Sexual Assault Myths Acceptance in University Campus: Construction and Validation of a Scale

Abstract

The study related to the myths of rape has been carried out since the 1980s at different scales. However, the interaction between the acceptance of these rape myths (myths related to sexual abuse) and the nightlife scene—where alcohol consumption becomes the epicenter of this particular context—has not been specifically evaluated. In this work, a questionnaire has been developed considering different scales. It has been tested online in a population of 367 first-year undergraduate students at the University of Alcalá (Spain). The results of the exploratory and confirmatory factor analysis, as well as the reliability ones, indicate the adequacy of the scale construction and validation process for the university student population. In addition, the results obtained, in line with the specialized literature, indicate that the consumption of alcohol and other drugs appear as justifying elements of sexual violence, exonerating the aggressors and perpetrating the victim.

1. Introduction

The aim of this research is to develop our own instrument to better measure the acceptance of myths of sexual assault on university campuses, paying special attention to those aspects related to partying and nightlife. For this, the existing scales and those items that, according to the literature, were more representative of the discourses and beliefs of this type of violence in the context of nightlife, have been reviewed. The new scale has been applied to a sample of 367 first-year students of various degrees at the University of Alcala, during the 2020–2021 academic year, through an online questionnaire conducted through the SurveyMonkey platform. The article is presented as follows: first a review is made on the measurement of the rape myths, and then the methodology followed during the construction of the scale and the field work is presented. Third, the results of the validation are presented to end with the conclusions.

Measuring the Rape Myths

Any myth can be defined as a set of schemes, or ideas of meanings that a major percentage of the population recognises and has internalised, and therefore circulates as a discourse of authority (Barjola 2018). Among these beliefs and stereotypes, the existence of rape myths stand out, which refer to false, stereotypical beliefs about the victims, the aggressors and the situational conditions that derive in a social judgement of an act such as a sexual aggression or a consensual sexual encounter (Saldívar Hernández et al. 2015; Sinko et al. 2020). Rape myths were introduced as a concept in a sociologic and a feminist context since the 70s (Payne et al. 1999), although it was not until the next decade, when Martha Burt created the first instrument to measure it (Burt 1980). In the case of rape myths, there are a shared set of beliefs, constant in time, that deny or justify the sexual aggression that took place, supporting and perpetuating male sexual violence against women (Bohner 1998; Lonsway and Fitzgerald 1994; Payne et al. 1999; Romero 2012; Saldívar Hernández et al. 2015). Furthermore, these myths in particular move the focus point of aggression and the aggressor toward the victim and their behaviour, judging these last two aspects to a greater extent (Sipsma et al. 2000; Velte 2019). They add to their definition that they are wrong myths in an ethical sense (Gerger et al. 2007).

Rape myths can be descriptive or prescriptive (Bohner 1998), and depending on the object to which they are addressed, they can be classified into four broad categories. First, there are those who blame the victim, for example, how she was dressed or how much she had drunk; secondly, they highlight those related to the aggression itself, define what is understood by rape and determine what requirements the aggression must meet in order to truly be considered an aggression; thirdly, there are those who exonerate the aggressor; and fourth and last, those that determine the types of women and their respectability and credibility (Gerger et al. 2007; Romero 2012). The persistence of many of them over time is unquestionable. For example, regarding the exoneration of the aggressor, rape has been justified since the 18th century through the belief that the male must calm his sexual needs, a model that has been called “steam boiler” (Sanyal 2019). In fact, others derive from the myth of male sexual incontinence, such as those related to provocative clothing or the amount of alcohol ingested by the victim, since by having these behaviours, the victim exposes themself to male sexual desire, which can be uncontrollable by their primary/ primal needs (Sanyal 2019). As can be seen, these examples transfer the guilt to the victim and exonerate the aggressor. Examples of myths about “what a rape must look like to really be rape” include those who claim that for it to be a sexual assault, the victim must resist or there must be physical contact. Finally, in those related to the types of women and their behaviours, those who defend that a woman who has had many sexual partners does not have as much credibility of that when women says no, they actually mean yes (Romero 2012; Sipsma et al. 2000).

The scientific measurement of the acceptance of these myths began in the 1980s, from Burt’s work (1980). She developed an instrument for this purpose, the Rape Myth Acceptance Scale (RMAS). However, it should be noted that there was a previous scale, which instead of measuring beliefs, it was focused on attitudes: the Attitudes Toward Rape Scale (ATR), created by Feild (Field 1978; Schlegel and Courtois 2019). Given the moderate quality of this scale, and the non-use of myth terminology, its use was not as widespread as it was with Burt’s work (Schlegel and Courtois 2019). In fact, since the publication of Martha Burt, many others have been developed, replicating it, adapting it to other geographical and temporal contexts or analysing the relationship between these myths with a wide range of variables related to other beliefs, attitudes, and behaviours (Briere et al. 1985; Costin 1985; Expósito et al. 2014; Janos and Espinosa 2018; Malamuth 1981; Megías et al. 2011; Payne et al. 1999; Quackenbush 1989). In this way, there has been an important development of RMAs in the last forty years (Gerger et al. 2007).

Among the variants derived from Burt’s first work, the Illinois Rape Myth Acceptance (IRMA) stands out for its extensive use. A scale developed by Payne et al. improves the psychometric properties and the validity of the construct and includes a less ambiguous language, which facilitates its application (Bergenfeld et al. 2020; Payne et al. 1999). Payne et al. developed two versions of the same scale: one made up of 45 items; and a shorter and more adaptable one, with 20 items (Payne et al. 1999). A second scale, widely used and more current, is the one developed by Gerger, Kley, Bohner and Siebler, the Acceptance of Modern Myths About Sexual Agression (AMMSA) (Megías et al. 2011; Tardón 2017). These authors observed in the scales used that the scores were located in the lowest options. As the distributions did not maintain the shape of the normal, therefore, they made their statistical analysis difficult. These results lead them to think that the participants are increasingly aware of what is politically correct (social desirability) in relation to sexual assaults (Gerger et al. 2007; Schlegel and Courtois 2019). In addition, they consider that, as stated by Swim et al., attitudes and beliefs towards racism or sexism are becoming more subtle, leading to a modernisation of beliefs (Swim et al. 1995). Faced with such hypotheses, they elaborated the AMMSA scale (e.g., Acceptance of Modern Myths about Sexual Aggression), achieving considerable improvements in its results in the distribution of scores among the participants (Gerger et al. 2007). The psychometric properties of this scale have been tested in English, German, Greek and Spanish versions (Hantzi et al. 2015; Megías et al. 2011; Schlegel and Courtois 2019).

In recent years, as a result of Illinois Rape Myth Acceptance, McMahon and Farmer have updated the item language of this scale to build a new one that measures the most subtle myths, finally made up of a total of 22 items. Their results, even with limitations, indicate that men have a higher rate of acceptance of these myths (McMahon and Farmer 2011). Their work is validated in the Italian context by Martini, Tartaglia and de Piccoli in 2021, however, these authors exclude items related to alcohol intoxication from the scale (Martini et al. 2021). Another very interesting scale to highlight is the Alcohol and Sexual Consent Scale, an instrument composed of 12 items, focused on the dimensions of alcohol consumption, sexual habits and sexual victimization in the specific environment of university campuses (Ward et al. 2012). These researchers designed this scale with the intention of measuring the attitude towards sexual consent in the presence of alcohol. Among their results, it is observed that a more permissive attitude towards the acceptance of sexual consent under the influence of alcohol is associated with a higher level of rape-supportive attitudes.

In the Spanish context, three different scales have been adapted and implemented on the subject matter of this study. First, Megías et al. adapted the AMMSA scale designed by Gerger and his team into Spanish, applying it to students at the University of Granada. For the validation of the instrument, two studies were carried out with 305 and 263 students respectively. Their results show high internal consistency, and, in addition, they achieve high scores on the scales, with distributions close to normal, which suggests the correct adaptation and good measurement adjustment of the AMMSA scale in the Spanish context took place (Megías et al. 2011). Of the 30 items included by Megías et al., two of them are linked to sexual assaults in festive settings and alcohol consumption: “If a woman invites a man to have a drink at home after having gone out at night, it means she wants to have sex and alcohol is often the cause of a man raping a woman.”. Secondly, Expósito et al. adapted the Illinois Sexual Harassment Myth Acceptance (ISHMA) using a double sample: on the one hand university students (339 participants), on whom the exploratory analysis is carried out; and, on the other, the general population (326 participants), a sample that is used to confirm the results found in the exploratory analysis (Expósito et al. 2014). The ISHMA scale was originally developed by Lonsway et al. in 2008, and focuses on sexual harassment mythology, giving exceptional weight to sexual harassment in the work environment (Lonsway et al. 2008). However, statements are made that reproduce situations and beliefs that do occur in nightlife contexts (Velte 2019): “If a woman is sexually harassed, she had to have done something to provoke it”. Finally, in 2017 the Sociological Research Center (CIS, Centro de Investigaciones Sociológicas, in Spanish) carried out a study on the social perception of sexual violence in a sample of 2465 participants over 16 years of age, from all over Spain. In the questionnaire, several questions are included in a battery (Likert type 0–5) on common beliefs related to sexual violence (Centro de Investigaciones Sociológicas 2017). Specifically, question number 6 sets out beliefs about the responsibility of women in the event of being sexually assaulted. Among the 6 items that make up this question, this explicitly related to the DFSA study, it is stated: “If a woman is sexually assaulted while drunk, she is partly to blame for having lost control”. In Peru, Janos and Espinosa developed the scale Acceptance of Myths about Sexual Violence, which is brought up for not only its usefulness in research in Spanish-speaking contexts but also its recent application (2018) and the breadth of the focus to be studied, due to the fact that it does not only focus on rape myths, but also considers them in a general context, on false beliefs associated with sexual violence. This scale is made up of 20 items with Likert-type response options (1–4), constructed from a qualitative research carried out by the same authors (Janos and Espinosa 2015, 2018).

Despite the diversity of scales and contexts in which they have been applied, there is an agreement that men present a higher degree of acceptance of this type of myth than women do (Expósito et al. 2014; Megías et al. 2011). However the results obtained in the investigations do not reach robust conclusions in relation to other sociodemographic variables (Payne et al. 1999). According to the literature, an association has been found between RMA scores and the probability of: perpetrating sexual violence; not intervening as a witness; not reporting the assault; or not feeling danger or vulnerability to rape in the case of women. The higher the score on the RMA scale, the greater the probability of having the aforementioned behaviours (Bergenfeld et al. 2020; Bohner et al. 2009; Bohner and Lampridis 2004; Megías et al. 2011).

After this review, there is a lack of measurement instruments that specifically reflect the acceptance of myths about sexual assault in university campus contexts, where the consumption of alcohol and other drugs in parties becomes a central element. For this reason, the aim of this research is to develop our own instrument to better measure the acceptance of myths of sexual assault on university campuses, paying special attention to those aspects related to partying and nightlife. To do this, the existing scales have been reviewed and selected those items that were more representative of the discourses and beliefs of this type of violence in the context of nightlife, where there is a strong presence of alcohol and drug consumption. The new developed scale has been applied to a sample of 367 first-year students of various degree programmes from the University of Alcalá, during the 2020–2021 academic year, through an online questionnaire conducted through the SurveyMonkey platform. In addition, the ultimate goal of this work is to promote the achievement of Sustainable Development Goal number 5—equality for women and girls—to achieve a more just, peaceful, and democratic world, according to the United Nations (General Assembly of the United Nations 2015).

2. Method

2.1. Scale Construction

For the elaboration of our instrument, five scales related to the acceptance of sexual assault myths have been used, all of them listed in detail in

Table 1. Different scales used in the elaboration of the instrument ¹.

Authorship	Year	Participants	Scale	Items	Likert’s Range	Cronbach’s α	Country
Payne et al.	1999	University students	IRMA-SF	20	1–7	0.87	United States
Megías et al.	2011	University students	AMMSA	30	1–7	0.91	Spain
Expósito et al.	2014	University student and general population	ISHMA	20	1–7	0.91	Spain
CIS	2017	Resident population	–	6	0–5	0.72	Spain
Janos and Espinosa	2018	Resident population	SVMA	20	1–4	0.90–0.76	Peru

¹ Note: IRMA-SF = Illinois Rape Myth Acceptance—Short Form, AMMSA = Acceptance of Modern Myths about Sexual Aggression, ISHMA = Illinois Sexual Harassment Myth Acceptance, SVMA = Sexual Violence Myths Acceptance. Source: Own elaboration.

. In several cases, the scales share identical or similar items, although the response options vary and Likert-type responses of 4, 5 or 7 values are used, as shown in Table 1.

The recommendations for the application of the expert judgment in the selection of items were incorporated in our instrument (Cabero Almenara and Barroso Osuna 2013). Therefore, 4 academic experts in the specific content to be measured collaborated –the acceptance of drug-facilitated sexual assault myths–, as well as an expert in social research methodology. To reach the expert judgment, the consensus method as nominal group technique was chosen, which favours reaching a consensus among the participants in a small group on the identification of the problem, the development of solutions and the establishment of priorities. This method is widely used in health care and research, as well as in education and social services (Manera et al. 2019; Potter et al. 2004).

The design of the instrument was carried out between the months of November 2019 and April 2020. The items incorporated in the scale are listed in

Table 2. Items incorporated in the scale ¹.

E1—A woman who dresses provocatively should not be surprised if a man tries to force her to have sex.

E2—If a woman is sexually assaulted while drunk, she is partly to blame for having lost control.

E3—A woman who has had many sexual partners has less credibility if she reports an assault.

E4—If a woman does not intend to have sex with a man, she should not flirt with him.

E5—A woman is more likely to be raped by a stranger than by an acquaintance.

E6—For men it is a biological need to release their accumulated sexual tension from time to time.

E7—Although women like to be shy, that does not mean they do not want sex.

E8—Alcohol is often the cause of a man raping a woman.

E9—Many women tend to interpret well-intentioned gestures as sexual harassment.

E10—If a woman is sexually harassed, she must have done something to provoke it.

E11—Almost all types of sexual harassment would end if the woman simply told the man to stop.

E12—For an act of sexual violence to take place, there must always be physical contact.

E13—When a woman allows herself to be invited to drinks at a disco by a man, it means that she wants sex with him.

¹Source: Own elaboration.

. The scale consists of 13 items, measured in a Likert-type format of 4 values (4 = strongly agree; 1 = strongly disagree). The intermediate option (neither agree nor disagree) has been eliminated to constrain the student body under study to position themselves and avoid what Schuman and Presser call “intermediate alternative floats” (Schuman and Presser 2017). In all the items, the option “no answer” has been given, thus respecting the participant’s voluntary nature (Bisquerra and Pérez-Escoda 2015; Díaz de Rada 2001; Gilljam and Granberg 1993).

Of the 13 items incorporated, three of them explicitly mention alcohol consumption as a justifying or contextual element of the aggression, clearly linked to the DFSA cases. The rest of the items that make up the scale, even not explicitly mentioning the festive atmosphere or the consumption of alcohol or other drugs, the experts agreed on the importance of their incorporation in the study. Hence, on the one hand, the link between these and social discourses in cases of high impact and controversial drug-facilitated sexual assaults in Spanish society, such as the well-known case of La Manada (Aurrekoetxea-Casaus 2020; Larrondo et al. 2019; Navarro and Coromina 2020; Robles et al. 2019; Velte 2019), expressed by Sanyal as the Spanish #MeToo moment (Sanyal 2019). On the other hand, the work carried out in Spain by the Noctámbul@s Observatory has been key in the selection of items. From 2014 to 2018, the date of the last published report, this institution investigated the relationship between drug use and sexual violence in nightlife spaces, releasing figures on the incidence of the phenomenon in Spanish youth (Observatorio Noctámbul@s 2014, 2015, 2016, 2017, 2018).

2.2. Instrument Validation

2.2.1. Participants

The scale developed has been tested in a convenience sample made up of 367 first-year undergraduate students from the University of Alcalá (Alcalá de Henares, Madrid, Spain) through the implementation of an online questionnaire carried out on the SurveyMonkey platform. Regarding the sociodemographic traits that characterise the sample, 126 are men (34.3%) and 241 (65.7%) are women. The mean age for the sample as a whole is 18.8 years (SD = 2.0) and its ideological position is 4.6 (SD = 2.3), having been measured on a scale where 1 is extreme left and 10 extreme right. 67.7% of the participants (243 cases) come from the health sciences areas of study and 32.3% (116 cases) from the social sciences. Regarding nationality, 326 (89.3%) have Spanish nationality, 22 (6.0%) have dual nationality (Spanish and other) and 17 (4.7%) have another nationality. Finally, 40 (11.6%) declare themselves practicing Catholics, 77 (22.3%) non-practicing Catholics, 16 (4.6%) believers of another religion, 61 (17.7%) agnostic, 82 (23.8%) atheist and for 69 (20.0%) religion is an indifferent matter to him/her.

Participation has been on a voluntary and anonymous basis, providing students with a link through which to access the questionnaire. In order to access the survey, they had to read and accept basic information about the study, such as the nature of the survey, the average duration, contact information, voluntary participation, etc. In this way, the informed consent of all the participants was obtained. In addition, the ethical aspects of the research have been supervised and approved by the Committee of Research Ethics and Animal Experiments of the University of Alcalá (CEI/HU/2020/21).

2.2.2. Statistical Analysis

For the statistical analysis, the IMB SPSS v.25 Statistical Package and the equation modelling software, AMOS™ 22.0 were used. An exploratory factor analysis (EFA) was performed on the sample, using the maximum likelihood extraction method (ML) and the Varimax rotation method. The selection of the final model after the EFA was agreed based on the following indicators: (a) size of the determinant of the correlation matrix; (b) measure of Kaiser, Meyer and Olkin (KMO) sample adequacy; (c) significance from Bartlett’s sphericity test;(d) values of the diagonal of the anti-image correlation matrix; (e) values of the commonalities of each variable; and, (f) the total variance explained by the model; along with the principles of parsimony and interpretability (Díaz de Rada 2018; Méndez Martínez and Rodón Sepúlveda 2012). Through this analysis, the aim was to discover the number of factors underlying the scale, as well as which variables or items on the scale were indicators of what was to be measured (Fernández Aráuz 2015).

After obtaining a relevant model, the reliability of the scale that they make up was checked using Cronbach’s alpha coefficient measure (Cronbach 1951). The Cronbach coefficient is a measure of the internal consistency of the scale, which indicates the covariance of the items and the presence of the object of study in these items (Oviedo and Campo-Arias 2005; Ventura-León 2017; Ventura-León and Caycho-Rodríguez 2017). This statistic tries to collect the capacity of the scale to be designed to measure consistently and accurately the characteristics that it claims to measure (Pérez López 2009). To consider that a scale has good consistency, an alpha value greater than 0.7 is required (Martín Arribas 2004).

Finally, we wanted to know the fit of the factorial model obtained by performing a confirmatory factor analysis (CFA) using equation modelling software (AMOS™ 22.0) on the same sample (Damar et al. 2020). The aim of exploring a structure with EFA and successively conduct a CFA in the same sample is to test the validity of those restrictions implied by the CFA which were not part of the EFA (Doğan et al. 2017; Steinmetz et al. 2020).

The CFA indicators that determine the fit of the model are: (a) the model chi-square and degrees of freedom; (b) the comparative fit index (CFI); and (c) the root mean square error of approximation (RMSEA) (Bollen 1989; Browne and Cudeck 1993; Custers and McNallie 2016; Plaza-Vidal et al. 2020). For a good fit of the model, scores lower than 0.06 are required for the RMSEA and values above 0.90 for the CFI (Custers and McNallie 2016; Plaza-Vidal et al. 2020).

3. Results

3.1. Exploratory Factor Analysis

In the exploratory factor analysis, together with the model’s goodness-of-fit indicators, there are two principles that should govern the researcher’s decisions: (a) the principle of scientific parsimony and (b) that of interpretability. The first refers to the fact that the best solution is the simplest, that is, that the phenomenon studied is possible to explain with the least number of factors. In addition, solution must be understandable, have a meaning based on the theory on which the understanding of the phenomenon is based (Díaz de Rada 2018; Pérez López 2009). After factorial exploration, in which up to five models with different numbers of factors and variables have been tested (see Table 2), according to the indicators and principles of factor analysis set out above, a model composed of 10 of the 13 variables was obtained (items of the original scale). Three factors were eliminated: (E7), (E11) and (E13) (see

Table 3. Communalities ¹.

Items	Communalities
E1—A woman who dresses provocatively should not be surprised if a man tries to force her to have sex.	0.645
E2—If a woman is sexually assaulted while drunk, she is partly to blame for having lost control.	0.730
E3—A woman who has had many sexual partners has less credibility if she reports an assault.	0.574
E4—If a woman does not intend to have sex with a man, she should not flirt with him.	0.490
E5—A woman is more likely to be raped by a stranger than by an acquaintance.	0.578
E6—For men it is a biological need to release their accumulated sexual tension from time to time.	0.400
E8—Alcohol is often the cause of a man raping a woman.	0.638
E9—Many women tend to interpret well-intentioned gestures as sexual harassment.	0.562
E10—If a woman is sexually harassed, she must have done something to provoke it.	0.778
E12—For an act of sexual violence to take place, there must always be physical contact.	0.587

¹ Source: Own elaboration.

). The first two due to their low communalities (less than 0.4) and the third because they belong to a factor composed solely by itself.

With the applied factorial model, a KMO equal to 0.792 was obtained, a statistically significant Bartlett’s sphericity test (Chi-square = 685.557; p-value < 0.05) and which explained 59.83% of variance. The determinant of the correlation matrix was low (0.075), indicating the intercorrelation of the variables and the possibility of reducing their dimensions (Pérez López 2009). On the other hand, the values of the diagonal of the anti-image correlation matrix, indicators that follow the same logic as the KMO statistic (the closer to the unit, the better fit of the model) and indicate the adequacy of each variable to the model (Díaz de Rada 2018), were high and adequate, all of them being between 0.700 and 0.889. The communalities, reflected in Table 3, had a sufficiently high value to admit the model, although it is true that two items are obtained with communalities with values between 0.5 and 0.4.

Table 4. Rotated Component Matrix ¹.

Item	FACT1	FACT2	FACT3
E1—A woman who dresses provocatively should not be surprised if a man tries to force her to have sex	0.784	0.159	0.077
E2—If a woman is sexually assaulted while drunk, she is partly to blame for having lost control	0.813	0.207	0.161
E3—A woman who has had many sexual partners has less credibility if she reports an assault	0.39	−0.044	0.647
E4—If a woman does not intend to have sex with a man, she should not flirt with him	0.197	0.671	0.042
E5—A woman is more likely to be raped by a stranger than by an acquaintance	−0.138	0.413	0.623
E6—For men it is a biological need to release their accumulated sexual tension from time to time	0.045	0.629	−0.044
E8—Alcohol is often the cause of a man raping a woman	0.02	0.064	0.796
E9—Many women tend to interpret well-intentioned gestures as “sexual harassment”.	0.243	0.674	0.219
E10—If a woman is sexually harassed, she must have done something to provoke it	0.84	0.261	−0.062
E12—For an act of sexual violence to take place, there must always be physical contact	0.282	0.68	0.214

¹ Source: Own elaboration.

shows the results of the matrix of rotated components, where the saturation of each variable in each factor is represented. After the Varimax rotation, the first factor (FACT1) explains 23.33% of the variance of the phenomenon, the second (FACT2) 20.76% and the third (FACT3) 15.74%. The higher the saturation value of the variable in the factor –in absolute terms, since the +/− sign indicates the direction of the variable-factor relationship–, the more said variable will be identified or belong to the factor (Comrey 1985; Díaz de Rada 2018).

Based on these criteria, it is stated that the first factor (FACT1), which explains 23.33% of the variance of the phenomenon being measured, is made up of the variables:

• (E1) “A woman who dresses provocatively should not be surprised if a man tries to force her to have sex”;
• (E2) “If a woman is sexually assaulted while drunk, she is partly to blame for having lost control of her”;
• (E10) “If a woman is sexually harassed, she had to do something to provoke it”.

It was decided to call this factor “Responsibility of the Victim”, since they are the a group of items on the scale that pose the responsibility of women, who are victims of sexual violence, based on different behaviours that she demonstrates, such as the clothes she wears (E1) (Gerger et al. 2007; Romero 2012).

Regarding the second factor (FACT2, 20.76% of total variance), it is made up of the following items:

• (E4) “If a woman has no intention of having sex with a man, she should not flirt with him”;
• (E6) “For men it is a biological need to release their accumulated sexual tension from time to time”;
• (E9) “Many women tend to exaggeratedly interpret well-intentioned gestures”;
• (E12) “For an act of sexual violence to take place there must always be physical contact”.

The grouping of items in FACT2 leads researchers to name this factor as “Social Conventions of Sexuality”, when collecting recommendations as well as descriptions related to sexuality and sexual relations (Sanyal 2019; Velte 2019).

Finally, in the third factor (FACT3, 15.74% of the total variance) the following items of the scale are grouped:

• (E3) “A woman who has had many sexual partners has less credibility if she reports an assault”;
• (E5) “A woman is more likely to be raped by a stranger than by an acquaintance”;
• (E8) “Alcohol is often the cause of a man raping a woman”.

The set of variables grouped in the FACT3 lead to it being named by the research team as “Context of Sexual Assault”, since the items that it comprises are related to contextual elements (Gerger et al. 2007; Romero 2012; Sanyal 2019; Tardón 2017; Velte 2019).

3.2. Reliability Analysis

The reliability analysis included 270 cases out of 367 in total, excluding those that had some missing value. For the analysis, Cronbach’s α was calculated for the sample as a whole and for subsets of that, in order to observe if the value of the statistic remains constant. The complete results are given in

Table 5. Cronbach’s Alpha study ¹.

Item	Total Sample	H.S. Students	S.S. Students	Women	Men
Valid cases	270	175	90	176	91
Cronbach’s Alpha	0.738	0.752	0.715	0.675	0.764
Cronbach’s Alpha if item is removed
E1—A woman who dresses provocatively should not be surprised if a man tries to force her to have sex	0.725	0.741	0.702	0.659	0.754
E2—If a woman is sexually assaulted while drunk, she is partly to blame for having lost control	0.711	0.726	0.7685	0.634	0.745
E3—A woman who has had many sexual partners has less credibility if she reports an assault	0.723	0.729	0.722	0.674	0.729
E4—If a woman does not intend to have sex with a man, she should not flirt with him	0.712	0.734	0.663	0.647	0.748
E5—A woman is more likely to be raped by a stranger than by an acquaintance	0.724	0.740	0.703	0.664	0.746
E6—For men it is a biological need to release their accumulated sexual tension from time to time	0.731	0.748	0.697	0.664	0.761
E8—Alcohol is often the cause of a man raping a woman	0.735	0.747	0.726	0.659	0.769
E9—Many women tend to interpret well-intentioned gestures as “sexual harassment”.	0.691	0.707	0.663	0.635	0.720
E10—If a woman is sexually harassed, she must have done something to provoke it	0.723	0.740	0.690	0.654	0.752
E12—For an act of sexual violence to take place, there must always be physical contact	0.685	0.700	0.657	0.628	0.704

¹ Note: H.S.= Health Science; S.S. = Social Science. Source: Own elaboration.

. A high value of Cronbach’s α is obtained, 0.738, which is why the scale is considered to have good internal consistency, as it is higher than 0.700 (Martín Arribas 2004). In general terms, the statistic maintains values above 0.700 for the proposed sample subsets, with the exception of women, where it decreases slightly. In the case of Health Sciences students, Cronbach’s α is 0.752; for Social Sciences students it is 0.715; for women 0.675 and for men 0.764. Given these results, it can be affirmed that the reliability of the scale obtained after the exploratory factor analysis is high and therefore is suitable for the purposes of the research. In addition, the reliability of each of the factors found is analyzed. In the case of FACT1 (E1, E2 and E10), a high Cronbach’s alpha (0.746) is obtained, however, in FACT2 (E4, E6, E9 and E12) and in FACT3 (E3, E5 and E8) the indices of reliability decreased slightly, obtaining 0.666 and 0.544 respectively.

3.3. Confirmatory Factor Analysis

In order to evaluate the factorial structure, a confirmatory factor analysis (CFA) was performed considering three first-order dimensions. The fit of the model of three correlated latent factors was evaluated: Responsibility of the Woman Victim, Social Conventions of Sexuality and Context of Sexual Assault. After adjusting the three-factor model, the following results are observed: CMIN/DF = 1338; CFI = 0.954; GFI = 0.985 and RMSEA = 0.036. According to the scientific literature, the values of the adjustment indices are within theoretically expected (Bollen 1989; Browne and Cudeck 1993; Byrne 1989; Custers and McNallie 2016; Kaplan 2009; Plaza-Vidal et al. 2020). A significant value of the Chi-Square test is presented, which allows CMIN/DF < 2 (Byrne 1989), the mean square error RMSEA is less than 0.05 (Kaplan 2009; Mulaik 2009), which is recommended, and the Comparative fit index (CFI) is within the range > 0.95 (Geiser et al. 2012). Based on these indices, we can affirm that the model developed is good and fits well with the empirical data (Cole and Maxwell 1985). Thus, the results confirm the construct validity and allow us to affirm that the model is relevant to verify the objectives proposed in this work, as can be seen in Figure 1, where the composition of the model obtained in the confirmatory factor analysis is represented.

Figure 1 shows the relationship between the factors (FACT1, FACT2 and FACT3), the scale items in each factor (E1, E2, E3, etc.) and the measurement errors of each of the items (e1, e2, e3, etc.). For a better understanding of Figure 1, the ovals correspond to the latent variables (factors), the rectangles to the observed variables (items) and the circles to the residual error. The pathways connecting the ovals represent interrelationships of latent variables; the paths from the ovals to the squares represent relationships of the latent variables over the observed ones. The direction of the arrows shows the causal relationships between the variables predicted by the model. The numbers above the arrows correspond to the standardized parameters (in which the effect that other variables may have on that relationship has been extracted). The numbers above the rectangles correspond to the communalities (the part of the variance explained by the factor associated with the variable).

4. Conclusions

A scale to measure the acceptance of drug-facilitated sexual assaults myths is an indicator of the beliefs and stereotypes that a population has around this type of violence. Therefore, the scale is a useful instrument to diagnose those areas in which a greater effort must be made to change such beliefs, which make it difficult to report and eradicate sexual violence in contexts in which there is consumption of psychoactive substances (Isorna and Rial 2015; Krug et al. 2002; Prego-Meleiro et al. 2020). As stated at the beginning, this work seeks to build a specific instrument for the measurement of sexual assault when there has been substance use by the victim and/or aggressor, an instrument that is in turn not very extensive, and can be included in larger questionnaires.

The results obtained support the construction and validation of a new scale for measuring the acceptance of drug-facilitated sexual assaults myths in the universitary population. In general terms, the scale shows adequate properties: high reliability and construct validity. The exploratory factor analysis demonstrates the existence of dimensions or latent variables of a greater degree of abstraction present in the scale, whose composition is theoretically consistent with the specialised literature by highlighting dimensions related to the responsibility of the victim, the beliefs about how it is a “true violation” and the influence of exogenous factors (Gerger et al. 2007; Romero 2012; Sanyal 2019). In addition, the reliability of the scale that results from the factorial model obtained is high, both for the sample as a whole and for the subsamples. On the other hand, the confirmatory factor analysis ends up indicating the goodness of fit of the model, supporting the construction of the scale.

According to the literature, the consumption of alcohol and other drugs is an element that, when present in a sexual assault, is used discursively and even judicially, as an element that exonerates the aggressor or aggressors (Menéndez 1991) at the same time as blames or holds the victim responsible (Observatorio Noctámbul@s 2018; Schuller and Wall 1998; Velte 2019). The results obtained in this research point in the same direction since those items that explicitly mentioned the consumption of alcohol or the state of intoxication of the victim are located in the factors “responsibility of the victim” and “context of sexual assault”. Precisely, the myth that alcohol is the cause of the aggression appears represented in the same factor as the item that indicates that it is more common for the aggressor to be an unknown person than known to the victim. This is another one of the major issues reported in DFSA cases, where most attacks are committed by someone close to the victim, highlighting a partner, ex-partner or friend (Barreto 2018; Isorna and Rial 2015).

The results of this research should be interpreted with caution due to the following limitations: it is a convenience/convenient sample in a specific context, since it involves young university students in the first year of their studies at the University of Alcalá, (Madrid, Spain); therefore, they cannot be generalized for the country as a whole. Furthermore, the contribution made by Ward et al. (2012) on alcohol-mediated sexual consent provides a new dimension that should be incorporated in future improvements to this scale. Given these limitations, it is necessary to replicate the results for other regions of the country to be able to definitively validate this scale for the Spanish case, considering adding items incorporated into the Alcohol and Sexual Consent Scale (Ward et al. 2012).