Current domestic violence theory indicates that abuse is a learned behavior, patterned after behaviors seen in the family of origin or the society at large (Gortner et al., 1997; Henning et al., 1996; Mihalic & Elliot, 1997). However, the majority of this research has focused on males as perpetrators. Those that have explored female perpetrators frequently do not examine the context of the violence, therefore placing the primary aggressors and those acting in self-defense in the same category. This study is intended to explore the histories of violence among male and female perpetrators of domestic violence. Additionally, the impact of these experiences on feelings about the relationship, specifically feeling of fear, weakness, and powerlessness in the relationship, will be explored.

LEARNING DOMESTIC VIOLENCE

Learning theories, such as socialization, social learning, and elements of feminist theory, assert that behaviors are learned throughout our lives through our interactions with others. These interactions teach individuals, in mostly subtle ways, what behavior is and is not appropriate, as well as what rewards and consequences will be brought about due to these actions (Mihalic & Elliot, 1997). In this way, an individual learns the emotional and physical tactics of domestic violence and incorporates them into his or her behavior. Learning and generational models claim that emotional, physical, and sexual violence are learned behaviors, most often modeled after witnessing violent behaviors of family members, whether the violence was inflicted against the child or just observed being inflicted on a parent (Cappell & Heiner, 1990; Gortner et al., 1997).

Postmodern thinking supports these claims, stating that individuals “create the social reality for their behavior according to how they interpret norms in particular milieus” (Markward, 1997). For children from violent homes, use of violence can be interpreted as the norm in intimate relationships. It is unlikely that children from abusive homes believe that physical violence is always acceptable. Most understand, at least to some degree, that abuse is wrong by the time they reach adulthood. However, children from violent homes may be more likely to be able to find rationalizations for their physically abusive behaviors (Wagar & Rodway, 1995). Additionally, attitudes about relationships may allow an individual growing up in an abusive home to view the emotional tactics, such as extreme jealousy, intimidation, and monitoring the behavior of his or her partner, as normal and acceptable. To the children, some violent and abusive behaviors may be seen as appropriate and normal expressions of emotions.

Children and adolescents not only learn that violence is acceptable, but also the specific abusive tactics that can be used to control a partner. It is theorized that children learn that conflict is resolved through violence, family interactions involve violence, and this violence is an acceptable means of stress management and conflict resolution (Wagar & Rodway, 1995). In this way, an individual learns the emotional and physical tactics of domestic violence and incorporates them into his or her behavior.

A number of factors contribute to the learning and replication of violence. For example, it is possible that individuals who has been abused or witnesses abuse lose faith in the fairness of the world and see violence as a means to avoid being further victimized (Straus, 1990). Studies of abusive men have found that men believe it is their spouses who wield tremendous power in the relationship (Claes & Rosenthal, 1990). Such men see relationships as continual power struggles in which there must always be a winner, the dominant partner, and a loser, the dominated partner. Violence is used to regain power and self-esteem (Campbell, 1993).

As children, abusive men learned that they must identify with either the abused or the abuser and model that behavior. Fear of being hurt may propel these individuals to choose to model the behavior of the abuser (Mihalic & Elliot, 1997). Another possibility is that children from abusive homes see the effectiveness of the abusive behaviors in getting the abuser what he or she wants (Hotaling et al., 1990). The utility and rewards of this behavior appear to outweigh the consequences; therefore, the behavior is learned and repeated.

Along with learning the skills to abuse, it is hypothesized that witnessing domestic violence has profound emotional impacts that carry though to adulthood, which may put the individual at greater risk of becoming an abuser (Hoglund & Nicholas, 1995; McNeal & Amato, 1998; Wagar & Rodway, 1995). Increased levels of feelings of shame and guilt have been linked with witnessing violence as a child, often from feeling responsible for the abuse and unable to stop it (Wagar & Rodway, 1995). Shame can have the effect of causing the person to feel inferior, defective, and helpless, all of which have been shown to be characteristics of domestically abusive adults. Shame and guilt have also been linked with hostility, including anger arousal, tendency to blame others and irritability (Hoglund & Nicholas, 1995).

Once an individual has learned to use abusive behaviors, either through societal support or childhood experiences with abuse, the individual is faced with the choice of whether or not to use these tactics in intimate relationships. As not all children from abusive homes nor all men raised in a sexist society become abusive (Kolbo, 1996; McCall & Shields, 1986; O'Keefe, 1998), factors must exist that allow some to choose to use violence while others do not.

Few studies have examined the factors that mediate the relationship between witnessing domestic violence and perpetrating it. The majority of research exploring the intergenerational cycle of violence does not control for possible mediating variables, such as substance abuse or marital stress (Julian et al., 1999; McNeal & Amato, 1998). However, some risk factors for perpetration have emerged in the literature.

A commonly reported intervening variable in the relationship between witnessing domestic violence and adult perpetration is experiencing child abuse (Domestic Violence Advisory Council, 1998). Due to the high correlation between child abuse and domestic violence (Hughes et al., 1987; McKernan McKay, 1994), it has been difficult to determine how these two types of violence interact to impact future violent behavior. In addition, because memories of abusive incidences are frequently distorted over time (Williams, 1994), it is likely that some experience with adult violence may either increase or decrease the likelihood of remembering and reporting childhood violence. This has the effect of introducing systematic error that retrospective studies have not been able to resolve.

Marital satisfaction was also believed to be related to violence (Julian et al., 1999; Mihalic & Elliot, 1997). Although marital conflict and satisfaction frequently appear in domestic violence research as independent variables for predicting abuse, these variables could be highly problematic. The difficulty is in the unproven presumption that conflict or dissatisfaction is a cause of violence and not a result of the abuse.

Few longitudinal studies are available to examine risk factors for violent behavior. One 12-year study, involving interviews with both parents and their adult children, explored the impact of non-violent marital conflict, child abuse, divorce, and substance abuse on the relationship between witnessing spousal abuse and later perpetration (McNeal & Amato, 1998). The study found that none of the hypothesized mediators were significantly related to violence perpetrated by the offspring. This may be due in part to the simplistic nature of some of the variables. For example, as a measure of child abuse, parents were only asked if the mother or father “is sometimes abusive toward the children.” (McNeal & Amato, 1998, p. 129) A second study (Mihalic & Elliot, 1997), following a sample for 13 years, found that non-traditional sex role attitudes, marital dissatisfaction, and stress were related to violent perpetration among males. Among females, factors such as criminal perpetration in adolescence, prior victimization, and child abuse were also important predictors.

The applicability of learning theory and the intergenerational transmission of violence (ITV) has been supported by the finding that the majority of males who use violence in intimate relationships are not violent with others outside the family (Saunders, 1992). Although some have posited that battering is a result of biological or psychological disorders (Dutton & Starzomski, 1993; Maiuro & Avery, 1996), such disorders would likely lead to violence outside the home in addition to the family violence. In actuality, individuals who are found to be generally violent comprise only 20% of batterers (Saunders, 1992; Walker, 1995). Early socialization into violent gender roles has also been used to explain male use of violence. Boys growing up in violent homes are three times more likely to become abusive men in their intimate relationships than boys from non-abusive homes (Mihalic & Elliot, 1997).

Theories of the intergenerational transmission of violent perpetration have primarily examined males. Females typically are examined only as potential victims. These theories are often based on traditional gender role assumptions that girls identify with the victim while males identify with the perpetrator. Evidence is mixed regarding whether girls from abusive homes are more likely to be victims of abuse as adults (Henning et al., 1996; Mihalic & Elliot, 1997). Most studies have failed to support the gender-role socialization theory, which states that girls are more likely to be abused as adults because they model passive, victimized behaviors of their mothers. Instead of being socialized to accept male domination and becoming self-sacrificing, abused women generally perceive themselves to be more liberal (Mihalic & Elliot, 1997).

While great strides have been made to explain male use of domestic violence in heterosexual violence, these theories have not generally been tested for their applicability to female perpetrators. Social learning and gender socialization theory, describe a path in which females exposed to violence in childhood should learn to be victims, not perpetrators. In each of these categories of theories, there are clear deficits in these theories' ability to account for female-perpetrated domestic violence.

White and Humphrey (1994) identify an alternative path in which women choose to use violence. The study found that more than one-quarter of all women using violence had experienced parental aggression and nearly half had experienced sexual assault as an adolescent. More than 85% had experienced verbal aggression and nearly half were physically abused in a dating relationship. They hypothesize that due to these high levels of past victimization, these young women may be less trusting and more alienated from others, leading them to feel more threatened by intimidating behaviors by their partners. This heightens the perceived need for self-defensive behavior by women.

GOALS AND HYPOTHESES

While a great deal of research has examined the relationship between family of origin violence and adult perpetration among males, few studies have examined this history among female perpetrators. This study will examine if female respondents have a history of violence in their childhood and previous dating relationships. The study expands on previous research on the transmission of violence by including sexual abuse in childhood. Research has repeatedly found a link between victimization and perpetration of sexual abuse (Falshaw et al., 1996), but is not typically included in examinations of the intergenerational transmission of domestic violence. Williams (2003) asserts that the many forms of violence examined in isolation do not accurately represent learning and emotional responses to trauma. Therefore, while they are generally not included in discussions of the path to learning violence, sexual abuse and prior adult victimization have been included in the analysis. The differences in the relationship of this history on attitudes and behaviors among males and females will be examined.

Some have posited that battering is a result of biological or psychological disorders instead of social learning (Dutton & Starzomski, 1993; Maiuro & Avery, 1996) However, such disorders would likely lead to violence outside the home in addition to the family violence. In actuality, individuals who are found to be generally violent comprise only 20% of batterers (Saunders, 1992; Walker, 1995). Due to this, violence perpetrated against individuals other than partners or family members will also be measured in this study. This is intended to indicate if it is the family violence that is learned or more generalized violent behavior.

Two hypotheses were developed to examine the impact of family of origin violence on females arrested for domestic violence. First, it was expected that both males and females would report high levels of abuse and violence in the family of origin. The second hypothesis describes the expected impact of the abuse on women's decisions to use violence, similar to the work of White and Humphrey (1994). Individuals who have experienced abuse, either in childhood, prior adult relationships or both, were believed to be more likely to use violence in response to real or perceived threats than those who did not. These individuals were expected to be more likely to report using violence in self-defense and were expected to report feeling powerless and scared at the time of the incidences in which they are violent.

METHODS

A one-time written questionnaire was administered to a sample of English-speaking participants in batterer intervention counseling programs in Los Angeles County, California. All participants in each of the 15 selected counseling groups were invited to participate. Agencies were selected from the urban regions of Los Angeles County in an attempt to reflect the diversity of the Los Angeles metropolitan area. Three (2.6%) individuals refused to participate. Eight additional surveys (6.4%) were excluded from the sample because they were only partially complete. Of the surveys that were excluded from the study, seven were from males and four from females, resulting in a sample that was 52.6% male (n = 60) and 47.4% (n = 54) female.

The questionnaire was administered by the researcher at the counseling site during the first hour of a normally scheduled counseling meeting and took 45 min to 1 hr to complete. Group facilitators were not present at the time of the administration, but introduced the researcher to the group at the beginning of the session. The researcher read the survey instrument aloud while the respondents completed the instrument in order to minimize literacy barriers. Subjects were assured that their participation was voluntary and anonymous. Both the Los Angeles Probation Department and General Campus Human Subject Protection Committee approved these procedures prior to beginning the study.

INSTRUMENTATION

Prior experience with violence was measured in 10 variables. Subjects were asked if they experienced abuse in a previous relationship, were the victim of a sexual assault, witnessed abuse in the family of origin, or were the victim of abuse by a parent or other primary caregiver as a child. Emotional, sexual, and physical abuse were included. Given a list of several behaviors that would constitute each form of abuse, respondents were asked to rate the frequency each occurred on a Likert-type scale, with one Violence perpetrated against people other than a partner in adulthood and childhood was also measured to determine the degree to which the respondent is violent outside intimate relationships. Only physical violence was measured on these four perpetration variables.

Recent domestic violence victimization was measured using an abuse inventory comprised of a modified version of the Psychological Maltreatment of Women Scale (Tolman, 1989, 1999), and a modified Conflict Tactics Scale (Straus, 1979). Together, the scales examined the spectrum of abuse, including emotional, sexual, and physical. The scale consisted of 24 items measured on a five point Likert-type scale, ranging from “never” (1) to “very frequently” (5). Subjects were asked how often they have experienced abusive behaviors in the past 6 months. The scale was found to have high internal consistency (α = .96)

In addition, subjects were asked how often they thought their partners were fearful of them and how frequently they were afraid of their partners. The measure of fear was included to indicate the degree of threat the individual felt, since the same acts may not be perceived or experienced the same way by different people. As dating violence literature has found, the impacts of similar types of violence is different for males and females (Molidor & Tolman, 1998; O'Keefe & Treister, 1998). This item, measured on a five point Likert-type scale where five indicates very frequent fear, measures generalized fear of one's partner. Participants were also asked about emotions they may have felt at the time they perpetrated violence, measured on a five point Likert-type scale. Three items (afraid, powerless, and weak) from the scale were used to measure the concept of context-specific disempowerment and fear. This scale had high internal consistency (α = .90)

Self-defense was measured in only one question. Participants were asked, of all the times they had used violence in the past 6 months, how often they used violence to protect themselves. This item was measured on a five point Likert-type scale ranging from “never” to “always.”

Several measures were considered possible confounding variables and their effects were examined in the analysis. These included the amount of time the subject has been receiving counseling, whether or not participation is court ordered or voluntary and social desirability response bias. Demographics such as age, sexual orientation, race, income, and education were also measured.

SUBJECTS

The sample was drawn from the 59 batterer intervention service providers that provide services for both women and men in Los Angeles County. The sample was demographically similar to the County population and no significant differences were found between males and females in the sample. The sample was racially diverse. Nearly half of the sample (46.6%) identified as Chicano/Latino, 33% as Caucasian, 9.7% as African American/Black, 5.8% as biracial, 3.9% Asian American/ Pacific Islander, and 1% Native American.

Average age of respondents was 34, ranging from 19 to 70 years old (SD = 9.73). Most of the respondents had completed high school or obtained a GED. The median education level was “some college, but no degree.” Partners of the respondents had slightly less education than the respondents with a median education of “high school diploma or GED” (39.4%, n = 39). Reported annual family income of the sample ranged from no income to $750,000, with a median income of $37,000. However, this analysis needs to be viewed with caution due to the high level of missing data on this item (36.8%). The percent of individuals living below the poverty level was 16.7%.

Most of respondents committed violence against a partner they were involved with at the time of the incident (72%), reporting being either married (46%), dating (32%), or engaged (20%). All but two of the remaining respondents committed violence against a former partner. Those arrested for violence against a person other than a partner or former partner reported the violence occurred in the context of an abusive incident with a partner. Most respondents (71.9%) were living with their partner at the time of the incident. Abuse occurred in both new and long lasting relationships, ranging in length from 1 month to 20 years. The mean length of the relationships was just under 6 years (5.8) with a median length of 4 years and 8 months. Two thirds of the respondents (n = 67) have children under the age of 18. The average household size was 3.13, ranging in size from one to seven individuals.

At the time of the survey, the mean number of counseling sessions attended was 26, or approximately half of the 52 week program. A small percentage (9.6%) had previously attended a batterer intervention program at a different agency. The median number of session attended at another agency was 34, with one woman completing two 52 week programs for a total of 104 sessions. Nearly all of the respondents (94%) had been ordered into the counseling program, including criminal court or probation, the Department of Child and Family Services or family court. Only seven respondents were voluntarily attending the program.

RESULTS

Prior experience with family and dating violence was extremely common among the sample. Nearly all respondents reported some previous exposure to violence. Only 2.9% (n = 3) indicated that they had never experienced any of the forms of violence prior to their current relationships. None of the demographic or counseling variables was significantly related to experience with violence in childhood, indicating that experiences with the different forms of abuse were similar across the sample.

Nearly three-quarters (74%) had witnessed some form of domestic violence as children (n = 77). Emotional abuse was more common than physical, with 70% reporting witnessing emotional abuse at least occasionally, compared to 61% for physical abuse. Most (68%, n = 71) reported that they had been victims of child abuse. Like witnessing domestic violence, emotional abuse was more common than physical. Just over half (53%) reported experiencing physical abuse and 64% reported emotional abuse. Differences in child abuse victimization were not significantly different for males and females (t = 1.883, df = 102, p = .062). Child abuse and witnessing domestic violence were highly correlated (r 2 = .76, p = .000).

Rates of sexual assault and abuse were also quite high, with 43% reporting some sexual victimization as a child or adult. Females were significantly more likely to have experienced sexual abuse (t = −3.698, df = 102, p < .001), with 59% reporting some victimization compared to 29% of males Table I.

Table I. Sexual Assault and Abuse by Gender
Full size table

Domestic violence was also common in previous relationships, with more than half (60%) reporting prior victimization. Physical abuse was reported in at least one prior relationship by 58% of females and 24% of males and 56% of females and 51% of males reported emotional abuse. Females were significantly more likely than males to report prior physical domestic violence victimization (t = −2.402, df = 103, p = .018).

The majority of respondents reported that they were not generally violent outside the family, supporting the theory that most perpetrators of domestic violence are only violent in the home. Approximately half of the sample (52% of females and 46% of males) reported never getting in physical fights with adults outside the home. There were no significant differences in the rates of males and females fighting outside the family (F = .144, df = 104, p = .705).

Both males and females reported low levels of fear of their partner, with a response of É1É indicated no fear and “5” indicated very frequent fear. However, the difference was statistically significant (t = −2.046, df = 109, p = .045). Females (M = 2.21, SD = 1.39) reported more fear of their partners than males (M = 1.71, 1.18). Fear of one's partner was significantly correlated with the level of abuse victimization reported for both males and females (r 2 = .71, p = .000), indicating that fear of partner was directly related to the frequency and severity of abuse.

Most of the demographic and counseling variables were not significantly related to fear of one's partner, with the exception of race and income. Race was a significant predictor of fear of partner (F = 2.821, df = 99, p = .020). Those identifying as biracial reported significantly more fear than Chicano/Latinos (p = .047). Income was also a significant predictor (F = 10.656, df = 82, p = .002) with individuals reporting higher incomes reporting more fear of their partners.

PRIOR ABUSE AND PERCEPTION OF THREAT

It was hypothesized that both males and females, who have experienced abuse in prior adult relationships or childhood would be more likely to use violence in response to real or perceived threats than those who did not experience abuse. To test the hypothesis, regression analyses were used to determine the relationship between abuse history and feelings of fear and powerlessness as well as identification of self-defense as a motivation violence.

The three forms of prior abuse were entered into the regression model. The model included witnessing domestic violence, child physical and emotional abuse, and sexual abuse or assaults as an adult or child. This model was not found to significantly predict reports of self-defense (r 2 = .073, p = .056). The model was then re-examined using a stepwise approach to determine which of the variables were significantly related to self-defensive behavior. Only prior sexual abuse emerged as a significant predictor (r 2 = .060, p = .012). Those who reported more sexual abuse were more likely to report self-defense as a motivation for using violence in a relationship. When the effects of gender were controlled, this model remained a significant predictor of self-defense (p = .042), but explained only 6% of the total variance.

The prior abuse model was examined for its relationship to perception of threat. The model was then tested for its ability to predict feelings of fear, powerlessness, and weakness. The model significantly predicted all of the emotions at the time of the abusive incidences as well as generalized fear of one's partner. Generalized fear of one's partner was significantly predicted by the model (p = .012), explaining 10% of the variance. Prior domestic violence victimization, and sexual abuse were significant predictors in the model. When controlling for the abuse history variables, gender was not a significant predictor (p = .187) indicating that both males and females who had greater exposure to previous abuse indicated more fear of their current partner.

Fear at the time of abusive incidents was also significantly predicted by the abuse model (r 2 = .134, p = .002). Only prior sexual abuse was significant when controlling for the other abuse variables (p = .003), with those experiencing more abuse reporting being more afraid. A significant interaction was found between gender and sexual abuse (p = .015), with the impact of sexual abuse being stronger among male survivors of sexual abuse than among females.

Feelings of powerlessness were also predicted by the model (r 2 = .166, p < .000). Again, only prior sexual abuse emerged as a predictor when controlling for the other variables (p = .004). Gender was not a significant predictor in this model. Feeling emotionally weak at the time of an abusive incident was also significantly predicted by the prior abuse model (r 2 = .131, p = .003). Like the other models, prior sexual abuse emerged as the only significant predictor when controlling for the other variables (p = .029). Gender did not significantly contribute to the model.

Prior exposure to, and experience with violence was found to have a profound impact on feelings of fear and weakness in relationships for both males and females. Those who have histories of abuse in childhood and previous relationships experience greater fear in relationships. Prior sexual abuse was the most significant predictor.

LIMITATIONS OF THE STUDY

The study is limited in some ways due to the sampling design and methodology. First, the cross-sectional design is an important limitation of the study. Reliance on self-report and retrospective data may bias the findings because of social desirability, memory distortions, personal re-definition of events, or biased recall. The ability to collect data longitudinally, gathering information from a variety of sources, such as parents and partners, would provide for a much richer picture of the experiences and context of the abuse.

Some limitations to the study are foreseen due to the sampling design and methodology. First, the decision to select participants from a clinical population limits the generalizability of the findings. Domestic violence is a highly underreported crime in which only a small proportion are arrested and a few are convicted (Tolman & Weisz, 1995). Therefore, the primarily court-mandated sample is likely to exhibit more violent behavior than the general population of abusive individuals. These factors may result in a potentially biased sample in which the results will only be generalizable to other individuals in counseling for domestic violence perpetration. However, the findings may have the greatest usefulness with a clinical population as they may provide direction for treatment programs for those arrested for domestic violence. Additionally, the findings show that abuse in the family of origin has a profound impact on experiences in adult relationships, indicating that prevention programs may be more effective if they also address these issues. However, other limitations, such as small sample size and the limited geographic area of the study may limit generalizability to other clinical populations.

The inability to control for time in counseling was another important limitation of the study because the counseling intervention may influence attitudes and behavior. Since many of the questions involved current attitudes or behavior in the past six months, they may not adequately reflect the dynamics of the abusive relationship prior to intervention. To account for this, the study statistically controlled for time in counseling for its relationship to key variables in the whole sample and was not found to significantly impact responses. This provides some information on the extent to which this limitation impacts the findings.

DISCUSSION

According to theories of the intergenerational transmission of violence, children, typically operationalized as male children, learn abusive behavior in the family of origin (Gortner et al., 1997; Mihalic & Elliot, 1997). While the current study found that participants reported high levels of exposure to family violence in the family of origin. Less than 3% of the participants reported no prior exposure to violence, including child physical or emotional abuse, witnessing domestic violence or sexual abuse or assault.

The impact of sexual assault and abuse, as an adult or child, on feelings in adult relationships, was found to be significant. When controlling for other forms of prior abuse, sexual abuse predicted feelings of powerlessness, fear and weakness as well as generalized fear of one's partner. Additionally, those who experienced sexual assault or abuse were also more likely to report using violence in self-defense.

Supporting prior research (Wyatt et al., 2000), these findings seem to indicate that those who have been sexually abused may be hyper-vigilant in relationships, sometimes responding violently to possible abuse. Additionally, physical abuse victimization in previous relationships was reported significantly more often by females than males and was significantly related to generalized fear of one's partner. This may be due to the first-hand knowledge of the escalation and consequences of abuse in a relationship.

Males and females in the current study reported similar levels of abuse in childhood, with the exception of sexual abuse. The impact of abusive experiences on feelings of fear and powerlessness in a relationship was, in most cases, similar for males and females. However, a significant interaction between gender and sexual abuse emerged in predicting fear at the time of an abusive incident. The impact was higher among males who had experienced sexual abuse than among female survivors of sexual abuse. This means that males who had experienced sexual violence reported more fear than females who experienced similar levels of abuse.

These findings suggest that prior victimization, particularly as it relates to sexual abuse, is likely to be an important component for exploration in treatment. Survivors of sexual abuse frequently experience after-effects of the abuse, including fear of intimacy, low self-esteem, sexual promiscuity, depression, suicidal ideation, and eating disorders (Freeman & Morris, 2001). Many of these factors have been found to be interrelated with domestic violence perpetration (Julian et al., 1999).

Due to these emotional impacts of prior exposure to violence, some have indicated that violence would be focused primarily in the home. Research indicates that about 20% of batterers are violent outside the home (Saunders, 1992; Walker, 1995). The current study found similar results, with 20% of males and 14% of females reporting getting into physical fights sometimes to very frequently. While only 12% of females in the current study reported fighting with strangers and 18% with acquaintances, this number represents a population whose needs are likely to be significantly different in treatment from those who are not using violence outside the family. For example, a small portion of the sample indicated gang affiliation. This frequent exposure to, and possible participation in, community violence among the gang-affiliated group is likely to impact perceptions about the acceptability of, and tolerance for, violence.

These findings support, to some extent, the theory that violence may be passed from one generation to the next through a learning or socialization process. Further exploration of this process among females is needed to fully understand how violence may be learned. At least two paths are possible in the acquisition of violent behaviors. It is possible that females, like males, are directly modeling the behavior of an abusive parent, who may have been either male or female (Mihalic & Elliot, 1997; Wagar & Rodway, 1995). Alternatively, exposure to violence may make these women more likely to choose to retaliate when violence is used against them, because of hyper-vigilance, awareness of the consequences of abuse, or determination not to be a victim of violence again.

The need to address victimization in batterer intervention treatment is highlighted by the findings of this study. Whether the perpetrator is acting in an attempt to control a partner, in self-defense, or in retaliation for abuse in the current relationship, important issues arise regarding victimization. Specifically, lasting effects of sexual abuse and assault have a strong impact on feelings and beliefs about domestic violence. Addressing these issues may increase the effectiveness of intervention in preventing future violence. Some integration of process psycho-dynamic approaches may be helpful in creating a supportive environment to explore these issues. These less structured groups may allow for more open, in-depth exploration into the underlying issues in the choice to use violence in a relationship (Gondolf, 1987; Sakai, 1991). Research of differences in recidivism among those who completed group intervention programs has shown no significant differences between the two primary models of treatment, although a greater percentage in the psycho-dynamic group completed the program (Browne et al., 1997). However, this study did not examine differences in abuse history as a factor related to program completion or recidivism.

Because experiences as a victim of violence, both as a child and in current relationships, differ in degree and severity, it may be necessary for counseling programs to develop assessment tools to determine the needs of each individual entering the program. It is possible that concurrent individual counseling to address the adult manifestations of childhood abuse may strengthen the impact of batterer intervention treatment. However, further research is necessary to evaluate the impact of this type of intervention.