Profiles of Men’s Rape Myth Beliefs and the Association With Rape Proclivity

Abstract

The two constructs of rape myth acceptance (RMA) and rape proclivity are associated with sexual violence (SV) perpetration. Further understanding these constructs can help improve prevention efforts aimed at reducing SV perpetration. Latent profile analysis was conducted to examine typologies of RMA among 474 incoming college men and found that male college students can be categorized into four profiles. Some groups endorsed lower or mid-levels of rape myths (RMs) and others endorsed higher levels of some or all RMs, indicating the heterogeneity of RM beliefs. And within each subgroup of college men’s RMA, intention to join an all-male sports team and/or a fraternity (two risk factors) and bystander attitudes (a protective factor) were examined as covariates in the model. Bystander attitudes appear to act as a protective factor as they are higher among profiles of men with lower RMA. Furthermore, this study examined the four subgroups (latent profiles) of college men based on their RMA to examine whether membership within each subgroup/profile is differentially associated with rape proclivity. The findings indicate that subgroups of men with high levels of RMA have higher mean rape proclivity scores compared to the subgroup of men with the lowest level of RMA. Implications for prevention programming tailored for high-risk groups of men, based on their RMA beliefs, as well as possible future research within this area are discussed.

Keywords

sexual assault prevention offenders rape myths

Introduction

Campus sexual violence (SV) is a well-documented problem nationwide: nearly 20 to 35% of men perpetrate SV in college (Thompson & Morrison, 2013; Zinzow & Thompson, 2015). To reduce these rates, prevention efforts have been implemented on campuses nationwide. However, most prevention programs targeted at potential perpetrators have not demonstrated effectiveness in reducing rates of SV (DeGue et al., 2014). The lack of effectiveness in SV perpetration prevention efforts is attributable to several factors, including failure to account for individual differences among those who may perpetrate SV regarding attitudes related to SV. Prevention responses might target high-risk groups to limit the potential of individuals within these groups of developing into offenders (Welsh & Farrington, 2012). Because students enter college with differing beliefs, programming efforts not accounting for individual differences might be poorly received and be less effective at reducing beliefs related to SV perpetration.

One set of widely researched attitudes related to SV perpetration is rape myth acceptance (RMA) which falsely attributes blame for SV on victims, not perpetrators (Brownmiller, 1975; Burt, 1980). Rape myths (RMs) are more likely to be held by men who are sexually aggressive or violent (Tharp et al., 2012; Yapp & Quayle, 2018) and are associated with other beliefs related to SV perpetration such as hostility toward women (Russell & King, 2016). Another frequently researched risk factor for SV perpetration is involvement in all-male peer groups such as fraternities or all-male sports teams, and a review study found that belonging to these types of groups was associated with increased SV perpetration (Tharp et al., 2012). Less studied are factors that might reduce the risk of an individual committing an act of SV, protective factors (Tharp et al., 2012). The current study will examine factors related to SV perpetration to assess if heterogeneity in RMA among college men is associated with risk and protective factors.

Attitudes and Beliefs Associated With SV Perpetration

RMA, as first conceptualized by Burt (1980), incorporates a range of beliefs that suggest victims, namely women—not perpetrators—are to blame for SV. For example, women might be blamed for “causing” an assault because of their style of clothing or their use of alcohol. Overall, RMA is common among men who have committed SV (Tharp et al., 2012). Several types of RMs have been identified including (1) She Asked For It, (2) It Wasn’t Really Rape, (3) He Didn’t Mean To, (4) She Wanted It, (5) She Lied, (6) Rape Is A Trivial Event, (7) Rape Is Deviant Event, and (8) He Did Not Mean To Due To Intoxication (Lonsway & Fitzgerald, 1994; McMahon & Farmer, 2011). While a few studies have investigated these individual types, or subscales, of RMs, little research has been conducted into the association of perpetration risk factors or related variables with individual types of RMs. Despite scholars’ suggestions that examining individual types of RMs can further our understanding of perpetration (McMahon & Farmer, 2011; Mouilso & Calhoun, 2013), with a few exceptions (Carroll et al., 2016; McMahon, 2010, 2015), most researchers have examined mean RMA averaged across subscales of RMs, thus ignoring individual types of RMs.

A concept related to perpetration of SV is rape proclivity developed by Malamuth (1981) who showed that many men demonstrate a likelihood of SV perpetration, and such proclivities are associated with rape-related beliefs, attitudes, and sexual aggression against women. While attitudinal in nature, rape proclivity measures participants’ self-reported likelihood of committing SV under certain circumstances, usually if the participant knew they would not be caught. Researchers have found an association between RMA and rape proclivity (Bohner et al., 2010; Seabrook et al., 2018). Studies examining rape proclivity found that one in three men affirm they would commit SV, including rape, under certain circumstances (Malamuth, 1981; Untied et al., 2013). Although these studies do not prove that men who demonstrate rape proclivity will indeed go on to perpetrate SV, it can be argued that any manifestation of rape proclivity is a risk marker for actual SV perpetration.

Factors That Influence Perpetration

The Centers for Disease Control and Prevention’s socioecological framework for preventing SV outlines both risk and protective factors for victimization and perpetration that might be targeted in prevention interventions (Dahlberg & Krug, 2002) and suggests factors that might be associated or covary with perpetration and be targeted in SV prevention. However, few scholars have examined both perpetration risk and protective factors and associated covariates in primary prevention efforts of SV (Tharp et al., 2012). The current study includes variables that might be addressed in efforts to reduce attitudes associated with SV perpetration and can be viewed as protective or risk factors for perpetration.

Bystander attitudes

Tharp et al. (2012) in a review of risk and protective factors for SV perpetration found many studies identified perpetration risk factors, but few studies examined protective factors. The authors suggest this gap in the research limits our understanding of SV perpetration and prevention. Bystander attitudes might serve to protect against perpetration. Although no known research has examined bystander attitudes as a protective factor against SV perpetration, there is preliminary evidence that suggests bystander attitudes may reduce RMA. Bystander attitudes indicate how willing a person would be to intervene before, during, or after an incident of SV. In relationship to RMA, bystander attitudes are negatively related such that an increase in RMA or rape-supportive attitudes predicts lower bystander attitudes (Fleming & Wiersma-Mosley, 2015; Orchowski et al., 2016; Powers et al., 2015). These studies suggest an association between RMA and bystander attitudes such that bystander attitudes may lower RMA. However, despite the need for a greater understanding of protective factors for SV perpetration, there has been little research examining how bystander attitudes might protect against attitudes related to SV perpetration such as RMA.

All-male peer groups

While men’s attitudes and beliefs have been associated with SV perpetration, these individual-level factors are not the only factors linked to SV perpetration. Another extensively researched area is the relationship between all-male peer groups and campus SV perpetration, with a focus on fraternity members and male athletes. Scholars have postulated that such all-male peer groups create a “rape-prone culture” in which attitudes that support SV are normalized and even encouraged (Boswell & Spade, 1996; Schwartz & DeKeseredy, 1997). Finally, some research documents higher endorsement of RMs among men belonging to these all-male peer groups (Murnen & Kohlman, 2007).

Heterogeneity of RMs

Although research examining heterogeneity among RMA has been limited, there are individual studies that examine subscales of the RMA in relation to other factors (Carroll et al., 2016; McMahon, 2010, 2015). Research on RMs has traditionally analyzed these beliefs as a monolithic construct. Given the wide variety of types of beliefs that RMs encapsulate, some researchers argue that RMs are multidimensional and as such should be measured with subscales that capture the multiplicity of the types of existing RM beliefs (Lonsway & Fitzgerald, 1994; McMahon & Farmer, 2011). Even when subscales have been used in research, few scholars have examined differential RM subscale endorsement and its implications. However, RM subscales may be useful in distinguishing different types of myths that call for tailored prevention interventions. For example, two college students may have similar overall RMA levels and yet vary vastly on the individual types of RMs (subscales) they endorse. One participant might strongly endorse the subscale He Didn’t Mean To, while another might strongly endorse She Lied as an RM. These are both RMs, yet interventions used to target these beliefs may differ, with one targeting RMs about the attitudes of those who perpetrate and another targeting the belief that women falsely accuse men of rape. In addition, given the importance of types of RMs, it is likely that type of RM endorsement may also be differentially associated with other variables such as rape proclivity or SV perpetration.

Current Study

The current study examines four research questions: (1) Are there meaningful patterns across individuals in terms of the types (subscales) of RMs they endorse? (2) If there are meaningful patterns across individuals, how can men be categorized into profiles based on the types of myths individuals endorse? (3) What are the characteristics of the subgroups related to demographic factors? Specifically, what demographic factors, including membership in all-male peer groups, are associated with the subgroups of men? (4) Is membership within subgroups, based on men’s RMA, differentially associated with rape proclivity?

Methods

Procedures

The data from this study are drawn from a dataset that focused on prosocial bystander behaviors and campus SV. The study was conducted at a large public Mid-Atlantic university and was approved by the institution’s IRB. Incoming first-year students were invited to participate in the parent study during new student orientations. Students who participated were entered into a raffle for a television or iPad. Only the first wave of data from the parent study was used in the current paper. As part of the original study, participants watched a peer theater performance addressing the issue of campus SV after the baseline wave of data was collected. Afterward, participants were randomized into groups. For a full description of the methods and the intervention, see McMahon et al. (2015).

Participants

Incoming first-year students participated in the study between June 2010 and September 2011 and at five time points thereafter. For the parent study, 1,390 students made up the final sample (see McMahon et al., 2015 for information on the sample). Since the rape proclivity questions are geared toward men, only the male sample was used in this study. The sample, after eliminating those who did not consistently identify their gender, consisted of 513 men. An additional 39 participants who failed to answer a reliability check question were removed from the analytic sample for a final sample of 474 men. Of the participants, 59% intended to join an athletic team, and 33% a fraternity (Table 1). The RM subscale with the highest mean score was He Didn’t Mean To (M = 3.1), and the lowest score was for Was Not Really Rape (M = 1.7).

Table 1.

Descriptive Statistics of the Sample and for RM Subscales (N = 474).

	Percentage/Mean	SD
Intent to join an athletic team	59.0%
Intent to join a fraternity	33.6%
Intent to join student government	33.5%
High school student council	19.7%
High school varsity athlete	45.4%
Race/ethnicity
White	48.2%
Asian	34.5%
Other	7.0%
Latino	6.8%
Black	3.6%
RM
RM1: She Asked for It	2.9	0.8
RM2: Was Not Really Rape	1.7	0.8
RM3: Intoxicated, Didn’t Mean To	2.1	0.9
RM4: Didn’t Mean To	3.1	0.9
RM5: She Lied	3.0	0.7
Rape proclivity	1.35	0.7
Bystander attitudes	3.62	0.7

Note. RM = rape myth.

Measures

RM beliefs

This study used a modified 17-item version of the Illinois Rape Myth Acceptance scale (McMahon & Farmer, 2011). Participants indicated agreement on a 5-point scale on questions such as “If a girl goes to a room alone with a guy at a party, it is her fault if she is raped,” response options ranged from 1 (Strongly Disagree) to 5 (Strongly Agree). Items were reverse coded as needed so higher scores indicate higher RMA. This scale has the following subscales: (1) RMA 1: She Asked for It (4 items α = .70); (2) RMA 2: Was Not Really Rape (3 items α = .77); (3) RMA 3: Intoxicated, Didn’t Mean To (2 items α = .64); (4) RMA 4: He Didn’t Mean to (3 items α = .66); and (5) RMA 5: She Lied (5 items α = .81)

Demographic correlates

This study measured intention to join an all-male peer group including intention to join a fraternity through a yes/no question asking participants if they intended to pledge a fraternity in college. Men’s athletic team participation and college student government were gauged in the same method as fraternity participation with a question asking if students intended to join. In addition, race (dichotomized as white compared to non-white students), participation in high school varsity athletics, and/or student council were added as covariates to the model.

Bystander attitudes were measured using a modified version of the Bystander Attitude Scale (McMahon et al., 2014). The scale contains 18 questions. An example question is, “In the future, how likely are you to confront a friend who plans to give someone alcohol to get sex.” Participants indicated how likely they would be to perform such behaviors in the future from 5, Unlikely to 1, Very Likely.

Rape proclivity

This item averaged together two questions for a mean score on rape proclivity (α = .76). The first question asked, “How likely would you be to force another person to do something sexual even if she didn’t want to?” (Malamuth, 1989). The second question reads, “How likely would you be to have sex with another person who was too intoxicated to resist your sexual advances?” This second question was created from the Perpetrator History Scale (Lisak et al., 2000) and a question from Malamuth’s (1989) study. Both questions included the caveat of “if you were assured that no one would know and that you could in no way be punished.” The response options ranged from 1, Not at All Likely to 5, Extremely Likely.

Analyses

Research questions 1–3: Latent profiles and covariates

LPA on the RM subscales was used to derive the profiles corresponding to the patterns in the data and demonstrate the prevalence of each profile. The subscales of RMs were used to examine the patterns of subscale endorsement. LPA tests a series of models with differing numbers of profiles, starting with one profile and increasing the number of latent profile in each model. LPA was performed with Mplus Version 8.0 (Muthen & Muthen, 2017), using full-information maximum likelihood estimation to identify models. Fit criteria were used to compare the model fit. These fit criteria included (1) the Akaike’s information criterion (AIC); (2) the Bayesian information criterion (BIC); (3) the consistent Akaike’s information criterion (CAIC); and (4) the approximate weight of evidence criterion. For all of these criteria, the lowest value indicates the best-fitting model (Masyn, 2013). Other indices included those where each model is compared to the prior model (e.g., a three-profile compared to a two-profile model); a statistically non-significant p value indicates a better fit. The adjusted Lo–Mendell–Rubin likelihood ratio test was such a test in this study. The Bayes factor (BF) was also examined: values of 10 or greater indicate a better model fit. After the latent profiles were identified, covariates were entered into the model to conduct a multinomial logistic regression analysis using the three-step approach (Asparouhov & Muthén, 2014) (Research Question 3).

Research question 4: Predicting rape proclivity

To test differences in membership in the latent profiles of RMs on rape proclivity, an outcome, rape proclivity, was entered into the model with the recommended Bolck et al. (2004) BCH (Bolck, Croon, and Hagenaars) method (Asparouhov & Muthén, 2018). The BCH method estimates the latent profile structure before including outcomes as is current best practice and accounts for the uncertainty in profile membership (Nylund-Gibson & Masyn, 2016). In addition, the BCH method is recommended for continuous outcomes and may help prevent biased parameter estimates (Bakk & Vermunt, 2016). The results of the outcome analysis demonstrate differences in mean rape proclivity across profiles using a Wald test while controlling for any significant covariates from the prior model.

Results

Research Questions 1 and 2: Latent Profiles

Table 2 presents the fit indices for the solutions fitted for a non-diagonal (unrestricted)/profile-varying model. The fit criteria did not indicate a clear best model instead, several different models were suggested as a good fit to the data. For the AIC, BIC, and CAIC, the values fluctuated, first decreasing and then increasing slightly and then decreasing again. The points at which the values first decreased before increasing again are bolded in Table 2. Line graphs of the values were also examined for an “elbow” (or point at which the values level off with decreasing gains) from adding additional profiles (Masyn, 2013). No clear pattern appeared from the examination of these graphs.

Table 2.

Fit Indices for Latent Profiles.

Model (K-profiles)	LL	n-par	AIC	BIC	CAIC	AWE	Adj LMRT p value	BF	Entropy
1	−3089.72	10	6199.43	6241.80	6251.80	6334.16	.016	<0.01
2	−2962.73	16	5957.46	6025.24	6041.24	6173.03	.013	<0.01	0.738
3	−2891.66	22	5827.31	5920.51	5942.51	6123.71	.097	<0.01	0.844
4	−2592.65	28	5241.31	5357.52	5875.95	6106.57	.119	<0.01	0.793
5	−2804.36	34	5676.73	5820.76	5854.76	6134.80	.234	<0.01	0.807
6	−2773.42	40	5626.84	5796.29	5836.29	6165.75	.020	<0.01	0.852
7	−2735.25	46	5562.50	5757.38	5803.38	6182.25	.931	<0.01	0.913
8	−2724.68	52	5553.35	5773.64	5825.64	6253.93	.457	>0.10	0.909
9	−2714.85	58	5545.69	5791.40	5849.40	6327.11	.2056	<0.01	0.878
10	Not well-identified

Note. Bolded values indicate the best value with the fewest number of profiles. LL = log-likelihood; AIC = Akaike’s information criterion; AWE = approximate weight of evidence; BF = Bayes factor; BIC = Bayesian information criterion; CAIC = consistent Akaike’s information criterion; LMRT = Lo–Mendell–Rubin likelihood ratio test.

Additional indicators of model fit also did not point to a single best model. The adjusted LMR p value reached non-significance at a three-profile model, indicating that any model of three or more profiles fit the data better than prior models. The BLRT reached a value of 1 at the six-profile model. The BF value was 10 for both the four- and six-profile solutions. Finally, the cmP value never reached above a value of 1 but, in the six-profile model, the value increased closer to one compared to prior models. The final factor in determining the best model was profile size and interpretation. Profiles with a small percentage of the sample may indicate profile over-extraction and model breakdown (Masyn, 2013). For all models with five or more profiles, the percentage of the sample in some of the profiles decreased drastically, perhaps indicating profile over-extraction. A four-profile model was selected based on the BIC and CAIC values (Table 2) and because only a single profile had a smaller percentage (3%) of the sample.

Latent profile characteristics

Each of the four profiles of the selected model are distinguished by differing mean scores on the five RM subscales (the highest mean scores for each subscale are bolded; Table 3). In addition, Table 3 shows the average posterior profile probabilities for most likely profile membership by profile. Values bolded on the diagonal are all above .8, indicating adequate separation and classification precision (Nagin & Tremblay, 2005). The findings below address Research Questions 1 and 2, demonstrating heterogeneity across the sample as students meaningfully differ on the types of RMs subscales they endorse.

Table 3.

Four-Profile Model Profile Prevalence, Size, Results (Means and Standard Deviations) and Classification Probabilities.

	Profile 1: Low RMs		Profile 2: Medium RMs		Profile 3: Elevated RMs		Profile 4: Mid/High RMs
Profile prevalence	15%		57%		3%		25%
Estimated Profile sizes	68		269		13		119
She asked for it	2.04	(0.12)	2.96	(0.09)	3.40	(0.32)	3.26	(0.09)
Was not rape	1.16	(0.03)	1.41	(0.07)	4.42	(0.19)	2.50	(0.11)
Intoxicated, didn’t mean to	1.35	(0.07)	1.91	(0.1)	3.88	(0.22)	2.51	(0.08)
Didn’t mean to	2.37	(0.18)	3.18	(0.08)	3.18	(0.27)	3.35	(0.09)
She lied	2.23	(0.20)	3.06	(0.05)	3.28	(0.31)	3.44	(0.07)
Profile 1	0.83		0.17		0.00		0.00
Profile 2	0.07		0.87		0.00		0.06
Profile 3	0.00		0.00		0.98		0.03
Profile 4	0.00		0.12		0.00		0.88

Note. Estimated means and standard deviations on each subscale for each profile are provided. Average posterior class probabilities for most likely profile membership by profile are provided. RM = rape myth.

Profile 1 (low RMs; 15% of the sample; n = 68)

This profile is made up of men who have the lowest mean scores on all the five RM subscales across all four profiles, meaning low RMA. Students with this profile have lower mean scores, compared to other profiles, on every subscale of the RMs, and all mean scores are below the midpoint of the scale.

Profile 2 (medium RMs; 57% of the sample; n = 269)

Participants with this profile had mid-range mean scores on the RM subscales when compared to the other profiles. All mean scores were higher than those in Profile 1, Low RMs, and either less than or equal to those in Profile 3, Elevated RMs and Profile 4, Mid/High RMs. This was the most common participant profile.

Profile 3 (elevated RMs; 3% of the sample; n = 13)

Students with this profile had mean scores on subscales that were between Profile 2 and Profile 4 for two of the subscales and overall higher levels of RMs. Importantly, participants in this profile had much higher scores, compared to all other profiles, on two RM subscales: RMA 2, Was Not Really Rape and RMA 3, Intoxicated, Didn’t Mean To, making these the distinguishing beliefs for this profile, as well as high scores on RMA 1, She Asked for It. Although the rarest profile of participants, participants with this profile had the highest subscale mean scores for specific RMs. This is particularly true for the mean score for RMA 2, Was Not Really Rape (M = 4.42), near the endpoint on this scale, indicating an elevated level of endorsement.

Profile 4 (mid/high RMs; 25% of the sample; n = 119)

Participants with this profile had the highest scores on two RMs subscales (Didn’t Mean To and She Lied). For the other RM subscales (She Asked for It; Was Not Really Rape and Intoxicated, Didn’t Mean To), only Profile 3 had higher scores.

Research Question 3: Latent Profile Covariates

The third research question asked if profile membership for the subgroups was related to demographic factors including membership in all-male peer groups. Table 4 presents the odds ratios (ORs) for the covariates. Compared to Profile 1, low RMs, white students were less likely to be in every other profile: Profile 2, medium RMs (OR = .25, p < .01), Profile 3, elevated RMs (OR = .19, p < .05), and Profile 4, mid/high RMs (OR = .27, p < .01). Men who had been in student council in high school were less likely to be in Profile 2, medium RMs (OR = .32, p < .05) and Profile 4, mid/high RMs (OR = .26, p < .01) than in Profile 1, low RMs. Those who had been in varsity athletics in high school were more likely to be in Profile 4, mid/high RMs, (OR = 2.07, p = .05) than in Profile 2, medium RMs. Bystander attitudes were reduced within profiles with high RMA. For example, compared to Profile 1, low RMs, those in Profile 2, medium RMs (OR = .37, p < .05), Profile 3, elevated RMs (OR = .29, p < .05), and Profile 4, mid/high RMs (OR = .32, p < .05) had lower bystander attitudes than those in Profile 1. Intention to join a fraternity, an athletic team, or student government were not related to membership in the profiles (p > .1).

Table 4.

Odds Ratios for Covariates.

	Compared to Profile 1			Compared to Profile 2			Compared to Profile 3			Compared to Profile 4
	Profile 2	Profile 3	Profile 4	Profile 1	Profile 3	Profile 4	Profile 1	Profile 2	Profile 4	Profile 1	Profile 2	Profile 3
White students	0.25**	0.19*	0.27**	3.94**	0.74	1.06	5.30*	1.34	1.43	3.70**	0.94	0.70
Student council	0.32*	0.20	0.26*	3.08*	0.63	0.80	4.90	1.59	1.28	3.84*	1.25	0.78
Varsity athlete	0.89	1.98	1.86	1.12	2.21	2.07⁺	0.51	0.45	0.94	0.54	0.48⁺	1.07
Intention to join fraternity	2.65	2.23	1.96	0.38	0.84	0.74	0.45	1.19	0.88	0.51	1.35	1.14
Intention to join athletics	1.29	0.53	1.06	0.78	0.41	0.82	1.88	2.42	1.99	0.95	1.22	0.50
Intention to join student government	1.76	0.92	1.21	0.57	0.52	0.69	1.09	1.93	1.32	0.83	1.46	0.76
Bystander attitudes	0.37*	0.29*	0.32*	2.68*	0.77	0.84	3.46*	1.29	1.09	3.17*	1.19	0.92

Note. Profile 1: low RMs; Profile 2: medium RMs; Profile 3: elevated RMs; Profile 4: mid/high RMs. ORs = odds ratio; RM = rape myth.

p = .05. *p < .05. **p < .01.

Research Question 4: Predicting Rape Proclivity

The final research question was addressed by examining if membership in the subgroups, based on men’s RMA, differed in relation to rape proclivity as an outcome. All subgroups were compared to each other. Men within some subgroups with higher levels of RMA, those in Profiles 3 and 4, had significantly higher rape proclivity levels compared to men in Profile 1, low RMs. Specifically, compared to men in Profile 1, low RMs (M = 1.14, standard error [SE] = .06), men in Profile 3, high specific RMs (M = 1.97, SE = .34) had higher mean rape proclivity scores (Wald X² = 34.59, df = 1, p = .000) as did men in Profile 4, mid/high RMs (M = 1.52, SE = .08), (Wald X² = 4.28, df = 1, p = .039). However, men in Profile 2, medium RMs (M = 1.31, SE = .05) did not have significantly higher rape proclivity scores (Wald X² = 0.57, df = 1, p = .450). Finally, those with the highest rape proclivity scores, men in Profile 3, high specific RMs had significantly higher rape proclivity than those in Profile 4, mid/high RMs (Wald X² = 15.47, df = 1, p = .000) and those in Profile 2, medium RMs (Wald X² = 35.35, df = 1, p = .000).

Discussion

The results of this study demonstrate heterogeneity among male college students on RMA, addressing Research Question 1 that RMA will vary from person to person on the five types of RMs. Specifically, four profiles of RMs were identified with varying levels of RMA. Representing 15% of the sample, men in Profile 1, low RMs had the lowest level of RMA. Over half of all the participants fell into Profile 2, medium RMs, and tended to have moderate levels of RMA. Participants in Profile 3, elevated RMs were rare but had extreme levels of RM beliefs on two of the five RMs subscales (Was Not Really Rape and Intoxicated, Didn’t Mean To). The final profile, Profile 4, mid/high RMs, generally had mid to high levels of RMA and one in four men were in this profile.

Attending to Research Question 2, this study found that some men endorse RMs at higher rates than others: some students endorse few, or none, of the RM subscales while others will endorse most or all of the RM subscales. Examination of Profile 3, elevated RMs, and Profile 4, mid/high RMs, suggests that some subsets of college men have elevated and problematic levels of RMA. Students within Profiles 3 and 4 had the highest rates of RMA. Compare this to the men in Profile 1, low RMs, who scored below these profiles on all subscales of RMA. While those in Profile 4, mid/high RMs, had overall mid to higher rates of RM subscale endorsement, men in Profile 3, elevated RMs, were notable for their elevated, and perhaps problematic levels of endorsement on two subscales: Was Not Really Rape and Intoxicated, Didn’t Mean To, indicating a concerning pattern with men in this subgroup in terms of these particular RMs. Interestingly, proportionally, this profile was very small indicating that while men in this subgroup have problematically high levels of Was Not Really Rape and Intoxicated, Didn’t Mean To, these individuals are not among the majority of the population. Notably, the profile with the largest percentage of the sample was Profile 2, medium RMs. Combining those in profiles 1 and 2 demonstrates that over two in three of the men within this sample, while still endorsing RMs, do so at lower rates and thus, might not need as intensive interventions as other subgroups to address RMA.

Addressing Research Question 3, the results were mixed. Profiles of RMA did not differ on intention to participate in athletics, a fraternity, or student government. However, other demographic factors were related to membership within the profiles. White students were less likely than non-white students to be in profiles with higher RMA (profiles 2–4); those who had been in student council in high school were also less likely to be in profiles with higher RMA (profiles 2 and 4) but not less likely to be in profile 3 with elevated RMs. In addition, bystander attitudes acted as a protective factor as men in profiles with higher RMA had lower bystander attitudes, less intention to intervene, compared to men in Profile 1, low RMs.

The findings from this study do not support the idea that subgroups of men are differentially associated with intention to join a fraternity or an athletic team. While some research indicates that members of all male-peer groups are more likely to endorse RMs and/or perpetrate SV (Kingree & Thompson, 2013; McCray, 2015; Seabrook et al., 2018), there have been contrary findings in this area. Some studies suggest that fraternity men do not differ from other men in sexual coercion (Thompson & Morrison, 2013), and a meta-analysis found no association between fraternity status and RMA (Suarez & Gadalla, 2010). Likewise, research on athlete status and SV perpetration is mixed with several research studies finding no association between athletic team membership and sexual aggression perpetration (Boeringer, 1996; Thompson & Morrison, 2013).

The findings from the current study indicate that men within subgroups with higher levels of RMA endorse rape proclivity at higher rates than men with lower levels of RMA. Compared to men with the lowest levels of RMs, Profile 1: low RMs, men within the two of the three (Profiles 3 and 4) subgroups of men with higher levels of RMA had higher rape proclivity. In addition, men in Profile 3, elevated RMs, had significantly higher rape proclivity when compared to all other profiles, indicating the importance of this subgroup. These findings affirm Research Question 4: men’s membership within the subgroups of men, based on their RMA, is differentially associated with rape proclivity.

While the relationship between RMA and rape proclivity has been examined previously (Bohner et al., 2010; O’Connor, 2020; Seabrook et al., 2018), this is the first study to investigate men’s membership within subgroups with differing levels of RMA in relation to rape proclivity. These findings suggest that not only is RMA associated with rape proclivity, as has been found previously, but also that some subgroups of men who endorse higher levels of some or all RMs have higher rape proclivity compared to the subgroup of men with the lowest levels of RMA.

In addition to the relationship between men’s membership in subgroups with differing levels of RMA, this study also sheds light on an important covariate related to RMA and rape proclivity: bystander attitudes. Past research indicated that bystander attitudes, which are negatively associated with RMA, might serve as a protective factor (Fleming & Wiersma-Mosley, 2015; McMahon, 2010; Orchowski et al., 2016; Powers et al., 2015). The findings from the current study indicate that bystander attitudes are lower in subgroups of men with higher RMA and may help fill the gap surrounding protective factors related to perpetration of SV. Also, students who participated in high school student council, but not those who intend to do so in college, were less likely to be in profiles with higher RMA, supporting research indicating that positive peer networks can help buffer the risk of SV perpetration (Valido et al., 2022). In addition, race/ethnicity was associated with being in profiles with higher RMA, similar to some, but not all, other research studies (for a review, see O’Connor & McMahon, 2022). This may indicate important considerations regarding the legacy of racism or important differences in the measurement of RMs which have not been accounted for in scale development.

Limitations

While this study examines important factors related to subgroups of men based on their RMA, there are limitations. First, sparseness is a concern in LPA when one or more of the profiles has a small sample size, such as Profile 3 (n = 13) in this paper. Sparseness refers to the cell counts of each latent profile by each covariate. To the extent that the cell counts are small, it can affect regression analysis estimations within LPAs (Collins & Lanza, 2010) and can cause “small sample bias” or “spare bias” even within large samples when conducting regression (Greenland et al., 2016). However, Collins and Lanza (2010) suggest that such problems are likely to affect results when the estimation of the model fails, and no results are obtainable. In this case, although the data are sparse within one profile, the model estimation did not fail suggesting that sparseness did not affect the model estimation in a meaningful way. Also, the rape proclivity measure used within this study was based on only two items averaged together. Other methods of measuring rape proclivity using scenarios have been used in past research (Bohner et al., 2010) and might be better at capturing rape proclivity. In addition, the similarity of some of the profiles might appear to suggest that RMs exist on a continuous scale. However, a one-profile solution, indicating homogeneity, was tested and rejected, indicating that the profiles ultimately differ enough to warrant attention. This is particularly true for Profile 3, a subset of men with elevated scores on two subscales. In addition, some scholars assert that RMs have become more subtle over time (McMahon & Farmer, 2011; O’Connor et al., 2018), with decreasing endorsement of the traditional RM measures. Given this, the differences between the groups may appear small but are meaningful especially when considering the finding regarding covariates and rape proclivity. Finally, some of the RM subscales had low reliability which may have affected how these measures performed. Likewise, the three covariates that measured intention to join a group (fraternity, athletic team, and student government) only measured intention, not actual behaviors, which may explain the lack of significant associations with these variables.

Implications and Future Research

This study’s findings suggest several possible implications for interventions to address campus SV perpetration with the end goal of reducing rates of SV among college students. The first point of intervention involves tailoring prevention efforts to men with higher RMA levels. Among the prevention programming efforts that have been implemented aiming to reduce the perpetration of SV, none, including bystander interventions, are effective among men at high risk for SV in reducing outcomes such as rape-supportive beliefs (Malamuth et al., 2018).

Given the failure of many prevention programs to reduce SV perpetration rates (DeGue et al., 2014), the results of the current study may provide one possible avenue for improving prevention efforts through tailoring prevention programs for men who enter college with differing pre-existing beliefs and attitudes related to SV perpetration. If men within those higher RMA subgroups (Profiles 3 and 4) receive appropriate education to reduce RMs, such programming should also reduce rape proclivity. Men come into college with varying levels of RMA: some subgroups with lower levels of RMA and others with higher levels on some or all of the RMs. Likewise, those subgroups with higher levels of RMs also have higher rape proclivity. As such, tailoring prevention programming efforts for men with higher levels of RMA may be a more cost-effective use of programming; in addition, programming might be tailored toward reducing specific RMs endorsed within specific subgroups. Profile 3 with its higher endorsement of Was Not Really Rape and Intoxicated, Didn’t Mean To might be targeted with messages to target these myths in particular. For example, Orchowski et al. (2018) found that among heavy-drinking college men, an intervention to reduce SV perpetration was associated with lower RMA overall, but not among men who perpetrated SV after the intervention. Given the heavy alcohol use among these men, perhaps individual RMs, including those related to alcohol use, should be targeted to reduce endorsement of certain types of RMs. Ultimately, such efforts will work to reduce beliefs and attitudes related to perpetration, such as RMA and rape proclivity, with reducing SV rates being the ultimate end goal. Studies examining tailored prevention programming for men at high risk of perpetration are rare (Bosson et al., 2015; Stephens & George, 2004, 2009). The findings from this current study might inform future interventions aimed at reducing RMA, specifically within subgroups of men with higher RMA, to see whether these types of interventions also reduce SV perpetration and rape proclivity.

Importantly, some scholars have suggested and found that traditional prevention programming does not work with high-risk men (Stephens & George, 2004) or even has a “backlash effect” or “boomerang effect” increasing sexual aggression among high-risk groups perhaps as a result of resistance to prevention interventions (Bosson et al., 2015; Malamuth et al., 2018; Stephens & George, 2009). In addition, high-risk men may surround themselves with like-minded peers. For example, within a single fraternity, men may hold rape-supportive attitudes and reinforce these beliefs with one another; thus, increasing these types of beliefs in the group as a whole, making such groups more resistant to prevention programming. One of these studies (Bosson et al., 2015) demonstrated that men with high hostile sexism increased sexually aggressive behavior after a prevention intervention, while men with lower hostile sexism did not demonstrate this increase. These findings indicate that uniform prevention programming, not tailored to high-risk groups, can be ineffective and even harmful if these types of programs increase SV perpetration. Perhaps tailoring programming to students’ pre-existing beliefs will help lower resistance among high-risk groups and increase receptiveness. More research is needed to examine tailored interventions and high-risk groups.

Another implication of this study stems from the findings regarding covariates associated with RMA and rape proclivity. As few studies focus on protective factors related to SV perpetration (Tharp et al., 2012), this study suggests a further reason to implement bystander programming beyond the existing evidence suggesting that bystander interventions reduce SV rates (Mujal et al., 2021). Only a few studies have examined the relationship between bystander attitudes and RMs (Fleming & Wiersma-Mosley, 2015; McMahon, 2010; Orchowski et al., 2016; Powers et al., 2015), and no known studies have examined the relationship between RMA and rape proclivity using person-centered methods. This study indicates that programs aimed at increasing bystander interventions might also serve as a protective factor for RMA among some subgroups of men. Future research might both investigate the relationship between RMs and bystander attitudes and also the relationship between rape proclivity and bystander attitudes.

Conclusions

The results of this study can be used to understand men at high risk of perpetrating SV, those with either moderate or high levels of RMA, and factors that might serve as protective or risk factors to tailor prevention efforts to reduce SV perpetration with these higher risk men. In this way, prevention interventions might target men with higher levels of RMs to decrease attitudes such as RMs and rape proclivity to ultimately lower SV perpetration rates. Similarly, prevention efforts might build upon existing bystander intervention efforts with the specific aim of increasing bystander attitudes coupled with decreasing RMs and rape proclivity.

Footnotes

Acknowledgements

The author would like to thank Sarah McMahon for her invaluable feedback and assistance with this manuscript. Judy Postmus, N. Andrew Peterson, and Lindsay Orchowski also provided feedback on previous versions of this manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interests with respect to the authorship and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research and/or authorship of this article: Preparation of this article was supported by a Grant from the Centers for Disease Control and Prevention (Grant 1R01CE001855-01, principal investigator: Sarah McMahon, coprincipal investigator: Judy Postmus). Its contents are solely the responsibility of the author and do not necessarily represent the official views of the Centers for Disease Control and Prevention.

ORCID iD

Julia O’Connor

Author Biography

Julia O’Connor, PhD, MSW, MPH, is an assistant professor at the University of Central Florida in the School of Social Work and a member of the Violence Against Women faculty cluster. Her research interests include violence against women focused on primary prevention of violence, those who perpetrate violence, and marginalized populations. Her practice experience includes many years as a domestic and sexual violence advocate.

References

Asparouhov

Muthén

(2014). Auxiliary variables in mixture modeling: Three-step approaches using Mplus. Structural Equation Modeling: A Multidisciplinary Journal, 21(3), 329–341. https://doi.org/10.1080/10705511.2014.915181

Asparouhov

Muthén

B. O.

(2018). Auxiliary variables in mixture modeling: Using the BCH method in Mplus to estimate a distal outcome model and an arbitrary second model [Webnote No. 21, Version 3]. https://www.statmodel.com/examples/webnotes

Bakk

Vermunt

J. K.

(2016). Robustness of stepwise latent class modeling with continuous distal outcomes. Structural Equation Modeling: A Multidisciplinary Journal, 23(1), 20–31. https://doi.org/10.1080/10705511.2014.955104

Boeringer

S. B.

(1996). Influences of fraternity membership, athletics, and male living arrangements on sexual aggression. Violence Against Women, 2(2), 134–147. https://doi.org/10.1177/1077801296002002002

Bohner

Pina

Viki

G. T.

Siebler

(2010). Using social norms to reduce men’s rape proclivity: Perceived rape myth acceptance of out-groups may be more influential than that of in-groups. Psychology, Crime & Law, 16(8), 671–693.

Bolck

Croon

Hagenaars

(2004). Estimating latent structure models with categorical variables: One-step versus three-step estimators. Political Analysis, 12(1), 3–27.

Bosson

J. K.

Parrott

D. J.

Swan

S. C.

Kuchynka

S. L.

Schramm

A. T.

(2015). A dangerous boomerang: Injunctive norms, hostile sexist attitudes, and male-to-female sexual aggression. Aggressive Behavior, 41(6), 580–593. https://doi.org/10.1002/ab.21597

Boswell

A. A.

Spade

J. Z.

(1996). Fraternities and collegiate rape culture: Why are some fraternities more dangerous places for women? Gender and Society, 10(2), 133–147.

Brownmiller

(1975). Against our will: Men, women, and rape. Simon & Schuster.

10.

Burt

M. R.

(1980). Cultural myths and supports for rape. Journal of Personality and Social Psychology, 38(2), 217–230. https://doi.org/10.1037/0022-3514.38.2.217

11.

Burt

M. R.

(1980). Cultural myths and acquaintance rape. In Parrot

Bechhofer

(Eds.), Acquaintance rape: The hidden crime. Wiley.

12.

Carroll

M. H.

Rosenstein

J. E.

Foubert

J. D.

Clark

M. D.

Korenman

L. M.

(2016). Rape myth acceptance: A comparison of military service academy and civilian fraternity and sorority students. Military Psychology, 28(5), 306–317. https://doi.org/10.1037/mil0000113

13.

Collins

L. M.

Lanza

S. T.

(2010). Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. Wiley.

14.

Dahlberg

L. L.

Krug

E. G.

(2002). Violence-a global public health problem. In Krug

E. G.

Dahlberg

L. L.

Mercy

J. A.

Zwi

A. B.

Lozano

(Eds.), World report on violence and health (pp. 1–56). World Health Organization.

15.

DeGue

Valle

L. A.

Holt

M. K.

Massetti

G. M.

Matjasko

J. L.

Tharp

A. T.

(2014). A systematic review of primary prevention strategies for sexual violence perpetration. Aggression and Violent Behavior, 19(4), 346–362. https://doi.org/10.1016/j.avb.2014.05.004

16.

Fleming

W. M.

Wiersma-Mosley

J. D.

(2015). The role of alcohol consumption patterns and pro-social bystander interventions in contexts of gender violence. Violence Against Women, 21(10), 1259–1283. https://doi.org/10.1177/1077801215592721

17.

Greenland

Mansournia

M. A.

Altman

D. G.

(2016). Sparse data bias: A problem hiding in plain sight. Bmj, 352, i1981. https://doi.org/10.1136/bmj.i1981

18.

Kingree

J. B.

Thompson

M. P.

(2013). Fraternity membership and sexual aggression: An examination of mediators of the association. Journal of American College Health, 61(4), 213–221. https://doi.org/10.1080/07448481.2013.781026

19.

Lisak

Conklin

Hopper

Miller

Altschuler

Smith

B. M.

(2000). The abuse-perpetration inventory: Development of an assessment instrument for research on the cycle of violence. Family Violence & Sexual Assault Bulletin, 16(1–2), 21–30.

20.

Lonsway

K. A.

Fitzgerald

L. F.

(1994). Rape myths: In review. Psychology of Women Quarterly, 18(2), 133–164. https://doi.org/10.1111/j.1471-6402.1994.tb00448.x

21.

Malamuth

N. M.

(1981). Rape proclivity among males. Journal of Social Issues, 37(4), 138–157.

22.

Malamuth

N. M.

(1989). The attraction to sexual aggression scale: Part one. The Journal of Sex Research, 26(1), 26–49. https://doi.org/10.1080/00224498909551491

23.

Malamuth

N. M.

Huppin

Linz

(2018). Sexual assault interventions may be doing more harm than good with high-risk males. Aggression and Violent Behavior, 41, 20–24. https://doi.org/10.1016/j.avb.2018.05.010

24.

Masyn

K. E.

(2013). Latent class analysis and finite mixture modeling. In Little

T. D.

(Ed.), The oxford handbook of quantitative methods (Vol. 2, pp. 551–611). Oxford University Press.

25.

McCray

K. L.

(2015). Intercollegiate athletes and sexual violence: A review of literature and recommendations for future study. Trauma Violence Abuse, 16(4), 438–443. https://doi.org/10.1177/1524838014537907

26.

McMahon

(2010). Rape myth beliefs and bystander attitudes among incoming college students. Journal of American College Health, 59(1), 3–11. https://doi.org/10.1080/07448481.2010.483715

27.

McMahon

(2015). Participation in high school sports and bystander intentions, efficacy to intervene, and rape myth beliefs. Journal of Interpersonal Violence, 30(17), 2980–2998. https://doi.org/10.1177/0886260514555009

28.

McMahon

Allen

C. T.

Postmus

J. L.

McMahon

S. M.

Peterson

N. A.

Lowe Hoffman, M. (2014). Measuring bystander attitudes and behavior to prevent sexual violence. Journal of American College Health, 62(1), 58–66. https://doi.org/10.1080/07448481.2013.849258

29.

McMahon

Farmer

G. L.

(2011). An updated measure for assessing subtle rape myths. Social Work Research, 35(2), 71–80.

30.

McMahon

Winter

S. C.

Palmer

J. E.

Postmus

J. L.

Peterson

N. A.

Zucker

Koenick

(2015). A randomized controlled trial of a multi-dose bystander intervention program using peer education theater. Health Education Research, 30(4), 554–568. https://doi.org/10.1093/her/cyv022

31.

Mouilso

E. R.

Calhoun

K. S.

(2013). The role of rape myth acceptance and psychopathy in sexual assault perpetration. Journal of Aggression, Maltreatment & Trauma, 22(2), 159–174. https://doi.org/10.1080/10926771.2013.743937

32.

Mujal

G. N.

Taylor

M. E.

Fry

J. L.

Gochez-Kerr

T. H.

Weaver

N. L.

(2021). A systematic review of bystander interventions for the prevention of sexual violence. Trauma Violence Abuse, 22(2), 381–396. https://doi.org/10.1177/1524838019849587

33.

Murnen

S. K.

Kohlman

M. H.

(2007). Athletic participation, fraternity membership, and sexual aggression among college men: A meta-analytic review. Sex Roles, 57(1–2), 145–157. https://doi.org/10.1007/s11199-007-9225-1

34.

Muthen

L. K.

Muthen

B. O.

(2017). Mplus user’s guide (8th ed.). Muthen & Muthen.

35.

Nagin

D. S.

Tremblay

R. E.

(2005). Developmental trajectory groups: Fact or a useful statistical fiction? Criminology: An Interdisciplinary Journal, 43(4), 873–904. https://doi.org/10.1111/j.1745-9125.2005.00026.x

36.

Nylund-Gibson

Masyn

E. K.

(2016). Covariates and mixture modeling: Results of a simulation study exploring the impact of misspecified effects on class enumeration. Structural Equation Modeling: A Multidisciplinary Journal, 23(6), 782–797. https://doi.org/10.1080/10705511.2016.1221313

37.

O’Connor

(2020). The longitudinal effects of rape myth beliefs and rape proclivity. Psychology of Men & Masculinities, 22(2), 321–330. https://doi.org/10.1037/men0000324

38.

O’Connor

Cusano

McMahon

Draper

(2018). Students’ articulation of subtle rape myths surrounding campus sexual assault. Journal of College Student Development, 59(4), 439–455.

39.

O’Connor

McMahon

(2022). The role of rape myths in contributing to sexual aggression: Theory, research and implications for prevention. In Orchowski

L. M.

Berkowitz

A. D.

(Eds.), Engaging boys and men in sexual assault prevention: Theory, research, and practice (pp. 1–27). Oxford University Press.

40.

Orchowski

L. M.

Barnett

N. P.

Berkowitz

Borsari

Oesterle

Zlotnick

(2018). Sexual assault prevention for heavy drinking college men: Development and feasibility of an integrated approach. Violence Against Women, 24(11), 1369–1396.

41.

Orchowski

L. M.

Berkowitz

Boggis

Oesterle

(2016). Bystander intervention among college men: The role of alcohol and correlates of sexual aggression. Journal of Interpersonal Violence, 31(17), 2824–2846. https://doi.org/10.1177/0886260515581904

42.

Powers

R. A.

Leili

Hagman

Cohn

(2015). The impact of college education on rape myth acceptance, alcohol expectancies, and bystander attitudes. Deviant Behavior, 36(12), 956–973. https://doi.org/10.1080/01639625.2014.982747

43.

Russell

T. D.

King

A. R.

(2016). Anxious, hostile, and sadistic: Maternal attachment and everyday sadism predict hostile masculine beliefs and male sexual violence. Personality and Individual Differences, 99, 340–345. https://doi.org/10.1016/j.paid.2016.05.029

44.

Schwartz

M. D.

DeKeseredy

W. S.

(1997). Sexual assault on the college campus: The role of male peer support. Sage.

45.

Seabrook

R. C.

McMahon

O’Connor

(2018). A longitudinal study of interest and membership in a fraternity, rape myth acceptance, and proclivity to perpetrate sexual assault. Journal of American College Health, 66(6), 510–518. https://doi.org/10.1080/07448481.2018.1440584

46.

Stephens

K. A.

George

W. H.

(2004). Effects of anti-rape video content on sexually coercive and noncoercive college men’s attitudes and alcohol expectancies. Journal of Applied Social Psychology, 34(2), 402–416.

47.

Stephens

K. A.

George

W. H.

(2009). Rape prevention with college men: Evaluating risk status. Journal of Interpersonal Violence, 24(6), 996–1013. https://doi.org/10.1177/0886260508319366

48.

Suarez

Gadalla

T. M.

(2010). Stop blaming the victim: A meta-analysis on rape myths. Journal of Interpersonal Violence, 25(11), 2010–2035.

49.

Tharp

A. T.

DeGue

Valle

L. A.

Brookmeyer

K. A.

Massetti

G. M.

Matjasko

J. L.

(2012). A systematic qualitative review of risk and protective factors for sexual violence perpetration. Trauma, Violence, & Abuse, 14(2), 133–167. https://doi.org/10.1177/1524838012470031

50.

Thompson

M. P.

Morrison

D. J.

(2013). Prospective predictors of technology-based sexual coercion by college males. Psychology of Violence, 3(3), 233–246.

51.

Untied

A. S.

Orchowski

L. M.

Lazar

(2013). College men’s and women’s respective perceptions of risk to perpetrate or experience sexual assault: The role of alcohol use and expectancies. Violence Against Women, 19(7), 903–923.

52.

Valido

Rivas-Koehl

Espelage

D. L.

Lawrence

T. I.

Robinson

L. E.

(2022). Latent class analysis of victimization patterns and associated protective factors among LGBTQ youth. International Journal of Environmental Research and Public Health, 19(16), 9953.

53.

Welsh

B. C.

Farrington

D. P.

(2012). Science, politics, and crime prevention: Toward a new crime policy. Journal of Criminal Justice, 40(2), 128–133. https://doi.org/10.1016/j.jcrimjus.2012.01.008

54.

Yapp

E. J.

Quayle

(2018). A systematic review of the association between rape myth acceptance and male-on-female sexual violence. Aggression and Violent Behavior, 41, 1–19. https://doi.org/10.1016/j.avb.2018.05.002

55.

Zinzow

H. M.

Thompson

(2015). Factors associated with use of verbally coercive, incapacitated, and forcible sexual assault tactics in a longitudinal study of college men. Aggressive Behavior, 41(1), 34–43. https://doi.org/10.1002/AB.21567