Barriers to sexual harassment bystander intervention in Ecuadorian universities

ABSTRACT

Previous research, mainly in the United States, has identified several barriers to acting as a bystander in sexual harassment at university campuses. Despite the high frequency of harassment in Latin America, there is a dearth of studies investigating barriers to bystander behaviour in this context. In this pilot study, we report findings exploring harassment and bystander behaviour in university staff and students in Ecuador, a Latin American country characterised by masculine social norms and high levels of gender-based harassment. In an on-line survey, 129 staff and students from universities in different regions of Ecuador answered questions about perceptions of seriousness of harassment, rape myth acceptance, actual incidences of being a perpetrator, victim, or a bystander, and the likelihood and difficulties of bystander action. Women and those who scored higher in rape myth acceptance reported more intervention difficulties. In addition, women and those who had previously perpetrated harassment rated their likelihood of intervening lower. Finally, perceptions of harassment as a serious problem in campuses related to a higher likelihood of intervening as a bystander. We discuss the results in terms of practical applications in devising culturally appropriate bystander intervention workshops.

KEYWORDS:

• Sexual harassment
• bystander intervention barriers
• rape myth acceptance
• Ecuador
• Latin America

Introduction

Gender-based harassment in universities is a serious global problem and has been identified as the ‘single most widespread educational hazard’ (Rudman et al., 1995, p. 519), affecting approximately one in four female students (Bondestam & Lundqvist, 2020). Harassment in higher education takes different forms, ranging from sexualised comments to blackmail and rape (Bondestam & Lundqvist, 2020; Fedina et al., 2018), with serious negative mental health consequences for the victims (McGinley et al., 2016). Although the prevalence rates and characteristics of victims and perpetrators vary from one study to another, it is safe to say that women (intersecting with sexual orientation, age, race, disability, and social class) are at the highest risk of being harassed (Bondestam & Lundqvist, 2020; Fedina et al., 2018). To address the problem, many institutions have adopted bystander intervention workshops as a primary prevention strategy, based on the premise that harassment is often witnessed by others who have the power to intervene before, during, or after the event (e.g. Labhardt et al., 2017). To date, most of the work has been conducted in North America with a paucity of data from other regions of the world. Of importance, the barriers and enablers may be different in other parts of the world based on the culture of gender violence.

In order to develop culturally sensitive bystander intervention workshops, we need to know more about the context in which harassment and bystander behaviour occurs (Banyard, 2011). Each culture has different norms around helping behaviours, masculinity, and the acceptability of harassment, which together influence the likelihood of being an active bystander (e.g. Kamimura et al., 2016; Zietz & Das, 2018). We currently have very little knowledge of bystander behaviours and attitudes in Latin American cultures. These cultures are characterised by relatively inflexible gender norms, strong patriarchal orientations, and high levels of violence towards women and girls (Janos & Espinosa, 2015; Maldonado-Maldonado & Acosta, 2018; UN women, 2013). Indeed, masculine gender norms in the Latin American region may perpetuate gender inequality, and ‘downplay the duty of bystanders to intervene in situations of abuse, with wide variations both among and within individual countries’ (Bott et al., 2012, p. 114). In order to develop culturally appropriate bystander intervention workshops, it is important to investigate bystander barriers that individuals in Latin American countries may face.

Masculine gender norms could be a barrier for bystander action for several reasons. For example, endorsement of masculine gender norms relates to the acceptance of several inaccurate myths around rape (e.g. blaming the victim rather than the perpetrator), which could impede bystander intervention (Labhardt et al., 2017; Leone et al., 2020; Martini & De Piccoli, 2020). Those who endorse rape myths may fail to take responsibility for intervening (Yule et al., 2020), and downplay the seriousness of sexual assault (LeMaire et al., 2016). There is a notable lack of research investigating rape myth acceptance outside the U.S (Labhardt et al., 2017; Yapp & Quayle, 2018), with a paucity of studies from Latin America too (although see Hernández et al., 2015; Janos & Espinosa, 2015). Moreover, we could not find any studies focusing on the relationships between rape myths and bystander behaviour in Latin America. As rape myth acceptance is one of the largest barrier to bystander behaviour in sexual harassment, it is important to widen the research to the Latin American context.

In the present study, we aim to add to the sparse literature by investigating bystander barriers in universities in Ecuador, a country in South America in the Andean region. Although several women's movements are fighting for a social change to promote gender equality (Glidden & Shaffer-Cutillo, 2017), both men and women in Ecuador often endorse machismo and traditional gender norms (Morris, 2019). The high prevalence of sexual harassment in diverse faculties in various regions of Ecuador has been well established (Agustín Bosch, 2018; Altamirano Zabala, 2020; Guarderas et al., 2018; Morales Cobos et al., 2020; Villarreal Cueva, 2019). In the University context in Ecuador, harassment is often not reported further because victims lack knowledge of their legal rights (Arias Saa & López Acero, 2018; Gordillo Ipiales, 2020), perpetuating the culture of harassment. Alternatively, victims may fear adverse consequences if they report abuse (López Baño, 2019; Morales Cobos et al., 2020), especially when the perpetrator is a member of staff and the victim is a student (Morales Cobos et al., 2020). However, less is known about the factors that present as a possible barrier or enabler for bystander action.

In this study, we investigate some of the known factors from prior research in the U.S that might explain bystander behaviour in Ecuador. As well as rape myth acceptance, there may be other potential barriers (e.g. perpetration of sexual violence; Yapp & Quayle, 2018) and facilitators (e.g. victimisation in sexual violence; Kim et al., 2019) that aid or prevent bystander action. We expect this pilot study will shed light on an important issue, providing future research directions for bystander behaviour in this cultural context.

Methods

Participants

Participants were 129 (90 women, 39 men) staff and students from eleven universities in Ecuador who completed an on-line survey advertised through the Ecuadorian research teams’ social networks and via e-mails to universities. We used non-probability convenience and snowball sampling, which are commonly used in investigations of sexual harassment experiences in academia (e.g. Clancy et al., 2014). Twenty-seven participants were staff members and 102 were students. Sixty participants studied or worked in the coastal region of Ecuador and 69 in the mountains/highland region. The mean age of the participants was 27.90 years (SD = 10.30).

Materials and procedure

All materials were back-translated from English to Spanish by three members of the research team, all fluent in both languages. In addition, two team members (MM and MA) have extensive experience of gender issues in Ecuador via their work in the Women’s science network. In order to investigate the incidence and perception of harassment, we devised a 28-item questionnaire using questions adapted from Espelage et al. (2014), Fitzgerald et al. (1995) and Jagsi et al. (2016). The three Ecuadorian team members, familiar with the University context, reviewed the questions to ensure that only those that were relevant were chosen. Participants were asked about their direct experiences in perpetrating sexual harassment, being a victim, and being a bystander in various scenarios. The questions ranged in severity from sexist jokes and comments to sexual coercion (see Appendix 1). Example items include ‘Commenting in a sexual manner about the way somebody looks; Making inappropriate sexual gestures; Spreading sexual rumours about somebody; Forcing someone to kiss him or her; Forcing someone to do something sexual other than kissing’. After each of the 28 acts, participants answered the following questions: Have you ever done this?; Has someone ever done this to you?; Have you ever observed/heard anybody doing this to somebody? (Yes/No/Not sure). For analytical purposes, the Yes answers were coded as 1, and the No/Not sure answers as 0. The questions were summed together to form an index of (i) perpetration, (ii) victimisation, and (iii) bystander opportunity. In addition, after each question, we asked about perceived seriousness (How serious a problem do you think this is in University campuses? 1 = Not serious at all, 7 = Extremely serious) and likelihood of intervening as a bystander (How likely is it that you would intervene if you observed this happening to another person? 1 = Not likely at all, 7 = Extremely likely). These items were summed together across the 28 questions to form an index of perceived seriousness and bystander likelihood.

To investigate the extent that participants endorsed rape myths, we used the 20-item version of the Illinois Rape Myth Acceptance Scale (IRMA; Payne et al., 1999). The questionnaire consisted of 20 items (example item; ‘A woman who dresses in skimpy clothes should not be surprised if a man tries to force her to have sex’; 1 = Not at all agree, 7 = Very much agree). IRMA is a widely used, validated psychometric measure to gauge participants’ attitudes around rape. In the present study, we used the Spanish translation of IRMA (Expósito et al., 2014), and this showed good internal consistency (Cronbach’s α = .89).

For bystander difficulty, we used 12 items from a 51-item measure (Banyard et al., 2005) asking about difficulty in acting as a bystander (1 = very difficult, 7 = not difficult at all) in scenarios such as ‘Confront your friend who says he plans to get a girl drunk to have sex’, or ‘Cooperate with the police or campus security in an investigation of sexual assault that your friend committed’. The items were summed together to form an index of bystander difficulty, with higher scores indicating less difficulties in being a bystander. The scale had excellent internal consistency (Cronbach’s α = .96).

Ethics approval

The ethical approval for the study was granted by the University of Liverpool Central Ethics Research Committee (REF: 5886).

Data analysis plan

First, we examined the distribution of responses and performed data checks to ensure the data were normally distributed with no outliers. Next, we performed zero-order correlations among the main measures. Spearman’s rank correlations were conducted for categorical measures to determine if measures of gender, location (highlands versus coast), or status (student or staff), were related to the main continuous measures. Only gender had consistent, significant relationships with the continuous variables. Location was unrelated to the other variables. Status was correlated only with perpetration of harassment, to the effect that students were more likely to report perpetration than staff were (Spearman’s R = .20, p = .02). Out of the categorical variables, we decided to include only gender in further analyses.

We then analysed if variance in bystander difficulty was statistically explained by several predictor variables (i.e. having previous experience of sexual coercion as a perpetrator or a victim, bystander opportunity, rape myth acceptance, perceived seriousness of sexual harassment) via multiple regression. For bystander likelihood, we performed another regression in the same way, but included bystander difficulty as an additional predictor since the perceived ease of intervening might relate to one’s likelihood to intervene. We performed a series of data checks and transformations (see Appendix 2). The transformations resulted in a reflection of the likelihood and difficulty variables, so that higher values were indicative of less likelihood and more difficulty.

Results

In Table 1, we present the descriptive statistics for the study variables. Table 2 shows the zero-order correlations.

Table 1. Descriptive data for main continuous measures.

	N	Mean	Standard deviation	Min	Max	Skewness	Std. error skewness	Kurtosis	Std. error kurtosis
Rape Myth	114	1.92	1.20	0.94	7.18	2.45	0.27	6.71	0.45
Seriousness	119	184	77.40	0	270	−0.62	0.21	−0.89	0.42
Bystander Opportunity	129	10.10	5.94	0	26	0.46	0.21	−0.40	0.42
Bystander Difficulty	119	82.30	19	18	96	−1.84	0.22	2.88	0.44
Bystander Likelihood	129	177	59.90	0	260	−0.88	0.21	0.487	0.42

Table 2. Zero-order correlations among main study variables.

	1	2	3	4	5	6
1. Age	–
2. Rape Myth	−.09	–
3. Seriousness	−.12	.02	–
4. Bystander Opportunity	−.17	.14	.06	–
5. Bystander Difficulty (Reflected)	−.05	.26**	.08	.02	–
6. Bystander Likelihood (Reflected)	−.03	.22*	−.48***	.04	.23*	–

*p < .05, **p < .01, ***p < .001.

Bystander difficulty was significantly correlated with rape myth acceptance, such that those with higher rape myth acceptance were less likely to perceive bystander intervention as easy. In addition, rape myth acceptance had a significant and negative correlation with bystander likelihood, indicating that those higher in rape myth acceptance were less likely to report that they would intervene. There was a moderately strong and positive correlation between perceived seriousness and bystander likelihood, suggesting perceived seriousness of sexual harassment relates to a greater likelihood of intervening as a bystander. Finally, we found a positive correlation between bystander difficulty and bystander likelihood, indicating that the ease of intervening relates to a higher reporting of doing so. Bystander opportunity was unrelated to these measures.

As for perpetration of sexual coercion, Table 3 shows that people who perpetrated sexual coercion held accepting views toward rape myths. People who reported greater perpetration also reported more opportunities to intervene as a bystander (i.e. were more likely to witness inappropriate behaviour). People who experienced sexual coercion as a victim also reported more opportunities to intervene as a bystander; again, they were more likely to witness inappropriate behaviour. Correlations with gender showed that women were less likely to hold accepting views of rape myths and were less likely to feel at ease in intervening as a bystander. Thus, gender was controlled in subsequent analyses. Age was unrelated to all measures, so it was not considered further in the regressions.

Table 3. Spearman’s rank correlations with demographic measures and sexual coercion.

	Gender (female = 1)	Student/Staff (student = 1)	Location (highland = 1, coast = 0)	Perpetration	Victimization
2. Rape Myth	−.22*	.16	−.11	.33***	.05
3. Seriousness	.06	.13	−.06	−.09	−.01
4. Bystander Opportunity	−.11	.12	−.03	.36***	.66***
5. Bystander Difficulty (Reflected)	.20*	.01	.08	.08	.07
6. Bystander Likelihood (Reflected)	.10	.06	−.06	.15	.03

*p < .05, **p < .01, ***p < .001.

To investigate how the predictor variables related to bystander difficulty, we conducted a simultaneous multiple regression with gender, rape myth acceptance, seriousness of sexual harassment, perpetration, victimisation, and bystander opportunity as the predictor variables. Together, these variables explained 15% of the variance, which was significant, R = .39, F (6, 107) = 3.16, p = .007. The unstandardised and standardised coefficients along with the standard errors are shown in Table 4. Gender and rape myth acceptance were significant predictors, such that men and those low in rape myth acceptance reported less bystander difficulties.

Table 4. Simultaneous multiple regression for bystander difficulty and bystander likelihood.

	Bystander Difficulty (Reflected)			Bystander Likelihood (Reflected)
Predictor	beta	SE	Stand. beta	Beta	SE	Stand. beta
Intercept	0.90	0.39	8.10	1.06
Gender (1 = female)	0.71	0.24	.28**	1.34	0.65	.19*
Rape Myth	0.81	0.25	.31***	1.34	0.69	.18
Seriousness	0.01	0.01	.09	−0.02	0.00	-.41***
Perpetration	0.13	0.15	.09	0.81	0.39	.20*
Victimization	0.00	0.18	.00	−0.13	0.46	-.03
Bystander Opportunity	−0.01	0.02	-.04	0.02	0.06	.04
Bystander Difficulty (Reflected)	0.47	0.25	.16
R^2	.15**	.31***

*p < .05, **p < .01, ***p < .001.

To identify the predictors that were most important for explaining bystander likelihood (i.e. likelihood of intervening), we conducted another regression with the same predictor variables as above, with the addition of bystander difficulty. Together, the predictors explained 31% of the variance. The full model was significant, R = .55, F (7, 106) = 6.70, p < .001. Women and those who reported higher previous perpetration were less likely to intervene. Those who perceived sexual harassment as a serious problem reported greater likelihood of intervening as a bystander. This unique effect was significant over and above rape myth acceptance, bystander difficulty, and having prior opportunities to be an active bystander.

Discussion

To our knowledge, this is one of the first studies to examine bystander behaviour in relation to sexual harassment in the Latin American context. We found that amongst university staff and students in Ecuador, rape myth acceptance was associated with perceptions that bystander actions are difficult to perform. Further, perceiving sexual harassment as a serious problem related to greater self-reported likelihood of intervening in various harassment scenarios. Yet, those who reported greater perpetration sexual harassment themselves reported less likelihood of intervening. Finally, in this sample, gender influenced bystander behaviour to the effect that women perceived intervention as more difficult, and were also less likely to intervene.

Consistent with previous research assessing rape myth acceptance and bystander behaviour in the U.S (Ortiz & Thompson, 2017), rape myths related to bystander difficulties in Ecuadorian participants too. Beliefs in rape myths may result in intervention difficulties for several reasons. For instance, those who blame the victim may have lower empathy towards them, resulting in a perception that intervention is not their responsibility. Further still, they may perceive harassment as causing minimal negative consequences for the victim (Banyard, 2011). Future research should investigate the mechanisms by which rape myth acceptance functions as a barrier in Ecuador.

Our findings also indicated that those who perceived specific acts of harassment as a serious problem reported greater likelihood of intervening as a bystander, while those who had perpetrated harassment reported less of a likelihood of intervening. Failing to perceive harassment as serious has been identified as a bystander barrier in U.S. studies (Banyard, 2011), and the same patterns applied in our Ecuadorian sample. Playing down the consequences of harassment has also been related to perpetration (Moore & Mennicke, 2020), which speaks for the importance of bystander interventions that emphasise the seriousness of sexual harassment on the victims. Tackling the perceptions that play down the seriousness of harassment could be one of the first important steps towards changing the culture of misogyny (Aguayo & Nascimento, 2016).

In addition to the variables discussed above, we found gender differences in bystander behaviours, to the effect that women reported lower bystander efficacy and less likelihood of being an active bystander than did men. There could be several explanations for this finding. For example, U.S studies have suggested that men’s bystander action may stem from benevolent sexism, the perception that women need protection (Burns et al., 2019; Hoxmeier et al., 2020). Latin American cultures are characterised by embracement of benevolent sexism (Aguayo & Nascimento, 2016), men typically supporting these views more than women (Cárdenas et al., 2010). In order to discover whether bystander behaviour stems from chivalry or from strive to increase social justice, future research should investigate men’s motivations to act as positive bystanders in the context of Ecuador.

Women, in turn, may be prevented from intervening due to high perceived risk of injury (Witte et al., 2017), or skills-deficits (Yule & Grych, 2020). The perceptions of risk may be even more prevalent in cultures such as Ecuador, where violence against women is relatively common (Ibáñez, 2017). In the U.S context, intervention programmes targeted at women have successfully increased bystander efficacy and willingness to intervene (e.g. Foubert et al., 2010). Interventions adapted for Latin America may need to rely more heavily on promoting less direct forms of bystander intervention. These include alerting someone with perceived authority (e.g. bar manager or professor) so they can intervene, or offering support to the victim after the event. Our results suggest that interventions that increase awareness of the existence of indirect bystander methods would also be beneficial in Ecuador.

Limitations and future research

Although the present findings provide useful data for the development of future studies, our research has several limitations. Firstly, we used positivist, quantitative questionnaires that have previously been used mainly in Western cultures. To understand cultural scripts around experiences relating to harassment and bystander behaviour, further qualitative research is necessary (Madriz, 1998). Qualitative research gives more nuanced ideas about the specific barriers that bystanders face (e.g. Hoxmeier et al., 2019), which is especially important in new contexts of investigation (e.g. see Hennelly et al., 2019, for a study in the UK). To understand the bystander barriers in Ecuador, it is important to focus on participatory qualitative research and use the results to design and validate quantitative tools that may be more appropriate in the Ecuadorian and Latin American context (see also Liamputtong, 2010).

Second, reflecting the prevalence of sexual harassment that occurs in academia (e.g. Guarderas et al., 2018; Klein & Martin, 2019), the present study focused on bystander behaviour in the context of Higher Education in Ecuador. However, other contexts are also important, and may present different barriers to bystander intervention. For example, sexual harassment in public spaces, streets, and public transport (see López et al., 2020 for ‘acoso callejero’) are problematic in Ecuador. The barriers and facilitators for intervening as a bystander in these contexts should be investigated in future research since they may have an effect on community life more generally.

Third, our pilot study had small, self-selected sample, which is not representative of the wider University population in Ecuador. Although convenience and snowball sampling are an easy way to obtain pilot data, this sampling method limits generalisability of the findings (Tyrer & Heyman, 2016). Future research should focus on obtaining a more representative sample of staff and students in Ecuador. Fourth, the small sample size also meant we could not investigate differences between academic disciplines. For instance, medical students may experience harassment from patients or patients’ families (Vargas et al., 2020), which may result in different barriers for bystanders. The issues could be rectified in future studies by utilising random sampling and comparisons between disciplines.

Fifth, we did not consider how intragroup dynamics and social hierarchies influence the difficulty and likelihood of acting as a bystander. Bystander behaviours are more challenging to perform when the perpetrator has more power in the social hierarchy (see Wamboldt et al., 2019). Although Ecuadorian law prohibits the use of social position in soliciting sexual favours (Article 166; Secretaria de Education Superior, Ciencia, Tecnologia, e Innovacion, 2018), the use of power imbalance as a tool for harassment is sadly not uncommon among professors (see Morales Cobos et al., 2020). The risks for the bystander can be especially pronounced when the perpetrator holds power over the bystander. For example, a female professor in Mexico was recently dismissed for helping her students to report harassment from a male member of staff (Matias, 2020). This does not only demonstrate that staff may face repercussions from university administration, but also shows that power imbalances exist even in horizontal hierarchies – that among peers. In addition, professors have the power over student grades (e.g. Morales Cobos et al., 2020), which could risk the future academic success of those who take a stance as a bystander on behalf of a fellow student. Deeper knowledge of how vertical and horizontal hierarchies influence behaviours in Latin America would be beneficial for future interventions.

Finally, our research did not consider sexual orientation or gender identity. Bystander action in sexual harassment situations involving individuals from the LGBTQ+ communities may depend on attitudes and social contact (Dessel et al., 2017). Although globally, Ecuador is a relatively inclusive country (Lamontagne et al., 2018), discrimination and violence is still part of the lived experiences of individuals in the LGBTQ+ communities (Zúñiga-Salazar & Pillajo, 2017). We currently have no empirical knowledge of how sexual orientation of the victims and perpetrators influence bystander behaviour in Latin American context. This is an important gap that future research should aim to target.

In summary, sexual harassment remains an important issue in the Higher Education, with significant consequences for the health and wellbeing of survivors of such behaviour (e.g. Bastiani et al., 2019). Although there is a large body of knowledge of bystander behaviour in the U.S, a lack of research in Ecuador and elsewhere in Latin America prevents the design of culturally sensitive interventions. Our results suggest that in the context of Higher Education in Ecuador, women may have less confidence to intervene. In addition, those who consider harassment as a serious issue, and those who have low levels of rape-myth acceptance may be more likely to intervene as active bystanders. These results present the first step towards understanding the context of bystander behaviour in gender-based harassment in Latin America and could be used to give directions for future research and intervention development.

Acknowledgements

We would like to thank Anna Forrest-Redfern for her research assistance at early stages of the project, and British Embassy in Quito for funding the project.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Appendices

Appendix 1

Unwanted and unwelcome sexual or romantic attention is behavior that interferes with the life of the person who receives the attention. Unwanted attention is behaviors that the recipient of these behaviours does not like or want. (for example kissing, touching or flirting)

Have you ever observed, perpetrated, or directly experienced any of the following either on University campus, or via electronic means (social media, phone messages, e-mail) with someone from university? Below, we will present a series of scenarios, and ask you to answer the questions keeping in mind it is about behaviours in University.

1. Have you ever done this? Yes/No/Not sure

2. Has someone ever done this to you? Yes/No/Not sure

3. Have you ever observed/heard of anybody doing this to somebody? Yes/No/Not sure

4. If you observed this happening to someone, how likely it is that you would intervene and do something? 1 = Not likely at all, 7 = Extremely likely

5. How serious of a problem do you think this is on University campuses? 1= Not serious at all, 7 = Extremely serious

6. Please use this box if you like to explain more about the event (who was the person who did it, who was the person who was the receiver, what happened before/during/after)

Repeat these six Q’s after each question below

1. Commenting on a sexual manner about the way someone looks

2. Commenting on how someone has a lesser ability to do something because they are female

3. Making sexist jokes

4. Looking at someone in an inappropriate way

5. Making inappropriate sexual gestures

6. Talking about what sexual advances a person would do on another person

7. Whistling at someone

8. Sending sexual materials (sexual pictures, photographs, web pages, illustrations, messages or notes electronically) (e.g. e-mail, social media, phone)

9. Sending nude pictures of oneself electronically (e.g. e-mail, social media, phone)

10. Posting sexual messages about someone on the Internet (e.g. websites, blogs)

11. Showing someone sexual materials on a computer

12. Using bribery (e.g. money, help, better grades) in order to get sexual advantage

13. Using threats (e.g. harming someone’s career) in order to engage in sexual behaviour

14. Spreading sexual rumours about someone

15. Calling someone a gay or a lesbian or a homophobic name (such as faggot, dyke or queer)

16. Spied on someone as they dressed or showered (e.g. in a dorm, in a gym, etc.)

17. Flashed their genitals at someone

18. Touched, grabbed, or pinched someone in a sexual way

19. Intentionally brushed up against someone in a sexual way

20. Asked someone to do something sexual in exchange for something (e.g. a better grade, a recommendation, class notes, etc.)

21. Pulled at clothing in a sexual way

22. Pulled off or down clothing

23. Blocked the way, cornered, or followed someone in a sexual way

24. Forced someone to kiss him or her

25. Forced someone to do something sexual, other than kissing

26. Repeated requests for drinks, dinner, etc, even after rejection

27. Retaliated in some way if sexual or romantic advantages were refused

28. Used alcohol or drugs to get someone incapacitated in order to sexually exploit them

Appendix 2

Data checks: First, we examined the data to ensure that there were no suspicious answers. We examined completion time and responses that indicated answering with a particular response bias (all answers of ‘1’ with no indication of dynamic responses (e.g. completing free text). There was one participant who identified as a male student who met these criteria and was removed. The remaining 129 participants were retained for analysis. Table 1 shows the distribution of the measures. Four outliers were identified by looking at boxplots and violin plots for rape myth acceptance; these were scores over 5. These were reduced to the next highest scores of 4.9 (for one score over 5), 5.0 (for two scores over 6), and 5.1 (for one score over 7). Bystander difficulty had two scores identified as outliers (scores of 18 and 19) which were brought one point closer to the lowest score of 21). Further, rape myth acceptance and bystander difficulty were skewed and leptokurtic so log transformations were conducted; bystander difficulty was reflected first since the skew was negative. This brought the z-scores for skewness from 10.88 and −8.37–3.86 and 1.40 respectively. After bringing two outliers (scores of 0 and 1) closer to the next lowest score of 21, bystander likelihood was reflected and subjected to a square root transformation, bringing skewness z-score down from −4.34 to −0.83. Perpetration and victimization were count variables ranging from none to 20 and 25, respectively. The variables were overdisbursed with standard deviations (3.13 & 5.29) as high as the means (3.94 & 6.74, respectively). Thus, tertiary splits were made for perpetration at 2.00 and 4.00, and for victimization at 3.00 and 7.33.