Gay Community Stress in Sexual Minority Men and Women: A Validation Study in the Netherlands

ABSTRACT

Intraminority gay community stress theory posits that social stressors within sexual minority communities of men may be risk factors for mental health problems in gay and bisexual men. The recently developed 20-item Gay Community Stress Scale (GCSS) is a valid and reliable measure of gay community stress, but was not yet validated in the Netherlands. This study developed a Dutch-translated version of the GCSS and validated this scale in sexual minority men and sexual minority women, as it was hypothesized that sexual minority women may also experience intraminority stress. Exploratory and confirmatory factor analyses were subsequently performed in independent samples of men and women, and produced a 16-item GCSS for men and a 12-item GCSS for women. The four-factor structure of the original GCSS was replicated in men and women, and encouraging support for discriminant and concurrent validity of the GCSS was found in both men and women. The total scale and subscales were internally consistent in men (α and ω ≥ .87) and in women (α and ω ≥ .78). The Dutch-translated GCSS seems to offer a valid and reliable way to assess intraminority stress in Dutch-speaking sexual minority men and sexual minority women, although further validation is warranted.

KEYWORDS:

• Gay community stress
• intraminority stress
• sexual minority
• scale validation
• mental health
• social pressure
• LGBT community

It has been well documented that members of sexual minority groups experience more mental health problems, suicidality, substance use disorders, and other mental health problems than their heterosexual counterparts (e.g., Bos et al., 2008; Rodriguez-Seijas et al., 2019). The sexual minority stress theory is the leading theoretical model to explain these health disparities. It states that sexual minority individuals are, in addition to general stressors, exposed to unique forms of stress that stem from a society’s prejudice and stigma directed at sexual minorities (Meyer, 2003). Consequently, sexual minority individuals experience more negative life events that increase the risk for stress-sensitive mental health problems. Feeling connected to a sexual minority community may buffer the negative consequences of minority stress (Meyer, 2003), as sexual minority individuals may experience support from the community (Frost et al., 2016). However, sexual minority individuals may also experience pressure from the minority community to behave in certain ways and conform to social rules and norms. As such, the community may pose an additional stressor for sexual minority individuals (e.g., Pachankis et al., 2020).

Intraminority stress in communities of sexual minority men

Several researchers have described that sexual minority individuals may experience stress from the sexual minority community itself. Green already revealed in (2008) that gay men with a low sexual status face intraminority stressors such as avoidance of others, stigmatization, and rejection. It was Pachankis et al. (2020) who formally theorized and defined the construct of “gay community stress” (GCS), which refers to the unique type of stress that gay men experience within their sexual minority community. The Gay Community Stress Scale (GCSS) that these researchers developed to assess GCS was constructed and validated in the United States (Pachankis et al., 2020), but has not yet been validated in the Netherlands. Therefore, this study aimed to validate a Dutch-translated version of the GCSS in gay and bisexual men in the Netherlands. As sexual minority women may also experience intraminority stress, the second aim was to validate a Dutch-translated GCSS in sexual minority women in the Netherlands.

Pachankis et al. (2020) conducted a series of studies to formulate their intraminority GCS theory, which states that the gay male community comprises multiple elements that pose mental health risks to gay and bisexual men. In short, this theory posits that due to the unique nature of gay and bisexual men’s social and sexual interactions, they experience status-based social and sexual competition in a community with clear status hierarchies. The stress following these unique elements is not at all or somewhat experienced in interactions between heterosexual men. This intraminority GCS can be defined in four factors that each capture stress resulting from the perception that the gay community is strongly focused on sex, status, social competition, and exclusion of diversity (Pachankis et al., 2020). The focus on sex refers to the gay community’s hypersexuality and risky sex, which comes at the expense of romantic relationships. The status factor refers to the community’s valuing of wealth and prestige. The social competition factor describes the perceived competitiveness of the community, which includes fighting, gossiping, being materialistic, and being judgmental. Finally, the exclusion of diversity factor refers to the perception of the gay community being exclusive, in the sense that the community is racist and excludes HIV-positive gay and bisexual men.

In one of their studies, Pachankis et al. (2020) showed that intraminority GCS comprising these four factors predicted mental health symptoms over and above minority stress and general life stress. As such, the intraminority GCS theory complements the more traditional minority stress model (Meyer, 2003) that explains the reduced levels of mental well-being in members of sexual minority groups relative to the heterosexual majority. Thus, the GCS theory not only sheds more light on the etiology of mental health symptoms experienced by members of the gay male community, it also offers new grounds for clinical practice aimed at alleviating these symptoms and promoting mental well-being in gay and bisexual men.

The intraminority GCS theory was initially developed to understand the mental health of gay and bisexual men. In developing their theory, Pachankis et al. (2020) built upon three former theoretical frameworks that each describe how men, in general, may experience stress by the status hierarchies in their male communities: the intrasex competition theory (Anderson et al., 2001), the sexual field theory (Green, 2014), and the theory of precarious manhood (Vandello et al., 2008). The researchers reasoned that the stress experienced by men might exacerbate in communities of gay and bisexual men who are particularly compelled to show and defend their social status.

Intraminority stress in communities of sexual minority women

However, sexual minority women may also face intraminority stress that affects their mental health (Mahon et al., 2021). For instance, empirical research has shown that lesbian and bisexual women may experience within-community pressure to look like authentic lesbian and minority community members regarding clothing and appearance (Clarke & Spence, 2013; Clarke & Turner, 2007). Further, Lyle et al. (1999) reported that nonwhite lesbian women might experience tension between the appearance norms of their own ethnic culture and the White beauty ideals of mainstream lesbian communities. Women of color may experience racism and not only in White sexual minority communities but also in sexual minority communities of color (Lehavot et al., 2009). Other studies have also revealed that social norms of female sexual minority communities can be strong and that such strict norms may threaten the mental well-being of sexual minority women (Heath & Mulligan, 2008). For instance, Boyle and Omoto (2013), who collected correlational data in their survey, found that young sexual minority women find it challenging to meet the standards of the strict social norms they perceive in lesbian communities and that this disparity is associated with feelings of depression and anxiety. This empirical evidence suggests that lesbian and bisexual women—like gay and bisexual men—may experience unique intraminority community stress that is to a lesser extent or not at all experienced by heterosexual women. Therefore, examining and validating the concept of GCS in sexual minority women is highly relevant.

The need for cross-national validation

To our knowledge, the GCSS developed by Pachankis et al. (2020) has been validated outside the US to only a limited extent and not yet in the Netherlands. However, it is important that the GCSS is more extensively validated cross-nationally, as Pachankis and colleagues point out that GCS might comprise different factors across global regions. Moreover, the degree to which intragroup marginalization occurs and members of sexual minority groups experience intracommunity stress may be affected by the cultural norms and values of the societal context in which these groups reside. Although the USA and the Netherlands are both Western countries and quite comparable in terms of the cultural dimensions power distance, individualism, and uncertainty avoidance, the USA is considerably more masculinist in its national culture (Dustdar & Hofstede, 1999). Further, when it comes to attitudes toward sexuality, Haney (2016) noted that both countries were sexual traditional until the 1960s, after which a sexual revolution took place that occurred much faster in the Netherlands than in the USA. Consequently, the Netherlands adopted policies and legislation promoting gay rights earlier than the United States, and moreover, same-sex activity between two consenting adults was a criminal offense in multiple states of the US until 2003, whereas laws prohibiting same-sex activity were abandoned almost two decades earlier in the Netherlands (Haney, 2016). This course of events may explain that more conservative attitudes about sexual expression have consistently been reported in the US with the US showing higher levels of social intolerance toward same-sex sexual orientation than the Netherlands and other European countries (Hadler, 2012). On the other hand, there is evidence that the Dutch are tolerant of same-sex sexual orientation only from a comfortable distance (Buijs et al., 2011), and that they may accept sexual minority orientations to a limited extent because it is the norm to do so (Keuzenkamp & Bos, 2007). There is also evidence that anti-gay sentiment in the Netherlands is on the rise despite progressive policies that support gay rights (Collier et al., 2015; Huijnk et al., 2022). Still, the differences in national contexts make it important to validate the GCSS in the Netherlands.

Study aims

This study aimed to seek cross-national evidence for the psychometric quality of the recently developed Gay Community Stress scale (GCSS; Pachankis et al., 2020) in Dutch sexual minority men. We hypothesized that the four factor structure underlying GCS as identified by Pachankis and colleagues in US and Swedish samples of gay and bisexual men can also be identified in Dutch populations of sexual minority men. A second aim was to examine the psychometric properties of the GCSS in Dutch sexual minority women, as they may also perceive within-community pressure to behave and look like according to the community’s norms which may induce stress. As differences between male and female sexual minority communities can be expected in terms of the social norms and values they hold, it is likely that conceptualizations of GCS between sexual minority men and sexual minority women differ as well. Moreover, the items of the original GCSS were formulated based on interviews with only men. We therefore did not hypothesize that the four factor structure underlying GCS can be identified in Dutch populations of sexual minority women, although we did expect that the concept of GCS can be identified in sexual minority women. Further, we hypothesized that scores on the Dutch-translated GCSS are weakly or not at all associated with scores on a scale for general life stress, as general life stressors (e.g., not feeling able to cope with important life changes, being upset because of unexpected events, and feeling constantly overwhelmed) are conceptually similar to, but distinct from, stressors within sexual minority communities that are uniquely experienced by sexual minority individuals and not by heterosexual people. Finally, we hypothesized that scores on the Dutch-translated GCSS positively associate with scales measuring stress-sensitive mental health symptoms.

In pursuing these aims, three consecutive studies were performed. All analyses were performed separately for sexual minority men and sexual minority women, as we did not assume that the psychometric properties of a Dutch-translated GCSS are equal between men and women. Stratifying the analyses by sex assigned at birth enabled capturing potential differences in GCS processes between sexual minority men and sexual minority women. Approval for this research project was obtained from the Institutional Review Board (IRB) of the Faculty of Social and Behavioral Sciences of the University of Amsterdam in the Netherlands (#2020-CDE-12894).

Exploratory factor analysis (Study 1)

Study 1 aimed to translate the original GCSS in Dutch, and to perform exploratory factor analyses (EFA) to determine whether the factor structure as identified by Pachankis et al. (2020) could be reproduced in a sample of Dutch-speaking sexual minority men and a separate sample of Dutch-speaking sexual minority women.

Materials and methods

Participants

Participants in Study 1 were 16 years or older, Dutch-speaking, and mostly identifying themselves as lesbian, bisexual, or gay. However, participants who not labeled themselves as lesbian, bisexual, or gay, but who did experience same-sex attraction could also participate. In total, N = 152 respondents with a male birth sex filled out all questions of the Dutch-translated GCSS (further referred to as sample of “men”), and N = 173 respondents with a female birth sex filled out all these questions (further referred to as sample of “women”). The minimum sample size for the EFA model structure was 100 according to Soper’s (2022) sample size calculator for structural equation modeling (with the significance level set at .05, statistical power set at .80, an anticipated medium effect of 0.3, twenty observed variables, and an anticipated number of latent variables set at four). Thus, we considered the male and female samples of sufficient size for EFA. Table S1 presents characteristics of the sampled men and women.

Measures

Demographic characteristics

The following demographics were assessed: age in years, sex assigned at birth (based on the survey question “What sex were you assigned at birth” with response categories female or male), gender identity (based on the survey question “I mainly identify as … ” with response categories female, male, not female nor male, unable to tell, and other), sexual orientation (based on the survey question “Which of the following best describes your sexual orientation?” with response categories lesbian/gay, bisexual, and other), highest level of education (categorized into (applied) university, vocational college, and secondary education or lower), race/ethnicity (based on the survey question “How would you describe your race/ethnicity” with responses dichotomized into Western and non-Western), whether or not participants have a relationship, and if so, the duration of that relationship (in years).

Gay community stress

GCS was measured with a Dutch-translated version of the Gay Community Stress Scale (GCSS; Pachankis et al., 2020) that comprises one total scale, four subscales, and 20 items in total. Across the four subscales, the items measure stress related to perceptions of the gay community’s strong focus on sex (e.g., “The mainstream gay community is overly focused on sex”), status, (e.g., “The mainstream gay community overly values men who are powerful and high status”), competition (e.g., “The mainstream gay community has a culture of competition and jealousy”), and exclusion of diversity (e.g., “The mainstream gay community is racist”). For each item, participants were asked to indicate the extent to which they believed the item was true on a 5-point Likert-scale ranging from strongly disagree (1) to strongly agree (5). A total score was created by summing item scores and calculating an average. Total scores can range from 20 to 100, and higher scores indicate higher levels of stress.

As the GCSS was developed in the United States, all items had first to be translated. In fall 2020, the first author conducted a forward translation of the GCSS from the original language (American English) into Dutch. Each translated item was checked independently against the original item by the second author after which discussion was held until full consensus was reached on the accuracy of all item translations. For most items, a literal or close-to-literal translation was found to be accurate and appropriate for the Dutch language and cultural context. In translating the sixth GCSS item (“In the mainstream gay community, there is a lot of risky sex;” see Appendix 1), we found it appropriate to add a short explanation of “risky sex” to the Dutch-translated item as we reasoned that some people might not know what risky sex exactly refers to. In the translation of item 19 (“The mainstream gay community sexually objectifies men of color”), we decided not to literally translate “[…] of color” as that would produce uncommon Dutch language. Instead, we translated the phrase as “[…] of non-Dutch descent” which is more appropriate for the Dutch context. After the original GCSS items were translated into Dutch, all Dutch-translated items were pilot tested for wording and comprehension by eight native Dutch-speaking third-year Bachelor’s students who were blind to the original GSCC items and who independently assessed each item, the response categories, and the instruction of the questionnaire. The follow-up discussion between the assistants and both study authors did not lead to further adjustment of the items. After the pilot, the students were trained in recruiting participants according to the guidelines of the IRB.

As the original GCSS was specifically developed for gay and bisexual men, the items were somewhat adapted so that items could be applicable to both the male and female sexual minority community (e.g., “The mainstream lesbian, gay, or bisexual community that I feel most connected to values sex over meaningful relationships” instead of the original item “The mainstream gay community values sex over meaningful relationships.”). Also, the fill-out instructions for the Dutch-translated GCSS stated that all items pertain to a participant’s experiences of the lesbian, gay, or bisexual community he/she feels most connected to.

Procedure

Between February 2021 and May 2021, the two authors of this manuscript and eight third-year Bachelor’s students conducted an internet-based recruitment of Dutch participants through sexual minority organizations’ websites and social media accounts, as well as personal social media accounts. All recruitment advertising mentioned that the study focused on stress and health in lesbian, gay, and bisexual young adults of 16 years and older. It was not explicitly described that the survey also assessed GCS to avoid a priming effect. Recruitment posts mentioned that a voucher of 50 Euro was awarded to ten participants by means of a lottery. Interested participants were taken to a Qualtrics survey after clicking on a link in the recruitment advertisement and were first presented with information about the target population, the aims of the study, and the procedure of the online survey. They were also informed that participation is voluntary, that the survey is anonymous, that they can stop filling out the survey at any time, how confidentiality is assured, and how the principal investigators and the IRB can be contacted by e-mail in case they have questions or wish to file a complaint based on any question that was presented in the survey. When participants had read all information and agreed to participate they gave their electronic consent by replying to the first question of the survey. When participants did not agree to participate the survey was closed and no further questions were presented.

At the beginning of the Qualtrics survey, participants answered two screener questions on their current sexual orientation and the degree to which they feel attracted to women and men. Next, a block of demographic questions was presented on age, gender identity, level of education, race/ethnicity, and relationship status. The remainder of the survey comprised blocks of questions to assess stress-related and health-related constructs including GCS. After completing the survey, participants could enter their e-mail address if they wanted to participate in the voucher lottery, but this was not mandatory which was explicitly described. In the end, information on how to contact the principal investigators and/or the IRB was presented, and participants were thanked for their participation. The full survey took approximately 45 minutes to complete, and during the survey participants were encouraged to take short breaks whenever they felt in need of a break. Participants could stop and resume filling out the survey at any time, although partly filled out questionnaires were closed and registered 72 hours after a participant started the survey.

Statistical analyses

An exploratory factor analysis was performed in the JASP statistical software package (version 0.16; JASP Team, 2022), and separately in the samples of men and women. The Kaiser-Meyer-Olkin Measure of Sampling Adequacy and Bartlett’s Test of Sphericity were used to assess whether the data were suitable for factor analyses. The EFA was based on an analysis of a correlation matrix and performed with Promax rotation, as we expected that the factors of the (Dutch-translated) GCSS are not independent and thus intercorrelated at least to some extent (see also Costello & Osborne, 2005). Maximum likelihood was used as the estimation method, and statistical extraction of factors was primarily based on parallel analysis that is regarded as one of the most suitable methods for factor extraction (e.g., Zygmont & Smith, 2014).

Results

Sexual minority men

Table S2 presents the mean, standard deviation, kurtosis, and skewness for all items of the Dutch-translated GCSS. The Kaiser-Meyer-Olkin measure of sampling adequacy was .88, and Bartlett’s test of sphericity was significant, χ² (190) = 1682.67, p < .001, indicating that the associations between GCSS items were sufficiently strong to conduct EFA. Based on recommendations of Kahn (2006), a model was chosen following results of parallel analysis (Horn, 1965), visual inspection of a scree plot (Cattell, 1966), and conceptual clarity of the factor loadings and underlying structure. The results of the parallel analysis clearly suggested a four-factor model that was in line with the scree plot that showed a break after the fourth factor (see Figure S1). The structure revealed by the parallel analysis strongly resembled the four-factor-structure that Pachankis et al. (2020) found in US and Swedish men, and therefore, retaining four factors was also deemed appropriate from a conceptual viewpoint.

Model fit indices indicated that the four-factor solution was a relatively good fit. The Root Mean Square Error of Approximation (RMSEA) value of .051 (90% CI [.032; .070]) is below the .06 criterion for good model fit (Hu & Bentler, 1999), with the upper bound of the 90% confidence interval below the .08 criterion for good model fit (Hooper et al., 2008). The Tucker-Lewis Index (TLI; also known as the Non-Normed Fit Index [NNFI]) reached a value of .947 which is only slightly below Hu’s and Bentler’s (1999) criterion of .95 for good model fit. Based on the EFA results and the fit indices, the four-factor model was determined to be a good fitting model and accounted for 58.1% of the total variance.

Inspection of the factor loadings revealed that item 10 (i.e., “The mainstream gay community overly values stylish clothes and up-to-date fashion”) loaded on both the factor Status (with factor loading 0.48) and the factor Competition (with factor loading 0.40). The difference between the factor loadings was only .08 and below the threshold of .15 (Worthington & Whittaker, 2006), and as such, item 10 was deemed to insufficiently discriminate between stress invoked by the community valuing wealth/prestige and stress invoked by the community being competitive. We therefore decided to remove this item from the factor solution and reperform the EFA to ensure that the factor structure would not substantially be affected by this item removal. The EFA results again showed that a four-factor solution fitted the data relatively well, and that 58.6% of the total variance was explained by this final model (RMSEA = .055; 90% CI [.035; .074], TLI = .942).

The four-factor solution comprised the four factors that are reflective of the underlying GCS construct, as described by Pachankis et al. (2020). The first factor Competition (items 11–17) accounted for 19.7% of the total variance and represents the community’s competitiveness. The second factor Sex (items 1–6) accounted for 17.1% of the total variance and reflects the community’s strong focus on sex at the expense of romantic relationships. The third factor Status (items 7–9) accounted for 11.3% of the total variance and represents the community’s strong value of wealth and prestige. The fourth factor Exclusion (items 18–20) accounted for 10.4% of the total variance and reflects the community being socially exclusive. All items showed factor loadings on their respective factors of .40 or higher, which is a cutoff recommended by for instance Brown (2015).

The four Dutch-translated GCS subscales were internally consistent with Cronbach’s alpha’s of .88 (Competition), .87 (Sex), .85 (Status), and .82 (Exclusion). McDonalds Omega was .89, .87, .86, and .85, respectively. The factor correlations revealed that each of the four factors is significantly and positively associated with all other factors (see Table 1), which supported the choice for an oblique rotation in the EFA. Table S3 lists the factor loadings of each item on the four factors with factor loadings of at least .40 highlighted in bold.

Table 1. Factor correlations of the final EFA models in sexual minority men (N = 152) and women (N = 173) (Study 1).

	Competition	Sex	Status	Exclusion
Sample of men
Competition	[.70]
Sex	.58**	[.72]
Status	.46**	.27**	[.82]
Exclusion	.51**	.35**	.31**	[.77]
Sample of women
Competition	[.66]
Sex	.54**	[.65]
Status	.50**	.42**	[.83]
Exclusion	.44**	.45**	.33**	[.81]

Factor correlations are expressed in Pearson’s r. Average unstandardized factor loadings are presented on the diagonal.

**p < .01.

Sexual minority women

Table S2 presents means, standard deviations, skewness, and kurtosis of the items in the sample of women. Both the Kaiser-Meyer-Olkin measure of sampling adequacy (.88) and the significant test of Bartlett’s test of sphericity, χ² (190) = 1644.11, p < .001, showed that the sample data were suitable for EFA. Once again, the parallel analysis clearly suggested a four-factor model that was in line with the scree plot showing a break after the fourth factor (see Figure S2). This four-factor solution strongly resembled the four-factor structure described by Pachankis et al. (2020) in men, and thus retaining four factors seemed also conceptually logic. Model fit indices showed an acceptable fit of the four-factor structure on the sample data with an RMSEA of .053 (90% CI [.036; .070]) and a TLI of .934. This factor solution accounted for 53.4% of the total variance.

Item 10 (i.e., “The mainstream gay community overly values stylish clothes and up-to-date fashion”) did not produce a factor loading of at least .40 on any of the four factors and showed cross-loadings on the factors Sex (.33) and Status (.28) with only a small difference in factor loadings (.05). As in men, we decided to remove this item and reperform the EFA to ensure stability of the four-factor solution. The EFA results again revealed that a four-factor solution fitted the data relatively well, and accounted for 54.5% of the total variance (RMSEA = .042; 90% CI [.019; .062], TLI = .960). The four factors reflected the underlying GCS construct that was identified in men (see Pachankis et al., 2020) with the factor Competition (items 11–17) accounting for 17.0% of the total variance, the factor Sex (items 1–6) accounting for 14.8% of the total variance, the factor Status (items 7–9) accounting for 11.4% of the total variance, and the factor Exclusion (items 18–20) accounting for 11.2% of the total variance. All items loaded on their respective factors with a loading of at least .40. The four Dutch-translated GCS subscales were internally consistent with Cronbach’s alpha’s of .86 (Competition), .83 (Sex), .85 (Status), and .85 (Exclusion). McDonald’s Omega was .86, .83, .86, and .86, respectively. Factor correlations indicate that each factor is significantly and positively associated with all other factors (see Table 1), supporting the choice for an oblique rotation in the EFA. Table S3 lists the factor loadings.

Discussion

The EFA results suggest that the Dutch-translated GCSS captures four factors in both sexual minority men and sexual minority women, and that each subscale showed strong internal consistency in both men and women. The factor loadings and factor correlations indicate that in both men and women the subscales of the translated GCSS are sufficiently distinctive to consider them as separate subscales. The results of this study offer first support for the feasibility of assessing GCS with the items of the Dutch-translated GCSS in both sexual minority men and sexual minority women. We therefore proceeded with further validating the translated GCSS.

Confirmatory factor analysis (Study 2)

Study 2 sought to examine whether the factor structure identified in the EFAs could be replicated in confirmatory factor analysis (CFA). We performed separate CFAs in new samples of Dutch-speaking sexual minority men and sexual minority women that were independent from the EFA samples analyzed in Study 1.

Materials and methods

Participants

As in Study 1, participants were 16 years or older, Dutch-speaking, and mostly identifying themselves as lesbian, bisexual, or gay. Participants who were not using these labels, but who did experience same-sex attraction could also participate. A total of N = 131 respondents with a male birth sex filled out all questions of the Dutch-translated GCSS (further referred to as sample of “men”), and N = 315 respondents with a female birth sex filled out all these questions (further referred to as sample of “women”). According to Soper’s (2022) sample size calculator for structural equation modeling, the minimum sample size for the CFA model structure was 91 for a first-order CFA with four latent variables, and 112 for a second-order CFA with five latent variables (with the significance level set at .05, statistical power set at .80, an anticipated medium effect of 0.3, and nineteen observed variables). Therefore, we considered the sample of men and the sample of women of sufficient size for CFA. Table S1 presents the characteristics of these samples.

Measures

Demographics and GCS were assessed in the same way as in Study 1.

Procedure

The same procedure as in the EFA study (Study 1) was followed to recruit new samples of sexual minority men and women that were independent from the participants that were sampled for the EFAs. To ensure that the samples examined in the CFAs were fully independent of the EFA samples, an additional screener question was added to the Qualtrics survey in which we asked respondents whether they had participated in Study 1. Four participants had also participated in Study 1 and were therefore removed from the dataset. The internet-based recruitment was once again organized by the two authors of this manuscript with eight third-year Bachelor’s students who did not participate in organizing the recruitment for Study 1. Data were collected between February 2022 and May 2022.

Statistical analyses

A robust maximum likelihood confirmatory factor analysis was performed in the JASP statistical software package (version 0.16; JASP Team, 2022), and separately in the samples of men and women to further evaluate the construct validity of the Dutch-translated GCSS. A CFA is a more rigorous test of the underlying factor structure than an EFA in the sense that items are restricted to load on only one factor, with item loadings on other factors fixed to zero. As an oblique rotation was used in the EFA, it was assumed in the CFA that factors are correlated. Based on the results of Study 1, four factors were specified with 6 items loading on Factor 1 (Sex), 3 items on Factor 2 (Status), 7 items on Factor 3 (Competition), and 3 items on Factor 4 (Exclusion). Scaling of latent variables was done with the effects coding method, which is a relatively new method introduced by Little et al. (2006) that maintains the metric of the scales of the observed indicators and is therefore a non-arbitrary scaling method (opposed to the more traditional fixed-factor and marker-variable methods that are arbitrary in nature). The following criteria of model fit indices were used to determine good model fit: TLI > .95, CFI (Comparative Fit Index) > .95, RMSEA < .06 with upper bound of 90% CI not exceeding .08 and p value > .05, SRMR (Standardized Root Mean Squared Residual) < .08 (Hooper et al., 2008; Hu & Bentler, 1999). In addition, χ²/degrees of freedom [df] ≤ 2 was used to determine good model fit (Tabachnik & Fidell, 2013). For completeness, we also report on the model χ² and its p value, but due to its high dependency on sample size (e.g., Browne et al., 2002) and because it insufficiently takes model complexity and item characteristics into account, it should not be relied upon as a basis for model acceptance or rejection (e.g., Schermelleh-Engel et al., 2003).

Results

Sexual minority men

Table S4 presents the mean, standard deviation, kurtosis, and skewness for each item of the Dutch-translated GCSS. The initial CFA produced a significant factor model, χ² (146) = 267.05, p < .001, and multiple fit indices did not reach the threshold for good model fit: χ²/df = 1.83; RMSEA = .080 with 90% CI [.064; .095] and p < .01; TLI = .905; CFI = .919; and SRMR = .067. To identify elements of strain in the model, modification indices of cross-loadings and residual covariances as well as R squared values were evaluated. To determine whether items should be removed or retained, these statistics were weighed against the formulation of individual items and the construct they (should) represent. Accordingly, several items were identified as candidates for removal. Item removal occurred in a stepwise manner in which the CFA was reperformed after removal of individual items to check how the model fit is affected.

Substantial residual covariance between items 11 and 12 were detected with a high modification index. After a conceptual evaluation of both items we concluded that item 11 describes two competition components instead of one, and uses rather old-fashioned wording that may be difficult to grasp for younger respondents. Item 11 was therefore removed. Next, item 20 on HIV/AIDS discrimination explained the least variance with an R² value (.34) which was less than the suggested lower bound of .40 (Awang, 2012). Research has revealed that HIV prevalences are relatively low in the Netherlands, and quite lower compared to the US (UNAIDS, 2022). Further, although people with HIV still have to deal with HIV-related stigma (Pitasi et al., 2018), a decline in stigmatizing attitudes over time has been reported (Herek et al., 2002), that may be due to advances in antiretroviral therapy and a greatly reduced risk of HIV-transmission. In light of this, we reasoned that item 20 insufficiently contributes to the social exclusion factor in Dutch communities of gay and bisexual men. We therefore removed item 20.

Next, item 14 showed the lowest R² value (.35) that did not reach the threshold of .40. This may be due to problems in translating “cliquey” that was used in the original English item, as the Dutch language doesn’t have a single and unequivocal equivalent for this word. We therefore removed item 14. Finally, substantial residual covariances between items 1 and 3 were detected with a relatively high modification index, but individual standardized factor loadings (0.80 and 0.84) and R² values (.65 and .71) were also substantial. As both items show semantic similarity, but from a conceptual viewpoint refer to distinct aspects of the Sex factor, we decided to retain both items and to add one within-factor correlation to the model. The CFA following all modifications produced a significant factor model, χ² (97) = 130.73, p = .013, but with model fit indices reaching the thresholds for good model fit: χ²/df = 1.35; RMSEA = .052 with 90% CI [.025; .073] and p = .439; TLI = .966; CFI = .973; and SRMR = .054. Figure 1 depicts the CFA structure.

Figure 1. First-order confirmatory factor analysis structure of the Dutch-translated gay community stress scale in sexual minority men.

Numbers in rectangles represent item numbers (see Appendix 1 for all items).

Fc1 = Sex factor; Fc2 = Status factor; Fc3 = Competition factor; Fc4 = Exclusion factor.

1.00 = pathway constrained to one. Standardized coefficients are depicted.

To make inferences about the convergent validity of the factor model, the average variance extracted (AVE) was examined for each of the four factors (see Table 2). An AVE value measures the amount of variance captured by a factor in relation to the amount of variance attributed to measurement error (Fornell & Larcker, 1981). The convergent validity is deemed to be adequate when AVE values equal or exceed 0.50, which is the threshold at which the variance captured by a factor exceeds the variance due to measurement error. The results revealed that the AVE values ranged from .72 to .92 and all exceeded the threshold of .50, indicating adequate convergent validity. To evaluate discriminant validity of the factor model, the square root of the AVE for each factor (i.e., the Fornell-Larcker criterion) was compared with the correlations between that factor and all other factors. When all Fornell-Larcker criteria exceed the relevant cross-correlations, the discriminant validity of the model is deemed to be adequate. The results showed that for each of the four factors, the Fornell-Larcker criterion exceeded the cross-correlations of the factor (see Table 2), and therefore, the discriminant validity of the four-factor model was judged to be adequate.

Table 2. Factor correlations of the final CFA models in sexual minority men (N = 131) and women (N = 315) (Study 2).

	Competition	Sex	Status	Exclusion
Sample of Men
Competition	[.76]/[.87]
Sex	.59**	[.72]/[.85]
Status	.56**	.32**	[.86]/[.93]
Exclusion	.58**	.42**	.40**	[.92]/[.96]
Sample of women
Competition	[.71]/[.84]
Sex	.48**	[.67]/[.82]
Status	.52**	.34**	[.83]/[.91]
Exclusion	.42**	.26**	.24**	[.89]/[.94]

Factor correlations are expressed in Pearson’s r. On both diagonals, the Average Variance Extracted (AVE) is presented followed by the Fornell-Larcker criterion (i.e., the square root of the AVE).

**p < .01.

Finally, we conducted a second-order CFA to evaluate whether the overall general factor of Gay Community Stress predicts the four latent factors Sex, Status, Competition, and Exclusion (see Figure S3 for the CFA structure). This model fitted the data well (χ² (99) = 132.59, p = .014; χ²/df = 1.34; RMSEA = .051 with 90% CI [.024; .072] and p = .457; TLI = .967; CFI = .973; and SRMR = .056), suggesting that it is appropriate to examine either the subscales or the overall scale to assess (aspects of) GCS in sexual minority men. The four subscales as well as the total scale were internally consistent with Cronbach’s alfa’s of .88 (Sex; items 1–6), .89 (Status; items 7–9), .87 (Competition; items 12, 13, and 15–17), .91 (Exclusion; items 18–19), and .92 (Total scale; items 1–9, 12, 13, and 15–19). McDonald’s Omega was .87, .90, .88, .90, and .92, respectively. See Appendix 1 for the final Dutch-translated GCSS for gay/bisexual men.

Sexual minority women

See Table S4 for the mean, standard deviation, kurtosis, and skewness for each item of the Dutch-translated GCSS. The initial CFA model was significant, χ² (146) = 392.95, p < .001, and fit indices did not indicate good model fit: χ²/df = 2.69; RMSEA = .073 with 90% CI [.065; .082] and p < .001; TLI = .892; CFI = .908; and SRMR = .068. In identifying elements of model strain, we followed the same iterative procedure as we followed for sexual minority men. Item 2 showed clearly the lowest R² value of all items (.23) and a factor loading (.48) not reaching the suggested threshold of .60 (Awang, 2012), and therefore we removed this item. In the following steps of the iterative procedure we consequently removed items 11, 20, and 14, as we observed similar problems with these items as we did in the sample of men (see previous section). Next, we found a relatively low R² value for item 6 (.39) that also showed a rather strong cross-loading on the Exclusion factor with a high modification index and relatively high residual covariances with items 1 and 3. Conceptually, we know that sexual risk taking behavior is far more prevalent under men than under women (e.g., Skakoon-Sparling & Cramer, 2021), and therefore we decided to remove this item.

A further strain in the model was now identified for item 7 that showed substantial residual covariances with items 8 and 9 (within the same factor Status), but also with item 12 of the Competition factor. We reasoned that “having a high status job” overlaps with the content of item 8 on being wealthy and item 9 on being powerful and having a high status, we decided to remove this item. Next, the results showed that item 17 on materialism cross-loaded substantially on the Status factor (with a modification index of 12.51). Materialism can be regarded as the desire for material wealth, and as such, item 17 shows conceptual similarity with item 8 on valuing being wealthy. Therefore we decided to remove item 17. Finally, and similar to sexual minority men, we observed substantial residual covariances between items 1 and 3, with both items showing good factor loadings (>.70) and R² values (>.40). The items show semantic similarity but conceptually refer to distinct aspects of the Sex factor, and therefore we retained both items and added one within-factor correlation to the model. In the end, the CFA model was trend significant χ² (47) = 61.92, p = .071 with fit indices showing good model fit (χ²/df = 1.32; RMSEA = .032 with 90% CI [.000; .052] and p = .932; TLI = .986; CFI = .990; and SRMR = .032), indicating that the four-factor solution fitted the data well (see CFA structure in Figure 2). The AVE values all exceeded .50 and the Fornell-Larcker criteria were larger than the relevant cross-correlations (see Table 2), and thus the convergent and discriminant validity of the four-factor model were judged to be adequate.

Figure 2. First-order confirmatory factor analysis structure of the Dutch-translated gay community stress scale in sexual minority women.

Numbers in rectangles represent item numbers (see Appendix 1 for all items).

Fc1 = Sex factor; Fc2 = Status factor; Fc3 = Competition factor; Fc4 = Exclusion factor.

1.00 = pathway constrained to one. Standardized coefficients are depicted.

The second-order CFA also revealed a trend significant model, χ² (49) = 62.429, p = .094, with fit indices reaching thresholds of good model fit (χ²/df = 1.27; RMSEA = .029 with 90% CI [.000; .050] and p = .955; TLI = .988; CFI = .991; and SRMR = .032) indicating that the subscales and the overall scale can be used to assess (aspects of) GCS in sexual minority women (see Figure S4 for the CFA structure). The subscales and the total scale were internally consistent with Cronbach’s alfa’s of .78 (Sex; items 1, 3–5), .88 (Status; items 8–9), .80 (Competition; items 12, 13, 15, and 16), .88 (Exclusion; items 18–19), and .86 (Total scale; items 1, 3–5, 8, 9, 12, 13, 15, 16, 18, and 19). McDonald’s Omega was .78, .80, .81, .88, and .85, respectively. Appendix 1 presents the final Dutch-translated GCSS for sexual minority women.

Discriminant and concurrent validity (Study 3)

The aim of Study 3 was to examine the discriminant and concurrent validity of the Dutch-translated GCSS by calculating correlations with several other scales. Prior to these calculations, we combined the EFA (Study 1) and CFA (Study 2) samples separately for men and women.

Materials and methods

Participants

To further examine the discriminant as well as the concurrent validity of the Dutch-translated GCSS in men and women, we combined the EFA and CFA samples for men and women separately. Because not all participants in Study 1 and Study 2 completed the full Qualtrics survey, there were missing values on one or more items of the five measures used in this study. We excluded participants that did not fully complete any of these five measures, implying that the total sample sizes for Study 3 were N = 203 sexual minority men and N = 342 sexual minority women. Table S1 shows characteristics of the combined EFA and CFA samples, separately presented for men and women.

Measures

Gay community stress

To assess GCS in the combined EFA and CFA samples, the Dutch-translated version of the GCSS was used following the results of Study 2 (see Appendix 1 for the GCSS for men and the GCSS for women).

General stress

Respondents filled out a Dutch-translated version of the Perceived Stress Scale (PSS; S. Cohen et al., 1983) that measures general life stress during the month prior to study participation. This self-report scale comprises 14 items and measures the extent to which participants experienced stress in a variety of situations during the past month. The items are rated on a 5-point Likert-scale ranging from never (1) to very often (5) (Cronbach’s alpha in the sample of men = .92 [ω = .91] and in the sample of women .91 [ω = .91]). Examples of items are: “In the last month, how often have you been upset because of something that happened unexpectedly?” and “In the last month, how often have you felt nervous and ‘stressed’?.” A total score was calculated by first reversing the scores on each of the positively formulated items after which all scores were summed. Total scores can range from 14 to 70 and a higher score is indicative of higher levels of stress. The items of the original scale were translated into Dutch by the first author after which the translated items were checked by the second author. Disagreements on any of the translated items were discussed between the first and second authors until full consensus was reached on the final translation. Finally, all translated items were checked independently by eight research assistants (third-year Bachelor’s students), which did not lead to further adjustment of the items. The same translation procedure was applied to the three measures below.

Psychological distress

A Dutch-translated version of the Kessler-6 Psychological Distress Scale (Kessler-6; Kessler et al., 2002) was used to assess psychological distress. Items asked how often in the past 30 days participants had felt “nervous,” “hopeless,” “restless or fidgety,” “so depressed that nothing could cheer you up,” “that everything was an effort,” and “worthless.” Participants answered on a 5-point Likert scale ranging from (1) “none of the time” to (5) “all of the time.” Individual item scores were added into a total scale score with higher values indicating more psychological distress. Cronbach’s alpha in men was .87 [ω = .87] and in women .91 [ω = .91].

Social anxiety

The Social Interaction Anxiety Scale (SIAS: Mattick & Clarke, 1998) was translated into Dutch and used to assess social anxiety during the week prior to study participation. This self-report scale comprises 20 statements describing how participants may feel in a variety of social situations (e.g., “I have difficulty making eye-contact with others,” and “When mixing socially I am uncomfortable”). For each statement respondents were asked to “indicate the degree to which you feel the statement is characteristic or true of you.” Response options are on a 5-point Likert scale ranging from (0) “Not at all” to (4) “Extremely.” Individual statement scores were added up to calculate a total score with higher values indicating more social anxiety. Cronbach’s alpha in men was .94 [ω = .94] and in women .94 [ω = .94].

Interpersonal sensitivity

A Dutch-translated version of the Interpersonal Sensitivity Measure was used to assess over-sensitivity to the interpersonal behavior of others, social feedback, and negative evaluations of others (IPSM; Boyce & Parker, 1989). The scale has 36 items that are rated on a 4-point Likert scale ranging from (1) “very unlike me” to (4) “very like me,” and comprises the subscales interpersonal awareness, need for approval, separation anxiety, timidity, and fragile inner self. Example items are “I worry about the effect I have on other people” and “My value as a person depends enormously on what others think of me.” Individual items scores were added up to calculate a total score with higher values indicating more interpersonal sensitivity. Cronbach’s alpha in men was .90 [ω = .91] and in women .91 [ω = .90].

Statistical analyses

To assess discriminant validity, Pearson correlations between all scales of the Dutch-translated GCSS and the PSS were calculated. Negligible to small correlations were expected, as GCS is uniquely experienced by sexual minority individuals and comprises stressors that differ in nature from general life stressors. To assess concurrent validity, correlations between the GCSS and Kessler-6, SIAS, and IPSM were calculated. Small-to-moderate correlations were expected, as psychological distress, social anxiety, and interpersonal sensitivity are stress-sensitive mental health symptoms that may be induced by the experience of GCS. The correlations were calculated separately for sexual minority men and women, tested two-sided, and interpreted using J. Cohen’s (1992) criteria: r < .10 represent negligible correlations; 0.10 ≤ r < .30 represent small correlations; .30 ≤ r < .50 represent moderate correlations; and r ≥ .50 suggests large correlations.

Results

The correlations are found in Table 3 with significant correlations in bold. As for discriminant validity, three of the four GCSS subscales are not associated with the PSS in neither men nor women, whereas the Competition subscale and (thus) the GCSS total scale show small positive correlations with the PSS in both men and women. These results provide moderate evidence for discriminant validity of the GCSS in sexual minority men and women.

Table 3. Discriminant and concurrent validity of the Dutch-translated gay community stress scale (GCSS) in Dutch samples of sexual minority men (N = 203) and sexual minority women (N = 342).

GCS (sub)scale^a	General Stress	Psy. Distress	Social Anxiety	Int. Sensitivity
Sex (6/4 items)	.028/.043	.079/.050	.110/.073	.091/.128*
Status (3/2 items)	.095/.061	.051/.088	.105/.099^+	.145*/.091^+
Competition (5/4 items)	.265**/.263***	.244***/.281**	.281***/.307**	.301***/.257**
Exclusion (2/2 items)	.134+/.095^+	.162*/.100^+	.106/.174**	.159*/.091^+
Total scale (16/12 items)	.166*/.167**	.175*/.184**	.206**/.224**	.226***/.205**

Pearson correlations were estimated to evaluate the association between the (sub)scales of the Dutch-translated GCSS and the Perceived Stress Scale (measuring general stress), the Kessler-6 Psychological Distress Scale (measuring psychological distress), the Social Interaction Anxiety Scale (measuring social anxiety), and the Interpersonal Sensitivity Measure (measuring interpersonal over-sensitivity). Psy = Psychological; Int = Interpersonal.

Correlations highlighted in bold are significant at the 95% significance level.

^aThe numbers refer to the number of items comprising a GCS (sub)scale for men/women.

⁺p < .010; *p < .05; **p < .01; ***p < .001.

As for concurrent validity of the GCSS in men, the GCSS subscales except the Sex subscale are positively and significantly associated with at least one of the Kessler-6, SIAS, or IPSM measures (mostly small correlations; see Table 3). The GCSS total scale showed small positive correlations with all three measures. In women, all GCSS subscales except the Status subscale are positively correlated with at least one of the Kessler-6, SIAS, or IPSM measures (mostly small correlations). The GCSS total scale shows small positive correlations with all three measures. Overall, these results provide evidence for concurrent validity of the GCSS in sexual minority men and women, with the strongest evidence for the Competition subscale and the total scale in both men and women.

Discussion

To our knowledge, this study is the first to develop a Dutch-translated version of the Gay Community Stress Scale (GCSS; Pachankis et al., 2020) and to examine its psychometric properties in not only Dutch sexual minority men but also in Dutch sexual minority women. The findings reveal that a four-factor model best fits the data in both men and women, implying that the factorial model adequately reproduced the theoretical underpinnings of the scale as outlined in the pioneering work of Pachankis et al. (2020). As such, four factors of gay community stress (GCS) can be distinguished in sexual minority men and sexual minority women in the Netherlands: Stress stemming from the perception that the minority community is strongly focused on (1) sex, (2) status, and (3) social competition, as well as (4) stress stemming from social exclusion of diversity. The results also revealed that a general GCS factor predicts the four specific factors of GCS, in both men and women. We also found moderate support for discriminant and concurrent validity of the male and female versions of the translated GCSS in both men and women. Finally, we saw satisfactory internal consistencies of the Dutch-translated GCS total scale and subscales in men (with α ranging from .87 to .92) and women (with α ranging from .78 to .88).

Similarities and differences in GCS across sexual minority men and women

Our study shows that both Dutch sexual minority men and women may experience GCS, and that GCS with its four factors Sex, Status, Competition, and Exclusion, can be validly and reliably measured with the Dutch-translated GCSS. In men, 16 of the original 20 GCSS items can be used to assess overall GCS and its four subtypes, whereas 12 of the original 20 GCSS items can be used to assess GCS in women (see Appendix 1). The difference in scale length suggests that intraminority stress processes differ between sexual minority men and women. This may not be surprising as the 20 items of the original GCSS (Pachankis et al., 2020) were derived from in-depth interviews with gay and bisexual men only. Some of these original items may reflect experiences of intraminority stress that are more prevalent in men than in women or only prevalent in men.

For instance, the item measuring the perception of “a lot of risky sex” in the community may not be relevant for measuring intraminority stress in sexual minority women because sexual risk taking is less prevalent in women than in men (Skakoon-Sparling & Cramer, 2021). The item measuring the perception of the community being “overly materialistic” may not (equally) relevant or applicable to female sexual minority communities, as studies found that men score generally higher than women on materialism and conspicuous consumption (Kamineni, 2005; Segal & Podoshen, 2013) indicating that possessions seem to matter more in communities of men. These findings support the notion that women value their social roles in society more than men (e.g., Gould, 1987; Vleeming & Engelse, 1981). However, these differences between men and women were not found in recent research (Keech et al., 2020), suggesting a decline in differences over time.

There may also be aspects of intraminority stress that are unique to sexual minority communities of women and are not assessed in the current female GCSS. Qualitative research has revealed that sexual minority women may have negative experiences within their minority community based on their appearance (Dahlenberg et al., 2021). Sexual minority women that chose to dress in a more feminine manner may be disapproved of by the community because of adopting heteronormative norms, whereas women choosing to dress in a more masculine way or “butch” style may not be taken seriously and also be rejected. In female sexual minority communities there may be pressure to reject heteronormative standards to a certain degree, which may induce stress in women seeking connection to these communities.

A potential stressor that might be relevant to assess in sexual minority women but is not captured by the current (Dutch-translated) GCSS is stress resulting from attitudes of female sexual minority communities toward motherhood, parenting, and boy children. Qualitative research indicated that bisexual mothers with a male partner may experience social exclusion from lesbian communities (Tasker & Delvoye, 2015), and that communities of women may hold anti-male sentiments toward boy children (Bradford et al., 2013). Earlier research in the nineties has revealed that non-parenting lesbians may see parenthood as an artifact from a previous heterosexual life (Slater, 1995), and that lesbian communities may question the authenticity of the sexual orientation of sexual minority mothers because they have children (Siegenthaler & Bigner, 2000). As a result, sexual minority mothers may feel rejected by their minority community. On the other hand, more recent research reported that sexual minority mothers spoke favorably about participating in the LGBTQ-parent community (Gartrell et al., 2019), implying that attitudes of female sexual minority communities toward motherhood have become more positive and that negative attitudes may only be present in subcommunities of sexual minority women. To what extent these negative attitudes are present in Dutch minority communities and induce stress in Dutch sexual minority women is unknown and important questions for future research.

Comparing the original GCSS with the Dutch-translated GCSS

The difference in scale length between the Dutch-translated male GCSS and the original GCSS (Pachankis et al., 2020) might also indicate cross-national differences in intraminority stress processes. For instance, social exclusion or rejection of men with HIV/AIDS may be less pronounced in Dutch sexual minority communities than in US sexual minority communities. Compared to the US, there are less stigmatizing attitudes toward people living with HIV/AIDS in the general population in the Netherlands (Du et al., 2018), which in turn may affect the strength of stigmatizing attitudes in male sexual minority communities. Dutch sexual minority men may perceive or experience lower levels of discrimination against men with HIV/AIDS in their communities than their US counterparts, so that measuring discrimination against men with HIV/AIDS may therefore not be appropriate when assessing GCS in Dutch sexual minority men. On the other hand, the Dutch-translated GCSS for men does comprise all eight items of the Abbreviated Intraminority Gay Community Stress Scale developed and evaluated by Maiolatesi et al. (2023) in US samples of gay and bisexual men. This overlap in items indicates that GCS is similarly experienced by gay and bisexual men in different countries, providing cross-national support for the factor stability of the GCSS to a substantial extent.

Another reason for the discrepancy in scale lengths may be grounded in language differences. The original GCSS comprises an item with the English word “cliquey,” but there is no single and unambiguous Dutch equivalent to this word. For this item, the Dutch-translation may not have captured the essence of what “cliquey” means for gay and bisexual men in the US. A similar issue might have occurred in the Dutch translations of “bickering” and “cattiness” that are used in a different item of the original GCSS, although a more important issue is that three instead of only one concept is measured with this item which is detrimental to the item’s quality. Finally, the original GCSS item on the gay community overly valuing stylish clothes and up-to-date fashion was not found to be a significant item for assessing GCS in men nor women. It seems that Dutch-speaking sexual minority men and women do not have the perception that wearing stylish clothes or keeping up with fashion is set as an expectation or social norm in their communities that induce stress.

Discriminant and concurrent validity of the Dutch-translated GCSS

The correlational analyses revealed moderate support for discriminant validity of the Dutch-translated GCSS. In sexual minority men and women, three of the four GCSS subscales did not correlate with the PSS that measures general stress, suggesting that these subscales assess different aspects of stress than the PSS does. However, the GCSS Competition subscale did share variance with the PSS in men and women although the correlations represented small associations. It may be, as this result suggest, that intraminority stress experienced by the sexual minority community’s focus on competition may make sexual minority men and women feel “angered,” “upset,” and “stressed” which are explicitly measured by the PSS.

There was moderate and encouraging support for concurrent validity of the Dutch-translated GCSS. The Competition subscale and the GCSS total scale correlated as expected with three stress-sensitive mental health symptoms (i.e., psychological distress, social anxiety, and interpersonal over-sensitivity) in both sexual minority men and sexual minority women. This indicates that GCS and particularly stress from competition in sexual minority communities of men and women may induce stress-sensitive mental health problems. More modest support for concurrent validity was found for the other three GCSS subscales that correlated with only one mental health symptom in either sexual minority men or sexual minority women (i.e., the Sex and Status subscales), or with all three examined mental health symptoms but either in men or women (i.e., the Exclusion subscale). At first, this might indicate that the sex, status, and exclusion aspects of GCS play a less prominent role in mental health symptoms, but this may be premature to conclude. Future research on concurrent validity examining other stress-sensitive mental health symptoms, such as hostility/aggression, risky sexual behavior, drug/alcohol (ab)use, loneliness, and feelings of disconnection, is warranted.

Study limitations

Some study limitations should be acknowledged. The test-retest reliability as well as the predictive validity of the scale could not be assessed due to the cross-sectional design of the current studies. Further, voluntary response bias might have affected the results, as particularly participants interested in stress and/or mental health may have been inclined to participate. Another limitation refers to how the instructions for participants were presented in the Qualtrics survey. All participating men and women received the same instructions for filling out the Dutch-translated GCSS and were asked about their perceptions of the “lesbian, gay, or bisexual community that they feel most connected to.” It might be that, for instance, a participating lesbian woman feels more socially connected to the gay male community than to the lesbian female community, or vice versa. This might have affected the results of the EFA and CFA, and the validity of the male and female versions of the Dutch-translated GCSS. We therefore recommend for scale implementation and future research that the instructions for participants are specifically tailored to sexual minority men or sexual minority women.

Also, we examined the convergent validity of the Dutch-translated GCSS to only a limited extent. However, no other measures are yet available for assessing intraminority stress in sexual minority communities and we were therefore unable to further examine the convergent validity as we did in assessing the discriminant validity. Finally, it has been recommended that diagonally weighted least squares (DWLS) estimation or its robust variant weighted least square mean and variance adjusted estimation (WLSMV) is used when modeling categorical or ordered data in CFA (e.g., Brown, 2015; Li, 2016). However, in the procedure for identifying elements of strain in our CFA models, using DWLS or WLSMV resulted in a non-converging model, warnings stating that the number of observations is too small to compute the Gamma statistic, or model evaluation indices suggesting perfect fit. Therefore, we decided to use robust maximum likelihood (RML) estimation—and not maximum likelihood (ML) estimation as applied by Pachankis et al. (2020) -, as RML has been shown to be superior to ML when modeling ordinal observed variables (Li, 2016; Satorra & Bentler, 1994; Yuan & Bentler, 1998). Compared to (robust) DWLS, robust ML requires samples of smaller size to produce stable structural coefficients and standard error estimate, and produces generally more accurate coefficients and less biased standard error estimates when the data are symmetric than (robust) DWLS (Li, 2016). Also, recent simulation research has revealed that DWLS estimation may inflate the RMSEA, CFI, and TLI indices when analyzing ordered categorical data (Xia & Yang, 2018).

Directions for future research

An important direction for future research is to further validate the Dutch-translated GCSS in sexual minority women, as the original GCSS was developed in gay and bisexual men and processes of intraminority stress are likely to differ between sexual minority men and women. Further, there may be differences in perceptions and experiences of GCS across age cohorts, as the importance of the gay community to sexual minority individuals may change over time. For instance, young people just starting to identify as a sexual minority may seek connection to the gay community to build a social network of like-minded friends, and may be more sensitive to GCS than older sexual minority people who already have a secure like-minded social network. On a similar note, sexual minority people with a satisfactory and enduring romantic relationship may experience less GCS, as they are not actively looking for a romantic partner and may therefore experience less social pressure to conform to the community’s social norms and rules.

Further, competition, which is an aspect of GCS, is positively correlated with education levels (Buser et al., 2014, 2017, 2018), and the participants in our samples were relatively highly educated. This raises the question whether GCS and specifically stress from competition is experienced differently in sexual minority men and women with lower levels of education. Also, participants with a gender identity other than male and female, and participants with a sexual orientation other than gay, lesbian, and bisexual were underrepresented in our samples. It cannot be merely assumed that intraminority stress processes are the same across (sub)communities of people with different identities and orientations. We thus recommend that future research examines GCS across groups of minority individuals that differ in age, relationship status, educational background, gender identity, and sexual orientation. Finally, this study confirms the notion of Pachankis et al. (2020) that intraminority stress is an identifiable type of stress that may play a key role in mental health problems of sexual minority men and women. How GCS contributes to sexual orientation disparities in the mental health of sexual minority men and women, and how intraminority stress theory complements minority stress theory are important endeavors for future research.

Conclusion

Until today, mental health problems of sexual minority individuals are primarily explained by minority stress theory (Meyer, 2003) that focuses on the stress coming from the society’s prejudice and stigma directed at sexual minorities. Pachankis et al. (2020) showed in their pioneering work that intraminority community stress can also be an important risk factor for mental health problems of gay and bisexual men, and developed the Gay Community Stress Scale (GCSS) to assess this unique type of intraminority stress. This study developed a Dutch-translated version of the GCSS and validated this scale in not only sexual minority men, but also in sexual minority women as the latter may also experience intraminority stress resulting from their community’s social rules and norms. The results following exploratory and confirmatory factor analysis revealed that a 16-item Dutch-translated GCSS for men and a 12-item Dutch-translated GCSS for women reproduced the factor structure of the original 20-item GCSS of Pachankis et al. with good psychometric properties. The results suggest that the 16-item and 12-item GCSS are sufficiently valid and reliable self-report instruments for assessing intraminority stress in Dutch-speaking sexual minority men and women, respectively. As such, our expectation that intraminority stress can be present in both sexual minority men and women was confirmed. Future research and clinical practice might use the Dutch-translated GCSS to shed more light on the role of intraminority stress in mental health problems of sexual minority men and women in the Netherlands, and to advance interventions addressing these problems.

Acknowledgments

We thank Sjams Barra, Femke Begeman, Mirthe Bommer, Emma Bos, Dilara Bükülmez, Megan Burgstra, Sterre Doest, Kay Holvast, Sam Jansen, Dirkje Schoon, Charlotte Tijburg, Anne Tummers, Isabeau van Velsen, and Leonie Wijnen for their assistance in collecting the data.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/00918369.2023.2231119.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.