Rape myth acceptance reflects socially shared beliefs about rape that serve to justify men’s sexual aggression against women (Lonsway & Fitzgerald, 1994), therefore rape myths acceptance and conservative gender relations in society are closely connected (Viki &
Abrams, 2002). According to the global gender gap index, Hungary holds the 105th position in equality of the positions of men and women in society (The Global Gender Gap Report, 2020, suggesting that gender equality is lower than in most of the western world. Estimations suggest that unreported rape cases are 415 times higher than reported ones in Hungary (Wirth
& Winkler, 2015). While rape can disrupt the harmony between men and women and draw attention to gender inequalities (Searles, 1995), rape myths can hinder the recognition of the structural aspects of rape (Chapleau, Oswald, & Russel, 2007), that the perpetrators are mostly men, whereas victims are overwhelmingly women (European Union Agency for Fundamental Rights, 2014).
36
Rape myths create a normative environment in which labeling a case not as rape, blaming a victim for it, and excusing the perpetrator is more acceptable than in social contexts in which rape myths are refuted (Bohner, Siebler, & Schmelcher, 2006). Based on previous results (European Union Agency for Fundamental Rights, 2014) it is reasonable to think that low gender equality and the associated high level of rape myth acceptance creates an
environment in which victims do not trust the police and other authorities, and do not think they will be treated fairly (Sable et al., 2006; Wirth & Winkler, 2015). They therefore do not report the rape to the police (McMahon & Farmer, 2011), which explains why latency is higher in less gender equal countries that are also likely to endorse rape myths more widely (European Union Agency for Fundamental Rights, 2014). Therefore, understanding the phenomenon of rape myth acceptance has great relevance in the normative context of
Hungary, both to check whether the construct is identical in this societal context compared to the data collected mostly in Western countries too, and to examine its practical implications, e.g. how people react to rape cases and victims.
The importance of examining reactions to various rape scenarios is twofold: on the one hand, people’s reactions reflect the normative context of rape in society, therefore it affects whether perpetrators do or do not think that rape is a serious crime, or what constitutes a rape, and on the other hand, it affects whether victims report the case to the police or seek help at all (Bohner, Siebler, & Schmelcher, 2006).Evaluation of rape cases is affected by the stereotypicality of the rape, which affects the perceived certainty of rape. Previous studies found that a rape scenario is perceived as stereotypical if the perpetrator is a stranger to the victim and a deviant person (Greenberg & Ruback, 1992), he is armed or uses physical force during the rape (McGregor et al., 2000), and the victim immediately reports the case to the police and cooperates with them (Bongiorno, McKimmie, & Masser, 2016). Moreover, the gender of the victim and the perpetrator and their prior relationship affect whether a case fits
37
into a stereotypical rape scenario, which in turn affects evaluations of the rape. Participants were more likely to blame the victim and believe that it was not rape when the case was perceived counter-stereotypical, that is, when the victim did not fight against the perpetrator physically and did not cooperate with the police (Sheldon & Parent, 2002).
Group membership of the perpetrator and the victim can also produce bias in how a rape case is perceived and evaluated (Bal & Van den Bos, 2010; George & Martinez, 2002;
Harrison et al., 2008; Masser, Lee, & McKimmie, 2010; McKimmie, Masser, & Bongirono, 2014). This can be explained by social identity theory suggesting that people are motivated to see members of their ingroup more positively than members of the out-group (Tajfel &
Turner, 1986). Previous research has shown that people blame an out-group perpetrator more than an ingroup perpetrator (Bal & Van den Bos, 2010; George & Martinez, 2002; Harrison at al., 2008), and blame an ingroup victim less than a victim belonging to an out-group (Harrison et al., 2008). However, putting together the effects of stereotypicality and group belonging, Bongiorno and colleagues (2016) found that the perpetrator’s out-group membership did not affect the evaluation of a stereotypical rape case, however, the ingroup perpetrator was more likely to be excused and the victim blamed for the rape when the rape was
counter-stereotypic.
Research Aims and Hypotheses
The main purpose of Study 2 was twofold, on one hand to examine the connection between rape myth acceptance and the evaluation of uncertain and undisputable rape cases and how group membership affects the evaluation of uncertain rape scenarios in a highly gender unequal country. Again, we tested the validity of UIRMA, this time using an online sample that is demographically similar to the Hungarian population in terms of gender, age, and settlement in Hungary. Similarly, to Study 1, we aimed to check the five-factor solution
38
of the UIRMA-SF scale and expected that men accept rape myths more than women (H1); a strong positive relationship between UIRMA-SF and hostile sexism (H2); and moderate positive correlations with benevolent sexism and with social dominance orientation (H3).
Based on previous research (e.g. Chapleau, Oswald, & Russel, 2007; Suarez &
Gadalla, 2010) in case of the stereotypicality of the cases we expected, that rape myths acceptance will predict victim blaming and labelling the case as rape (H4) and people with higher rape myth acceptance would blame the victim more and label the case less as rape, especially when the case is uncertain (H5). While in case of group membership, we expected, that participants will blame the victim more (H6) and label the case as rape less (H7), when the victim is an outgroup member.
Materials and methods
Participants
We recruited participants with the help of an opinion poll company (SoliData) who relied on an online pool of respondents that are demographically similar to the Hungarian society in terms of gender, age, and type of settlement, but participants had a higher than average education (N = 1007, 49.2% male 50.8% female). We did not calculate sample size based on previous results which is nowadays often required to prevent Type 2 error due to a larger sample size that would be ideal to test the hypothesized relation, and targeted N = 1000 which is typically used in representative opinion poll surveys in Hungary (see Poll of Polls, 2018). Mean age was 41.52 years (SD = 13.05) ranging from 18 to 64 years, level of education and type of settlement are presented in Table 8.
Table 8 Level of education and type of settlement of participants
N %
39 Education
Primary degree or less 8 0.8
Secondary degree 464 46.1%
Another type of degree 118 11.7%
College/university degree or higher 417 41.5%
Settlement
Capital 191 19%
County capital 212 21.1%
Town 331 32.9%
Village 273 27%
Measures and procedure
Data was collected in 2016 following the regulations of IRB approval of Eötvös
Loránd University. After giving their informed consent, participants were presented with three scenarios: (1) an uncertain scenario with a victim with different group membership, (2) an undisputable scenario, and (3) an uncertain rape scenario in this order (for the exact wording of the cases, see the A1 in the Appendix). For technical reasons, randomization was not possible in the data collection, therefore, we conducted a complementary analysis in order to check whether the responses given to the first rape vignette influenced the responses given to the following one. We run a moderation model (Model 1) using the Process macro (Hayes, 2017) and found that rape myth acceptance influenced the uncertain rape labelling and victim blaming beyond and above the labelling and victim blaming in the uncertain rape scenario
40
(see the results of the moderation analysis in in the Appendix, A2). which suggest that results are not consequences of the order effect only.
We established the level of certainty based on Bongiorno, McKimmie, and Masser’s (2016) research, in the undisputable rape scenario the victim physically resisted to the perpetrator, and she fully cooperated with the police (e.g. “Éva [the victim] said that she screamed and tried to escape but she couldn’t. At the same night Éva went to the police and reported the case.”), but in the uncertain scenarios she did not (e.g. “She said many times that she does not want to have sex with him, but physically she did not resist” „Szilvia [the victim]
went to the police and reported the case but it was really hard for her to work with them.”).
We measured victim blaming with one item (“I think Éva/Szilvia [the victim] is responsible for what happened.”) and participants labelled the case whether they considered it a rape or not, and both were measured on a 7-point scale from 1 = it was certainly not rape to 7 = it was certainly rape.
We varied the victim’s groups membership between Slovenian and Hungarian in the other scenarios. In the outgroup condition we added the nationality of the victim and used the Slovenian version of a well-known name in Hungary in the description (“…accused to committing rape against the Slovenian Julija…”). To test whether the scenarios were equivalent in stereotypicality, we run a pilot test (N = 25) on the scenarios, using only
Hungarian names. Participants were presented with both scenarios in a randomized order. We found no difference in the evaluation of the scenarios in stereotypicality (see Hiba! A
hivatkozási forrás nem található.)
Participants completed the short form of the Ambivalent Sexism Inventory (5 items hostile sexism scale, α = .84; 5 items benevolent sexism scale α = .79; Glick & Fiske, 1996;
Szabó, 2008), Social Dominance Orientation (8 items, α = .78; Ho et al., 2015, Faragó &
Kende, 2017) and the Hungarian version of the Updated Illinois Rape Myths Acceptance
41
Scale (18 items, α = .90; McMahon & Farmer, 2011). The data collection was a part of an omnibus survey. Besides the mentioned variables, we measured modern sexism (Swim &
Cohen, 1997), but we do not discuss the findings related to this variable within the presentation of this study.
Results
To check whether the five-factor solution can be identified, we performed a
confirmatory factor analysis again based on the factor structure suggested by McMahon and Farmer (2011) and tested in Study 1. Again, due to non-normality of the distribution of several ratings, we used MLR estimator. CFA were performed with MPlus 8 (Muthén &
Muthén, 2015).
The five-factor model showed good fit to the data (χ2 = 421.850, df = 124, CFI = .944, TLI = .931 RMSEA = .054 [.048; .060], SRMR = .049). Standardized factor loadings of the general factor ranged from 0.31 to 0.90 (see Fig 2.). Measurement invariance (see Chen, 2007), and scalar invariance of the UIRMA scale was established across gender groups (see A4).
Fig 2. Confirmatory factor analysis of the Updated Illinois Rape Myths Acceptance Scale (SF).
42
All the factor loadings are standardized and significant (p < .01)
Convergent and Discriminant Validity. Because of non-normal distribution
(skewness of .307, SE = .077 and kurtosis of -.130 SE = .154) we used Mann-Whitney test to compare the UIRMA scores of men and women (see Table 9). Men scored significantly higher than women on rape myths acceptance and on every subscale of UIRMA, except on subscale He didn’t mean to (intoxication), however these differences were weak.
Table 9 Difference between men and women on UIRMA and on its subscales
Men (N = 495) Women (N = 512)
43 Main
rank
Mean (SD) Main rank
Mean (SD) Mann-Whitney U
P η2
Rape myths acceptance 529.25 3.50 (1.11) 479.59 3.30 (1.23) 114223.00 p = .007 .007 She lied 557.35 3.80 (1.57) 452.42 3.22 (1.61) 100310.50 p < .001 .033 She asked for it 530.13 3.84 (1.49) 478.74 4.11 (1.71) 114994.00 p = .011 .006 Wasn’t really rape 532.01 2.90 (1.46) 476.92 2.64 (1.43) 113785.50 p = .005 .008 He didn’t mean to 520.72 3.82 (1.29) 487.84 3.60 (1.50) 112856.00 p = .003 .009 He didn’t mean to
(intoxication)
529.25 2.38 (1.37) 479.59 2.27 (1.40) 118445.00 p = .069
As predicted, UIRMA correlated moderately positively with hostile sexism (r = 49, p
< .001) indicating convergent validity (see Table 10) and correlated weakly with benevolent sexism (r = .25, p <.001) and with social dominance orientation (r = .24, p <.001).
Table 10 Correlation between benevolent sexism, hostile sexism, social dominance orientation, UIRMA and its subscales
1 2 3 4 5 6 7 8
1 Benevolent sexism
2 Hostile sexism .14
3 Social Dominance Orientation
.02 .32
4 Rape myths acceptance .26 .50 .24
5 “She lied” .18 .55 .23 .80
6 “She asked for it” .22 .39 .18 .78 .52
44
7 “Wasn’t really rape” .16 .36 .18 .81 .60 .50
8 “He didn’t mean to” .26 .23 .11 .66 .36 .41 .38
9 “He didn’t mean to (intoxication)”
.14 .26 .16 .68 .42 .42 .52 .55
Significance is below p < .001 in each cell.
We tested whether people evaluated the uncertain and undisputable scenarios
differently with a paired sample t-test. We found that participants blamed the victim more and labelled the case as rape less in the uncertain rape scenario (victim blaming M = 3.36 SD = 1.68; rape labelling M = 4.96, SD = 1.92) than in the undisputable case (victim blaming M = 1.42 SD = 1.02; rape labelling M = 6.70, SD = 0.97). We run a hierarchical linear regression to test the role of RMA in the evaluation of both an uncertain and an undisputable rape case (see Table 11 and Figure 3). We controlled for gender and age in the regression. We found that benevolent sexism and RMA were significant positive predictors of rape labelling both in case of an uncertain rape scenario and in case of an undisputable rape scenario. In the case of victim blaming, hostile sexism and RMA were significant predictors both for the uncertain and undisputable rape scenarios. However, RMA better predicted the evaluation of the rape case, that is that predicted with a stronger effect size and explained a greater part of variance when the case was uncertain, then when it was undisputable.
Table 11 Hierarchical linear regressions on rape labelling and on victim blaming in an uncertain and undisputable rape case
Outcome variable: Rape labelling
Uncertain case Undisputable case
B SE β p ΔR2 R2 B SE β p ΔR2 R2
45
Step 1 .008 .00
8
.006 .006
Constant 4.13 0.27 6.38 0.14 < .001
Gender 0.26 0.12 0.07 < .001 0.12 0.06 0.06 .042
Age 0.01 0.01 0.07 0.029 0.00 0.00 0.04 .174
Step 2 0.026 .035 .04
3
.004 .010
Constant 5.48 0.39 6.15 0.120 < .001
Gender 0.02 0.13 0.00 < .001 0.14 0.06 0.07 .034
Age 0.01 0.01 0.06 0.891 0.00 0.00 0.04 .246
Hostile sexism -0.26 0.04 -0.20 0.060 0.01 0.02 0.01 .801 Benevolent
sexism
0.02 0.04 0.02 < .001 0.04 0.02 0.06 .049
Step 3 0.628 .106 .14
9
.018 .028
Constant 5.81 0.37 6.22 0.20 < .001
Gender 0.15 0.12 0.04 < .001 0.16 0.06 0.09 .011
Age 0.01 0.00 0.09 0.210 0.00 0.00 0.05 .113
Hostile sexism -0.01 0.05 -0.00 0.002 0.06 0.03 0.09 .020 Benevolent
sexism
0.12 0.04 0.09 0.901 0.06 0.02 0.09 .005
RMA -0.63 0.06 -0.39 0.005 -0.13 0.03 -0.16 p <
.001
Outcome variable: Victim blaming
Step 1 .001 .00
1
.003 .003
46
Constant 3.56 0.24 1.64 0.14
Gender -0.02 0.11
-0.006
-0.10 0.06 -0.05
Age -0.00 0.00
-0.032
-0.00 0.00 -0.02
Step 2 .084 .08
5
.044 .047
Constant 1.58 0.33 0.75 0.21
Gender 0.32 0.12 0.094 0.05 0.07 0.02
Age -0.00 0.00
-0.017
-0.00 0.00 -0.01
Hostile sexism 0.36 0.04 0.306 0.16 0.02 0.22 Benevolent
sexism
0.01 0.04 0.008 0.01 0.02 0.01
Step 3 .085 .17
0
.010 .057
Constant 1.32 0.32 0.70 0.21
Gender 0.21 0.10 0.064 0.03 0.07 0.01
Age -0.01 0.00
-0.046
-0.00 0.00 -0.02
Hostile sexism 0.15 0.04 0.131 0.11 0.03 0.16 Benevolent
sexism
-0.07 0.04 -0.055
-0.01 0.02 -0.01
RMA 0.50 0.05 0.348 0.10 0.03 0.12
47
Figure 3 Difference in the evaluation of uncertain and undisputable rape cases
To examine the effect of victim’s group membership on the evaluation of rape cases, we run paired sample t-tests. We found that participants labelled the case less as rape (ingroup victim M = 4.96 SD = 1.92, outgroup victim M = 4.35 SD = 1.90 , t(1006)= -10.40, p<.001) blamed the perpetrator less (ingroup victim M = 5.18 SD = 1.69, outgroup victim M = 4.63 SD = 1.70, t(1006)= -10.72, p<.001) and blamed the victim more (ingroup victim M = 3.36 SD = 1.68, outgroup victim M = 3.67 SD = 1.54, t(1006)= 6.42, p<.001; see Figure 4).
Figure 4 Difference in the evaluation of rape cases with an ingroup or with an outgroup victim
Discussion of Study 2
48
We replicated and confirmed the original five-factor model of the UIRMA scale on a representative sample. In line with our hypothesis found that men accepted rape myths more, however these differences were small (H1). We found a positive relationship between rape myth acceptance and other oppressive believes people with higher rape myth acceptance tended to endorse oppressive beliefs like hostile sexism (H2), benevolent sexism and social dominance (H3) orientation to a greater degree.
Using a representative sample enabled us to generalize our findings regarding the scale validation to the Hungarian context. Both in Study 1 and 2 we found evidence for convergent and discriminant validity of the scale, suggesting that rape myth acceptance is part of a generalized hostility toward women (Amnesty International, 2012) and it is deeply embedded in the society’s belief system about gender roles and inequality. At the same time, measuring rape myth acceptance can offer a better understanding of rape related attitudes than more general ideologies about gender or about victim blaming in general (e.g. through just world beliefs).
Rape myths serve to justify men’s sexual aggression over women (Lonsway &
Fitzgerald, 1994) and mask the structural aspects of rape (Chapleau & Oswald, 2014). This explanation is supported by data suggesting that rape myths are more accepted in more conservative, and less gender equal societies (Aosved & Long, 2006; Foster & Kidd, 2014) and that rape myth acceptance is usually more accepted by men than by women (e.g. Lonsway
& Fitzgerald, 1994). Although this can be a reason that gender differences in rape myth acceptance can be found in more gender unequal societies like Hungary (Hanzi et al., 2016) we did not examine rape myths acceptance cross-culturally in the current study.
Furthermore, in line with previous findings (Suarez & Gadalla, 2010) we found that rape myth acceptance predicted how rape cases were evaluated. Participants with higher rape myth acceptance blamed the victim more and labeled the case less as rape (H4). We also
49
found that rape myth acceptance was a stronger predictor of rape evaluations than sexism in both cases, which supports the assumption that rape myth acceptance is a different concept than sexism and more relevant to understanding rape related attitudes than sexism in general (e.g. Lonsway & Fitzgerald, 1994).
In this study we also wanted to test whether the level of rape myth acceptance predicts rape labelling. Rape does not usually happen in dark alleys and by deviant perpetrators, therefore we examined whether the evaluation of counter-stereotypical and stereotypical (uncertain vs. indisputable) rape cases depended on the participant’s rape myth acceptance. In line with previous research (Chapleau, Oswald, & Russel, 2007), we found that people with higher rape myth acceptance label both cases less as rape and blame the victim more, furthermore people evaluate uncertain rape cases less harshly, that is, they blame the victim more and label the case less as rape.
However, rape myth acceptance explained greater variance and was a stronger predictor in the evaluation of the rape cases that was uncertain, but it was still a significant predictor when it was undisputable (H5). These findings align in line with previous research (Chapleau, Oswald, & Russel, 2007) that when rape cases are uncertain – i.e. they don’t fit to the stereotypical rape scenario which is in fact the case most of the time – rape myths affect people’s way of thinking about the case even more, resulting in stronger victim blaming and the excusing of the perpetrator.
Furthermore, we found that when the victim is an outgroup member, people tend to label the equally uncertain rape case less as rape, excuse the perpetrator, and blame the victim more for the rape, that is in case of uncertain rape cases, an irrelevant dimension as the
victim’s group membership effects the evaluation of rape (H6-7). In line with previous findings (e.g. Bongiorno et al., 2016) these results suggest that ingroup positivity and group membership has an effect on how people evaluate rape cases, when those are uncertain.
50
In conclusion, we found that when a rape case is uncertain in comparison when it is
undisputable, people use other information even more to fill the missing data like their general attitudes toward rape (rape myths acceptance) to be able to form a coherent opinion about the case. Furthermore, this assumption is supported by the result, that people used victim’s outgroup membership as an excuse for the perpetrator, when they were uncertain about the rape. In summary we can say that when a rape case is uncertain, prior attitudes are even more important, and irrelevant factors such as the victim’s group membership becomes a factor that affects the evaluation of the rape case.
However, because neither the scenarios nor the randomization of the order of their presentation was not available at SoliData, we had to use a not randomized within subject experimental design. Although we were able to identify differences between the scenarios in line with our predictions, in Study 3 we aim to test our findings in a randomized between subject designed experiment. We widen our scope and examine not only outgroup victims, but outgroup perpetrators affect the evaluation of a rape case, furthermore, we aim to examine the role of outgroup status with a perpetrator and a victim from a group with lower status.