The responses “never”, “mostly not” and “sometimes” were categorized as possessing
The responses “never”, “mostly not” and “sometimes” were categorized as getting condomless anal sex, whereas “mostly” and “always” were categorized as using condoms. The injunctive norm variable was dichotomized as follows: The responses “3” and larger have been categorized as disapproval of condom use, whereas the responses “” and “2” have been categorized as approval of condom use. Univariate and multivariate logistic regression analyses have been carried out to investigate the association amongst sort of internet site and norms. For the type of web site variable, sex venue was chosen because the reference category. Primarily based on a preceding study (Grov, 202), the multivariate model incorporated variables that were regarded possible confounders for the association between variety of web-site and norms: age, education level, nationality, and sexual orientation. For MSM who reported both the assumed behavior and their very own actual behavior (n262), we compared the scores around the norms variables using McNemar tests. Moreover, univariate and multivariate logistic regression analyses have been carried out to investigate the association amongst norms and one’s own condom use. All norm variables have been entered as continuous into the model. For these analyses, two multivariate models had been constructed. Model examined the associations involving norms relating to other guests and one’s personal condom use. Model two examined the associations between norms relating to the superior pal and own condom use, controlling for norms regarding other guests. Each models had been adjusted for variables that were regarded as possible confounders for the association amongst norms and own condom use (i.e age, education level, nationality, sexual orientation) as suggested in the literature (e.g (Ghaziani Cook, 2005)). For all analyses, we checked for interactions in between the principle variables of interest along with the other variables in the final multivariate models. Prior PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/25870032 to the logistic regression analyses, collinearity amongst all variables was checked by computing variance inflating aspect (VIF) for every determinant and by computing correlations amongst the determinants. VIF Tyrphostin NT157 values above 0 and correlations above 0.70 had been regarded as an indication of collinearity. For all analyses, a p 0.05 was considered statistically significant. Analyses were performed with the SPSS 9 statistical package (SPSS Inc Chicago, IL, USA).Author Manuscript Author Manuscript Author Manuscript ResultsEnrollmentA total of 2376 participants enrolled in the study at 56 web-sites: 376 participants at eight sex venues, 537 participants at two barsclubs, three participants at 9 socialsports gathering venues, 495 participants at six dating internet websites and 855 participants at 2 social network web sites. Sample characteristics The median age of the participants was 30 years (IQR 2243), 56 had been hugely educated, and 82 have been of Dutch background (Table ). General, sample characteristics differed among varieties of sites. At offline sex venues, participants were on average older, along with a higherAuthor ManuscriptHealth Psychol. Author manuscript; out there in PMC 206 August 0.van den Boom et al.Pageproportion was nonDutch. At all 3 forms of offline web-sites, participants were a lot more probably to be very educated than at web sites. Sites have been most often visited, socialsports gatherings the least (Table ). Norms and selfreported behavior per site At the 8 offline sex venues, 64 (242376) and 24 (69283) of participants assumed that other visitors and their good buddy would have condomless sex, r.