A sample of 1,820 adults divided into two independent subsamples was used: gay people and heterosexual people. Non-probabilistic convenience sampling was used to obtain a subsample of 910 participants (505 men and 405 women) who met the following inclusion criteria: (a) Spanish language as their mother tongue; (b) be 18 years of age or older; (c) have a gay orientation; (d) be cisgender; and (e) maintained a relationship with another cisgender and same-sex person for at least three months. In addition, there was a second subsample of 910 participants selected by incidental sampling (505 men and 405 women), who met the same inclusion criteria, except that they were exclusively heterosexual and maintained a relationship with a cisgender person of different sex for at least three months at the time of the evaluation. Table 1 shows the sociodemographic information of the participants.

Sociodemographic and Sexual History Questionnaire. This ad hoc questionnaire collects information on sexual identity, age, nationality, relationship, partner’s sexual identity, partner’s age, duration of the relationship, exclusivity of the relationship, cohabitation with the partner, sexual relations with the current partner, age of first sexual relation (anal, oral, or vaginal) and number of sexual partners.

Kinsey scale (Kinsey, Pomeroy, & Martin, 1998). Identifies sexual orientation using eight response options from exclusively heterosexual (1) to exclusively homosexual (7). An eighth option was included to account for asexuality. Only subjects who marked the exclusively homosexual option (gay subsample) or exclusively heterosexual (heterosexual subsample) were selected.

Spanish version of the IEMSSQ (Lawrance et al., 2011), validated by Sánchez-Fuentes et al. (2015) described above.

Spanish version of the Massachusetts General Hospital-Sexual Functioning Questionnaire (MGH-SFQ; Fava, Rankin, Alpert, Nierenberg, & Worthington, 1998), validated by Sierra, Vallejo-Medina, Santos-Iglesias, and Lameiras Fernández (2012). Its five items assess sexual functioning during the past month: interest, arousal, orgasm, erection (only for men), and overall sexual satisfaction (e.g., How has your interest in sex been over the past month?). Answers are given on a five-point scale from Has decreased completely (0) to Normal (4). Given that the possible range of scores is different for men and women, the mean score was used instead of the total score. Item 5 (Sexual satisfaction) was excluded from the calculation of the mean score to avoid the overlapping of the sexual functioning and sexual satisfaction measures; this item was then used independently as an additional measure of sexual satisfaction. Higher scores indicate better sexual functioning. This version has shown good reliability and convergent validity (Sierra et al., 2012). In the present study, a Cronbach’s alpha of .84 was obtained in gay men and .83 in lesbian women.

Reduced Spanish version of the Dyadic Adjustment Scale (DAS; Spanier, 2017), adapted by Santos-Iglesias, Vallejo-Medina, and Sierra (2009). Its 13 items (e.g., How often do you and your partner quarrel?) assess dyadic adjustment in the couple based on three dimensions (Satisfaction, Consensus, and Cohesion). The scale uses six-point response options (from Always disagree to Always agree) and five-point response options (from Never to Everyday), depending on the item. Higher scores indicate higher adjustment. The scale presents adequate reliability and validity (Santos-Iglesias et al., 2009). A Cronbach’s alpha of .79 was obtained in the present study, both for gay men and lesbian women.

Firstly, a linguistic adaptation of the RCC was carried out so that it could be answered by gay people. Subsequently, the draft was sent to three experts in Human Sexuality for the examination of the wording of the items. Nine items that explicitly allude to heterosexual people were modified (e.g., Ability/inability to have a child was replaced with Ability/inability to have a child [either adopted or biological]). After the expert judgment, a pilot study was conducted in which the questionnaire was administered to ten gay adults (five men and five women). After each item, they were asked whether the text was understandable, and if they expressed doubts, they were asked to specify. Given that all the items were reported as understood, no additional changes were made, which resulted in the final version of the RCC for people with a same-sex partner. No linguistic adaptation was required for the rest of the IEMSSQ scales. See Appendix 1.

The instruments were administered using both the traditional paper and pencil format, and online format; these two methods have been shown to present no differences in terms of the information on sexual behaviors that they can obtain (Sierra, Moyano, Vallejo-Medina, & Gómez-Berrocal, 2018; Velten, Scholten, & Margraf, 2018). The participants who completed questionnaires in paper and pencil format were contacted at educational centers, social centers, public and leisure places, as well as through lesbian, gay, bisexual, and transgender (LGBT) associations. The online version was distributed using virtual platforms (Facebook®, Twitter®, WhatsApp®, and e-mail), using the Limesurvey® software; the IP addresses of the respondents were controlled. To access the survey, the participant had to confirm the access by answering a security question consisting of a random sum. All participants completed the evaluation voluntarily, and their anonymity and the confidentiality of the data were guaranteed at all times. Before answering the questionnaire, the respondents were asked to read and accept the informed consent forms, which included a description of the purpose of the study and information on data confidentiality and privacy. The study was approved by the Ethics Committee in Human Research of the University of Granada.

The omega (McDonald, 1999; Ventura-León, 2018) and ordinal alpha coefficients (Zumbo, Gadermann, & Zeisser, 2007) were used to estimate the internal consistency of GMSEX and GMREL. The MBESS package (version 4.2.0; Kelley, 2018) was used for calculating omega, and the userfriendlyscience package (version 0.7.2; Peters, 2018) was used for calculating ordinal alpha. Both packages run in the R® environment (version 3.4.4; R Core Team, 2016) with its RStudio® integrated interface (version 1.1.447; RStudio Team, 2018). The analysis used the omega hierarchical, which gives the proportion of variance of the scores of each scale based on a general factor (McDonald, 1999; Zinbarg, Revelle, & Yovel, 2007). Pearson’s correlations were used to examine construct and concurrent validity. Due to the large size of the sample, a conservative alpha (p<.01), as well as Cohen (1988) criterion was used to establish the rank of the effect according to its size: small (0.10), moderate (0.30) and large (0.50). To determine whether GMSEX and GMREL assess two different but related constructs invariably in gay and heterosexual men and women, a Confirmatory Factor Analysis (CFA) was performed using the lavaan package (version 0.6-3; Rosseel, 2012). An analysis of multigroup invariance was carried out, in which four groups were defined: gay men, lesbian women, heterosexual men, and heterosexual women. The progressive invariance (configural, weak, strong, and strict) of the two-factor model was also tested. GMSEX and GMREL were established as the two factors; each factor had five observable variables between errors. The analysis included the Diagonally Weighted Least Squares (WLSMV) estimation method, a robust estimator in case of a lack of multivariate normality designed for ordinal data (Li, 2016). The following criteria were taken into account to assess the multigroup AFC fit: CFI and TLI>.90 and RMSEA<.08 (Hu & Bentler, 1999; Wang & Wang, 2012); in addition, it was established that if the difference between the values of two nested models in the CFI was higher than .01 in favor of the model with fewer restrictions (Bentler, 1990), the model with more restrictions would be rejected (Cheung & Rensvold, 2002). Ultimately, the invariance levels concerning the relationship between observable variables and latent variables were examined (Milfont & Fischer, 2010). Finally, for the analysis of the comparisons of GMSEX and GMREL by sex and sexual orientation, and given that their scores did not adjust to a normal distribution, the Mann-Whitney U was used.

An analysis was made of the median, mean scores and the standard deviation of the items to detect extreme scores (Guillot-Valdés, Guillén-Riquelme, & Buela-Casal, 2019). The means ranged between 5.85 and 6.64 (theoretical range of 1 to 7). Standard deviations ranged from 1 to 1.44. This indicates that the subjects have a response pattern with high frequencies at the upper end (Table 2).

In the case of GMSEX, gay men obtained an omega of .90 and an ordinal alpha of .93; in lesbian women, the coefficients were .88 and .92, respectively. In the case of GMREL, gay men obtained an omega coefficient of .94 and ordinal alpha of .95; in lesbian women, values of .92 and .94 were obtained, respectively. The reliability of internal consistency of both scales does not improve with the elimination of any of its items.

Evidence on the construct validity of the different IEMSSQ measures except for EQREW and EQCST reflects that they were significantly associated with each other in both gay men and lesbian women (Table 3). Regarding concurrent validity, all IEMSSQ measures except for EQREW and EQCST were significantly associated with the scores of an individual item of sexual satisfaction, sexual functioning, and dyadic adjustment, both in gay men and lesbian women, presenting sex-based differences in the strength of the associations. In men, EQREW was associated with the Consensus component of dyadic adjustment and EQCST with the Satisfaction component; in women, EQCST was correlated with the Satisfaction component of dyadic adjustment and the global dyadic adjustment score (Table 3).

A series of Fisher r to z transformations were performed to assess the magnitude of the differences between each pair of correlation coefficients and determine whether the magnitude of the correlations between each pair differed significantly between men and women (Sánchez-Fuentes et al., 2015). Due to a large number of comparisons (90), the Bonferroni correction was used to avoid the type I inflated error (Cabin & Mitchell, 2000) (α=.0005). Only two pairs of correlations (number of rewards-dyadic adjustment and the number of rewards-Satisfaction component of dyadic adjustment) were statistically different between men and women and significantly higher among men. In other words, the patterns of correlations between the IEMSSQ measures of dyadic adjustment and sexual functioning are similar for gay men and lesbian women.

The results obtained show the two-dimensional model (GMSEX and GMREL), evidencing a strict adjustment in the factorial invariance using four groups: gay men, lesbian women, heterosexual men, and heterosexual women. As no differences higher than .01 were obtained in the CFI, the maximum level of adjustment was reached (see Figure 1 and Table 4).

Figure 1.

Confirmatory Factor Analysis (CFA).

(0.15MB).

Once the strict invariance of GMSEX and GMREL was established, the mean scores of gay men and women were compared with those of heterosexual men and women. Significant differences were found between lesbian and heterosexual women, both in GMSEX (U=69950.50, p<.001, d=0.24) and in GMREL (U=71788, p<.01, d=0.21); higher scores were obtained by lesbian women in both cases (Table 5).

Table 6 shows the scores of gay men and women for each IEMSSQ measure. In general, gay men and lesbian women obtained high scores in all the measures. Given its factor invariance, GMREL and GMSEX measures were compared between gay men and lesbian women. Significant differences were found in GMSEX (U=83815, p<.001, d=−.30) and GMREL (U=89223, p<.01, d=.21); women scored higher in both cases.