women's acts that might be self defense in a particular context should not be criminalized. Dr. Cook noted that context has been a very difficult topic on which to construct a Likert scale. The study was based on actual actions of the women (not beliefs). However, for many people, distinctions on the scale (like "some of the time" versus "less than half of the time") might have been too abstract and easily confused. Both open- and closed-ended supplementary questions are important and help to code the violence. The qualitative part of the survey took the longest amount of time.
There are overarching contexts within which other "sub-contexts" appear. Aggression experienced in conflicts is likely to be understood differently by each party than that which happens in a coercive context or randomly. There are certain relationships in which isolation forms a framework within which violence, control, and intimidation emerge. Even within coercive control relationships, violence occurs in various circumstances.
One participant said that criminologists often target questions weakly in connection with conflict behavior that has not been "thought to be a crime." Using a "crime context" (like the NCVS) elicits a dramatically different response than asking from a context that is more "sociological." Martin Schwartz recommended Jody Miller's book on girls involved in gangs (Oxford University Press). Among these girls, 100 percent described situations of extreme gender inequality, but in screening questions, this was not admitted.
An important context is how attached the woman is to the partner, how "invested" the girl or woman feels with regard to the male's behavior. An aspect of this is the effect of motherhood on experience with violence. Over 80 percent of Dr. Cook's study population were mothers. The woman may try to "put a different context" on coercive action. Dr. Krishnan said she was very interested in more longitudinal study of this kind of "deconstruction" of violence, particularly in a cultural context. Dr. Cook also suggested looking at the correlations of motherhood to help-seeking behavior.
How are people "hearing" the instrument questions: do they say, "talk to us about conflict behavior, even if you don't think it is a crime?" It could be valuable to have different descriptive wording for the violent act in different parts of the survey. The most serious batterers may not respond in the context described in the CTS. Both women and men probably hear the questions differently than "Ph.D. researchers." For example, responses to questions about "traumatic life events" bring out descriptions of incidents that were overlooked when the word "assault" was used. Dr. Jeffries noted that, though it may be valuable to "change gears" in questionnaire wording, the reader should be privy to the methods and reasoning behind the survey strategy. How questions are introduced to the interviewed individuals is itself a legitimate research topic. Perhaps two studies should be done, one using more "behavioral" questions and one mimicking the NCVS victimization framework. Dr. Stark asked whether the same surveyed person was reporting on multiple contexts. He commented that without baseline data showing duration, prevalence, and incidence, it was hardly possible to evaluate the need for services or the effects of intervention.
Summation: Measurement Issues
Dr. Tjaden said victimization has to be seen as multidimensional and criminal justice functions and the CDC have different paradigms concerning violence. They still need to "learn to talk together." She had wanted, with the NVAWS, to compare findings over time, to get lifetime and annual prevalence. But other research attempts to show a victimization rate for family violence that can be compared to the NCVS. Although there is some stability in lifetime prevalence rates, annual and incidence rates are very unstable, relating particularly to people's unwillingness to discuss the phenomenon. Even combining instruments, it is really difficult to determine a "beginning" or "end" to the violence. Dr. Campbell added that the controlling behavior complicates this: it may continue beyond and between periods in which actual violence has happened. Past-year prevalence can be interesting to compare with lifetime prevalence. An hypothesis that also introduces factors relating to class/employment, or race/ethnicity, would perhaps be more able to look at the ability women have to extricate themselves from abusive relationships, using a large data set over time.
Dr. Tjaden discouraged the group from being "falsely scientific." This pattern of violence is different from traditional public health case definitions. Health problems, resources, and injuries do not (singly) define the psychological pattern. People using the survey instruments "translate" a great deal. For example, she found that, to the question, "How many times?" a person might say "12" when they actually meant "once or twice a month approximately." This can be very problematic for measurement. A participant suggested trying to use first-time hospitalizations or first-time arrests to estimate incidence. Professional interviewers who will hold to a predetermined script can help achieve a consistent approach. There are significant conflicts between research attempts based on event measurement and those based on a typology (within a variety of frameworks).
The context in which surveys have been administered can cause disparity in findings for prevalence and symmetry (as found with the NVAWS). More information is needed before this particular instrument is propagated. Cultural/ethnicity issues and timing of questions (during the interview) should be examined. Fatigue starts to be a factor; for example, a three-hour phone interview would be impossible. Questions framed in a context of personal safety receive different responses in screening than those worded with reference to "crime." Women seem generally more willing to talk than men, about their own use as well as experience of violence. This affects statistics on women's use of violence. Also, interviewers sometimes miss important qualitative information when they feel pressured to get a "completed instrument."
Women's Use of Violence
Presentation and Group Discussion
Suzanne Swan gave the presentation on women's use of violence in intimate relationships. She described her preliminary contextual model for addressing this issue. The model is based on a study of 108 women who used some form of physical violence against a male partner in the previous six months, as well as the existing literature on women's violence. The variables addressed the context of women's violence relative to victimization, motivations for violent behavior, different coping strategies as related to use of violence, effects of childhood trauma on use of violence, and psychological outcomes in the framework of women's violence.
First, the model portrays the relationship between women's violence and victimization (i.e., their male partners' violence against them). The model's bidirectional path relates the woman's violence to her partner's violence; that is, as the violence of one partner increases, the violence of the other partner will increase as well. Evidence suggests that women's commission of physical or psychological violence and injury of partners is significantly correlated with their male partners' commission of these behaviors. While women in the sample used equivalent levels of emotional abuse and more moderate physical violence against male partners than the partners used against them, the women were more often victims of more serious types of violence, including sexual violence, injury, and coercive control behaviors. Even for women who were primary aggressors, significant victimization occurred, according to their self-reports.
The second part of the model portrays women's motivations for using violence, categorizing these into (1) reactive motivations, i.e., reactions to the male partner's violence (self-defense, protecting children, and fear) and (2) proactive motivations, predictive of women's violence but not necessarily reactions to the partner's violence (control, jealousy, and retribution). Dr. Swan's study particularly assessed self-defense (a reactive motivation), and retribution and control (proactive motivations). Self-defense was predicted by the extent of women's victimization. Retribution and control, in contrast, did not correlate with women's victimization, but did correlate with women's use of violence.
The model next examines women's coping methods for abusive relationships and how coping style relates to women's violence. Within the coping literature, coping behavior is often grouped into avoidant, problem solving, and support seeking. In her 1999 study, Dr. Swan assessed how much women used these three types of coping strategies to deal with relationships conflicts. As expected, the more problem solving strategies women employed, the less violence they used; and the more avoidant strategies women used, the more violent they were. Unexpectedly, social support was also positively associated with violence. That is, the more social support the women received from friends and family, the more violence they used. Some women's social support networks may have encouraged the use of violence to avoid coercion.
The next part of the model examines the relationship between childhood trauma and women's violence. Evidence indicates that rates of childhood trauma and abuse are very high among women who use violence. Dr. Swan asked participants in her 1999 study about a variety of traumatic events, ranging from natural disasters to sexual abuse. For the current model, she counted only events of major or extreme impact on the participant's life; 36 percent of the women experienced childhood physical abuse, 39 percent experienced emotional abuse, 50 percent were sexually abused, and 38 percent witnessed violence or fighting between parents while growing up.
Childhood trauma predicted women's violence and victimization as well. Women who witnessed parental violence and/or were victims of childhood abuse also experienced greater levels of violence from the male partner. Victims of emotional abuse as children used more avoidant coping strategies in relationships. Victims of childhood physical abuse were more likely to injure their partners, and sexual abuse victims used more coercive control tactics and showed more anxiety and depression as adults.
Finally, Dr. Swan's model examines the relationship between women's use of violence and outcomes. Some studies have found that women's violence was a risk factor for severe retaliatory violence by their partners, but this does not appear to be a straightforward relationship. In the present study, two-thirds of the women who used violence in self-defense said that it was at least sometimes successful in getting the abuser to stop. Fighting back, then, is a problem solving strategy associated, in some cases, with more positive outcomes.
To examine this hypothesis, the sample was divided into relatively "low violence" and "high violence" relationships. These low and high violence relationships were again divided into those in which the woman was primarily the victim and those in which the woman was primarily the aggressor. Substance use, psychological outcomes, and injury were compared across the four groups. Women who were primarily aggressors in low violence relationships had significantly less depressive symptoms, and women victims in high violence situations were the most depressed. The other two groups fell between these scores. This same pattern emerged for PTSD and problem alcohol consumption. Furthermore, women who were aggressors in low violence relationships were injured significantly less than women in all three of the other groups.
Dr. Swan discussed women's violence in the context of "intersectionality," For African-American women, positive racial socialization (pride in heritage and culture), along with spirituality and extended family, may be resilience factors. Specifically, a more positive socialization experience may make entry into (or long exposure within) violent relationships less likely, but it is unclear how racial socialization affects women's own use of violence.
Some aspects of Hispanic cultures also affect responses to abuse and decisions to use violence. In one study comparing Hispanic, African-American, and Anglo women in domestic violence shelters, Hispanic women reported the longest duration of abuse and fewest attempts to seek help. Another study found no differences in the severity and frequency of abuse among Hispanic and Anglo women, but Hispanic women labeled behavior as abusive less often than Anglo women, unless it occurred more frequently. In research on patriarchal culture and Mexican women, those who adopt gender roles of "Marianismo" (submissiveness, self-sacrifice, and stoicism) may be particularly vulnerable to abuse. While Dr. Swan's study found support seeking coping behavior to be positively related to African-American women's violence; for Latina women, this pattern may be reversed. The Latina women's social support networks may be more likely to discourage use of violence.
Women living in poverty have been shown repeatedly to be at elevated risk for domestic and other forms of violent victimization. Severe economic disadvantage in a neighborhood could be related to a "cognitive landscape" in which violence is accepted as inevitable and normal. Dr. Swan predicted very strong paths between women's violence and victimization, in cases of poor women living in such disadvantaged neighborhoods. These women will be accustomed to using violence as a survival strategy, often responding violently to partners' violence against them.
The presentation stimulated a variety of questions (explored in varying degrees) among the workshop participants:
- How should "proactive" and "reactive" motivations be clearly distinguished? Would retribution be considered "reactive?" Could "protecting children" be considered proactive rather than reactive?
- Can antecedents be sorted well from outcomes? Is the coping behavior an outcome of victimization or a precursor? Is there any direct path from victimization to coping?
- How exactly were the questions to the women worded? How can motivation be "operationalized" in the questions?
- What is the role of male peer support in woman abuse?
- How is anxiety and substance use related to women's use of violence?
- What is the explanation for contradictory findings on "positive racial socialization" for African American women?
- Should adolescent development be studied in connection to later family violence patterns?
- What are the driving effects of societal and cultural messages about violence (especially for boys)?
Women's use of violence involves a complex set of adaptive relationships. A tendency to dichotomize in quantitative research often avoids the complication common to this subject. Dr. Swan said that self-defense, among the women respondents, correlated more with victimization rather than use of violence. Women who said they tried "to control the partner" or "get even" were considered to be using a "proactive motivation," and this motivation was also more predictive of violence from the woman. "Reactive" was defined as a response to the man's violence, while proactive motivation was more concerned with controlling the man's behavior. Dr. Swan said the motivational elements in the study were targeted with simple wording: "How often did you use violence to keep him from using violence against you?" This question would contribute to the "self defense" variables.
One person expressed concern about the choice of words. It may look at first like the man and woman are doing the same thing (in a violent action), but it must be seen in a "gendered" and cultural context. The woman's use of violence in a relationship may be largely controlled by how the man responds to that violence. The "common couple" violence situation may be an adaptive behavior that should be understood differently than other kinds of violence. A separate study of the subgroup of women who are proactively aggressive and violent might clarify this phenomenon. Female aggression may be better understood in the context of deprivation of liberty. This study approach should be distinct from stigmatizing psychological ideas like "learned helplessness," to avoid stereotyping.
Another participant said Beth Richie had found an inverse relationship of positive racial socialization to "staying in abusive relationships." In some African American communities, men have been perceived as too disproportionately affected by racism. The women in those areas somewhat "excuse" abusive behavior and do not want the law to "come down too hard." Also, the women feel strong responsibility for keeping the family together at all costs. Some disadvantaged urban areas are practically "riddled with absence of positive social support (for men and women alike)."
Work on female gang involvement in Australia and Boston has indicated that girls are using violence partially to defer further violence. In other words, perhaps, they prefer "being victimized" in a gang situation to victimization on the outside. In a University of Pennsylvania study of gang girls, many saw their own use of violence in a situation where tensions were mounting, particularly when they fought other girls at school or in community settings, as the best way to reduce the prospect of further and more severe violence, even if they lost. One participant referred to intriguing research about gang experience occurring in Hawaii. Gang affiliation might be a perceived solution for poverty, hopelessness, and violence in the family.
Dr. Straus objected to considering any form of violence as "part of a solution" to family violence. His own research found that women who responded with violence ran the greatest risk of escalating violence and injury. In the long term, Dr. Malcoe said, antagonism (gender hostility) has not worked well for girls according to their own reports, although it might look like an adaptation. Their own violence "scared them" over time, and they would initiate change (such as ending the relationship). Use of violence blocks development of nurturing relationships and has been connected to PTSD symptoms in women. Childhood victimization may help to understand hypervigilance, readiness to attack, etc. in girls; but "adaptive violence" is certainly not a positive approach to the world. To study this more deeply, a model would need to look at developmental aspects, using time-ordered cross-sectional inferences and a longitudinal dataset. There are similarities in the difficulties of young boys and girls alike in some inner city areas.
David Ford asked the group to think as well about general sources of violence in society and cultural messages about violence. Girls showing this "adaptive violence" should be regarded as serious physical and sexual assault victims. What developmental factors are involved among these young women who take such a different path than the gendered norm? Some of the behavior is pointing to problems in the systemic responses, such as pro-arrest policies recently attempted, etc. The woman may find it as helpful to "fight back" as to call the police. Glamorized violence has been a driving influence generally, especially among young boys. Perhaps it is surprising that women are not more violent.
There are, however, different kinds of men and different kinds of families, Dr. Malcoe said. Longitudinal research needs to collect broad data on institutional, family, and structural factors, including the degree to which various kinds of physical aggression have been "criminalized." When women are acting violently, is it really "equal?"
Dr. Stark said research efforts must get beyond thinking that "physical fighting of any kind" is the same as coercive control. When a demonstrably "oppressed person" fights back, it is not "equal demonstration of violence." There are race and class issues connecting to the larger society and material well being. Recognizing that this struggle has to do with subjugation and dominance can put important issues "on the table." Broader research may help social responses to get beyond the "second-level-misdemeanor revolving door" for domestic violence offenses. The group should be using its knowledge and expertise to frame the issues surrounding women's use of violence. Dr. Stark recommended establishing baseline information on depression and "phobic" behavior in certain communities, referring to research by Ericson on "community traumatization." PTSD, as it is used in court situations, means a chronic, debilitating, not easily remedied psychiatric condition. Labeling for this condition can be extraordinarily stigmatizing for women.
Violence is certainly not only inter-gender. There has been such a rush to criminalize forms of violence that some attempts to change policies (dual arrest situations, as a possible example) have become ineffective. Juvenile dating pairs often show (even severe) violence relating to fidelity and behavior showing respect. Developmental paths (as a context) lead to different kinds of violence, and this cannot be seen if the research just counts injuries. Context has to be understood also in relation to men's violence. We are all exposed to the same kind of cultural messages, Dr. Ford said. For batterer intervention, most failure means the person never showed up. This alone would not be considered "failure" in substance abuse interventions. But for substance abuse, the social response does not try to give "everyone the same" treatment. Dr. Tjaden said that the message is not to condone any violence, but to reduce it overall.
Summation: Women's Use of Violence
Gender makes a difference, noted Claire Renzetti opening her summation of the discussion on Suzanne Swan's presentation. Women are not just "acting like men." A common thread in the discussion has been that the object of research involves violent interactions rather than just violent incidents. Longitudinal, qualitative research, with rich detailed interview data on context, meaning, and motive would be desirable. Existing studies have had inadequate categories for diversity (for example variations in responses of Latin American women depending on specific country - Puerto Rico, Mexico, Dominican Republic, etc.) Better study methods require a layered model with social, ethnic, and other factors intersecting. Findings controlled for neighborhood disadvantage in past studies have shown that many racial and ethnic differences disappear. Dr. Swan's paper raises more questions than it answers, but points toward factors that cannot be neglected in the effort to understand women's violence. It asks whether male and female violence have different causes and forms. Should "counts" be performed in the context of differently defined types of violence? Also, the woman's use of violence may be influenced by the community and justice system response to her requests for help.
To study victimization in the life experience, Dr. Renzetti said, we have to conceptualize other factors that intersect and characterize important pathways to violence, including factors like juvenile development of roles, sexual assault, and harmful behaviors like substance abuse. These are often different for women than men. What is specifically meant by "common couple" violence, a kind of "catch-all" phrase? How can survey instruments better "get at" context and motivation (especially over time)?
Collaborative research, which places advocates and clinicians on an equal footing, may help achieve clearer perspectives on meaning and context. The group today has considered those aspects of Johnson's typology that involve the most violence, but for gender symmetry, further study would have to bring in all other parts. Dr. Swan's study aimed to find women who might be classified as batterers, but she found only 16 of the 108 women she studied (all of whom admitted to violent acts against intimate partners) who could fit into the "intimate terrorist" typology. Logically, one might expect even more women to murder or use non-lethal violence. Rather than just "dichotomizing" inter-gender violence, research should be concerned as well with specific forms and context of the violence.
Walter DeKeseredy discussed and summarized directions of work illuminated by the day's presentations and exchange. Pain and suffering are multidimensional in relationships. Conflict, meaning, and motive have been discussed a lot in this meeting, together with strategies for measurement that can capture these dimensions. Path-breaking work by Dan Saunders on these issues maybe an important direction to take. There is a sense that the research community has been divided into two camps: criminologists (some with feminist roots) and family conflict researchers. However, academic work has moved beyond conflict of the 1980s and, as reflected in today's conversation, can support new theoretical and empirical approaches. Integration of feminist concerns with main stream survey research was also addressed in a recent CDC conference. There is a need to examine variation in female and family violence across different ethnic groups and in specific geographic locales. Multiple measures are needed to address underreporting (among both men and women), and studies should highlight societal structures that may be failing women who have been abused. Examination of the heightened risk reported in connection with marital separations would also be valuable.
In trying to arrive at standards for policymakers, the research has to avoid using particular "world views" toward intimate relationships. Clearly, it is time to move past simple counts of injury, seeing the dangers associated with some symmetrical interpretations of numbers from conflict tactics instruments. We are seeing a climate characterized by backlash, said Dr. DeKeseredy; in Canada, certainly. While the public sector cannot hold researchers responsible for what people do with data; how data is generated, the context of the research itself, and research publications have serious implications. We must be careful about how we disseminate our research, he noted. There are problems with using data and "hierarchies of seriousness of violence" in political arenas. If a small group (with little money) that was providing services to women and batterers can be considered "violating Alberta's men's human rights," said Dr. DeKeseredy, such a precedent tells researchers to be cautious about the use of their data, notes, and work. Government agendas are often pushed by conservative lobby groups, and independent academic work needs to be strengthened.
The complex nature of the work has to be communicated clearly to the media and the public in order to avoid distortion, using explanations that carefully interpret effects of gender,race, and social class using local survey strategies. There may be geographic variations in different cities (as already found in the U.K.)
Regardless of whether men or women are perpetrating violence, it does not exist in a vacuum. Too often, Dr. DeKeseredy said, we miss linkages between economic and political difficulties where families are living and bearing stresses that are internal to the family. These are opportunities to connect "personal troubles to public issues" as Mills would say. The ultimate sociological question involves linking macro level forces with those acting on the individual level.
Dr. Jeffries thanked the group for the depth of commitment and expressed honesty. We have to think about what we write, he said; there are collateral consequences for criminalizing behavior and unintentional "social construction of crime." The discussion showed that even methodological research itself, as a behavior, draws criticism. Coping strategies and adaptive behavior were important topics. Spencer's research has worked on the way in which boys and girls in early adolescence identify "gender roles," and theories concerning what is maladaptive in that process. People can now speak more openly of gender behaviors as relational, meaning not just certain actions in men or certain behavior in women. The work of this workshop will help to move the research field forward.
. Reporter's note: prepared for the Violence and Victimization Division, Office of Research and Evaluation, National Institute of Justice, U.S. Department of Justice. Workshop was attended by 36 people, from 16 different universities, several private research organizations, and officials from the Department of Justice and the National Center for Injury Prevention and Control (Centers for Disease Control).
. Straus, M. A. (1999). "The Controversy over Domestic Violence by Women: A Methodological, Theoretical, and Sociology of Science Analysis." In X. Arriaga & S. Oskamp (Eds.), Violence in Intimate Relationships (pp. 17-44). Thousand Oaks, CA: Sage.
. Straus, M. A., Kaufman Kantor, G., & Moore, D. W. (1997). Change in Cultural Norms Approving Marital Violence: From 1968 to 1994. In G. Kaufman Kantor & J. L. Jasinski (Eds.), Out of the Darkness: Contemporary Perspectives on Family Violence . Thousand Oaks: Sage Publications.
. Simons, R. L., Lin, K.-H., & Gordon, L. C. (1998). "Socialization in the Family of Origin and Male Dating Violence: A Prospective Study. Journal of Marriage and the Family, 60(2), 467-478; Straus, M. A., & Yodanis, C. L. (1996). "Corporal Punishment in Adolescence and Physical Assaults on Spouses Later in Life: What Accounts for the Link?" Journal of Marriage and the Family, 58(4), 825-841.
. Straus, M. A., Hamby, S. L., Boney-McCoy, S., and Sugarman, D. (1999). "The Personal and Relationships Profile (PRP). Durham, NH: Family Research Laboratory, University of New Hampshire; Straus, M. A., and Mouradian, V. E. (1999, November 19, 1999). "Preliminary Psychometric Data for the Personal and Relationships Profile (PRP): A Multi-Scale Tool for Clinical Screening and Research on Partner Violence." Paper presented at the American Society of Criminology, Toronto, Ontario.
. Defined by Bograd, "Intersectionalities color the meaning and nature of domestic violence, how it is experienced by self and responded to by others, how personal and social consequences are represented, and how and whether escape and safety can be obtained."