Women’s Experiences of Adult Sexual Assault, by Perpetrator Gender and Participant Sexual Orientation

Abstract

The vast majority of theoretical and empirical research assessing sexual assault (SA) focuses on man-against-woman SA (MWSA), leaving other forms such as woman-against-woman SA (WWSA) understudied. Relatively simple questions, such as the relative frequency of WWSA for women of different sexual orientations, the tactics (e.g., coercion, force), or forms (e.g., groping, vaginal penetration) employed in WWSA remain unanswered. To address these issues, 268 women (approximately one-third of each bisexual, lesbian, and heterosexual) completed an online survey reporting on whether or not they had experienced a variety of nonconsensual sexual experiences in their adult lifetime, asking specifically about perpetration by men and women. Results indicated that although SA perpetrated by men was far more common, a substantial minority of participants reported having experienced WWSA. In general, tactics and forms employed were similar, regardless of the gender of the perpetrator or the woman’s sexual orientation, with two exceptions: bisexual women had experienced higher rates of MWSA, and lesbian women had experienced more SA in the form of oral sex compared to other groups. Implications for theory, empirical research, and especially community and clinical practice are discussed.

Imagine hearing that a woman acquaintance of yours has been sexually assaulted. What images or assumptions likely pop into your mind? You might envision a forcible rape, perhaps not realizing that sexual assault (SA) can actually take a wide variety of forms, ranging from unwanted sexual touching to penetrative assaults (LEAF, 2020), and can be obtained by diverse tactics, such as coercion (e.g., verbal pressure, telling lies, threatening to end the relationship), taking advantage of someone who is too intoxicated to provide consent, and/or physical force (i.e., threats of, or actual, physical harm to self or valued others; Struckman-Johnson et al., 2003). You might also immediately assume the SA was perpetrated by a man. But, might it have been perpetrated by a woman instead? How likely is that? Does the answer to that question vary, depending on whether your acquaintance’s sexual orientation is heterosexual, lesbian, or bisexual? Are the specific forms or tactics employed likely to differ if the assailant was a man versus a woman, and/or according to your acquaintance’s sexual orientation? These questions may appear fairly simple; perhaps surprisingly, though, existing research provides few clear answers to these questions.

Sexual Assault Perpetrated by Men Versus Women

At least one point is clear from the existing literature: your acquaintance is far more likely to have experienced SA perpetrated by a man than by a woman (Conroy & Cotter, 2017). Multiple studies (Basile et al., 2022; Gambardella et al., 2020) that asked women who had experienced SA about the gender of the perpetrator found that over 90% of SA was man-against-woman SA (MWSA), rather than woman-against-woman SA (WWSA). This predominance of MWSA extends to sexual minority women. For example, one nationally representative survey found that over 70% of bisexual and lesbian women surveyed reported having experienced only MWSA in their lifetimes (Chen et al., 2023).

Thus, we know that MWSA is more frequent than WWSA, but otherwise know relatively little about WWSA. As Turchik et al. (2016) persuasively argue, most theories and empirical research on SA either explicitly or implicitly assume MWSA only. Other forms of SA, including WWSA, therefore, receive considerably less theoretical and empirical research attention. ¹ Much of the existing research on women’s experiences of SA is therefore silent regarding whether it was MWSA or WWSA. Some studies assume that the perpetrator is a man without asking (Canan et al., 2021; Scheer et al., 2002) or separate perpetration by men versus women (Morgan & Long, 2018; Munroe & Shumay, 2022), but then do not report the gender of the participant, making it difficult to establish the relative rates of MWSA vs. WWSA.

Some studies have addressed WWSA: for example, Girshick (2002) investigated WWSA qualitatively, and Wolff et al. (2006) reported on WWSA within the prison system. Still, research on WWSA remains sparse, leaving basic questions incompletely addressed.

Sexual Orientation of the Target of SA

One issue requiring more study is whether women’s sexual orientations are associated with the frequency of past WWSA. Because intimate partners are frequently the perpetrators of sexual assault (Basile et al., 2022), it seems logical that in many cases the gender of the perpetrator would mirror the gender of an individual’s usual intimate partners (e.g., lesbian women experiencing the most WWSA, heterosexual women the least WWSA). Is that in fact the pattern seen in the literature?

Unfortunately, that question is difficult to answer. A number of studies (Canan et al., 2021) have found that sexual minority women (i.e., lesbian and bisexual women) have experienced more SA overall than heterosexual women, often (but not always) with bisexual women experiencing the highest rates of SA (Canan et al., 2021; Freedner et al., 2002). However, in many cases (Canan et al., 2021), these studies did not assess the perpetrator’s gender, making it difficult to disentangle where exactly the heightened risks of SA lie. A few other studies (Menning & Holtzman, 2014) have found no differences in rates of SA according to women’s sexual orientation, suggesting more information is needed to assess when, whether, how, and from whom any increased risks lie.

Some studies do assess the gender of the perpetrator, but statistical comparisons of women of different sexual orientations are either not conducted at all due to low Ns (Chen et al., 2023), or contain no follow-up tests comparing specific cells (Balsam et al., 2005). Visual inspection suggests that perhaps lesbian women experience fewer MWSA and heterosexual women experience fewer WWSA compared to the other two groups (Balsam et al., 2005; Chen et al., 2023); however, the lack of statistical tests makes it difficult to make these statements with confidence. Furthermore, these studies do not provide detailed analyses regarding the nature of the SA experienced. When women do experience SA, does it tend to differ in terms of tactics (e.g., verbal coercion vs. physical force) or forms (e.g., groping vs. vaginal penetration) depending on the gender of the perpetrator or the sexual orientation of the survivor?

Tactics and Forms of Sexual Assault

Again, current research makes it difficult to answer these questions with confidence. Some researchers do not ask about the perpetrator’s gender (Canan et al., 2021), while others do not ask about the respondent’s sexual orientation (Martin et al., 2022). A few researchers ask about both factors (Chen et al., 2023; Krahé & Berger, 2013), but then do not conduct full statistical comparisons. By visual inspection, coercion appears to be a more common tactic than force in both MWSA and WWSA (Fornicola & Peterson, 2022; Krahé & Berger, 2013), and across all sexual orientations (Balsam et al., 2005). However, information remains sparse on both tactics and forms. A more comprehensive study would ask women of different sexual orientations about their past experiences with both MWSA and WWSA, including information on both the form the SA took and the tactics used by the perpetrator. Such information would provide a more nuanced understanding of diverse women’s experiences with SA, benefiting both theory and support providers.

To address these issues, we conducted an online survey of women’s experiences of past adult MWSA and WWSA, intentionally oversampling lesbian and bisexual women to facilitate group statistical comparisons. We employed an adapted version of the Sexual Experiences Survey (Koss et al., 2007), which assesses various tactics and forms of SA.

Given past research, we hypothesized (H1) that overall, more past MWSA would be reported than WWSA. Given the scarcity of past research, however, our study primarily took an exploratory approach, posing the following research questions:

• RQ1: Will the overall rates of past MWSA versus WWSA differ according to respondents’ sexual orientation? If so, how?

• RQ2: Will similar patterns (i.e., any effects of gender of perpetrator and/or sexual orientation of respondent found when addressing RQ1) apply across different tactics and forms of SA?

Method

Recruitment and Participants

All study procedures were approved by Acadia University’s Research Ethics Board. Participants were recruited using Amazon’s MTurk, where the study was described as focused on sexual violence experienced by women. Compensation was 1 USD. To increase sexual minority representation, purposive sampling was used, posting recruitment information about the study to various 2SLGBTQIA+ social media groups. There, the study was described as focused on sexual violence experienced by sexual minority women, and participation was voluntary (no compensation). Participants had to be 19 or older and identify as women (inclusive of cisgender, transgender, nonbinary, agender, and Two-Spirit women); those who did not meet these criteria were piped out of the online survey via screening questions.

A total of 268 eligible participants completed the online survey. The average age was 30.6, ranging from 19 to 68 years. Participants were primarily college-educated North American cisgender women with modest incomes. The sample included a good representation of those who were currently single/not dating, in serious relationships, and married/common-law. As intended, there were relatively similar numbers of heterosexual (N = 84), bisexual (N = 98), and lesbian (N = 86) women (see the Supplemental Material for detailed demographic information).

Measures/Materials

Revised Sexual Experiences Scale—Short Form Victimization

In the present study, SA was defined as any unwanted and nonconsensual sexual activity. The original Sexual Experiences Scale—Short Form Victimization (SES-SFV) (Koss et al., 2007) was developed to assess a wide variety of different types of SA. Participants rated how often they had experienced each behavior on a scale of 0, 1, 2, and 3+ times. In previous research, the scale showed adequate internal consistency (Cronbach’s α of .74 for women, Koss & Gidycz, 1985).

For the current study, we modified the original SES-SFV. The age range was changed from 14+ to 19+, and the scale was revised to create two versions, one assessing past MWSA and one assessing past WWSA. Every participant completed both measures. See the Supplemental Material for the full text of the modified measures. In our study, Cronbach’s alphas for the full 0 to 3+ scales were excellent: .95 for WWSA, and .93 for MWSA.

Participant responses were highly skewed, with zero being the modal response, violating the assumption of multivariate normality. Therefore, rather than using sum scores, instead we created 11 new dichotomous variables, wherein 0 = the relevant event(s) had never occurred in the participant’s adult lifetime, and 1 = the relevant event(s) had occurred at least once in the participant’s adult lifetime). The “overall” score assessed the presence of any adult lifetime SA, regardless of tactic or form. Three scores then assessed experiences of three different tactics (i.e., coercion; taking advantage while the victim was drunk or otherwise incapacitated; force or threat of force) and seven different forms of SA (i.e., groping, attempted oral sex, oral sex, attempted vaginal penetration, vaginal penetration, attempted anal penetration, anal penetration).

Procedure

After clicking on a link in the online recruitment message, participants completed a self-administered online survey, consisting of a consent form, demographics, and then the two versions of the modified SES-SFV scale (assessing self-reported experiences of past MWSA and past WWSA in their adult lifetimes) in a random order, as well as additional measures not relevant to the current paper. Participants responded to several attention checks throughout the survey; only participants who passed all checks were included in the final sample. At the end of the online questionnaire, participants were asked to envision a scene that was relaxing to them while taking slow deep breaths, to help restore a positive mood. Participants were then shown a debriefing form, including information on online resources for participants experiencing any psychological distress.

Results

Descriptive Statistics

Table 1 shows descriptive statistics for all variables, that is, adult lifetime percentages for each type and form of SA, broken down by perpetrator gender and participant’s sexual orientation. Within each grouping, variables are displayed in order from the most frequent to the least frequent overall. Supplemental Table S1 in the online resources breaks down responses even further, showing every possible combination of tactic and form.

Table 1.

Percentage of Women Having Experienced Adult Sexual Assault, Overall and in Different Tactics and Forms, by Perpetrator Gender and Sexual Orientation of Respondent.

Dependent Variable	Assaults by Men				Assaults by Women
	Bisexual	Heterosexual	Lesbian	Overall	Bisexual	Heterosexual	Lesbian	Overall
Overall	93.9	77.4	75.6	82.8	39.8	14.3	41.9	32.5
Tactics
Coercion	76.5	58.3	68.6	68.3	31.6	9.5	29.1	23.9
Take Adv.	74.5	53.6	58.1	62.7	20.4	14.3	23.3	19.4
Force	65.3	48.8	41.9	52.6	19.4	9.5	19.8	16.4
Forms
Groping	83.7	65.5	70.9	73.9	29.6	10.7	34.9	25.4
Vaginal	69.4	52.4	40.7	54.9	15.3	9.5	18.6	14.6
Oral	64.3	35.7	50.0	50.7	10.2	10.7	26.7	15.7
Att. oral	51.0	41.7	43.0	45.5	18.4	11.9	17.4	16.0
Att. vaginal	54.1	39.3	36.0	43.7	13.3	9.5	17.4	13.4
Anal	38.8	20.2	26.7	24.3	8.2	9.5	5.8	7.5
Att. anal	31.6	19.0	20.9	29.1	7.1	8.3	7.0	7.8

Note. Numbers represent the percentage of respondents reporting each type of sexual assault. Ns = 98 bisexual; 84 heterosexual; 86 lesbian; 268 overall. Adv. = advantage; Att. = attempted.

Analysis Strategy

We conducted a series of 11 multilevel logistic regression models, 1 for each row in Table 1, using MPlus version 8.7. Multilevel models allow for nesting of individual responses within participants, with predictor variables at Level 1 (akin to within-subjects variables in ANOVAs) and/or at Level 2 (akin to between-subjects variables). Instead of assuming the usual multivariate normal distribution, multilevel logistic models employ the logit model, appropriate for dichotomous outcome variables (Sommet & Morselli, 2017).

Outcome variables were always scored 0 if the participant had not experienced that type of SA in their adult lifetime and 1 if they had. Predictor variables were (a) the gender of the perpetrator (Level 1; scored 0 for women and 1 for men), (b) the sexual orientation of the participant (Level 2; dummy-coded with bisexual women as the reference group, thereby creating indicator variables for heterosexual and lesbian sexual orientations), and (c) the interaction between perpetrator gender and sexual orientation. Results are shown in Table 2. The numbers shown are in log odds form. For main effects, zero indicates no effect; positive values indicate the risk of SA is greater for the category coded 1, compared to the category coded 0; negative values indicate the opposite. Interaction terms represent the additional increase (or decrease) in the likelihood of having experienced SA when both of the variables involved in the interaction are coded 1, relative to what one would expect when considering the two main effects alone.

Table 2.

Results of Multi-Level Logistic Models, Predicting Likelihood of Women Having Experienced Adult Sexual Assault, Overall and in Different Tactics and Forms, by Perpetrator Gender and Sexual Orientation of Respondent.

Dependent Variable	Perp. Gender	Het. Women (vs. Bisexual)	Perp. Gender × Het. Int.	Lesbian Women (vs. Bisexual)	Perp. Gender × Lesbian Int.	Random I. Variance
Overall	3.43*** (0.56)	−1.51*** (0.41)	−0.08 (0.62)	0.10 (0.34)	−1.80** (0.62)	0.55 (0.55)
Tactics
Coercion	2.22*** (0.39)	−1.63*** (0.45)	0.68 (0.52)	−0.13 (0.36)	−0.32 (0.51)	0.65 (0.52)
Take Adv.	3.76*** (0.65)	−0.66 (0.57)	−0.78 (0.67)	0.23 (0.53)	−1.38* (0.66)	3.51* (1.48)
Force	2.75*** (0.52)	−1.07* (0.53)	0.15 (0.60)	0.02 (0.48)	−1.32* (0.60)	1.97* (0.92)
Forms
Groping	2.85*** (0.47)	−1.39** (0.46)	0.28 (0.57)	0.28 (0.36)	−1.11* (0.53)	0.71 (0.56)
Vaginal	3.43*** (0.60)	−0.72 (0.56)	−0.27 (0.64)	0.29 (0.51)	−1.94** (0.66)	2.14* (1.03)
Oral	3.67*** (0.65)	0.01 (0.59)	−1.64* (0.70)	1.48** (0.56)	−2.29** (0.69)	2.12* (1.01)
Att. oral	2.51*** (0.52)	−0.80 (0.63)	0.17 (0.66)	−0.08 (0.59)	−0.46 (0.65)	4.32** (1.66)
Att. vaginal	3.21*** (0.60)	−0.60 (0.66)	−0.37 (0.69)	0.50 (0.60)	−1.70* (0.71)	3.94* (1.59)
Anal	3.70*** (0.99)	0.01 (0.89)	−1.92 + (0.98)	−0.70 (0.99)	−0.44 (1.05)	8.77* (4.15)
Att. anal	3.24*** (0.89)	0.02 (0.91)	−1.40 (0.94)	−0.08 (0.95)	−1.02 (1.03)	8.19* (3.71)

Note. Numbers outside parentheses represent b coefficients, in log odds form, except for the final column, where they represent variances. Numbers inside parentheses are standard errors. Perp. = perpetrator; Het. = heterosexual; Int. = interaction; I. = intercept; Adv. = advantage; Att. = attempted.

⁺P < .10. *P < .05. **P < .01. ***P < .001.

Primary Analyses

We first describe the overall results (i.e., Table 2, first row, analyzing the data from Table 1, first row), to address H1 and RQ1. We then briefly note any variations of this overall pattern across tactics or forms, addressing RQ2.

Overall

In the overall analysis, supporting H1, there was a significant positive main effect for perpetrator gender, indicating that participants were more likely to have experienced past MWSA than WWSA. See Table 1, in which the percentages in the first three columns (MWSA) are much higher than those in the last three columns (WWSA).

The remaining columns address RQ2, examining whether patterns of past SA vary by participants’ sexual orientation, alone or in combination with perpetrator gender. There was a significant negative coefficient for the indicator variable for heterosexuals, meaning that heterosexual women were less likely to have experienced past adult SA than bisexual women, overall. The nonsignificant interaction term indicates that this pattern held true for both MWSA and WWSA. In Table 1, one can see these effects in action: relatively more bisexual women have experienced adult SA than heterosexual women, whether they were MWSA or WWSA.

The nonsignificant coefficient associated with the indicator variable for lesbian women means that, in general, lesbian and bisexual women were similar in their experiences of past adult SA. However, this main effect is qualified by a significant negative interaction, indicating that the rate of adult SA experienced by lesbian women (compared to bisexual women) was lower for MWSA than WWSA. This pattern manifests itself in Table 1 in the fact that bisexual and lesbian women experienced similar rates of WWSA, but bisexual women experienced more MWSA than lesbian women.

The last column in Table 2 shows the variance of the random intercept term. If this term were significant, it would indicate that there was still significant individual variability in participants’ rates of SA left over, after all the variables in the model had been taken into account. The fact that the term is nonsignificant here shows that our model does a very good job of accounting for variability in participants’ rates of overall SA.

By Tactic

Addressing RQ2, the pattern of results across the three tactics appears quite similar to the overall pattern. In general, we see that regardless of the tactic used, past MWSA are more frequent than WWSA and that bisexual women report higher rates of past SA—both MWSA and WWSA, relative to heterosexual women, but MWSA only, relative to lesbian women.

There are only two exceptions to the usual pattern. First, for coercion, the nonsignificant effects involving the interaction variable comparing lesbian women to bisexual women show that the two groups experienced fairly similar rates of coercive SA, for both MWSA and WWSA. As can be seen in Table 1, however, the pattern for coercion is still the same as for the other tactics (i.e., bisexual women still experience more coercive MWSA than lesbian women, but the difference is not significant with this sample size).

The second exception is for taking advantage, where the nonsignificant effect of the variable comparing heterosexual and bisexual women indicates, in principle, that the two groups experienced this form of SA at similar rates, regardless of perpetrator gender. However, Table 1 shows that the pattern is actually quite similar to the overall one here, with bisexual women being at somewhat higher risk of SA in this form than heterosexual women, both for MWSA and for WWSA. The differences are just slightly smaller here, with larger standard errors, so statistical significance is not reached. In sum, there is no substantial deviation from the overall pattern of results, across the three different tactics considered.

By Form of Assault

When considering SA in different forms, the main effect of perpetrator gender is fully consistent: overall, participants reported higher rates of MWSA than WWSA for every single form of SA. At first glance, the effects comparing heterosexual and bisexual women appear quite variable across different forms of SA; however, examination of Table 1 shows the pattern is actually fairly consistent, and quite similar to the overall results in most cases. The percentages are in the direction of bisexual women experiencing more past MWSA than heterosexual women, across all forms. The pattern is fully consistent; it is just that the group differences are somewhat smaller and the standard errors are somewhat larger with this more fine-grained breakdown, so statistical significance is not always reached.

The results are also in the direction of bisexual women experiencing WWSA at higher rates than heterosexual women for most forms. The three exceptions are for WWSA in the forms of oral sex, anal sex, and attempted anal sex. Here we can see in Table 1 that heterosexual and bisexual women reported these events at similar rates. This pattern manifests in Table 2 via a significant heterosexual × perpetrator gender interaction for oral sex, and a marginally significant one for anal sex. (The pattern is similar for attempted anal sex, but does not reach statistical significance.) These forms of WWSA are quite rare for both bisexual and lesbian women, however, representing only a few participants. These deviations from the usual pattern can only be considered with great caution, therefore, and require further replication with larger samples before they can be assessed with certainty. Overall, the conclusion that bisexual women have experienced both more MWSA and WWSA than heterosexual women, across most forms of sexual assault, is the safest one.

Comparing lesbian and bisexual women, however, there appears to be one variable that deviates from the usual pattern. For most variables, although the effects do not always reach full statistical significance, the pattern is quite consistent: bisexual women consistently experienced more MWSA than lesbian women, but the two groups were quite similar in their rates of WWSA. Oral sex, however, showed a different pattern: lesbian women were substantially more likely to experience WWSA in the form of oral sex than bisexual women. Our sample size did not allow us to conduct statistical comparisons across the different forms of SA. Descriptively, however, looking down the columns, lesbians may also be at somewhat heightened risk of MWSA in the form of oral sex. SA in the form of oral sex is more common than SA in the form of vaginal penetration for lesbian women, whereas the reverse is true for bisexual and heterosexual women. Thus, the general pattern of findings seems to mostly hold across different forms of SA, with the exception of lesbian women being somewhat more likely to have experienced SA in the form of oral sex. Note given the question wording, such assaults could have come in the form of being coerced/forced into either receiving or giving oral sex.

Finally, looking at the random intercept variances, they are significant in most instances, suggesting there is still some additional variance available to be accounted for, beyond what is explained using the variables in our models.

Checking Covariates

To adjust for potential confounding variables, we re-ran all analyses including demographic or background variables on which the sexual orientation groups differed as covariates. The pattern of results remained unchanged, with or without the inclusion of these covariates. See the Supplemental Material for more details.

Discussion

A Call for a Broader Perspective on SA

As expected, H1 was strongly supported: MWSA was much more frequent than WWSA, both overall and for every tactic and form assessed, replicating previous research (Chen et al., 2023; Conroy & Cotter, 2017). As Turchik et al. (2016) note, several classic theories in the literature focus primarily on MWSA. MWSA may stem from men’s efforts to dominate women (Brownmiller, 1975), hostile masculinity (Malamuth et al., 1996), cultural scripts that define men as pursuers and women as gatekeepers (Denov, 2003), and/or evolutionarily driven urges for men to reproduce (Thornhill & Palmer, 2000). Any or all of these explanations might potentially help account for the high rates of MWSA relative to WWSA.

What is particularly notable in our study, however, is that WWSA was not especially rare. Approximately 40% of bisexual and lesbian women and almost 15% of heterosexual women had experienced WWSA in their adult lifetime, and these assaults took every possible form and tactic. Given that our recruitment material predominantly attracted those who had experienced SA, and given our representation of sexual minority women, these rates are almost certainly much higher than would be seen in the general population. Still, among women who had experienced SA, experiencing WWSA was not uncommon.

Furthermore, the patterns sexual assaults take—the tactics employed, the forms used— seem to be remarkably consistent, whether perpetrated by women or men. We arranged our results in Table 1 in descending order from most to least frequently used tactics and forms, overall. The pattern across columns is generally highly consistent, indicating that for the most part, the more common tactics and forms employed are generally similar, regardless of the gender of the perpetrator or the sexual orientation of the participant. ² WWSA is rarer than MWSA, but it does occur, and not that infrequently. When it does occur, it generally does not seem to be notably different in kind, relative to sexual assaults by men.

These findings suggest that more theoretical, empirical, and policy-based attention should be focused on WWSA. As Turchik et al. (2016) note, many theories of sexual assault focus primarily or solely on understanding MWSA, and thus as noted in our introduction, many empirical studies take a similar lens. Perhaps unconsciously, the question may often be framed more as “why do some men sexually assault women?” rather than the broader question of “why do some people sexually assault others?” A consistently broader theoretical and empirical lens would be beneficial. For example, Tharp et al. (2013) conducted a large systematic review assessing what was known at the time about risk and protective factors for sexual perpetration. Their review identified 67 potential factors, the vast majority of which could theoretically apply to both men and women perpetrators; however, empirical research on both women perpetrators and same-sex relationships was rare, making any group comparisons difficult. As Tharp et al. (2013) note, more information is required to assess whether the same or different risk factors apply when research extends beyond MWSA.

If more empirical research is conducted, it might well turn out that many of the same risk factors apply, regardless of perpetrator gender. For example, Bouffard et al. (2016) conducted a study of factors predicting sexual coercion by women (gender of target unspecified but assumed to be primarily men). Risk factors originally identified in the literature on MWSA (e.g., rape myth acceptance, risk-seeking, erotica use frequency) were all significant predictors of women’s sexually coercive behavior as well. In the current research, the fact that the most common SA tactic is verbal coercion rather than physical force, for men and women alike, suggests that many SAs may arise because of faulty communication patterns, the misperception of social cues as being more sexualized than intended (Shea, 1993), or awkward and ill-informed attempts to signal sexual desires without adequate attention paid to understanding the other partner’s wishes or desires. Those issues around communication and understanding consent could apply to both men and women. If risk factors do in fact turn out to be very similar across perpetrators and relationship types, it will allow for focus on the most promising ones for intervention across the board; if risk factors turn out to vary systematically, it will allow for more targeted and hopefully more effective interventions, based on group-specific risks.

Two Qualifications

Overall, our results suggest far more similarities than differences across groups. There were two instances, however, where results varied according to perpetrator gender and/or respondent sexual orientation. First, bisexual women in our sample reported more past SA than lesbian or heterosexual women, replicating some past research (Canan et al., 2021; Chen et al., 2023). However, our findings add additional information: at least in our sample, this heightened risk for bisexual women came primarily from men.

Bisexual women reported higher rates of MWSA than the other groups. Given that intimate partners are frequently the perpetrators of sexual assault (Basile et al., 2022), it is possible that bisexual women having experienced more MWSA than lesbian women may simply reflect the fact that bisexual women are more likely than lesbian women to have had sexual relationships with men. ³ However, the fact that bisexual women experience more MWSA than heterosexual women requires more explanation.

It is possible the heightened rate of MWSA stems from biphobia, i.e., anti-bisexual prejudice (McInnis et al., 2022). Biphobia may portray bisexuals as being hypersexual or promiscuous (Alarie & Gaudet, 2013; Dickson, 2016). They may therefore be falsely seen as actually desiring sex, despite protests (Hoskin, 2019). Bisexuality may also sometimes be viewed as “just experimenting,” rather than being an agentic sexual orientation (Alarie & Gaudet, 2013; McInnis et al., 2022). Some perpetrators may assume that bisexual women do not know what they truly want sexually, and may therefore be motivated to push sexual boundaries, to the point of coercion and beyond. Finally, if bisexual women begin to internalize these messages, that is, experience internalized biphobia (McInnis et al., 2022), they too might potentially believe they do not truly understand their own sexual needs, making them vulnerable to partners who purport to be teaching them what they “really” want sexually.

Although biphobia is prevalent and can be held by women as well as men (Zivony & Saguy, 2018), in our sample these potential explanations would seem to apply more to perpetrators who are men; for women perpetrators, rates of SA against bisexual women are no higher than those against lesbian women. More detailed research on the motivations behind SA should assess the extent to which group-specific beliefs and stereotypes such as biphobia may play a role.

The second exception to the general patterns of similarities in types of SA across groups is that lesbian women were particularly likely to have experienced SA in the form of oral sex. Lesbian women experienced WWSA in the form of oral sex far more frequently than the other groups. Here, it may be a matter of base rates. Women in same-sex relationships report giving and receiving oral sex from their partner more frequently than women in mixed-sex relationships, and report enjoying it more (Blair et al., 2018). SA may at times represent sexual interactions gone awry. If oral sex is a frequent form of sexual interaction for lesbian women, then it may also be a frequent form of maladaptive sexual interactions, that is, assaults, with the assailant coercing/forcing the giving or receiving of oral sex.

However, lesbian women’s increased risk of SA in the form of oral sex applies to MWSA as well. To our knowledge, this finding is novel, and we do not have enough information to fully understand it. One possibility again incorporates base rates. If women who currently identify as, but did not always identify as lesbian are particularly likely to enjoy giving and receiving oral sex, they may have also favored that sexual practice in past relationships with men, and as suggested above, more engagement in a given sexual practice affords more scope for it to go awry. Another possibility is that some men might employ this form of SA against lesbian women to “prove something,” that is, that they can engage in this behavior as well as a woman partner would.

These explanations are all highly speculative. This particular finding requires replication and more detailed data collection to interpret with confidence. Again, however, our general point holds: more detailed and nuanced information would be highly valuable in terms of assessing when, whether, and why certain groups may be at increased risk of SA of particular tactics or forms.

Strengths and Limitations

As is often the case in research, our study’s strengths also have balancing weaknesses. We focused on the understudied phenomenon of WWSA but neglected other understudied forms of SA (e.g., woman-against-man, man-against-man, SA perpetrated by nonbinary individuals). We took an inclusive approach to defining women but did not have sufficient numbers of trans or nonbinary women to understand their potentially unique experiences. We focused mostly on those who had experienced some form of SA, making our findings relevant to practitioners who serve that population, but it means that our rates of SA do not generalize to the population. We used a well-validated behaviorally based measure of adult SA but had to adapt it to be inclusive of WWSA in ways that remain unvalidated. Our measure provided highly detailed information; however, analyzing these details involved conducting many analyses and thereby heightening the risk of type I error. Our sample size is quite good if one considers the challenges of recruiting minority populations, but it could always be larger to provide more power, especially important when assessing statistical interactions. To address issues of intersectionality, future research should purposively sample additional variables of interest (e.g., participants’ age, relationship status, education; participants’ experience of childhood sexual assault) and include those variables as additional factors in the analyses. We considered these additional variables as covariates (see Supplemental Material) but did not have sufficient sample size when these variables were crossed with our focal variables to incorporate them fully into the analyses as additional factors.

Implications

Our findings drive home the point that more theoretical and empirical work is needed on all types of SA (Turchik et al., 2016). It is possible, as found by Bouffard et al. (2016) in their work on sexual coercion by women, that many of the same factors driving MWSA also apply to other types of assault; however, it is also possible that there are some unique dynamics at play for different configurations of SA. Collectively, the field will not know until more detailed and nuanced data are collected.

Meanwhile, our findings have implications for clinical practice. If a SA does not take the usual MWSA form, practitioners may be less able to provide informed and sensitive support. Similar issues may arise as those discussed by Donovan et al. (2014), in their study of available support for intimate partner violence (IPV) in LGBT+ populations. Donovan et al. (2014) found that IPV in those populations did not often follow the heteronormative script, making it more challenging for victims/survivors, perpetrators, and those providing services to recognize and label that abuse was occurring. When LGBT+ victims/survivors or perpetrators did recognize a problem, they were often reluctant to seek out standard services, as they were not sure they would fit well with their specific needs. Their hesitation was not unfounded, as Donovan et al. (2014) reported that service providers lacked training in LGBT+ issues and were uncertain that they could provide quality care. Without more awareness and education for both those experiencing SA and practitioners, individuals experiencing WWSA may similarly not receive the attention and care they deserve.

We also follow Donovan et al. (2014) in calling for treatment that goes beyond assuming that SA takes the form of one partner seeking power and domination over the other partner. In their study, they found that in many instances, physical or psychological aggression was mutual, and often stemmed from excessive stress, poor communication skills, and/or difficulties in negotiating complex relationship challenges. We suspect that similarly, some SA (particularly in the form of coercion) may arise because perpetrators do not know how to seek fulfillment of their needs for intimacy in a respectful and mutually pleasurable manner. These deficits could be addressed by providing better sexual education for all, as well as providing skills training for those who need to improve their coping and communication skills (Action Canada, 2019). Such an approach would benefit all relationships but may be particularly beneficial for LGBTQ+ individuals, who may experience more challenges finding resource materials that fully map onto their relationship experiences. Enlightened sex education and services could help them, and all individuals, imagine and enact healthy and mutually respectful sexual relationships.

Author Biographies

Alyssa J. MacDonald is completing her PsyD at the University of Prince Edward Island. Her research focuses on sexual violence and the 2SLGBTQ+ community.

Diane Holmberg, PhD, is a professor emerita in the Department of Psychology, Acadia University. Her work is primarily in the area of close relationships, inclusive of LGBTQ+ relationships.

E. Lisa Price, PhD, completed a PhD at the University of New Brunswick and is currently a professor in the Psychology Department at Acadia University.

Rhea Ashley Hoskin, PhD, is SSHRC postdoctoral fellow and an AMTD Global Talent Postdoctoral Fellow at the University of Waterloo and St. Jerome’s University, where she is cross-appointed to the departments of Sociology & Legal Studies, and Sexuality, Marriage, & Family Studies. Her work focuses on Critical Femininities, Femme Theory, and femmephobia. More specifically, her work examines perceptions of femininity and sources of gender-based violence rooted in the devaluation or regulation of femininity, as well as strategies for revaluing and rethinking femininity. She is the founding Editor-in-Chief of the Journal of Femininities.