Cybersex Addiction: A Study on Spanish College Students

The aim of this study was to determine the type and frequency of online sexual practices among Spanish college students, the prevalence of risk and pathological cybersex use profiles, and the correlates/predictors of this behavior. Participants were 1,557 males and females between 18 and 25 years old. Results showed that cybersex use is not as frequent as that documented in other Western countries. However, a significant percentage of participants with a risky (8.6%) or pathological (1.7% in men and 0.1% in women) profile was identified. Finally, we found a set of variables that, in interaction with gender, explains 58% of the variance for cybersex addiction scores.


INTRODUCTION
According to the latest survey by the National Institute of Statistics, 97% of Spaniards between the ages of 15 and 24 had used the Internet during the past three months, most daily (Instituto Nacional de Estadística, 2013). The applications, uses, and available online content are innumerable, but there is one that has grown exponentially: pornography. While in 1998 there were an estimated 100,000 pornographic websites all over the world (Rice-Hughes, 1998), in 2006 this figure had risen to 4.2 million (12% of total websites; Family Safe Media, 2006), with pornography thus becoming a very lucrative business whose earnings range between 1 and 97 billion dollars annually (Wondracek, Holz, Platzer, Kirda, & Kruegel, 2010). The availability of Internet access devices, the ease of finding online sex content, the anonymity, and low cost (Cooper, 1998) make the Internet an ideal medium for young people to experiment with their sexuality. Lewin, &Štulhofer, 2014). For this reason, our study is exclusively focused on cybersex access through PCs.

Cybersex: Positive Outcomes, Negative Outcomes, and Addiction
The main reason for resorting to Internet sex is the search for sexual pleasure. Daneback, Sevcikova, Mansson, and Ross (2013) found that most young people between 18 and 24 years of age stated that their online sexual activity satisfied their sexual needs, either moderately or completely (80% in males and 73% in females). More frequent masturbation during online sexual activities was associated with greater sexual satisfaction.
Another common argument to justify the practice of cybersex is its educational potential: For many young people, the Internet is a useful tool to address the lack of sexual knowledge and sexuality, or to "learn certain skills" by the viewing of pornography (Simon, Daneback, & Sevcikova, 2014).
However, not everyone values the benefits of Internet sex. In fact, there is debate about its possible consequences for psychosexual development, especially among teenagers and young people (Döring, 2009). One of the biggest risks attributed to cybersex is the possibility of being a victim of unwanted sexual solicitation or abuse (Jones, Mitchell, & Finkelhor, 2012;Mitchell, Finkelhor, & Wolak, 2007). In a review of the impact of online pornography on adolescents and youth, Flood (2010) also highlighted its potential to cause negative emotional reactions (from disgust to anger or shock), to encourage the adoption of certain sexual practices (e.g., anal sex), to promote more open sexual attitudes, to encourage early first-time sex and subsequent promiscuity, to establish gender-role sexism, to increase the likelihood of committing a sexual assault, and finally to lead to the onset of addiction.
Multiple investigations have demonstrated the potential harms associated with cybersex addiction. Cybersex addiction or online sexual compulsiveness is defined as excessive and uncontrolled use of cybersex leading to serious work, social, and personal problems (Cooper & Griffin-Shelley, 2002). Ross, Mansson, and Daneback (2012) found, under lax criteria, that 7.6% of their sample of Internet users showed problems in controlling their sexual behavior online, while under stricter criteria, this percentage fell to 1.7%. The main predictors of problematic Internet use, according to this same study, were the degree of religiosity (the more religious, the higher the risk), the frequency of viewing pornography, and the types of online sexual activities (greater risk for users watching pornographic films who share pornographic images or input pictures of themselves). Other studies (Cooper, Griffin-Shelley, Delmonico, & Mathy, 2001;Egan & Parmar, 2013;Meerkerk, Van Den Eijnden, & Garretsen, 2006;Orzack & Ross, 2000;Schwartz & Southern, 2000;Sinković, Stulhofer, & Bozic, 2013;Tsitsika et al., 2009) have suggested that predictors of cybersex addiction are the following: men who are homosexual or bisexual; individuals who practice high sexual frequency with a range of partners and types of sexual practices in their offline relationships; individuals with a sexually transmitted disease; individuals who suffered abuse or addiction to the Internet, or exhibited behavioral problems and social maladjustment; individuals who have had early exposure to sexual content; individuals with a greater tendency to engage in sexual sensation seeking, and finally, individuals who suffer a general vulnerability in controlling their own behavior in a variety of areas.

Importance of Cultural Perspective in the Study of Cybersex Addiction
Griffiths (2012) suggests two scenarios in which sociocultural context can influence the consumption of cybersex. On the one hand, people living in countries with positive attitudes toward sexuality may be more likely to consume cybersex as an alternative form of sexual expression. On the other hand, in a conservative context there are two possible ways to react toward sexuality in terms of cybersex consumption: individuals could be more likely to engage in cybersex as an anonymous alternative to what, in real life, would attract social or even legal rejection, or these rigid sexual norms could simply inhibit the practice of cybersex. Taking into account sociocultural context may help us to understand why the prevalence of pornography consumption can reach 93% in Hong Kong (Lam & Chan, 2007) or 85% in Canada (Shaughnessy et al., 2011), while in Taiwan it barely reaches 38% (Lo & Wei, 2005). It may also explain in part why the pattern of cybersex consumption in young Peruvians (most especially with respect to the role of gender) varies greatly when compared with the young in the United States (Velezmoro, Negy, & Livia, 2012). The case of Spain in terms of attitudes toward sexuality is complex: In comparison with other European countries, Spain is one of the most permissive with regard to homosexuality (Stulhofer & Rimac, 2009). However, attitudes toward casual sex, teenage sex, or extramarital sex are conservative in comparison with countries such as Sweden, Norway, or the Netherlands (Widmer, Treas, & Newcomb, 1998). This double sociosexual standard has also been documented in studies exploring attitudes and sexual risk behaviors for HIV/AIDS among youth from Spain and Mexico (Ballester-Arnal, Gil-Llario, Giménez-García, & Ruiz-Palomino, 2009;Giménez-García, Ballester-Arnal, Gil-Llario, Cárdenas-López, & Durán-Baca, 2013). It seems to be the result of attempts by mass media and social movements to foster sexual openness in a society with strong religious beliefs (Castro-Calvo, Ballester-Arnal, Gil-Llario, & Gimenez-Garcia, 2015). Thus, it is difficult to predict how Spain's double sociosexual standard could influence the probability to engage or not in online sexual activities.
As Griffiths (2012) asserted, "cross-cultural comparisons of Internet sex addiction need to be conducted in order to contrast and compare its prevalence epidemiologically and assess the experience of it on an individual level" (p. 122). Based on this requirement, an investigation began in Spain that culminated in the publication of two articles: one about the influence of having or not a stable partner on the practice of cybersex (Ballester-Arnal, Castro-Calvo, Gil-Llario, & Gimenez-Garcia, 2014) and the other about online sexual activities among adolescents (Ballester-Arnal, Giménez-García, Gil-Llario, & Castro-Calvo, 2016). However, as far as we know, no large-scale scientific study has yet been performed in Spain that comprises participants aged 18 years or older. This lack of empirical evidence has motivated the present study.
In this explorative study, we intend to investigate the prevalence of different behaviors related to the use of the Internet for sexual purposes, the prevalence of different consumption profiles (recreational, risk, and pathological), and the predictive role of certain variables on the severity of cybersex consumption. We hypothesize the following: 1. Cybersex consumption will be very common, especially among males. 2. We expected that approximately 2% to 6% of the sample will match a pathological profile, with a similar percentage matching a risky profile. 3. Finally, we hypothesize that the severity of cybersex consumption will be predicted by a combination of psychological and sexual variables (such as sexual frequency or Internet addiction), some of them in interaction with gender.

METHOD Participants
A total of 1,557 college students between 18 and 25 years of age participated (M = 20.37; SD = 1.983). All belonged to the middle class, were living in urban environments, and had Internet access. Females comprised 64.4% of the sample. In terms of sexual orientation, 94.2% considered themselves heterosexual, 3.6% bisexual, and 2.2% homosexual. More than half (55.5%) had a steady partner, but the majority lived with their parents or shared flats.

Measures
During the evaluation of participants, three instruments were used:

Ad Hoc Questionnaire
This questionnaire collected information on demographic variables (gender, age, and education), sexual life (sexual orientation, same-sex intercourse, current frequency of sexual activity, specific sexual behaviors ever experienced, current frequency of pornographic use through means other than online use), and Internet access and use (accessing computer at home, number of hours online per week, and number of hours online for sexual pursuits, also weekly).

Internet Sex Screening Test
The Internet Sex Screening Test (ISST) was adapted and validated for the Spanish population (Ballester, Gil, Gomez, & Gil, 2010) from the original Delmonico version (1997). The ISST consists of 25 dichotomous (true/false) responses, evaluating the degree to which the online sexual behavior of a person is or is not problematic. With respect to the psychometric properties, Ballester et al. (2010)

Spanish Adaptation of the Internet Addiction Questionnaire
This questionnaire (CAI; Young, 2004) was composed of 20 Likert-type questions (from 1 = rarely to 5 = always), which identified the extent to which Internet use was problematic. The score ranged from 20 to 100. A psychometric study of a Spanish-language sample showed a reliability of 0.89 (Puerta-Cortés, Carbonell, & Chamorro, 2012). Using the Cronbach's alpha, the reliability of the present study's questionnaire was assessed at 0.91.

Procedure
Questionnaires were administered by members of UNISEXSIDA (AIDS and Sexuality Research Unit) at the Universitat Jaume I of Castellón (Spain). The research team hosted a booth at the faculty entrance of the university where a member of the team actively approached each person who passed the booth, offering the possibility of collaborating in our research voluntarily. Those who accepted signed an informed consent form and then completed the pencil-andpaper questionnaires described above. To ensure response confidentiality, participants completed questionnaires anonymously and alone.

Data Analysis
Data were analyzed using SPSS version 21.0. We used chi-squares for categorical variables (items of the questionnaire) and t tests for continuous variables (factors and total score). The magnitude of the differences in these contrasts was estimated by calculating the effect size: specifically, Cohen's d estimated for the results of the t test and Cramer's Phi for the chi-square test (Sheskin, 2000). Pearson correlations coefficients were calculated to ascertain linear relationship between variables. Hierarchical linear regression was performed to determine the most effective predictors of dimensionally measured cybersex consumption-that is, the dependent variable was the total ISST score. Continuous and categorical variables (such as sex, sexual orientation, or specific sexual behaviors) were introduced into the first block following Hardy (1993) coding recommendations. In order to test the possible moderating effects of sex on the relationship between independent and dependent variables of the model, gender interactions terms were entered into the second block.

Ethics
The study procedures were carried out in accordance with the Declaration of Helsinki. The Institutional Review Board of the Jaume I University approved the study. All subjects were informed about the study and all provided informed consent.

General Sexual Behavior
In our sample, 98.9% of males and 95.2% of females reported having ever experienced some kind of sexual behavior. In males, the most frequent sexual activities were masturbation (90.1%), vaginal intercourse (84.4%), oral sex (80.1%), and mutual masturbation (78%). Females usually practiced vaginal sex (89.3%), oral sex (72.5%), mutual masturbation (71.9%), and masturbation (56.3%). Regarding current sexual frequency, 26% of males and 17% of females claimed to have a sexual frequency of once a week; 23.5% and 26.6% respectively reported a frequency of three times a week; and finally, 44.7% and 19.5%, respectively, reported having a sexual frequency of more than three times a week.

Accessibility and Use of the Internet and Cybersex
Almost all of the participants (98.6%) had a computer at home from which to access the Internet. The average amount of time per week spent on the Internet for males was 11.24 hours (SD = 15.68), significantly higher (t = 5.05, p < 0.001, d = 0.31) than the time reported by females (M = 7.54, SD = 9.39).
Regarding the use of the Internet for sexual pursuits, males (M = 86.25 min, SD = 225.05) spent more time weekly on online sexual activities than women (M = 11.28 min, SD = 92.96; t = 8,717, p < 0.001, d = 0.49). The great variance in both cases suggests an uneven distribution on the time devoted to cybersex. According to Table 1, the vast majority of females (90.5%) did not practice cybersex regularly, 7.3% spent less than one hour per week, and the percentage who responded with more than an hour a week was practically nil. In the males, again a majority (55%) claimed not to practice often, but a significant percentage spent less than one hour per week (22.9%) or between one and three hours (12.5%). Finally, it is significant that 4.8% confirmed spending six hours or more per week (1.9% of them over 11 hours).

Online Sexual Behavior
As shown in Table 2, the most common online sexual behaviors from the ISST are looking for sexual material on the Internet (35.5%), having a username for surfing the Internet (30.6%), masturbating while connected to the Internet (30.2 %), or having bookmarked sex sites (29.6%). Sexual behavior that involved interaction with other users (COSOC) was less frequent. In this sense, 24.8% communicated with other users while connected, 17.5% had been face-to-face for sex with someone they had met online, and 12.9% confirmed that they had participated in sex chat rooms. The prevalence of certain behaviors that are characteristic of problematic cybersex use was highlighted: 7.9% had promised themselves to stop using Internet for sexual purposes and, for 7.4%, cybersex had interfered with their lives. Finally, 4% believed they were addicted to cybersex.
The analysis of online sexual behaviors by gender revealed that the men reported significantly higher prevalence in 24 out of 25 ISST items explored (all with a significance < 0.001). Given the large sample size, it is more appropriate to analyze these differences from the effect size estimated by Cramer's Phi. According to the criteria established by Cohen (1988), the difference in prevalence between males and females reached a medium effect (0.30 < Phi < 0.50) for all the behaviors covered by the COSOL factor. For instance, the prevalence of students who reported    masturbating while connected to the Internet was four times higher in men than in women (59% vs. 14.3%). Similarly, the percentage of participants who reported having searched for sexual material through an Internet browser is also greater in men (55.9% in men and 24.2% in women). The difference in the proportion of males and females who manifest certain signs of pathological consumption (COMPULS) reached a small effect (0.10 < Phi < 0.30), and this is also similar for the majority of behaviors in the social sphere of cybersex consumption (COSOC), or perception of its severity (PERGRA). To cite just three examples, significantly more men reported spending more than five hours per week in OSA (10.3%), having participated in sex chat rooms (20.5%), or having promised themselves to stop using the Internet for sexual purposes (14.9%) compared to women (0.8%, 8.6%, and 4%, respectively). Similar results were found when the average score for males and females in the ISST (sub)scales were analyzed. Total ISST score in men is more than two times the score in women (t = 19.52; p < 0.001) and more than four times in the case of the COMPULS score (t = 11.40; p < 0.001). In both cases, as well as in PERGRA and COSOL scores, gender differences reached large-effect size magnitudes (d > 0.50).

Frequency of a Profile of Recreational, Risky, and Pathological Use of Cybersex
According to the criteria established by Carnes, Delmonico, and Griffin (2007), participants were classified into three groups (Table 3): recreational users (ISST total score between 0 and 8), risky users (between 9 and 18), and pathological users (between 19 and 25).
The majority of participants evaluated (90.7%) matched a recreational profile, followed by a significant percentage (8.6%) of risky users. Less than 1% matched a pathological profile. We found important differences between males and females, especially in the recreational and risk profiles (Cramer's Phi of 0.28 and 0.26, respectively). While the percentage of female recreational users (96.9%) was higher than the male percentage, the percentage of males having at risk and pathological profiles (18.8% and 1.7%, respectively) was higher than females. Gender differences reach statistical significance in all three cases (significance < 0.001).

Correlates and Predictors of Cybersex Use
As indicated in Table 4, most of the variables considered were related in a statistically significant manner, and with a positive sign with respect to one or more cybersex consumption aspects. Age, for example, correlated significantly with the total ISST score and with two of its five subscales (COSOC and GASTO). All variables related to offline sexual behavior (having sex with  a same-sex partner, frequency of sexual activity, or different sexual practices, such as masturbation, mutual masturbation, oral sex, anal sex, and other behavioral activities) were significantly associated with the total cybersex questionnaire score and with several of its subscales (especially COMPULS, COSOL, and COSOC). Finally, Internet-use variables (time online and the total score on the Internet addiction questionnaire) and pornographic-consumption characteristics (time spent practicing cybersex and the consumption of pornography through other means) showed the strongest positive relationship. Conversely, having a stable partner was negatively and significantly associated with the total cybersex questionnaire score and with four of its five subscales. Also the practice of vaginal intercourse was inversely related to some dimensions of cybersex consumption.
The results of the first block of the hierarchical multiple-regression analysis (Table 5) revealed that the most important predictor of ISST scores was Internet addiction, followed by pornography consumption by other means and the time devoted to cybersex. The model included six additional variables (gender, having participated in oral sex, bisexual orientation, having undertaken unusual sexual practices, vaginal intercourse, and sexual frequency) and explains 56.3% of the ISST score variance (F = 125.505, p < 0.001). Only female gender and having participated in vaginal intercourse were negatively related to ISST score.
In the second block of the hierarchical multiple-regression analysis, interactions terms between gender and variables included in the previous analysis were performed. In order to simplify the table, we include only significant interactions. First, we found a significant interaction between gender and unusual sexual practices: For men, having undertaken unusual sexual practices was Notes. ISST = Internet Sex Screening Test. * p < 0.05. * * p < 0.01. * * * p < 0.001. associated with higher levels of ISST scores, but this was not the case for women (Figure 1). On the contrary, the time spent in online cybersex or having participated in vaginal sex was more associated with higher levels of ISST scores in the case for women. Finally, sexual frequency is positively related with ISST scores in the case for women and negatively related in the case for men (Figure 2).

Types of Sexual Activity and Frequency Among Spanish College Students
Our results indicate that young college students show a clear preference for online sexual behavior that requires no interaction with other users, such as searching for and viewing pornography. This preference had been documented in most previous research (Hald, 2006;Johansson & Hammerén, 2007;Shaughnessy et al., 2011) and may be understood in terms of its cost-benefit: People seek sex on the Internet as a quick way to satisfy sexual desire (Daneback et al., 2013). In this sense, the viewing of pornography becomes the most rapid option for attaining pleasure (Johansson & Hammerén, 2007;Morgan, 2011). These results are also in accordance with the Triple-A engine theory, which states that the accessibility, affordability, and anonymity of the Internet provide the perfect medium to foster sexual behaviors (Cooper & Griffin-Shelley, 2002). Nevertheless, the prevalence for seeking pornography or the use of sex chat rooms is significantly lower than that found in samples with similar characteristics in other Western countries like Sweden (Daneback et al., 2005), the United States (Morgan, 2011), and Canada (Shaughnessy et al., 2011). This is also true when we compare the prevalence of pornography consumption or sex chatting in our study (35.5% and 12.9%, respectively) with that found by Döring et al. (2015) in a recent study among four countries' college students (76.5% and 30.8%). An exception is the Goodson et al. investigation (2001) among U.S. university undergraduates, where the percentage of pornography consumption (43.5%) and the use of sex chat rooms (12.8%) are quite similar FIGURE 2 Gender moderating the association between sexual frequency and ISST (Internet Sex Screening Test) score.
to those obtained in this study. These low prevalence rates can be explained by the manner of exploring the different types of online sexual behavior: While in other studies (e.g., Shaughnessy et al., 2011) cybersex definitions were much broader and considered more types of sexual practices (the use of dating web pages, sexting, etc.), we have taken into account only the viewing of pornography or participation in sex chat rooms. Another explanation for these results could be linked to the way in which Spanish sociocultural context influences cybersex activity. Instead of fostering online sexual activities as an anonymous and secure form of sexual expression and experimentation, Spanish conservative attitudes toward certain sexual issues (such as casual sex) could inhibit the practice of cybersex. This alternative explanation would go against the hypothesis of a "new globalized net generation that appropriates the Internet in similar ways regardless of their national cultures" (Döring et al., 2015, p. 7), suggesting that sex attitudes and values still have an important role in sexual behavior and development.
In females, frequency and type of cybersex use could have an explanation on the basis of the sexual script theory (Wright et al., 2013). Cybersex leaves in second place the emotional and relational aspects of sexual activity (most important in female sexuality), focusing instead on physical attributes and erotic pleasure. This explains the low pornography consumption by females and their preference for sex chat rooms. The results obtained in this investigation only partially validated this theory. As expected, cybersex consumption by females was much lower than that recorded for males; however, we did not find a preference for the use of sex chat rooms (like their male counterparts, females prefer to use online pornography). This preference is not as marked as in the male case, but more than twice as many women reported seeking Internet pornography (24.2%) as opposed to females who reported participating in sex chat rooms (8.6%). While this finding is not common, other studies have found similar results (Shaughnessy et al., 2011).
More specifically, the percentage of females who say they have undertaken some of the major activities mentioned herein regarding sexual exploration online was significantly lower than that found in previous studies. For example, while Morgan (2011) reported a prevalence of pornographic use by females with 33% and Shaughnessy et al. (2011) with 39%, in our study the figure barely reached 24%. In the study by Wright et al. (2013) in which 18,225 American women were evaluated between 1973 and 2010, one must regress to 1995 to find a figure of pornographic consumption equal to that obtained in our study. The same is true when analyzing the prevalence of the use of sex chat rooms: Whereas in our study it is 8.6%, in the Cooper et al. (2003) study, the figure is 32%. Finally, the percentage of girls who masturbated while searching online pornography (14%) remained well below that obtained in other studies, such as the 18% figure in Shaughnessy et al. (2011).
While it seems that to chat for sexual purposes is relatively uncommon in both males and especially females, to do so for romantic reasons or to move the relationship into real life is more common. This was well reflected, for example, in the percentage of respondents who had come face to face with someone they had only met online (25%).
The investment of time in online sexual activities is not one of the most frequently studied issues. An exception is the study of Daneback et al. (2005), which found that male pornography users spent an average of four hours a week in such sexual activity online. In our study, males spent an average of 1.5 hours weekly in online sexual activities, and females spent less than 15 minutes. In both cases, the weekly time reported by Daneback et al. (2005) far exceeded that documented in our research.

Cybersex User Profiles Among Spanish College Students
The time spent on cybersex may also be taken as an indicator of pathological consumption. According to Cooper, Delmonico, and Burg (2000), recreational users would spend approximately 1 hour weekly on cybersex, risky users between 1 and 10 hours, and finally those pathologically affected spend more than 11 hours a week. The latter group comprised 7.92% of their sample. In our case, only 1.9% of young men exceeded 11 hours a week consuming cybersex. This 1.9% of pathological users identified from the time criterion of Cooper et al. (2000) nearly matched the 1.7% obtained in this study from a completely different approach, namely, the total ISST score. Ross et al. (2012) found exactly the same percentage of pathological cybersex users (1.7%), which adds further credibility to our prevalence estimation. Nor should the percentage of 19% for male risky users be overlooked, especially considering that many may eventually develop an addiction (Carnes et al., 2007).
In females, we found a quite similar percentage of pathological profiles considering time criterion and ISST score criterion. When considering the time criterion, 0.2% of females could be classified as pathological users, a percentage virtually identical to the 0.1% obtained from ISST scoring criteria. This result is consistent with studies where a lower prevalence of problems related to cybersex is documented in females than in males (Ross et al., 2012).
Finally, the proportion of males and females who claimed they experienced symptoms of pathological consumption is significant. The percentage of young college students who valued their consumption as a problem far exceeded the real prevalence of pathological users: While the prevalence of pathological consumption is estimated at 1.7% in males and 0.2% in females, 9.3% and 1.1%, respectively, believed themselves addicted to cybersex. Furthermore, 13.4% of males and 4.1% of females believed that cybersex has sometimes interfered with certain aspects of their lives.

Predictors of Cybersex Consumption Among Spanish College Students
We have found a set of variables that explains 58% of the variance for the Internet Sex Screening Test score. The most powerful predictor was the Internet addiction score (Young, 2004), a factor that affects men and women in a similar way. The most recent studies suggest that Internet addiction is related to cybersex addiction, either as a precursor or as a related phenomenon (Egan & Parmar, 2013). As a precursor, it is understood that cybersex addiction constitutes a subtype of Internet addiction (Schiebener, Laier, & Brand, 2015); as a related phenomenon, the cybersex and Internet addiction relationship may be explained by the fact that both share a common cause, namely, a vulnerability to developing addictive disorders. More research is needed in order to confirm the kind of relationship between Internet and cybersex addiction, but our results highlight the key importance of this issue in the understanding of cybersex addiction.
Following Internet addiction, there are two variables related to the characteristics of cybersex consumption that predict the total ISST score: alternative ways to consume pornography and time spent in online sexual activities. In the case of time spent in online cybersex, the results indicated a significant interaction with gender, suggesting that the relationship was different for men and women. Men's level of cybersex addiction did not significantly differ, whether they spent more or less time in cybersex consumption. Conversely, women who spent more time in online cybersex reported high levels of cybersex addiction (higher scores on the ISST). As such, time spent online seems to be more relevant to women's experience of cybersex addiction than men's.
Finally, we found a significant association between certain aspects of sexual behavior and the consumption of cybersex. Having practiced oral sex was positively related with the severity of cybersex consumption in both males and females. In males, undertaking unusual sexual behavior (understood as any sexual activity other than masturbation, mutual masturbation, oral sex, and vaginal or anal sex) was positively associated with higher ISST scoring. Other studies have documented a link between online and offline sexual behavior (Hald, Kuyper, Adam, & de Wit, 2013;Luder et al., 2011), but there is no record of a relationship with unusual sexual behavior. Presumably, some of the sexual behavior in this category are of a paraphilic nature, thus an investigation should be undertaken to establish whether sexual deviants find on the Internet a way to experiment with paraphilic content. Other aspects such as lower frequency of sexual intercourse with a partner or not having participated in vaginal sex were associated with higher ISST scoring in the case for males but not for females. Conversely, a high sexual frequency or having participated in vaginal sex was associated with higher ISST scoring in the case for females, highlighting the key importance of gender in the understanding of online sexual activity.
In both males and females, self-identified bisexual arousal patterns allow us to predict a significant variance in ISST scores. This result supports the hypothesis that sex on the Internet has not led to an environment that only satisfies sexual desire, but rather a way to find and make contact with people with similar sexual orientation or interests (Ross & Kauth, 2002).

LIMITATIONS AND FUTURE DIRECTIONS
We recognize that this study has a number of limitations. First, results obtained from the examined sample (university students) may not be representative of all young Spaniards between 18 and 25 years of age or of the university student population in general. However, the fact that the percentage of young Spaniards studying at the university, or already graduated, is higher than the European Union average (Ministerio de Educación, 2009) increases the sample's representativeness. In this sense, the affordability of university studies in Spain allows anyone (regardless of his or her economic status) to study for a degree. In addition, sexual activity and frequency are influenced by age, so younger participants are probably more sexually active than older individuals (Shulman & Connolly, 2013). Second, the tool used for assessment (ISST), despite being the only one that has been validated in this study's context (to ensure reliability and validity), does not cover the wide range of situations comprising the cybersex phenomenon, a fact that conditions the (over or under) estimate of the prevalence of online sexual behavior. Finally, we analyzed cross-sectional data, so causality between variables should not be inferred.
Despite these limitations, this study provides evidence that enhances the understanding of online sexual activity in young Spaniards and, in turn, guides our research in new directions.