Exploratory factor analysis

An exploratory factor analysis (EFA) was carried out on the n1 subsample (n1 = 240). Different models were tested, considering two fundamental principles of exploratory factor analysis: (a) the principle of parsimony and (b) the principle of interpretability. According to both principles, the most appropriate factorial solution is the one that adequately explains the phenomenon with the least number of factors, but at the same time, with high interpretability (Díaz de Rada 2018; Méndez Martínez and Rodón Sepúlveda 2012; Pérez López 2009). Following these guidelines, nine different models were analyzed; finally, the three-factor, 15-item solution was chosen. The following variables were eliminated due to low communality (less than 0.3): I1–I8, I10, I13, I15–I17, I19–I23, I25–I27, I31–I34 and I40. The final model is shown in Table 2, with the commonality values obtained for the corresponding items.

Table 2 Three factor model

Regarding the adequacy of the Pearson correlation matrix, this model obtained a Kaiser–Meyer–Olkin (KMO) value equal to 0.846 (good) and a statistically significant Bartlett’s sphericity (chi-square = 992.3; df =105; p-value < 0.001), and the determinant of the matrix was 0.003. Furthermore, the measures of sampling adequacy (MSA) for individual variables were between 0.761 and 0.910, which indicated excellent results. These statistics ensure the adequacy of the model (Lorenzo-Seva and Ferrando 2021). The percentage of variance explained by the model was 55.21%.

In Table 2, the communalities of each item and the results of the rotated loading components are presented after varimax rotation. The first factor (F1) explained 15.63% of the variance, the second (F2) 20.63%, and the third (F3) 11.55%. The absolute values of item loadings indicate the belonging to the factor (the higher the absolute values, the more belonging of the item to the factor). The sign (+/−) points out the direction of the item–factor relationships (Comrey 1985). In Table 2, the elements that belong to a factor are marked in bold. According to the results, the first factor (F1, 15.63% of total variance) is composed of the following items:

I9- I think that if you turn another person on, you should finish what you started.

I11- I think that the way of dressing indicates how available a person is, sexually speaking.

I12- I think that if something bad happens to you (like a sexual assault) when you are very drunk, you have some responsibility for having drunk so much.

I18- I think that if a person reports having been sexually assaulted at a party months after it happened, it is because he/she is making it up.

I41- It bothers me that there are people who, when they drink and go out partying, dedicate themselves to flirting, but then they do not get to finish what they have started.

Because of the items comprising F1 (see Table 2), it was decided to call this factor “Sexual Duty”. This factor includes items about how someone is expected to act at nightlife concerning sexual interactions (I9, I11), but also how they are expected to act if they have been a victim of sexual assault (I12, I18, I41). It has been called "Sexual Duty" to highlight the tone of these items of responsibility for one's behavior and, on this basis, how it is the correct thing to behave at nightlife if you want to avoid social sanctions linked to the existence of patriarchal sexual double standards turned up in the main discourse (Bogren et al. 2022; Griffin et al. 2012; Gunby et al. 2020; Hunt et al. 2022).

Regarding the second factor (F2, 20.63% of total variance), it was defined by the following items:

I24- When I hook up with a new person for the first time, it is usually at a party and after drinking alcohol.

I28- When I'm partying, and like a person, I wait until he/she has drunk a little to hook up with him/her more easily.

I29- I usually take advantage of the drunkenness of the person I'm interested in to hook up with him/her.

I30- When I'm partying, I look for people who have been drinking a lot because it's easier to get hooked up with them.

I37- When I am partying and want to flirt with someone, I feel safer if he/she is drunk.

I38- When I go out partying and drinking alcohol, I feel the need to flirt with other people. I don't care if they are drunk.

I39- When I'm at a party and want to hook up with someone, I prefer that person to be drunk because I'm less embarrassed.

The grouping of items in F2 leads researchers to name this factor “Drunken Effervescence”, applying the theory of collective effervescence at nightlife based on Durkheim (1995) and developed by Tutenges (2022). According to Tutenges, the collective effervescence at nightlife could be split into different subtypes of effervescence. Among the typology that the author explains, he points out the relevance of drunken effervescence. The alcohol consumption in the party environment induces a sense of euphoria and boldness in those involved, allowing them to engage in more wild and thoughtless behavior. Drunken effervescence allows for breaking behavioral norms, including those related to sexual interactions. Thus, a person is more likely to engage in a sexual relationship when he or she is drunk because, at that moment, they feel invincible. For this reason, it is common that, in nightclubs, some people seek out others who are affected by alcohol with the idea that it is easier to hook up with them if they are already drunk or even pay for other people’s drinks to increase their drunkenness (Hunt et al. 2022).

Finally, the following items define the third factor (F3, 11.55% of total variance):

I14- I think that when you are at a party and two people want to make out, there is no need to talk a lot to make it happen; words are unnecessary.

I35- I have fun when a friend hooks up with drunk people.

I36- I am glad when a friend of mine hooks up with a drunk person at a party.

The set of items grouped in the F3 lead to it being named “Sexual Success”. The chosen name for the third factor is since, according to the scientific literature (Grazian 2007; Jensen et al. 2019; Jensen and Hunt 2020), partying and having sex at nightlife are understood as collective activities, where success is relevant, not only for the individual but also for the group of friends insofar as individual status and success are transferred to the group. Therefore, it is a "victory" or "success" to be celebrated. Indeed, some authors point out that, especially among men, how they interact when they want to flirt at nightlife could be called “the hunt”. This metaphor highlights the collective and predatory search of women that, in some cases, men do at nightlife (Grazian 2007).

Confirmatory factor analysis

In order to evaluate the factorial structure of the model obtained in the exploratory factor analysis, a confirmatory factor analysis was performed considering three first-order dimensions. For this purpose, the second subsample set (n2 = 245) of students was used. The model fit of three correlated latent factors corresponding to the three scales (Sexual Duty, Drunken Effervescence and Sexual Success) was evaluated.

After fitting the three-factor model, the following results were presented: CMIN/DF = 1.261; CFI = 0.980; GFI = 0.950; AGFI = 0.924 and RMSEA = 0.033. The fit index values are >0.9, within the 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 low chi-square test is presented, which allows CMIN/DF < 2 (Byrne 1989), the root mean square error (RMSEA) is less than 0.05 (Kaplan 2009; Mulaik 2009), which is the recommended value, and the comparative fit index is within the range >0.95 (Geiser et al. 2012; Schermelleh-Engel et al. 2003). These results prove that the model obtained in the EFA was adequately adjusted (Custers and McNallie 2016; Plaza-Vidal et al. 2020).

Figure 1 shows the relationship between the factors (Sexual Duty, Drunken Effervescence and Sexual Success), the scale items in each factor (I9, I11, I12, etc.) and the measurement errors of each of the items (e9, e11, e12, etc.). For a more profound comprehension of Fig. 1, the ovals correspond to the latent variables (the three factors), the rectangles to the observed variables (items) and the circles to the residual error. The lines represent interrelationships among the elements. The direction of the arrows shows the causal relationships between the variables predicted by the model.

Fig. 1

Confirmatory factor analysis

Based on these indices, we can state that the model is good and fits empirical data well (Cole and Maxwell 1985). Thus, the results confirm the construct validity and affirm that the model is relevant to ascertain the objectives proposed in this work. Consequently, the 15-item scale grouped into three factors obtained after the exploratory analysis (see Table 2) is confirmed as adequate and valid for studying DFSA attitudes. Next, and as a final point, the scale's reliability will be analyzed.

Reliability analysis

For the reliability analysis, Cronbach's α test is calculated for the whole sample and each subsample used in the factorial and confirmatory analyses. Cronbach's α test is also checked for each sex to observe whether the value of the statistic remains at acceptable values since the scale is considered to have a good internal consistency from 0.700 above (Martín Arribas 2004).

As can be observed in Table 3, a high Cronbach's α value is obtained for the total sample (0.877), a value very similar to that found in subsample n1 (0.880) and subsample n2 (0.874). When the reliability is analyzed according to the sex of the respondents, it is observed that, while remaining high, reliability decreases in the case of women. (Cronbach's α = 0.799) and increases slightly for men (Cronbach's α = 0.891). Furthermore, it can be observed that Cronbach’s α, if some item is removed, remains high values, supporting the results for the global Cronbach in each sample analyzed. Given these findings, it can be affirmed that the reliability of the scale obtained after the exploratory and confirmatory factor analysis is high and, therefore, is suitable for the purposes of the research.

Table 3 Reliability analysis