Kibernetinis naudojimas ir probleminis kibernetinis naudojimas tarp jaunų šveicarų vyrų: Asociacijos su sociodemografiniais, seksualiniais ir psichologiniais veiksniais (2019)

Data were drawn from the third-wave questionnaire of the Cohort Study on Substance-Use Risk Factors. In Switzerland, assessment for eligibility for military, civil, or no service is mandatory for all young men, offering a unique opportunity to enroll a non-selective sample of the country’s population of 19-year-old men. Between August 2010 and November 2011, all young men reporting to the recruitment centers of Lausanne (French-speaking), Windisch, and Mels (German-speaking) were invited to participate in the study. A total of 7,556 gave their written consent. The C-SURF study was independent of the military’s procedures: recruitment centers were used to inform and enroll participants, but they completed their questionnaires outside of the military context. Full information on enrollment procedures and the study in general have been reported previously (Gmel et al., 2015; Studer, Baggio, et al., 2013; Studer, Mohler-Kuo, et al., 2013). A total of 5,516 men (73.0% response rate) filled out the third-wave questionnaire between April 2016 and March 2018. Due to missing values for at least one variable of interest, 184 respondents (3.3% of respondents) were excluded. The final sample for analysis comprised 5,332 participants (96.7% of respondents). Mean participant age was 25.45 years old. There were 3,046 (57.1%) French-speaking and 2,286 (42.9%) German-speaking participants. A total of 173 (3.2%), 2,156 (40.4%), and 3,003 (56.3%) participants reported primary schooling, vocational training, and post-secondary schooling as highest level of education, respectively (Table 1).

1 lentelė. Descriptive characteristics of the sample (N = 5,332)

Matavimai

Descriptive statistics were used to characterize the sample. Reliability of each multi-item scale was examined using Cronbach’s α. FCU reflected the usual number of days of CU per month (non-users were coded 0), and PCU reflected the number of symptoms endorsed. FCU was analyzed using Hurdle models, which were preferred over the more habitual Poisson, negative binomial (NB), or zero-inflated count models because the same model allows an analysis of both cybersex users versus non-users and the FCU among cybersex users. In Hurdle models, the binary part – differentiating between non-zero and zero observations (i.e., cybersex users and non-users) – uses logistic regression, whereas the count part uses a zero-truncated count distribution (Poisson or NB). Based on the Bayesian Information Criterion (BIC), the zero-truncated NB distribution was retained. PCU was analyzed among cybersex users only (N = 4,190). Several different count distributions [i.e., Poisson, zero-inflated Poisson (ZIP), NB, and zero-inflated NB (ZINB)] were evaluated for fit using the BIC, and NB regression models were retained to analyze PCU. SPSS version 25 (IBM Corp., Armonk, NY, USA) was used for data coding and descriptive statistics and Stata 15 (StataCorp LP, College Station, TX, USA) was used for Hurdle and NB models.

Two models were tested for the FCU and PCU. Model 1 tested the bivariate associations of each predictor variable, whereas Model 2 tested the associations of each predictor variable, adjusted simultaneously for sociodemographic and sexual variables, that is, highest level of education, linguistic region, being in a relationship, sexual orientation, number of sexual partners, and age. Associations were reported as odds ratios (ORs), for the first parts of the Hurdle models analyzing cybersex users versus non-users. Incidence rate ratios (IRRs) were reported for the NB models. To enable a comparison of the strength of associations, continuous predictor variables were z-standardized (i.e., M = 0, SD = 1).

C-SURF was approved by Lausanne University Medical School’s Ethics Committee for Clinical Research (research protocol number: 15/07).

In Hurdle models, post-secondary schooling (vs. primary) and living in the German-speaking region (vs. French-speaking) were significantly associated with higher odds of CU, but not with FCU (Table 2). Age and being in a relationship were significantly associated with lower odds of CU and lower FCU. As opposed to a heterosexual orientation, bisexual and homosexual orientations were significantly associated with higher odds of CU and higher FCU. Reporting more than one sexual partner in the past 12 months (vs. one) was significantly associated with higher odds of CU and higher FCU. On the contrary, reporting zero sexual partners was significantly associated with higher FCU but not with CU. Dysfunctional coping strategies and all the personality trait variables except denial were significantly associated with CU and FCU. Specifically, self-distraction, behavioral disengagement, self-blame, neuroticism–anxiety, aggression–hostility, and sensation seeking were significantly associated with higher odds of CU and higher FCU. By contrast, sociability was associated with lower odds of CU and lower FCU. Adjustment (Model 2) did not alter the results.

2 lentelė. Hurdle models for associations with cybersex use (CU) and frequency of cybersex use (FCU)

NB models for PCU showed that living in the German-speaking region (vs. French-speaking) was significantly associated with lower PCU (Table 3). Bisexual orientation (vs. a heterosexual orientation) was significantly associated with more PCU, whereas association of homosexual orientation did not reach significance. Reporting four or more sexual partners in the past 12 months (vs. one) was significantly associated with higher PCU, whereas no significant associations were found for reporting zero and two or three sexual partners. Regarding the associations of psychological factors, all the personality traits tested for and all the dysfunctional coping variables were significantly and positively associated with PCU, except the sociability trait, which exhibited a significant negative association. Adjustment (Model 2) did not alter these results.

3 lentelė. Negative binomial regression models for associations with problematic cybersex use (PCU)

In line with the results of Træen et al. (2006), this study showed that more educated participants were more likely to use cybersex. One potential explanation is that more educated individuals (vs. less educated) are more prone to CU because they have greater computer skills (Stack, Wasserman, & Kern, 2004). However, no evidence of any associations between education and FCU or PCU was found. Interestingly, compared with French-speaking participants, although German-speaking participants reported less PCU, they were more likely to report CU. One potential explanation is that CU may be more socially accepted in the German-speaking region than in the French-speaking region. If so, German-speaking individuals may be more inclined to disclose their CU, but perceive their CU as less problematic. In addition, differences in the understanding of the questions may exist between French- and German-speaking participants. Further research is needed to replicate and better understand this finding. Older (vs. younger) participants were less likely to use cybersex and used it less often. As Daneback et al. (2005) revealed, this suggests that CU decreases after 18–24 years. No significant association was found between age and PCU. This finding is in contrast with the negative association between age and problematic pornography use reported by Grubbs, Kraus, and Perry (2019) in a representative sample of the US Internet users (M_amžius = 44.8, SD = 16.7). Possibly, the narrow age range of participants in the present study may be insufficient to capture age-related differences in PCU.

In line with the results of previous studies (Ballester-Arnal et al., 2014; Ballester-Arnal, Castro Calvo, Gil-Llario, & Gil-Julia, 2017), participants in a relationship had lower odds of CU and lower FCU (vs. those not in a relationship). Among cybersex users, being in a relationship was not significantly associated with PCU. This finding suggests that those not in relationships may use cybersex to satisfy their sexual needs and compensate for their lack of real-life sexual activity (Ballester-Arnal et al., 2014). This explanation is also consistent with the finding that reporting no sexual partners (vs. one) in the previous 12 months was associated with more frequent CU. In any case, reporting no sexual partners and not being in a relationship may not be problematic, since no significant association was found with PCU. Moreover, as shown previously (Braun-Courville & Rojas, 2009; Daneback et al., 2005), individuals reporting several sexual partners (vs. one) were more likely to use cybersex and used it more often. Those reporting four or more sexual partners also endorsed more than 20% more PCU statements. The associations of this variable were among the largest of all the predictor variables tested. As proposed by Daneback et al. (2005), this suggests that those with a high level of interest in all things sexual are more likely to engage in cybersex and to have more sexual partners in real life.

The associations of sexual orientation were also among the largest observed in this study. Homosexual or bisexual orientations (vs. heterosexual) were positively associated with CU and FCU – a finding consistent with the results of previous studies (e.g., Daneback et al., 2005; „Giordano & Cashwell“, 2017 m; Peter & Valkenburg, 2011 m; Træen et al., 2006). Since non-heterosexual individuals may be more subject to the vulnerability of social marginalization (Takács, 2006), they may also be more prone to using cybersex because it offers more opportunities to find partners than does real life (Benotsch, Kalichman, & Cage, 2002; Clemens, Atkin, & Krishnan, 2015; Lever, Grov, Royce, & Gillespie, 2008). This finding may also reflect the greater openness of homosexual and bisexual individuals to less traditional types of sexual activity, such as cybersex (Daneback et al., 2005) and their greater risk of engaging in hypersexual behaviors (Bőthe ir kt., 2018 m). A non-heterosexual orientation was also associated with the endorsement of more PCU statements, but this was only significant for bisexual individuals. Non-heterosexual individuals (Kingas ir kt., 2008 m), particularly those with a bisexual orientation (Gonzales, Przedworski, & Henning-Smith, 2016; Loi, Lea, & Howard, 2017), are generally more prone to reporting more mental health problems, including addictions, than heterosexual individuals. Thus, among individuals with a bisexual orientation, PCU may be a consequence of using cybersex to cope with the stress and negative emotions caused by social marginalization. This suggests that efforts to develop prevention measures targeting and adapted to individuals of a bisexual orientation may be promising.

The findings regarding the associations between various psychological factors and CU, FCU, and PCU were in line with the proposition of Grubbs, Wright, et al. (2019) that cybersex is used for two major reasons: pleasure and mood management. More specifically, the significant positive associations between sensation seeking and CU, FCU, and PCU were in line with the results of previous studies (Beyens, Vandenbosch ir Eggermont, 2015 m; Cooperis ir kt., 2000 m; Peter & Valkenburg, 2011 m). This supports the hypothesis that sensation seeking may predispose individuals to CU for pleasure, but also to PCU. Since high sensation seekers need high levels of stimulation to reach their optimal level of arousal (Zuckerman, 1994), interventions providing alternative sources of stimulation, promoting attractive, alternative activities to CU may be effective in preventing PCU among high sensation seekers.

On the other hand, all dysfunctional coping strategies were positively related to CU, FCU (although not significant for denial), and PCU. This finding is consistent with the results of Laier and Brand (2014), showing that using sex to cope with aversive affective states and stress may play an important role in the development and maintenance of PCU. This study extends this finding to other dysfunctional coping strategies, as previously shown by Antons et al. (2019). Moreover, the significant positive associations between the aggression–hostility (a personality trait almost the reverse of agreeableness) and the neuroticism–anxiety trait and CU, FCU, and PCU are in line with the results of previous studies showing cybersex’s negative associations with agreeableness (Beutel ir kt., 2017 m) and positive associations with neuroticism (Egan & Parmar, 2013; Shimoni et al., 2018). Since both the neuroticism–anxiety and aggression–hostility traits are part of a larger construct, namely negative emotionality (Zuckerman, 2002), this finding indicates that these traits may predispose individuals to CU for mood management purposes, but also to PCU. Interventions such as reducing levels of stress, providing alternatives to coping using cybersex, and building self-esteem through life-skills training may be effective means of preventing PCU among those using cybersex for mood-management purposes.

Furthermore, significant negative associations were found between the sociability trait and CU, FCU, and PCU. This finding is in line with the negative associations between extraversion (a personality trait close to sociability; see Zuckerman, 2002) and sexual addiction (not specifically related to the Internet) observed by Egan and Parmar (2013). However, it contrasts with the non-significant associations observed by Shimoni et al. (2018) and with the significant positive association between extraversion and CU observed by Beutel et al. (2017). Further research is needed to better understand associations between sociability and CU and PCU.

This study had several limitations. The cross-sectional design did not enable us to draw causal relationships or conclusions. The sample made up exclusively of young males renders any generalization of the findings to women and other age groups impossible. Several scales showed moderate reliability (.60 < α < .70; Robinson, Shaver, & Wrightsman, 1991), and the reliability of the Self-Blame Dysfunctional Coping Scale was suboptimal. Moreover, significant associations were at best indicative of small effect sizes (Olivier, May, & Bell, 2017). Finally, the use of self-reported measures may introduce some bias, especially given the sensitive nature of questions about CU. Further studies using longitudinal designs, also including females, considering the entire lifespan are needed to generalize the findings. Moreover, further studies are needed to investigate the associations of CU and PCU with mental health outcomes, substance use disorders, and other behavioral addictions.