Introduction

In spite of a plethora of research on interpersonal violence among heterosexual intimate partners, very little is known about intimate partner violence in gay and lesbian relationships. There is not much data on the frequency and incidence of same-sex intimate partner abuse. However, the few epidemiological studies available suggest that the rates of domestic violence in gay and lesbian relationships are at least as high, if not higher, than those among heterosexual couples (Greenwood, Relf, Huang, Pollack, Canchola, et al., 2002; Renzetti, 1989). These self-reported estimates are undoubtedly low, given the prejudicial and discriminatory climate many gay men and lesbians experience. Also, data suggest that the problem does not appear to be diminishing; recently released national data indicate that gay and lesbian partner violence has increased 25% since 1991 and increased 23% just between the years 1998 and 1999 (Greenwood et al., 2002).

Disempowerment Theory

Some theoretical work has been done attempting to integrate the social and psychological aspects of domestic violence which appears applicable to gay and lesbian intimate partner violence (Dutton, 1998; O’Neil & Harway, 1999). These efforts indicate that physical and emotional abuse has been effectively conceptualized in a power/control paradigm, but not in a strictly feminist model (Gelles, 1999). According to disempowerment theory, those who feel inadequate or lacking self-efficiency are at risk of using unconventional means of power assertion, including violence (Archer, 1994). These individuals overcompensate by controlling persons they perceive threatening or who expose their insecurities (Gondolf, Fisher, Fisher, & McPherson, 1988). In their review of literature, Malik and Lindahl (1998) found only a few notable exceptions of research that address the role power has in relationships even though it is generally considered to be part of the conceptualization and treatment of domestic violence. A related perspective, resource theory (e.g., Claes & Rosenthal, 1990; Goode, 1971), suggests that when people have few resources or little capacity to exert control in more normative or traditional ways, they may resort to physical violence as a means of influence. The application of disempowerment theory to explaining gay and lesbian intimate partner violence is discussed in terms of three clusters of factors: (a) individual, (b) family of origin, and (c) intimate relationship characteristics.

According to disempowerment theory, individual characteristics place persons at risk for perpetration of intimate partner violence based on personality-oriented factors such as self-esteem or degree of attachment. Family of origin factors occur in childhood, essentially serving as a model for conflict resolution for the children as adults and impacting coping mechanisms. Intimate relationship characteristics refer to attributes of romantic relationships that can place an individual at risk for using physical violence against an intimate partner.

Individual characteristics

Gender role orientation, specifically a high level of masculinity, has been found to be a salient predictor of domestic violence. McConaghy and Zamir (1995) found that the more a gay man or lesbian identified with masculine personality components, the more likely he or she was to become abusive. Burke and Folingstad (1999) found that lesbian, gay, and heterosexual male batterers were very similar in personality traits related to aggressive behaviors. Thus, it could be hypothesized that having a masculine gender role orientation is positively associated with violence.

Attachment, specifically insecure attachment, may also be an important variable in predicting gay/lesbian intimate partner violence (Renzetti, 1992). Although power may be more equally distributed in gay/lesbian relationships, one partner may be more dependent and more vulnerable to violence when perceiving a lack of power and control. In fact, Lockhart, White, Causby, and Isaac (1994) found high levels of emotional dependency or over-attachment among lesbian batterers. Dutton (1998) also found that gay and heterosexual male batterers equally suffered from attachment disorders as well as greater psychological symptoms.

Psychological symptoms, such as depression, anxiety, low self-esteem, inability to trust, guilt, and helplessness may place one in a disempowered position in relation to an intimate partner (Ferraro & Johnson, 1983; Tech & Lindquist, 1984; Walker, 1984). Recent inquiries among heterosexual males have found that psychological symptoms may exacerbate a man’s power and control issues and hence impact the tendency toward abuse (Dutton, 1998; Hamberger & Hastings, 1991; Hauser, 1985). Conversely, mental health has been related to nonviolent relationship experiences. High self-esteem has been positively associated with feelings of competence, whereas feelings of incompetence may involve the need to control others (Dutton, 1998; Renzetti, 1992). Self-esteem, specifically, has been found to be related to domestic violence among heterosexual men and women (Johnson & Ferraro, 2000; National Research Council, 1998). From a disempowerment perspective, low self-esteem represents feelings of powerlessness and worthlessness (Arias, Samios, & O’Leary, 1987; Hotaling & Sugarman, 1986). Low self-esteem is also related to greater risk of alcohol and drug usage, which have been found to be highly related to domestic violence (Gelles, 2000).

Perpetrators of violence often abuse alcohol to accentuate feelings of power and self-importance when disempowered; ironically, alcohol has the effect of diminishing a sense of security (Gibbs, 1986). Alcohol abuse is a salient factor in same-sex partner violence as gay men and lesbians have been found to have higher rates of alcohol abuse than heterosexuals (Bradford, Ryan, & Rothblum, 1994; Kus, 1990). Several authors (Diamond & Wilsnack, 1978; Schilit, Lie, & Montagne, 1990) have contended that alcohol use is a common means of dealing with stress exacerbated by homophobia.

Frequently overlooked and understudied, internalized homophobia may be one of the major individual factors distinguishing between gay/lesbian and heterosexual partner violence (Coleman, 1994). Internalized homophobia has been linked with lower self-esteem, feelings of powerlessness, and self-destructive behaviors such as substance abuse. At present, there is no research that empirically examines the relationship between internalized homophobia and partner abuse, either as a victim or perpetrator (Renzetti, 1997). However, clinicians and other practitioners suggest examining such identity issues because perpetrators often experience negative feelings about being gay (Byrne, 1996). In fact, Schilit et al. (1990) found that sociocultural alienation in a primarily heterosexual society could exacerbate stress, thereby leading to substance abuse and subsequent lack of affective and behavioral control.

Family of origin characteristics

For perpetrators, witnessing violence in their families of origin or experiencing abuse themselves as children may provide an aggressive coping mechanism for resolving interpersonal conflict. In an attempt to maintain control over later intimate adult relationships, individuals may theoretically resort to violence because it is their most readily understood and accessible coping strategy in dealing with loss of power (Coleman, 1990; Renzetti, 1992). Another explanation for this relationship between family of origin violence and later adult violence in intimate relationships is the impact of such an environment on psychological symptoms, such as hostility, depression, and anxiety, which increase the risk of abusive behaviors in problem solving (Julian, McKenry, Gavazzi, & Law, 1999).

Parental homophobia also has been suggested to be a predictor of domestic violence in same-sex relationships. Ongoing poor relationships with one’s family of origin as a result of heterosexism and homophobia may accentuate internalized homophobia and related psychological symptoms such as depression (Island & Letellier, 1991). This vulnerability in turn may leave the adult child in a disempowered position within an intimate relationship.

Similarly, family social support, or the role that family members currently play in the lives of same-sex couples experiencing domestic violence has not been thoroughly investigated; however, the literature on heterosexual couples shows that isolation from one’s family is an indicator of the control/dominance characteristic of domestic violence, whether victim or perpetrator. For example, perpetrators may seek to isolate victims from family support as a mechanism for exerting further control. From a disempowerment perspective, this isolation reduces the resources available to victims and perpetrators. Family members may not be a source of support for many gay men and lesbians, particularly if they are estranged from them or are not “out” to them.

Intimate relationship characteristics

Many same and opposite sex couples experience relationship distress or dissatisfaction at some point in their relationship. Relationship stress or dissatisfaction may lead to domestic violence if one has a strong need for control or power (Gelles, 1999; NRC, 1998). How relationship stress manifests itself in terms of violence may be closely linked to the degree of emotional dependency between partners. Frequently studied among lesbian scholars, emotional dependency refers to the level of enmeshment with one’s intimate partner. Violent behaviors may be utilized to correct imbalances in emotional dependency. In fact, Bartle and Rosen (1994) contended, “violence in intimate couple relationships is, in part, a distance-regulating mechanism that maintains a balance between separateness and connectedness in the relationship” (p. 222). Lockhart et al. (1994) and Miller, Greene, Causby, White, and Lockhart (2001) found that fusion, or a high level of enmeshment with one’s intimate partner, was a predictor of psychological abuse in their sample of 286 lesbian participants.

Several researchers have found domestic violence to be associated with low income status or low earning potential of men in heterosexual relationships; thus verbal, physical, and sexual violence is seen as a resource to gain power when men fail to attain and/or maintain the culturally expected dominant position in the family (Goode, 1971; Hornung, McCullough, & Sugimoto, 1981; Straus, Gelles, & Stenimetz, 1980). In their extensive review of the heterosexual domestic violence literature, Hotaling and Sugarman (1986) found that husbands’ occupational status, income, and educational level were consistently correlated with husband-initiated partner violence. In a related finding, disparity in economic positions could also explain conflict and abuse in gay and lesbian relationships, especially when it leads to feelings of inferiority (Renzetti, 1988; Rutter & Schwartz, 1995).

Other partner status differentials have been found to be highly associated with violence in heterosexual relationships (Pagelow, 1984; Stark & Flitcraft, 1988), but they have been largely untested among gays and lesbians (Waldner-Haugrud, Gratch, & Magruder, 1997). Numerous variables have been linked with status inconsistency in gay and lesbian relationships that may contribute to violence. These include physical size, attractiveness, and job status (Lockhart et al., 1994; Meyers, 1989; Renzetti, 1988; Rutter & Schwartz, 1995).

Hypotheses

Based on disempowerment theory, this study hypothesizes that the perpetration of domestic violence is a function of disempowerment, i.e., real or perceived challenges to possessions, authority, or control of the intimate partner. More specifically, variables from three conceptual domains related to disempowerment will be assessed to determine their influence on physically violent behavior among gay and lesbian intimate partners: (a) individual characteristics (gender role orientation, attachment, psychological symptoms, self-esteem, alcohol abuse, and internalized homophobia); (b) family of origin characteristics (violence in family of origin, child abuse, parental homophobia, socioeconomic status; and family and friend support); and (c) intimate relationship characteristics (relationships satisfaction, relationship distress, emotional dependency, lower income status, status inconsistency factors). It is also proposed that there would be variations by sexual orientation.

Table 1 Comparison of demographic data by gender group (N=77)
Full size table

Methods

Participants

The sample for this study (n=77) consisted of 40 gay men and 37 lesbians residing in a large Midwestern metropolitan city. Enrollment was facilitated by referrals from therapists, mental health centers, domestic violence treatment and advocacy groups, and gay/lesbian community organizations as well as responses to advertisements in gay and lesbian newspapers and agency newsletters. Advertisements in the gay and lesbian publications yielded the most participants (80%). In all cases, participants responded to flyers or ads that informed them that they were being sought to participate in a study of conflict between gay and lesbian partners. Eligible participants were offered compensation of $25 each.

Categorization of physical violence was determined by at least one incidence of violence as measured by the physical violence subscale of the Revised Conflict Tactics Scale II (CTS2; Straus, Hamby, Boney-McCoy, & Sugarman, 1996) in the past year. Participants were categorized as: 1) violence perpetrators, 2) dominant partner in distressed, nonviolent relationship, or 3) less-dominant partners in distressed, nonviolent relationships. Of the 36 gay men, 14 (39%) were judged to be physically violent in the past year, and of 34 lesbians, 12 (35%) were determined to be physically violent in the past year. Interestingly, in all but three cases, there was one perpetrator who instigated the majority of the violence; in most cases, this person perpetrated all the physical violence. Perpetrators were contrasted to the more dominant partner in the distressed nonviolent group. The individual within each nonviolent couple who scored higher on the dominance subscale of the Psychological Maltreatment of Women was, for this study, considered the dominant partner. The comparison group consisted of less dominant partners in distressed but nonviolent relationships.

Procedures

Data were collected in private offices by clinically trained graduate student researchers. Interviews began with the administration of a background demographic questionnaire, followed by a series of quantitative instruments. Because gender differences can impede understanding and rapport (Williams & Heikes, 1993), male interviewers conducted the interviews with male participants, and female interviewers interviewed female participants. All interviewers had advanced graduate and/or clinical training and were thoroughly trained by the principal investigators. Qualitative data that were collected for this study were not analyzed for this study.

Demographics by gender

Demographic characteristics of the gay and lesbian couples are presented in Table 1. There were no differences between persons in nonviolent and violent relationships in terms of the major demographic variables examined including age, length of relationship, presence of children, and race. There were, however, significant demographic differences by gender in terms of length of relationship, occupational status, and income. That is, males had longer current relationships and higher mean incomes; females had slightly higher status occupations based on the Hollingshead Index (Hollingshead & Redlich, 1958).

There was a significant interaction regarding education and perpetration (F=4.39, p < .05). According to scores on the Hollingshead, male perpetrators (M=3.63) were less educated than the male nonperpetrators (M=2.33); however, female nonperpetrators (M=3.50) did not differ from the female perpetrators (M=3.12). The lower education of male perpetrators would mirror the literature that has found that men’s lower socioeconomic status is highly related to physical violence in heterosexual relationships (Gelles, 2000; Moore, 1997). Finally, male and female perpetrators did not differ in extent of violence or in terms of injuries inflicted. The mean score for all violent persons was 2.5, which translated to an average of 2–5 incidents per year.

Instruments

A demographic questionnaire was used to determine age; ethnicity; length of relationship with partner; presence and age of children; and present relationship status. In addition, the following instruments were chosen for their psychometric properties or their relevance to power/control theory.

Individual factors

Gender role orientation. The Personal Attribute Questionnaire (PAQ; Spence, Helmreich, & Holahan, 1979) will be used to measure gender role orientation. The PAQ assesses dimensions of masculinity and femininity with 24-items presented in a five-point bipolar format. The PAQ includes three subscales: masculine-valued, feminine-valued and sex-specific. In this study, the Cronbach’s α for the PAQ was masculine (.79), feminine (.83), and sex-specific (androgynous) (.66).

Insecure attachment. Griffin and Bartholomew’s (1994) Relationship Style Questionnaire was used to assess attachment style. Factor analysis has indicated that this 30-item rating scale has four factor dimensions, representing four distinct attachment styles. The potential range of scores on this scale ranges from “not at all like me” (1) to “very much like me” (5). Sample items include: “I find it difficult to depend on other people,” “I find it relatively easy to get close to others.” Cronbach αs range from .78 to .89. Scores on the three subscales, indicating insecure attachment, were summed. The Cronbach α for this study was .89.

Psychological symptoms. The Psychiatric Symptoms Checklist 90/Brief Symptom Inventory (BSI) was used to assess psychological adjustment (Derogatis, 1992). The BSI is a multidimensional inventory designed to assess psychopathology in psychiatric and medical outpatients. Respondents were asked to rate the extent to which they have been bothered by each symptom in the last few months. The checklist was scored on nine primary symptom dimensions; only the global score were used in this study. The Cronbach α for the global severity index in this study was .95.

Self-esteem. Self-esteem was measured by the Self-esteem Scale (Rosenberg, 1965). The scale measure global feelings of self-worth and consists of 10 items with a response set ranging from “strongly agree” (5) to “strongly disagree” (1). Cronbach’s α for this study was .92.

Alcohol use. The Short Michigan Alcoholism Screen Test (SMAST) was used to identify alcohol use. The SMAST has demonstrated strong evidence of validity and reliability (Stelzer, 1971). The measure consists of 12 questions pertaining to alcohol use. It is scored by summing the responses from the differentially weighted items. Cronbach’s α for this study was .84.

Internalized homophobia. The Internalized Homophobia Scale (Ross & Simon Rosser, 1996) was used to measure internalized homophobia. The scale consists of 26 statements regarding negative attitudes and beliefs about homosexuality. The 5-point response set ranges from “strongly agree” (5) to “strongly disagree” (1). Factor analysis indicated four distinct factors: public identification as gay; perception of stigma associated with being gay; social comfort with gay men; and moral and religious acceptability of being gay. These subscales exhibited significant concurrent validity when correlated with criterion measures. Cronbach’s α for this study was .80.

Family of origin factors

Family of origin violence. The physical and emotional abuse subscales from the Revised Conflict Tactics Scale (CTS2; Straus et al., 1996) was modified to assess the extent of parental violence (emotional and physical) that the participant witnessed during his/her childhood, even if/particularly when the violence was not directed toward the participant him or herself. The response set was modified to a five-point set of responses: never, rarely, sometimes, fairly often, and very often. Cronbach’s α was .87 in this study.

Family of origin child abuse. The Parent–Child Conflict Tactics Scales (CTSPC; Straus, Hamby, Finkelhor, Moore, & Runyan, 1998) was used to measure the extent to which a parent has carried out specific acts of physical and psychological aggression, regardless of whether the child was injured. Only the physical and psychological abuse subscales were used. These subscales consist of 20 items that ask how often, as a child, various abusive techniques were used. The response set was modified to a five-point response set: never, rarely, sometimes, fairly often, and very often. The scale was administered twice in order to assess the participants’ recall of each parent or guardian’s behavior. The Cronbach α was .85 and .88, respectively, for the mother and father score in this study.

Parental homophobia. The Homophobia Scale (Bouton, Gallaher, Garlinghouse, Leal, Rosenstein et al., 1987) was used to measure participant’s perception of parental homophobia. The scale consists of seven statements (three positive and three negative) regarding attitudes toward homosexuals and homosexuality. The scale was administered twice—once in terms of the biological or adoptive father and once for the biological or adoptive mother. The Cronbach α for this study was .91 for both mother and father scores.

Socioeconomic status (SES). The Hollingshead and Redlich (1958) Two-Factor Index of Social Position was used to determine the SES of the perpetrator’s family of origin. This index is based on a 7-point scale with lower scores indicating higher levels of education and occupation of the head of the household in the family of origin.

Family of origin and friend support. Friend support and family support was measured globally using an adapted version of the Perceived Social Support-Friends (PSS-Fr) and Family (PSS-Fa) scales (Procidano & Heller, 1983). There are 20-items, 5-point Likert-type scales, designed to measure perceived social support from friends and family. Cronbach’s α for the PSS-Fr and PSS-Fa, was.79 and .89, respectively.

Intimate relationship factors

Relationship satisfaction. The Kansas Marital Satisfaction Scale was modified to assess the quality of the relationship with the same-sex intimate partners (Schumm, Paff-Bergen, Hatch, Obiorah, Copeland et al., 1986). The scale consists of three items describing components of the relationships; with a seven category response set ranging from “extremely satisfied” to “extremely dissatisfied.” Cronbach’s α for this study was .94.

Stress. The Family Inventory of Life Events and Changes (FILE; McCubbin, Patterson, & Wilson, 1982) was used to assess the pile-up of life events experienced by a family and serves as a measure of family stress. The measure was administered in terms of two time periods—(a) regarding the past 12 months and (b) regarding the entire length of the relationship prior to 12 months ago. Cronbach α for this instrument were .87 for the past 12 months and .83 for the time prior to that.

Emotional dependency. The emotional reliance subscale of the Interpersonal Dependency Inventory (Hirschfeld, Klerman, Gough, Barrett, Korchin et al., 1977) was used to measure emotional dependency. The Interpersonal Dependency Inventory consists of 48 items, descriptive of interpersonal dependency in adults. The Cronbach α for the emotional reliance subscale, used in this study, was .87.

Status differential. The difference in status between the participants and their partners was measured by asking participants to provide ratings of self in relation to their partner regarding: educational level, income, perceived attractiveness, and physical size. The 7-point response set ranges from: much greater than me to much less than me. A total summary score was used in the analysis.

Relationship dominance. The dominance-isolation subscale of the Psychological Maltreatment of Women Index (PMWI; Tolman, 1989) was used to assess relationship dominance. The PMWI consists of 58 Likert-type items describing psychologically abusive behaviors. This measure was adapted for GLBT individuals. Participants were asked (a) if the behavior ever happened in the past 6 months and (b) if it did, how often. Cronbach’s α was .90.

Physical and verbal abuse toward partner. The Revised Conflict Tactics Scale (CTS2; Straus et al., 1996) was used to indicate the extent of physical violence toward their partners. The instrument consists of 78 questions, asking the respondent to indicate the number of times that a particular relationship behavior occurred in the past year. Subscale reliabilities ranged from .79 to .95. The Cronbach alpha for the physical violence subscale in this study was .92.

Data analysis

The data for this study were analyzed using SPSS 12.0. Because of the small sample size, only descriptive statistics and two-way ANOVAs (gender by perpetrator of violence) were performed. When there was mutual violence, the responses of the partner with the higher perpetration score (the sum of his score versus the sum of his partner’s score on the CTS) were used. In addition, because of the increased chance of making a Type II error in using a small sample, the α level was set at ≤.10.

Results

Individual characteristics

The greatest number of differences between perpetrators and nonperpetrators was found in the category of individual characteristics.

Gender orientation. There was a significant main effect for perpetrators in terms of masculinity; perpetrators (M=30.19) had significantly higher masculinity scores than did nonperpetrators (M=26.75) (F=8.9, p < .05).

Insecure attachment. There was a perpetrator main effect for security of the attachment relationship (F=2.79, p < .10); perpetrators (M=3.05) tended to have less secure attachment styles than nonperpetrators (M=3.57).

Psychological symptomatology. There was a gender by perpetrator interaction (F=3.83, p < .05) for the global severity index subscale. Nonperpetrating females had higher scores (M=1.95) than perpetrating females (M=1.63); perpetrating males (M=1.26); and nonperpetrating males (M=.37). Although not at a level of statistical significance, in general, females (M=1.85) had higher psychological symptoms scores than did males (M=1.32). The means of all subgroups except nonperpetrating males were within clinical norms.

Self-esteem. There was a perpetrator by gender interaction for self-esteem (F=8.31, p < .01). Male nonperpetrators (M=23.01) had significantly higher levels of self-esteem than male perpetrators (M=14.54); female perpetrators (M=16.87); and female nonperpetrators (M=13.0).

Alcohol usage. There was a perpetrator by gender interaction for the use of alcohol (F=5.84, p < .05). Female nonperpetrators (M=10.67) drank more than the other three groups (female perpetrators [M=3.43]; male perpetrators [M=5.25]; and male nonperpetrators [M=3.1]).

Internalized homophobia. Data in this study did not indicate that perpetrators evidenced more homophobia than their nonviolent counterparts.

Family of origin characteristics

Family of origin domestic violence. There were no significant effects for family of origin domestic violence. However, women (both perpetrators [M=4.17] and nonperpetrators [M=3.77]) tended to experience more parental domestic violence than did either perpetrating (M=3.59) or nonperpetrating men (M=3.41).

Child abuse. A gender main effect was found for experiencing child abuse. Females (M=4.00) were significantly more likely to report having been abused by their parents than males (M=3.12; F=11.72, p < .001).

Socioeconomic status (SES). There was a significant interaction (F=5.83, p < .02) regarding SES in family of origin. Male perpetrators (M=4.22) grew up in significantly lower SES families than did male nonperpetrators (M=2.83).

Other factors in this domain were not significantly related to physical violence include homophobia of the mother or father and support from parents and family members, including friends.

Intimate relationship characteristics

Relationship satisfaction. There were no significant differences in partner relationship satisfaction as a function of gender or perpetrator status. However, perpetrators (M=14.06) had lower satisfaction scores than the nonperpetrators (M=16.33), but not at the level of statistical significance.

Relationship stress. There was a significant perpetrator main effect in terms of stress experienced prior to the last 12 months with perpetrators (M=8.81) experiencing more stress than nonperpetrators (M=2.50; F=4.56; p < .05). There was also a significant gender main effect in terms of stress experienced in the last 12 months (F=4.11; p < .05). Females (M=22.71) experienced more stress-related changes than did males (M=15.44).

Relationship power differential.   No significant differences in physical violence as a function of perceived power differential, outing, or interpersonal dependency were found.

Discussion and Conclusions

This research examined the characteristics of gay/lesbian violence in three ecological domains: individual, family of origin, and intimate relationship factors. We hypothesized that the perpetration of domestic violence among both gay men and lesbians is a function of disempowerment, i.e., real or perceived challenges to possessions, authority, or control of the intimate partner. The findings do provide some support for disempowerment theory and suggest directions for future research on gay and lesbian intimate partner violence. While the nature of this study’s sample limits firm conclusions and generalizations, the study’s ability to overcome some of the weaknesses of prior studies suggest that the findings should be seriously considered.

Regarding gender orientation, the findings support a disempowerment perspective in that higher masculinity is related to a greater tendency toward aggressive behaviors of control when threatened (McConaghy & Zamir, 1995). Those with higher masculinity scores might be more inclined to use aggression to resolve relationship problems (Gelles, 1999). It should be noted that some researchers have found masculine characteristics to be positively related to psychological functioning whereas others related to negative functioning. For example, Spence et al. (1979) have extended the PAQ to distinguish between positive and negative masculinity.

Both male and female perpetrators also differed from the nonperpetrators in terms of insecure attachment. This is consistent with disempowerment theory; i.e., a less securely attached person would logically feel that their possession and control of an intimate would be more tenuous. Male and female perpetrators also evidenced more long-term relationship stress. This finding would be consistent with Walker’s (1999) cycle of violence theory in which physical violence develops over time, progressing from less serious to more serious. Long-term dissatisfaction would also be a threat to self-esteem, as the couple is not able to relieve stress and have their needs met in the relationship. Further, the longer the stress occurs in a relationship, the more likely partners might be to resort to physical violence instead of trying to discuss what might appear to be a hopeless situation.

For males but not for females, lower self-esteem, educational level, and SES background also contributed to propensity of perpetrator violence. Although a prediction model was beyond the scope of this study, future research might reveal that the disempowerment model fits gay men better than lesbians who are in distressed or violent relationships. In this case, traditional indicators of masculinity, such as material success and status, may be salient in male homosexual relationships just as they are for men in heterosexual relationships. These factors reflect the utility of the disempowerment perspective in terms of the pressure on males to succeed in the traditional role of a good provider. The men scoring lower on these measures may not have been able to demonstrate their masculinity in traditional ways and may have employed physical violence as a substitute (Archer, 1994; Barnett, Miller-Perrin, & Perrin, 1997). Perhaps these factors were not as predictive of women’s violence because the good provider or economic role in general is not as salient to their identity.

In terms of educational level, the literature suggests that lower educated males tend to have fewer effective coping mechanisms and problem solving skills, which would also place them at risk, infusing more stress in their lives because of fewer resources as well as more traditional gender role attitudes (Gelles, 2000; Moore, 1997). This finding would also be mirrored in the significant difference in SES; in addition, those from lower SES backgrounds would have more traditional views of relationships where intimate partner violence would more likely be sanctioned.

It is surprising that nonperpetrating females in distressed but nonviolent relationships had higher scores on some of the measures of pathology, i.e., psychological symptoms, alcohol use). The female nonperpetrators had very high scores on psychological symptoms and used more alcohol. It would seem that many of the participants, male and female, self-medicated with alcohol as a result of the stress and psychological symptoms in their lives. Gibbs (1986) found that violent males used alcohol to accentuate their sense of power. Renzetti (1988) found the use of alcohol was a means for lesbians to handle dependency, yet it just increased the risk for violence. It should be recalled that these nonperpetrating, but dominant, female partners were chosen because of their relationship distress; thus they may have been more likely to use alcohol than those in nondistressed relationships.

Implications for Practice and Future Research

Although tentative, these findings have important implications for professionals working therapeutically with gay and lesbian individuals or couples. Therapists, physicians, and other medical personnel or community-based interventionists should be educated about the extent and increasing incidence of violence in gay and lesbian relationships. Such knowledge would increase their sensitivity to this problem in their client population. This information may be obtained from a variety of GLBT organizations or domestic violence agencies.

Because of the stigma associated both with being gay and being involved in a violent relationship (Byrne, 1996), gay and lesbian clients may be reticent to mention the issue unless the professional initiates discussion. Also, clients may not recognize aggressive interactions in their relationships as domestic violence. Thus, clients may underreport violence and not receive necessary treatment. For example, upon eligibility screening for this study, when potential participants were asked if there was violence in their relationship, many (both men and women) would say, “I wouldn’t really say violence, but there is some pushing and shoving.” In fact, gay men often have difficulty seeing themselves as victims of domestic violence and most likely conceptualize violence differently than lesbians (Letellier, 1994). Thus therapists should specifically screen and probe for the presence and extent of domestic violence.

The findings from this study and others would suggest some important factors to consider in our attempt to understand gay and lesbian violence. For example, it might be beneficial for clinicians working with lesbians to explore the extent of alcohol usage in the relationship. Based on the current findings, alcohol use among lesbians was high for both perpetrators and nonperpetrators. It is highly likely then, that alcohol use will play a significant role in the occurrences of domestic violence. As previously suggested, the perpetrator may use alcohol to heighten their sense of power and therefore equally possible that the nonperpetrator uses alcohol as a moderating effect of the abuse. Still, we are not suggesting that alcohol use causes violence or that alcohol is involved in all situations where domestic violence occurs.

Another salient area for clinicians to explore is attachment. Although power may be more equally distributed in gay/lesbian relationships, one partner can be more dependent and more vulnerable to violence when perceiving a lack of power and control. High levels of attachment or dependency may create an environment that supports violence. Findings from this research indicate that both male and female perpetrators tended to have less secure attachment styles than do nonperpetrators.

Professionals should also be aware of mental health issues for both gay and lesbians. Low self-esteem, feelings of powerlessness and worthlessness may make individuals more susceptible to violence. Another area for clinicians to consider is internalized homophobia. Internalized homophobia has been linked with lower self-esteem, feelings of powerlessness, and self-destructive behaviors such as substance abuse (Coleman, 1994). For many, the lack of self-acceptance, and feelings of alienation and self-loathing only served to exacerbate stress in their relationships.

Therapists should be aware of community resources available to gays and lesbians experiencing violence. Locating resources that cater to gays and lesbians or that are “gay friendly” is essential in encouraging individuals to follow through with treatment. Those experiencing violence may need both GLBT-specific (i.e., support group, advocacy) and general (i.e. financial, housing) services to assist them. For example, lesbians experiencing domestic violence and needing temporary shelter may be able to seek shelter at a heterosexually-oriented women’s shelter based solely on their gender. However, a gay man experiencing violence may not have this same alternative.

These data, though preliminary, provide some evidence for the utility of a disempowerment perspective in the understanding of gay and lesbian intimate relationship violence. Prior to application, future studies of the role of disempowerment should employ larger samples that include a typology or wide range of violent behaviors. Ideally, these studies would offer a means of comparing not only violent with nonviolent partners, by including a control group of individuals who have not experienced intimate relationship violence, but also acknowledge that various levels of violence exist within couples. For this study, individuals were classified as distressed violent or nonviolent. It became clear that in some couples the prevalence, frequency, and nature of the violence was much higher than that of others. Therefore, placing couples into categories based on the chronicity and severity of their violence as well as the primary motivation for this behavior would be ideal. For example, Johnson and Ferraro (2000) have theoretically conceptualized violence into two groups for heterosexuals—those who have experienced “common couple violence” or “intimate terrorism.” Common couple violence is generally characterized by a low frequency of violent episodes, usually does not involve severe violence, and tends to not escalate over time. The specific context of this type of violence is one argument, with emphasis placed on how each partner behaved during that argument. In contrast, Intimate Terrorism (IT) is connected with a general need to control a partner with violence as merely one method of control. In IT there tends to be serious injury involved, as well as a higher per-couple frequency. The violence tends to be less mutually perpetrated and tends to escalate over time. The distinguishing feature of intimate terrorism is that general pattern of pervasive control, exhibited through nonviolent and violent behaviors.

The inclusion of some additional variables should be considered in further testing of a disempowerment perspective. For example, jealousy can be a source of conflict that sometimes results in violence. Jealousy, whether perceived or real, may be a result of power differences or negative views of oneself compared to a partner.

Another public health issue that emerges from these data is the link between intimate partner violence and the presence of HIV or HIV-risk behaviors. This may warrant further investigation of the relationship between those at-risk for domestic violence and those at-risk for HIV. The link between domestic violence and the risk of HIV can be theorized three ways: (a) HIV can leave those infected vulnerable to domestic violence; (b) domestic violence can leave those people vulnerable to HIV; and (c) there can be an unrelated co-occurrence of HIV and domestic violence in an intimate relationship. Each is worthy of careful investigation.