Abstract
Prior research has indicated that childhood sexual abuse (CSA), alcohol use, and posttraumatic stress disorder (PTSD) symptoms are important risk factors for adult sexual assault (ASA). A notable limitation of this prior work, however, is that it has almost exclusively focused on heterosexual women. The present study sought to remedy this by examining the extent to which CSA, alcohol use, and PTSD symptoms related to ASA among lesbians (n = 122) and gay men (n = 117). Using structural equation modeling, we found that alcohol use was the best predictor of ASA among lesbians whereas CSA was the best predictor of ASA among gay men. These results suggest that certain risk factors may be differentially related to ASA among groups with different sexual orientations. Such findings deepen our current understanding of ASA and offer important directions for reducing the risk of ASA for lesbian and gay individuals.
Keywords
Recent research indicates that nearly one in five women experience adult sexual assault (ASA; Black et al., 2011). Prior studies also suggest that sexual assault may lead to numerous long-term negative outcomes including posttraumatic stress disorder (PTSD), depression, lower self-esteem, social difficulties, sexual dysfunction, and suicidal ideation (e.g., Campbell, Dworkin, & Cabral, 2009; Resick, 1993). As a result of the relatively high prevalence and associated comorbidities, researchers have focused on understanding the factors which may confer risk for ASA.
One well-documented risk factor for ASA is childhood sexual abuse (CSA; Filipas & Ullman, 2006; Messman-Moore & Long, 2000); this phenomenon is often referred to as sexual revictimization. Researchers have theorized that this robust association may be due to the emotional and behavioral sequelae of CSA, rather than the abuse itself (Arata, 2002; Ullman, Najdowski, & Filipas, 2009). More specifically, CSA may lead to excessive substance use, PTSD, and other forms of psychopathology, which in turn increase the vulnerability to ASA by (a) impairing active defenses and risk perception and (b) increasing the likelihood of being targeted by others (Marx, Heidt, & Gold, 2005; Risser, Hetzel-Riggin, Thomsen, & McCanne, 2006). Prior findings on the associations among alcohol use, PTSD symptoms, and ASA are consistent with this hypothesis (e.g., Messman-Moore, Brown, & Koelsch, 2005). In a longitudinal study of sexual assault among women, Ullman, Najdowski, and Filipas (2009) found that PTSD symptoms and alcohol use mediated the relationship between CSA and ASA, suggesting that CSA-related sequelae indirectly lead to subsequent ASA.
Although studies have identified risk factors for ASA among heterosexual women, very few studies have examined the extent to which CSA, alcohol use, and PTSD may increase the risk for ASA among lesbians and gay men (LG). A review of studies on sexual coercion among LG indicated that the greatest need for research in this area is “identifying predictors of sexual coercion and communicating these findings to at-risk gays and lesbians” (Waldner-Haugrud, 1999, pg. 147). Unfortunately, there is still very little empirical research since the review was published over a decade ago. This is particularly troubling since research has indicated that sexual victimization rates are higher among LG than among heterosexual men and women (Balsam, Rothblum, & Beauchaine, 2005; Rothman, Exner, & Baughman, 2011; Tjaden & Thoennes, 2000). Studies have reported higher rates of alcohol use among LG compared with heterosexuals, which may be an important risk factor to examine (Hequembourg, Parks, & Vetter, 2008). Furthermore, LG report lifetime and 12-month prevalences of depression, anxiety disorders, substance use disorders, and suicidal behavior to a greater extent than their heterosexual counterparts (King et al., 2008). Studies to date have not compared rates of PTSD between LG and heterosexuals; however, as one of the disorders most commonly associated with sexual assault, PTSD is an important risk factor to consider.
In one of the few studies of revictimization among LG, Balsam, Lehavot, and Beadnell (2011) indicated that CSA increased the likelihood of ASA across lesbians, gay men, and heterosexual women and that histories of both CSA and ASA were associated with higher levels of psychopathology and alcohol use. The current study seeks to provide additional findings regarding the extent to which CSA, PTSD symptoms, and alcohol use confer risk for ASA among LG. Based on the extant research indicating that CSA, alcohol use, and PTSD symptoms confer risk for ASA among heterosexual women, we hypothesized that these factors also confer risk for ASA among LG.
Method
Participants
Participants were 342 individuals who were enrolled in a larger study of sexual victimization among sexual minorities (Heidt, Marx, & Gold, 2005). As we were interested in examining individuals who identified as “female and lesbian” or “male and gay,” we excluded five individuals who did not match the inclusion criteria (e.g., male who identified as lesbian, transgender identity). In addition, participants who labeled themselves as bisexual or who chose not to identify a sexual preference were excluded on account of prior research suggesting that bisexuals have a more flexible sexual orientation and that they may be at higher risk for anxiety, suicidal ideation, depression, and negative affect than LG (Jorm, Korten, Rodgers, Jacomb, & Christensen, 2002).
The final sample consisted of 122 female lesbians and 117 gay men. The age range of the sample was 16 to 77, with a mean age of 33.56 years (SD = 12.47). Of the participants who reported ethnicity, Caucasians comprised 64.8% of the sample, followed by African Americans (8.5%), Hispanics (7.7%), Asians (2.8%), and Native Americans (1.2%). In addition, 1.6% of the sample identified as mixed, and 5.3% endorsed “Other.”
Measures
Child Sexual Abuse
We used the Life Experiences Questionnaire–Modified (LEQ; Long, 1999) to assess for the severity, frequency, and duration of CSA for all participants. CSA was coded as a single ordinal variable, with higher scores indicating greater severity; the possible range of scores ranged from 0 to 4, with values ranging from no CSA, noncontact, contact, attempted penetration, and penetration. Using a definition of CSA commonly used in the field (e.g., Aosved, Long, & Voller, 2011; Messman-Moore, Brown, & Koelsch, 2005), we considered participants to have experienced CSA if they reported sexual experiences prior to the age of 18 with (a) an individual at least 5 years older or (b) an individual less than 5 years older who had used threat or force. The definition of the term “intercourse” was broadened to include oral, genital, and anal penetration by a body part or object. To decrease participant burden, we omitted questions pertaining to the relationship between respondent and perpetrator from the assessment battery. Research has shown that the LEQ has strong internal consistency (Cronbach’s α = .89) and 2-week test-retest reliability (κ = 0.91 for the nature of the abuse, κ = 0.39 for severity of force, and κ = 1.0 for duration; Messman-Moore & Long, 2000).
Adult Sexual Assault
The Sexual Experiences Scale–Modified (SES; Koss, Gidycz, & Wisniewski, 1987) is a 10-item self-report measure used to assess ASA. In accordance with the LEQ, the current study expanded the SES’s definition of “penetration” to include oral, genital, and anal contact. ASA was coded as a single ordinal variable with higher scores indicating greater severity; the possible range of scores were coded 0 to 4 and ranged from no ASA, contact, coercion, attempted rape, and rape. The SES has moderately strong internal consistency (Cronbach’s α = .74 and .89 for men and women, respectively), and strong 1-week test-retest reliability (κ = 0.93; Koss & Gidycz, 1985).
PTSD Symptoms
The Posttraumatic Stress Diagnostic Scale (PDS; Foa, Cashman, Jaycox, & Perry, 1997) is a 49-item self-report measure used to assess for the presence and overall severity of PTSD symptoms, as defined by the criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV; American Psychiatric Association, 1994). Individuals report on the severity of PTSD symptoms and these responses are summed to produce an overall score, with higher scores suggesting more severe symptoms of PTSD. The scale also yields means for each PTSD symptom cluster (reexperiencing, avoidance, and hyperarousal), which were used as three indicators of PTSD. The PDS has an internal consistency coefficient of .92, and 2-week test-retest reliability of .74 for overall symptom severity and PTSD diagnosis.
Alcohol Use
The Cahalan Drinking Habits Questionnaire (DHQ; Cahalan, Cisin, & Crossley, 1969) is a 13-item self-report measure used to assess how often one consumes beer, wine, and/or liquor (frequency), and how many drinks are consumed in a given episode of drinking (quantity) during the past 30 days. This information is used to calculate a Volume Variability Index, wherein individuals were classified as lifetime abstainers/ infrequent drinkers (consuming less than one drink a month or never), light drinkers (consuming less than 17.5 oz per month and between 1-6 drinks in one sitting), moderate drinkers (consuming between 17.6 and 44.9 oz per month and between 1-6 drinks per sitting), and heavy drinkers (consuming more than 45 oz per month and between 1-6 drinks per sitting).
Procedure
Study staff recruited participants at community lesbian, gay, bisexual, and transgender (LGBT) organizations and at two gay pride festivals in the northeastern United States. The protocol and informed consent form were approved by the local Institutional Review Board and the leaders of each organization or event where data collection occurred. All participants decided to take part in the study after reading the informed consent form and completed the packet of questionnaires in a private seating area. More detailed information on the procedure of this study is published in Heidt, Marx, and Gold (2005).
Data Analysis
We used Structural Equation Modeling (SEM) to analyze the relations between CSA history, PTSD symptoms, alcohol use, and ASA. SEM provides a flexible statistical framework in which the relations between many variables can be examined while controlling for measurement error that affects traditional statistical techniques. All models were analyzed using Mplus 6.1 and maximum likelihood estimation procedures. Identical models were specified for both the lesbian and gay subsamples. A confirmatory factor analysis model was first specified in which the latent construct of PTSD was identified by using the mean of the reexperiencing, avoidance, and arousal symptoms as three indicators. The latent construct of PTSD was allowed to freely covary with CSA history, alcohol use, and ASA history, which were each identified by specifying the loading of the single indicators to be 1.0. The scale for all latent constructs was established by specifying the latent variance to equal 1.0. After evaluating the model fit for the CFA models in both samples, SEM models were examined in which PTSD, CSA history, and alcohol use were specified as predictors of ASA. Measurement and structural models were evaluated for goodness of fit using several common fit indices: the root mean square error of approximation (RMSEA; Steiger & Lind, 1980) and its 90% confidence interval (90% CI), the standardized root mean residual (SRMR; Jöreskog & Sörbom, 1996), and the comparative fit index (CFI; Bentler, 1990).
Results
Prevalence of Sexual Assault/Rates of Revictimization
As shown in Table 1, of the 122 lesbians, 70 (57.4%) participants reported at least one sexual victimization experience. Of the 117 gay men, 72 (61.5%) participants reported at least one sexual victimization experience. Among those who reported a history of CSA (n = 99), 2 (2.0%) reported noncontact, 18 (18.2%) reported contact, 6 (6.1%) reported attempted penetration, and 73 (73.7%) reported completed penetration. Among those who reported a history of ASA (n = 93), 20 (21.5%) reported contact, 6 (6.5%) reported coercion, 17 (18.3%) reported attempted rape, and 50 (53.8%) reported completed rape.
Victimization Experiences of Lesbians and Gay Men by Frequency and Percentages.
Note: Table modified from Heidt, Marx, & Gold (2005). Nonvictims = No victimization experiences; CSA only = Childhood sexual abuse experiences only; ASA only = Adult sexual assault experiences only; Revictimized = Both child sexual abuse and adult sexual assault experiences.
Examination of revictimization histories indicated that almost twice the number of gay men reported revictimization compared with lesbians (28.2% and 13.9% respectively). As shown in Table 2, LG also differed in the strength of associations among the variables of interest. Notably, the association between CSA and ASA was significant among gay men (r = .36, p < .001), but not significant among lesbians (r = .09, n.s.) in our sample. Furthermore, drinking was significantly associated with CSA (r = .21, p < .05) and ASA (r = .23, p < .05) among lesbians but not significantly associated with CSA (r = .08, n.s.) and ASA (r = −.04, n.s.) among gay men. Based on these preliminary differences, we divided LG into separate subsamples for subsequent analyses.
Bivariate Associations Among Variables (n = 239).
Note: Correlations for lesbians (n = 122) are below the diagonal and correlations for gay men (n = 117) are above the diagonal.
p < .05. **p < .001
Latent Variable Analysis
We first used confirmatory factor analysis to evaluate the appropriateness of the proposed measurement models. The model fit of the SEM model in the lesbian subsample was excellent: χ2 (df = 6, n = 122) = 6.0, p = .43, CFI = 1.0, RMSEA = .00, SRMR= .02. The completely standardized results of the SEM model for the lesbian sample can be seen in Figure 1. Current symptoms of PTSD were correlated with CSA in lesbians (r = .44, p < .01); however, PTSD and CSA history did not have a significant effect on ASA. Current alcohol use was the only statistically significant predictor of ASA. Current PTSD symptoms, CSA history, and alcohol use together accounted for 6.3% of the variance in ASA in lesbians. The model fit of the SEM model in the gay men sample was good: χ2 (df = 6, n = 117) = 7.38, p = .29, CFI = 0.99, RMSEA = .044, SRMR= .03). The completely standardized results of this model can be seen in Figure 2. Consistent with the findings in the lesbian sample, current symptoms of PTSD were correlated with CSA history in the sample of gay men (r = .36, p < .01). Among gay men, CSA significantly predicted ASA. There was a nonsignificant trend (p =.075) for PTSD symptoms predicting ASA in gay men. Current PTSD symptoms, CSA history, and alcohol use together accounted for 16.5% of the variance in ASA in gay men. An examination of the 95% confidence intervals of the effects in each sample revealed that the effect of alcohol use on ASA was greater in the lesbian sample (β = .21; 95% CI .04 : .39) than in the gay sample (β = −.10; 95% CI −.27 : .07), whereas the effect of CSA on ASA was greater in the gay sample (β = .31; 95% CI .14 : .48) than in the lesbian sample (β = .01; 95% CI −.19 : .20).

Structural equation model of the relationships between PTSD symptoms, CSA history, alcohol use, and ASA revictimization in the lesbian subsample (n = 122).

Structural equation model of the relationships between PTSD symptoms, CSA history, alcohol use, and ASA revictimization in the gay subsample (n = 117).
Discussion
Among lesbians, CSA and PTSD were significantly related to each other, but neither significantly predicted ASA. It is surprising that CSA did not confer risk for ASA in this sample, but given that this is one of the first studies to assess these variables among lesbians, we can conjecture that lesbians’ sexual orientation may add an additional dimension that increases the complexity of the relationship between CSA and ASA. It is possible that lesbians who were sexually victimized by men in childhood have fewer relationships with men in adulthood (Balsam et al., 2005), thereby decreasing their risk for ASA perpetrated by men. In one study of sexual victimization among lesbians, the majority of sexual assault perpetrators in both childhood and adulthood were men (Morris & Balsam, 2003). However, without assessing the gender and sexual orientation of perpetrators in this study, we can only speculate that the sexual assaults reported in this sample were perpetrated by men.
Balsam et al. (2011) and Morris and Balsam (2003) indicated revictimization effects among lesbians in their samples. However, their measurements of CSA were limited in the type and severity of experiences, and thus may not have captured the complexity of various factors related to the risk of sexual victimization over the life span among lesbians, as was assessed in the present study. Hughes, Johnson, and Wilsnack (2001) indicated that lesbians endorsed a significantly broader range of CSA experiences compared with heterosexual women, which may indicate that there are indeed qualitative differences between lesbians and heterosexual women on the basis of their history of CSA. The results from this study point to the need for further research using comprehensive measures of victimization experiences and a prospective study design to determine conclusive results. CSA may also lead to other negative outcomes, albeit ones that less strongly confer risk for ASA among lesbians compared with heterosexual females, which warrants further research as well.
In contrast to the other results, as hypothesized, we found that alcohol is an important risk factor for ASA among lesbians. Alcohol use may increase risk for sexual victimization through (a) inhibited cognitive ability in evaluating risk, (b) increased motor impairment in resisting effectively, and (c) exposure to perpetrators who use alcohol to facilitate assault (Abbey, Zawacki, Buck, Clinton, & McAuslan, 2001). Research studies indicate higher rates of drinking among lesbians compared with heterosexual women, which suggests that alcohol use may be a particularly strong risk factor for ASA for this group (Hequembourg et al., 2008; Hughes, McCabe, Wilsnack, West, & Boyd, 2010). Further research examining alcohol use among lesbians is warranted to investigate the settings in which sexual assault may occur.
Consistent with the literature on sexual revictimization among heterosexual women (e.g., Ullman et al., 2009), CSA significantly predicted ASA and CSA was associated with PTSD symptom severity among gay men in our sample. However, in contrast to results of prior studies with heterosexual women, among gay men in our sample, PTSD symptom severity and alcohol use were unrelated to ASA; CSA also was unrelated to alcohol use. These results are inconsistent with previous theories that CSA may promote vulnerability to ASA through the use of numerous maladaptive emotion regulation strategies (e.g., dissociation, avoidance and withdrawal, hypervigilence, substance use) among survivors in an effort to reduce an intense fear response related to the abuse and any associated stimuli (e.g., Marx, Heidt and Gold, 2005). Importantly, the research on which such theories were predicated only included heterosexual women. Gay men may employ different emotion regulation strategies not studied here (e.g., illicit substance use; Cochran & Cauce, 2006) than heterosexual women, which put them at increased risk for ASA. Additional research is needed to explore this possibility.
Alcohol consumption may not have conferred risk for sexual assault among gay men due to gender differences in alcohol use. Research indicates that alcohol has a more pronounced physiological effect on women than on men due to differing metabolic rates (Ward & Coutelle, 2003), which may explain why drinking significantly predicted ASA for lesbians but not for gay men in the present study (Ward & Coutelle, 2003).
This study corroborates previous findings that LG experience unusually high rates of sexual victimization (i.e., Rothman et al., 2011). In this study, 58.1% of lesbians and 60.8% of gay men reported some form of sexual victimization. This study expands previous work by indicating that CSA, alcohol use, and PTSD symptomatology differentially relate to ASA among sexual orientation groups. Additional research is needed to elucidate the mechanisms by which certain risk factors affect some groups and not others and to understand other risk factors that may contribute to ASA among LG but not among heterosexual women. In this study, CSA, PTSD symptoms, and alcohol use collectively predicted 6.3% of the variance in ASA for lesbians and 16.5% of the variance in ASA for gay men, suggesting that other factors may more meaningfully contribute to ASA among LG.
One important factor to consider in future research is the perpetrator’s sex and/or sexual orientation, which is particularly pertinent for a sample who may have had both same-sex and opposite-sex experiences. A study by Brand and Kidd (1986) indicated that a greater proportion of lesbians experienced attempted rape by male dates (16%) at some point in their lives than by female dates (5%). Other research has suggested that when heterosexual experiences are controlled for, LG are both likely to report high rates of sexual coercion in their same-sex relationships, which merits further research as well (Waldner-Haugrud & Gratch, 1997). Although this study did not assess for the gender and sexual orientation of perpetrators, further research in this area would illuminate the factors surrounding sexual coercion among LG.
Several limitations of our study merit discussion. First, due to the cross-sectional and retrospective nature of this study, we cannot determine the causality of our results with full confidence. Although it is certain that CSA precedes ASA, the point at which alcohol use and PTSD symptoms may have developed is uncertain. PTSD symptoms may have occurred as a result of any sexual assault experience or as a by-product of an entirely different traumatic life experience. Increased alcohol use could be considered a result of initial victimization or revictimization, or may simply serve as a mediator. Despite these limitations, we chose to include these variables as predictors in the analyses since there is strong evidence from prior research indicating that these variables are potentially important in explaining the association between CSA and ASA. Another potential limitation of this study is that over 70% of all participants were Caucasian, suggesting that the results of this study may not generalize to other racial or ethnic groups. Furthermore, participants who were recruited from LGBT organizations and community festivals may represent a particular subset of individuals who are comfortable disclosing their sexual orientation in public and may not be representative of all LG (see Heidt et al., 2005 for a discussion).
This study highlights the need for additional research on sexual violence among sexual minority populations. Overall, our data suggest that alcohol use may be the strongest predictor of ASA among lesbians and CSA may be the strongest predictor of ASA among gay men, although there are likely to be many other risk factors that contribute to ASA among LG. Given the high prevalence rates of sexual assault among LG, it is imperative to examine which factors confer risk of sexual assault in this population and to disseminate this information to at-risk LG populations. Mental health professionals should be aware that sexual identity may play a unique role in influencing risk factors for sexual assault. Lesbian women who present with alcohol-related problems and gay men who have experienced previous sexual assault may be at particular risk for victimization. Clinicians must be aware of this heightened risk and conduct a thorough examination for the presence of CSA, PTSD, and alcohol use to obtain a better sense of current victimization risk (Balsam, Rothblum, & Beauchaine, 2005). Interventions for preventing sexual assault should not be limited to reducing the risk of sexual assault for those who are at-risk, given that it is the perpetrators who are responsible for ASA.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by funding from the Alcoholic Beverage Medical Research Foundation and Temple University to Brian P. Marx.
