Introduction

Sexual violence and/or rape against women is widespread, with numerous beliefs, misconceptions, and myths that have urged researchers to create and adapt scales for measurement (Mennicke et al., 2021). However, in many societies, scales that measure rape myth are developing at a slower rate (Steele et al., 2022), for instance, despite high rate of rape in Pakistan (NDTV, 2022), empirical research is scanty (Kazmi et al., 2023b) largely because of non-availability of psychometric instruments. In addition, attitudes and beliefs about rape myths are theoretical speculations among experts in Pakistan and in other countries, rather than empirical data (Ansari et al., 2019; Tyson et al., 2019). Psychological instruments that measure rape beliefs and myths are few to come by (O’Connor, 2021). One of the earliest studies (Lonsway & Fitzgerald, 1994) on rape myths with a sizable sample of 645 people suggested that a large number of them believed that at least one rape myth existed; and believed that psychological and sociocultural factors impacted these myths. Numerous factors responsible for the high rate of sexual violence and rape in Pakistan include media, inadequate and unenforced laws, myths about sex, alcohol and drug use, pornography, and promiscuity in early and middle adulthood (Riaz, 2021).

One factor intimately associated with such violence include beliefs, misconceptions, and myths people hold about rape (Kazmi et al., 2023a), and are likely to originate because of lack of empathy for the victim, minimizing severity of their experiences, blaming them, not believing their reports, and exonerating the perpetrator across North and South American regions and Pakistan (Ansari et al., 2019).

Views About Rape Myths

Many societies even today continue to blame victims for their sexual behaviors when they claim or report rape. However, in some societies people think rape myths are irrational beliefs associated with sexual assault or rape (Burt, 1980). Lonsway and Fitzgerald (1994) report young and middle-aged Americans think rape myths are false beliefs or perceptions held widely by the masses, who use them as excuses for justifying acts of aggression carried out against women. Bohner et al. (2010) also agree, rape myths have a role in facilitating sexual aggression that contribute towards deactivating social prohibition leading criminal-minded men to commit acts against women (Burt, 1980). Rape myths are widely and persistently held beliefs and attitudes that permit, justify, and excuse male aggressors against women (Walfield, 2021). These ideas about rape myth are confirmed by some quasi-experimental studies where university students and general population expressed rapists had higher levels of rape myth acceptance than non-rapists (Eyssel & Bohner, 2011; Feild, 1978; Koralewski & Cogner, 1992; see other disconfirming studies: Epps et al., 1993; Overholser & Beck, 1986). Additionally, rape myths have been identified as prejudicial, stereotypical, or false beliefs about rape, which create a hostile climate for victims of rape (Garza & Franklin, 2021). It is pertinent to note, rape myths obstruct rape reports and present inaccurate assessments about causes, context, consequences, perpetrators, victims, and their interaction (Kunst et al., 2019). Therefore, communities with higher acceptance for rape myth are less likely to endorse events reported by victims; and blame them rather than the perpetrator (Kazmi et al., 2023a). This acceptance was assessed among 1123 prisoners who were incarcerated for various offenses including sexual violence towards women; findings suggested that prisoners with psychopathic personality traits and childhood abuse contributed towards violence-supportive attitudes and beliefs (Debowska et al., 2018). Similarly, among other populations, rape myths are prevalent including police officers (Hine & Murphy, 2019), lawyers (Klement et al., 2019) and jurors (Willmott et al., 2018). Daly et al. (2023) reported rape myths among jury members that can impact their decisions resulting in unfair outcomes; affecting believability of and decisional confidence towards rape victims (Willmott et al., 2018). For example, jurors endorsing rape myths were more likely to show skepticism and concerns regarding the credibility of the victim and were more likely to attribute the rape towards the victims and not on the perpetrator impacting decisions made by judges (Daly et al., 2023).

In the general public, reasons for blaming women who report rape resulted from decision of women to remain in sexually abusive relations, not reporting intimate partner rape, not perceiving trauma following rape and perceived male sexual entitlement in intimate relations (Bergenfeld et al., 2022). In US (Sowersby et al., 2022) and in Pakistan (Kazmi et al., 2023b) people in general misperceive sexual assault, rape and rape myths as reasons for discrediting and trivializing rape reported by affected women; or beliefs people have about sexual assault which may make it allowable under certain conditions (Jamshed & Kamal, 2021). Research on rape myths centers around assumptions about rape (Thelan & Meadows, 2022) and include personal, psychological, and cultural factors to explain this misjudgment among members of the general public (Owens et al., 2021; Payne et al., 1999). These factors include age, gender (for example, men are more likely than women to accept rape myths), shifting of blame (the male perpetrator is more likely to be absolved while the female victim is more likely to be blamed), race, men's supposedly insatiable sexual desires, negative views of women, gender roles, self-reported physical aggression, acceptance of interpersonal violence, adversarial sexual beliefs (the idea that sex is inherently exploitative), and violation of traditional gender roles and norms (Angelone et al., 2021; Berg et al., 1999; Reed et al., 2020). Shaw et al. (2017) indicated that police records often contain statements that deny or justify assault-based circumstances than characteristics of the victim (however see Klement et al., 2019 for contradictory data). A systematic review suggests that victim credibility, and not rape myth acceptance was the primary factor in classifying rape as consensual encounter by police, softening charges against perpetrator and their perceptions of guilt (Sleath & Bull, 2017), which shows how police officers who are expected to provide protection to rape victims may hold negative attitudes against them. Lack of knowledge in general public about rape and victim-blaming has severe consequences (Hobstetter et al., 2015). For instance, some women may not recognize their experience as rape due to the fact that labeling it as such may be beneficial to them (Ezugwu et al., 2017). This can lead to rape being overlooked and misunderstood in cultural and psychological terms by victims, perpetrators and observers (Barn et al., 2021).

Rationale to Adapt uIRMA Scale

Despite adverse sexual behavior towards women in South Asia and Pakistan little empirical work is available on attitudes and beliefs of general people and people closely involved with rape. Some qualitative-descriptive reports do talk about reasons of such behaviors that have religious, cultural, and psychological underpinnings (Nisar et al., 2021). Other studies have used unstandardized measures of rape and rape myths in Pakistan and provided qualitative reports and prevalence data rather than quantitative data that can be used for further analysis (Zahid & Rauf, 2021). Although the prevalence data is helpful, it usually is localized to a sample or locality and cannot provide objective comparisons that could be generalized across larger populations. Standard scales are therefore needed to measure constructs like rape myth acceptance, its prevalence and its effects in general life, legal matters and policy decisions. The matter is not so simple, a number of valid and reliable measures are available for assessing rape myths that include Attitudes Towards Rape scale (Ward, 1988), the Rape Myth Acceptance (RMA, Burt, 1980) scale, Acceptance of Modern Myths about Sexual Aggression (AMMSA, Gerger et al., 2007) scale, and Updated Illinois Rape Myth Acceptance (uIRMA, McMahon & Farmer, 2011) scale. Schlegel and Courtois (2019) report RMA scale lack conceptual, linguistic and psychometric qualities; similarly, AMMSA scale assesses stereotypical perceptions and beliefs about sexual aggression in subtle ways but may not be sufficient to document or assess rape myths in societies that hold traditional and conservative beliefs such as Pakistan, despite having good psychometric properties. Best scale to assess rape myth acceptance that could be used in Pakistan is the uIRMA scale, which has been widely used in a number of other countries (Fansher & Zedaker, 2022). In addition, when a scale needs to be used in a society it is necessary to evaluate its factor structure and measurement equivalence for respondents that could perceive its constructs differently (Beshers & DiVita, 2021), lack of knowledge of measurement equivalence (Reed et al., 2020) results in poor results. And finally, without validating a scale, interpretations of findings in cross-cultural contexts are often met with doubt (Martini et al., 2022).

Purpose and Objectives of the Study

This study adapted uIRMA (McMahon & Farmer, 2011) scale in Urdu (uIRMA-U) and used a confirmatory factor analysis (CFA) to examine if uIRMA-U scale had a four-factor structure. In addition, the authors wanted to assess if latent measures and construct validity cross gender could be measured. The study also measured convergent and discriminant validities for uIRMA-U scale and determined other psychometric properties of the scale. To measure convergent validity Hostile Sexism subscale (Glick & Fiske, 1996) was used and to assess discriminant validity, Belief in a Just World (Dalbert et al., 1999) was used.

Method

Sample

A purposive sample of 178 men and 238 women (N = 416, no transgenders) based on G-power analysis (Kang, 2021) with 95% confidence interval from four major universities of Lahore, Pakistan was taken. The participants ranged in age from 18 to 68 years with a mean age of 27.86 years. Majority (~ 72%) of the participants were between 18 and 25 years, followed by 26 to 35 years (~ 20%), 36 to 45 years (~ 4%), and 46 to 68 years (~ 4%). Approximately ~ 79% were single, ~ 20% were married and remaining were either divorced or separated. About 85% of the participants were either graduates or post-graduates and others were undergraduates or high schooled. About 71% identified themselves as liberals while ~ 29% viewed themselves as conservative.

Instrument

Updated Illinois Rape Myth Acceptance (uIRMA) Scale

The uIRMA (McMahon & Farmer, 2011) is a 22-item questionnaire that uses a 5-point Likert-type scale (1 = strongly agree, 5 = strongly disagree) divided into four subscales, She Asked for It, It Wasn’t Really Rape, She Lied, and He Didn’t Mean To, to assess rape-supportive beliefs which has cross-cultural relevance (Fakunmoju et al., 2019). Higher scores for the entire scale or subscales depicted higher acceptance of rape myths. The overall reliability of the scale is 0.93 with the subscale alpha ranging from 0.74 to 0.93 (Payne et al., 1999). The uIRMA-U (Kazmi et al., 2023a) was translated in Urdu and tested on the sample to determine its factor structure.

Ambivalent Sexism Inventory (ASI)

Glick and Fiske (1996) developed ASI which consists of 22 items with two 11-item subscales, Hostile Sexism (HS) and Benevolent Sexism (BS) respectively. The scale was translated into Urdu language and used in conjunction with the uIRMA-U for assessing convergent and discriminant validity of the scale. The ASI consists of statements about relationships between men and women, e.g., most women interpret innocent remarks or acts as being sexist (HS) and women should be cherished and protected by men (BS). Each statement is measured on a 6-point scale that indicates level of agreement-disagreement (0 = disagree strongly to 5 = agree strongly). Six (3, 6, 7, 13, 18, and 21) items were reversed scored. Higher scores indicated more hostile and benevolent sexist attitudes toward women. To assess the convergent validity, we used HS which showed good reliability (r = 0.78, Glick & Fiske, 1996).

Belief in a Just World (BJW) Scale

An Urdu translated version of Belief in a Just World Scale (Fatima & Khalid, 2007) was adapted from the original scale (Dalbert, 1999). The scale assesses the extent to which individuals believe in a just world, a world which is fair where people get what they deserve. It comprises a total of 13 items with higher scores showing greater beliefs about fair outcomes in a fair and just world. According to Dalbert (2009) the scale has good internal consistency (Cronbach’s alpha = 0.83). For the purpose of assessing discriminant validity, we assessed correlations between uIRMA-U (overall scores) with belief in a just world with the criterion being a correlation of less than 0.85 as suggested by Garson (2001) and Kenny (2016).

Procedure

The Institutional Ethics Review Board of Government College University, Lahore approved the study. A forward and backward (Urdu-English) translation of the scale was carried out in accordance with the translation guidelines provided by the World Health Organization (Menon et al., 2012). The scale was translated into Urdu by two bilingual experts with extensive experience of translating English scales into Urdu. Second, the two translated Urdu versions of the scale were translated backward into English by two additional linguists with forensic and legal knowledge. Moreover, a panel discussion was held among all translators and two subject specialists to ensure syntactical meanings of items in the scales were clear; some minor modifications were done, and translation process was deemed satisfactory. We published uIRMA-U on WhatsApp messaging distribution list and sent it to 492 participants for further processing. Seventy-six respondents did not complete the informed consent page and were removed, leaving 416 response sets for data evaluation and analysis. To deal with cases were there was some missing data (< 25%) ipsative mean imputation was used (Segerstrom, 2020).

A confirmatory factor analysis was conducted using SPSS 20.0 to identify factor structure of uIRMA-U using a four-factor solution to identify best model fit. Our sample was adequate to run factor analysis, determined by Kaiser-Myer-Olkin (KMO) test (0.88) which was higher the recommended value (0.60, Schreiber, 2021) and the correlation matrix was identical to the identity matrix (Barletts’s Test of Sphericity). We tested four (1 through 4) different models finalizing on a four-factor model (58.6% variance) where a majority of scale items were clustered.

Results

The distribution of the rape myth acceptance score data was normal assessed by Shapiro Wilk test. The mean (M = 50.13, SD = 14.21) of rape myth acceptance was slightly lower than the mean (55) of the composite range score of uIRMA-U. A CFA was performed to test the four-factor model (McMahon & Farmer, 2011). And after removing two error correlations from the 22-item first-order CFA (items 5, 15, and 16), the model fit improved, resulting in adequate indices of RMSEA (0.06), CFI (0.92), and TLI (0.90) for the model. The four factors were dubbed as, She Asked for It, It Wasn’t Really Rape, She Lied and He Didn’t Mean To.

Table 1 displays mean, standard deviation, score range, and internal consistencies among scales and subscales of uIRMA-U. Positive and significant intercorrelations among scales and subscales were low to moderately high (George & Mallery, 2019), which provide evidence for strong construct validity for the uIRMA-U. In addition, the reliabilities were also fair to moderately strong (see Table 1).

Table 1 Psychometric properties of the uIRMA-U scale
Full size table

Convergent and Discriminant Validities

Table 2 shows positive and significant correlations among scales and subscales of uIRMA-U scale for HS, and though these correlations were fair for subscales the correlation was strong for the overall scale. This we believe is due to the number of items in subscales and the total scale; the total scale had more items than its subscales. These correlations showed that individuals who endorsed rape myths also scored high on HS, marked by an overt hostility towards equality for women. Discriminant validity of uIRMA-U scale was tested against BJW scale (see Table 2), positive but weak correlations (r = 0.12 to 0.19) were identified between subscales and uIRMA-U scale, which was way below the threshold suggested by George and Mallery (2019) revealing good discriminant validities.

Table 2 Convergent and Discriminant Validities for uIRMA-U and its Subscales using Hostile Sexism, and Belief in a Just World
Full size table

After meeting the assumptions for conducting an independent sample t-test findings revealed, men and women did not differ significantly with regard to uIRMA-U full scale or its subscales except for the subscales It wasn’t Really Rape where mean rating for men (M = 12.40, SD = 4.44) was significantly, t(414) = − 2.03, p < 0.004, higher than women (M = 11.48, SD = 4.35) suggesting men significantly differed in their beliefs that rape was a myth and women possibly misperceived the event (as shown in Table 3). Similarly, mean rating for men (M = 12.81, SD = 4.23) was significantly t(414) = − 2.13, p < 0.003, higher than women (M = 11.74, SD = 4.02) for He didn’t Mean to subscale. Men believe they men should less likely be blamed for rape because of other factors like sexual drive (Fig. 1).

Table 3 Mean differences in men and women on uIRMA-U and its subscales
Full size table
Fig. 1
figure 1

Confirmatory factor analysis of the updated IRMA scale. All factors had been standardized and significant

Full size image

The Chi square was significant (χ2 = 412.04, df = 142, p < 0.001) to run CFA, AMOS, version 21 (Arbuckle, 2012) for uIRMA-U and check its model fit. The indices for the initial model revealed TLI = 0.88, CFI = 0.90, and RMSEA = 0.07, were good however, after modification the final model (χ2 = 395.39, df = 140, p < 0.001) improved even further with indices, TLI = 0.93, CFI = 0.92. and RMSEA = 0.07. These indices were in the acceptable range (Kenny, 2016) with factor loadings greater than 0.50 (see Table 4). Furthermore, item-total correlations of each item were above the cut-off (0.30) standard (Field, 1978).

Table 4 Factor loadings, item-total correlation of uIRMA-U
Full size table

Discussion

Factor Structure

We adapted uIRMA scale (McMahon & Farmer, 2011) in Urdu (uIRMA-U) with 22 items and four factors or subscales originally taken from a 45-item IRMA scale (Payne et al., 1999) with five factors or subscales. The subscale, He didn’t Mean to-Alcohol from IRMA scale was subsumed into uIRMA (McMahon & Farmer, 2011). After translation and factor analysis uIRMA-U also solved into four factors (see Appendix) with largest cumulative variance (59.93%) for the final CFA model and that the items matched perfectly across the two versions (English and Urdu) of the scales. The original IRMA scale (Payne et al., 1999) was divided into five factors, and others for example Balezina and Zakharova (2023) have also reported a five-factor structure of IRMA across various cultural and ethnic groups in Russia (also see Canan et al., 2022 for American students). Similarly, Skov et al. (2022) in a Danish sample (uIRMA scale) and Fakunmoju et al. (2019) in a Nigerian sample (IRMA scale) confirmed the five-factor structure for the two versions of the scale. However, Canan et al. (2022) assessed and validated psychometric properties of IRMA scale using a sample of American students and reported a four-factor solution. Payne et al. (1999) had pointed out that the true structure of IRMA scale was not cast in stone, that is why later research divided the scale into eight factors (Feild, 1978), five factors (Das & Bhattacharjee, 2021, see studies above) and four factors (Briere et al., 1985; McMahon & Farmer, 2011) etc. And though Das and Bhattacharjee (2021) divided the structure of uIRMA into five factors, our data could not separate the three items on alcohol (or intoxication) into a separate factor. However, the internal consistency of uIRMA-U subscales and the overall scale were similar to Das and Bhattacharjee (2021). Moreover, all subscales positively correlated and significantly with each other and with the overall scale signifying construct validity (and in some fashion convergent validities, Table 2). The correlations, however, were moderate, which suggests that it is best to use the uIRMA-U as a scale with four factors and not as a single scale measure as argued by McMahon and Farmer (2011). The convergent and discriminant validities were acceptable; individuals who endorsed rape myths scored higher on HS as evidence for convergent validity, and weak correlation between the uIRMA and BJW supported the discriminant validity (see Kenny, 2016; Nyúl & Kende, 2021 for details). Moreover, Hayes et al. (2013) found that those having weaker BJW scores had a lower tendency to accept rape myths. Some studies have also found an inverse relationship between JWB score and the level of blame attributed on women for their rape (Kleinke & Meyer, 1990; Sinclair & Bourne, 1998), and it may be argued that rape myths are often based on a random and unpredictable event which can undermine the overall sense of control in those that believe in just world. The relationship between BJW and rape myth acceptance is complex that needs more research to assess these relations effectively.

Moreover, CFA showed a four-factor structure of uIRMA-U for a Pakistani sample laid out by first-order CFA model. After removing two error correlations from the 22-item first-order CFA (items 5, 15, and 16), the model fit improved, resulting in adequate indices of RMSEA (0.06), CFI (0.92), and TLI (0.90) for the model. The four factors were dubbed as, She Asked for It, It Wasn’t Really Rape, She Lied and He Didn’t Mean To. These findings supported the facets of rape myths e.g., blame attribution to rape victims and inanimate objects [alcohol], exoneration of perpetrators, minimizing rape victimization, and believability of the dilemma of the rape victim that contribute to understanding rape-supportive beliefs in Pakistan (Jamshed & Kamal, 2021; Kazmi et al., 2023a). Some validation of the above is reported in other research (Ali & Khan, 2007) that reported that alcohol consumption may be a contributing factor behind the violence towards women. However, Murshed (2010) argues, sexual assault or violence against women under the influence of alcohol is not sufficient reason to cover all other aspects violence against them and that this sole belief pervades thinking because of the justice system in the country. He further argued that there are a growing number of cases of sexual assault in which no form of alcohol is involved. Thus, a separate fifth factor, He didn’t Mean to-Alcohol did not reveal for our data as a rape myth.

Implications for Theory and Practice

The uIRMA-U scale can be useful in examining the degree of acceptance of rape myths and other rape-supportive beliefs and behaviors in men and women of Pakistan. McMahon (2010) for example concluded that rape myths should be integrated into bystander intervention programs. Therefore, uIRMA-U scale can be employed to identify the effects of rape myths on the willingness to protect victims of rape, which may ultimately improve research and practice in rape prevention and intervention programs in Pakistan. Assessing rape myth acceptance can become the first step towards rape prevention programs in Pakistan.

Strengths and Limitations

Adaptation and validation of uIRMA-U scale for a Pakistani sample has implications for cross-cultural research, e.g., attitudes towards rape and rape myth acceptance in Pakistani men and women can be compared to men and women in other countries and nations. This is important for an underrepresented region such as Pakistan where data on rape myth acceptance is non-existent. It should be noted that rape myth acceptance is not only associated with personal attitudes it is also aligned with societal oppressive beliefs and sharp social boundaries between genders. We think one limitation of adapting the scale stems from the fact that sample was restricted geographically (Lahore), age (age range 20 to 32 years) and socioeconomic status (low to middle). We do understand most adaptations of IRMA and uIRMA scales like many other scales in social sciences are based on university students. However, the next step should be validating this scale with other respondents from the general population and other regions, especially rural areas of Pakistan so that better understanding could be established about rape myth acceptance. Investigating the prevalence of rape myths across the country could be beneficial in determining the extent to which they lead to rape-supportive behaviors. And some testing should be carried face-to-face because it is possible that the respondents may be different in significant ways who had access to the Internet and completed the scale online. In addition, authors believe discriminant validity could be measured against other scales (like, Passionate Love Scale, Hatfield & Sprecher, l986) that tap dissimilar constructs to rape myth acceptance.

Conclusion

Our data divided uIRMA-U scale into four factors which is consistent with previous studies on rape myth acceptance. The reason for not finding the fifth factor is largely due to incorporation of three items on alcohol in the uIRMA scale (McMahon & Farmer, 2011) and possibly religio-cultural in Pakistan where alcohol use is a taboo. We believe four factor structure of the uIRMA-U scale is well supported and the scale is a valid and reliable to measure of rape myth acceptance within the Pakistani context and can assess the prevalence of rape myth acceptance in this culture.