In order to achieve the aim, the study assessed a relatively large sample of participants (when compared with previous studies, eg, [2,36-38]) who used cyberporn at least once during the previous 6 months.

No precise hypotheses were formulated before data collection, as the investigation was exploratory and data-driven. Exploratory research permits adaptation and exploration of new pathways or new data analysis methodologies as the study progresses, in contrast to structured research with preconceived assumptions. This approach encourages the use of different analytical methods for extracting information from data without being constrained by particular ideas or by following the path carved by previous research [45]. This has the benefit of opening up novel opportunities for investigation and avoiding bias or errors from earlier research or established theories [45].

Participants were recruited through Prolific, a crowdsourcing platform widely used for academic research [46,47]. The study was conducted entirely online using the SphinxOnline platform. Eligible participants were English-speaking adults (aged 18 y or older) who reported having viewed pornography at least once during the past 6 months. After accessing the survey link, participants were presented with a detailed information sheet and eligibility questions. Only those who met the inclusion criteria were granted access to the full questionnaire.

The study relied on an anonymous SphinxOnline survey. The Research Ethics Committee of Tours-Poitiers in France (2020-04-05) approved the study. All procedures were conducted in accordance with the Declaration of Helsinki and relevant institutional guidelines. Participants provided informed consent online before participation, after being fully informed about the study’s aims, procedures, anonymity, and their right to withdraw at any time without consequences. No personally identifying data were collected. Data were collected, stored, and analyzed anonymously. Participants were recruited through Prolific, a platform widely used for ethical participant recruitment in academic research [46,47]. Inclusion criteria included being at least 18 years old and having watched pornographic material at least once in the past 6 months. The survey took approximately 15 minutes to complete. The participants received time compensation for their study participation via the Prolific platform (approximately £10 [approximately US $14] per hour of participation). This study was not registered on a public trial registry; however, the protocol was approved in advance by the ethics committee, and no deviations from the original procedures occurred. No adverse events were reported during data collection.

This study follows the recommendations of the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for reporting observational research [62]. Although STROBE was initially developed for epidemiological studies, it is broadly applicable to cross-sectional observational designs in psychology and social sciences. We ensured transparency and completeness in reporting, in line with the EQUATOR Network recommendations [62].

It must be noted that the data had 0.2%-0.7% of missing values in the sociodemographic variables.

First, we conducted descriptive analyses on the study’s predictor and criterion variables. These analyses were conducted separately on male versus female participants.

Second, we built a correlation matrix including all the study variables.

Third, we tested the normality of the dependent variable (AMMSA score) using the Shapiro-Wilk test for normality. The test was significant (W=0.878, df=1557; P<.001), indicating that the distribution of this variable was significantly different from a normal distribution [63]. In addition, we conducted multicollinearity tests on the independent variables. The results indicated no significant multicollinearity.

Fourth, and given that the assumption of normality had not been met, we built two robust multivariate linear regression models (one per gender group). These linear regressions used the above predictor variables (except for sexual orientation and relationship status) to predict AMMSA scores, the study’s criterion variable. Robust linear regression can be used to improve the accuracy of predictions when the residuals are nonnormal and when the data contain outliers [64].

Finally, we conducted an ANOVA to test the effect of sociodemographic variables (sexual orientation and relationship status) on the AMMSA score.

All statistical analyses were conducted using the software SPSS Statistics version 29 (IBM Corp), except for the regression analyses, which were conducted using the R package “robustbase” [65]. These analyses were conducted with different statistical software that have different ways of dealing with missing values (eg, SPSS allows analysis considering only the participants without any missing values in each variable; on the contrary, the algorithm used for regression requires missing values imputation using the nearest neighbor approach).

Overall, 1590 English-speaking individuals fully completed the online questionnaire. However, 1557 participants were included in this study, among them 64.2% (n=1000) were men, and 35.8% (n=557) were women. The age of participants ranged from 18 to 75 (mean 33.3, SD 10.9) years. During the study period, 497 (31.9%) participants were single; 663 (42.6%) were in a relationship, not married; 389 (25%) were in a relationship, married; and 8 (0.5%) were widows. The participants’ sexual orientation distribution was as follows: heterosexual 1228 (78.9%), homosexual 98 (6.3%), bisexual 200 (12.8%), and other 31 (2%). It must be noted that, given this study’s objective, outlined above, 33 participants were excluded; among them, 27 were nonbinary, and 6 did not indicate their gender. Participants’ countries of residence can be found in Multimedia Appendix 2.

Table 1 presents the descriptive statistics of all variables by gender (male vs female participants) and the respective mean comparison.

As can be seen in this table, male participants had significantly higher scores than their female counterparts regarding the following variables: ASAM (AMMSA total score), SES–Perpetration (total score), CCU (CIUS total score), aroused by “groups with several females” porn scenes, and age. Female participants had significantly greater mean scores than their male counterparts on the following variables: SES–Victimization (total score), aroused by “submission” pornographic scenes, aroused by “soft” pornographic scenes, and aroused by “groups with several males” pornographic scenes.

Table 2 shows the correlation coefficients (by gender) between the variables included in the study, two-by-two combinations. As shown in these two tables, the strongest correlations are (r≥0.20) between AMMSA and SES–Perpetration (+), AMMSA and CCU (+), SES–Perpetration and SES–Victimization (+), humiliation porn scenes and submission porn scenes (+), soft porn scenes and humiliation porn scenes (−), porn scenes with group of males and humiliation (+), UPPS-P negative and positive urgency and CCU (+), and age and submission porn scenes (−) with a far more important association among women than men.

Table 3 (regression model 1, R² [explained variance]=0.42, male participants; regression model 2, R²=0.41, female participants) displays the relationships between the independent variables and the dependent variable, resulting from the multivariate regression analyses.

The results from model 1 suggest that male participants were more likely to endorse the ASAM if they had higher scores of sexual coercion perpetration, CCU, arousal with pornographic scenes displaying humiliation, negative urgency impulsivity, sexual self-esteem, and moral incongruence. They were less likely to endorse the ASAM if they had higher scores of sexual coercion victimization and arousal toward pornographic scenes displaying groups of several males.

The findings from model 2 suggest that female participants were more likely to endorse the ASAM if they had higher scores of sexual coercion perpetration, CCU, sexual self-esteem, and older age. They were less likely to endorse the ASAM if they had higher scores of arousal toward pornographic scenes displaying groups of several females.

In addition, the ANOVA results show that among male participants, there was no statistical difference between the AMMSA mean score of single (n=338) and in-relationship (n=650) participants. On the contrary, the AMMSA mean scores of heterosexual (n=841), homosexual (n=61), and bisexual (n=92) participants differ statistically (3.14, SD 1.31; 2.84, SD 1.42; 2.75, SD 1.29, respectively; F_3,991=3.81, P=.02, η²p=0.008). In particular, the AMMSA mean score of heterosexual participants is statistically higher than the AMMSA mean score of bisexual participants (t₉₃₁=2.44, P=.04). There was no difference between the AMMSA mean score of heterosexual versus homosexual and homosexual versus bisexual participants (t₉₀₀=1.56, P=.26; t₁₅₁=1.21, P=.45, respectively).

Among female participants, there was no statistical difference between the AMMSA mean score of single (n=143) and in-couple (n=391) participants. On the contrary, the AMMSA mean scores of heterosexual (n=387), homosexual (n=37), and bisexual (n=112) participants differ statistically (2.47, SD 0.23; 2.02, SD 1.21; 1.74, SD 0.89, respectively; F_3,533=12.48, P<.001, η²p=0.045. In particular, the AMMSA mean score of heterosexual participants is statistically higher than the AMMSA mean score of bisexual participants (t₄₉₇=4.92, P<.001). There was no difference between the AMMSA mean score of heterosexual versus homosexual and homosexual versus bisexual participants (t₄₂₂=1.56, P=.26; t₁₄₇=1.21, P=.45, respectively).

The study’s limitations include its cross-sectional design and a small number of people endorsing nonbinary gender. It is also limited by some single-item measurements (eg, sexual self-esteem and moral incongruence). Although we employed measurements used in prior research, we are aware of the difficulties of generalizing these results and the need for more elaborate measures in future studies.

The strengths of the data-driven approach used in this study may have been limited by the absence of certain dimensions that should be included in future research. For instance, a more refined assessment of sexual self-esteem, along with other potentially important dimensions, such as sexual function and relational interactions, could enhance the comprehensiveness of future studies.

The study introduced the assessment of different kinds of porn-arousing scenes, considering relational aspects of the contents. We cannot exclude that some wordings, such as the group scenes, may include different kinds of relational interactions. Additionally, the arousal to pornographic content was measured through participants’ self-reported responses to scene type labels (eg, humiliation, submission), without detailed descriptions or audiovisual material. This may limit the ecological validity of the measure. In addition, all participants responded to the same list of scenes, regardless of their sexual orientation, which may have affected the relevance of some items. Future research could benefit from more tailored assessments based on individual preferences.

Furthermore, given the cross-sectional design of this study, causal inferences cannot be drawn regarding the relationship between arousal to specific pornographic content and the ASAM. It remains unclear whether exposure to or preference for certain types of pornographic scenes increases the ASAM or whether individuals with higher ASAM are more likely to find such content arousing. Future longitudinal studies are needed to disentangle these directional effects and better understand the dynamics at play.

As far as we are aware, this study is the first (or one of very few) that has examined the relationships between “preferred” cyberporn scenes, sexual self-esteem, moral incongruence, and the ASAM. However, this study tested only the main effects; as such, all predictor variables were set at the same level. Future research based on this study could explore some predictor variables’ mediating and moderating effects. For instance, it would be interesting to examine if sexual coercion experiences diminish sexual self-esteem and, thus, increase the ASAM, and if the arousal to humiliation/submission cyberporn scenes is associated with potential problematic cyberporn use and the ASAM, and so forth.

The regression models utilized in this study yielded important insights into the psychological and behavioral correlates of ASAM across genders. CCU, the perpetration of sexual coercion, and sexual self-esteem were found to be significant predictors of the ASAM among both male and female participants. Among men, impulsivity and moral incongruence emerged as particularly salient predictors, whereas age was a significant factor exclusively among women. Furthermore, marked gender differences were observed in the degree of arousal elicited by specific pornographic content and in the association between participants’ subjective appreciation of such content and their levels of ASAM endorsement.

The findings from this study can be used to develop interventions in the following directions: (1) gender-sensitive educational interventions focused on critical consumption of pornographic content, particularly addressing how arousal and appreciation may unconsciously reinforce acceptance of sexual aggression myths; (2) targeted interventions for reducing CCU, such as app-based self-monitoring tools and cognitive-behavioral strategies aimed at improving online sexual self-regulation; (3) given the role of moral incongruence and impulsivity in male participants, incorporate modules on ethical decision-making, impulse control, and empathy training in school- and community-based programs; (4) tailor sexual health and consent education for women by addressing developmental and age-related factors that may influence their beliefs around sexual aggression myths; (5) integrate findings into rehabilitation programs for individuals who have engaged in sexually coercive behavior, focusing on restructuring harmful beliefs, enhancing sexual self-esteem in healthy ways, and reducing impulsivity; and (6) launch broader societal campaigns to debunk sexual aggression myths and reduce their normalization, with particular attention to gender-specific messaging and outreach.