Introduction

In most societies throughout the world, men as a group enjoy social and institutional privileges over and above women and have greater decision-making power within heterosexual relationships (Connell, 1987; Fleming, DiClemente, & Barrington, 2016b; Gilmore, 1990; Messner, 1997; Wingood & DiClemente, 2000). In order to be perceived as masculine and thus achieve the higher social status and power afforded to “real” men, men are pressured to and rewarded for adopting certain characteristics such as aggression, virility, and risk-taking, which influence their sexual behaviors (Connell, 1995; Courtenay, 2000; Williams, 2003).

Though many researchers are interested in how masculinities influence men’s sexual behaviors (Bowleg, 2004; Dworkin, Fullilove, & Peacock, 2009; Fleming et al., 2016b), there is little consensus on the best way to conceptualize and measure this in survey research. Approaches include trait measures (e.g., Bem Sex Role Inventory [Bem, 1974]); norms/ideology (e.g., the Gender Equitable Men [GEM] Scale [Pulerwitz & Barker, 2008]); gender role conflict or stress (e.g., Gender Role Conflict Scale [O’Neil, Helms, & Gable, 1986]); and gendered behavior (e.g., Gender Diagnosticity [Lippa & Connelly, 1990]) (for a review, see Smiler & Epstein [2010]). Survey research on masculinity and men’s sexual behaviors has been limited almost exclusively to normative/ideology measures such as the GEM Scale (Pulerwitz & Barker, 2008), the Male Role Norms Scale (Thompson & Pleck, 1986), the Male Role Norms Inventory (Levant, Hirsch, Celentano, & Cozza, 1992), and the Hypermasculinity Scale (Archer, 2010). These measures all assess an individual’s ideology about the appropriate roles and behaviors for men and women. The available evidence from cross-sectional studies demonstrates an association between having a more traditional/inequitable gender ideology and less frequent condom use and having a greater number of sexual partners (Fleming et al., 2016b).

While the norms/ideology approaches are useful for exploring the relationship between masculine norms and sexual behaviors, there are limitations. Norms/ideology measures assume that men are motivated to adhere to the attitudes they express (Thompson, Pleck, & Ferrera, 1992). But, these attitudes do not always translate into behaviors. Prominent models of human behavior (e.g., theory of planned behavior) highlight that attitudes do matter, but that the degree of social pressure one feels to conform to certain behavioral norms also are crucial (Montano & Kasprzyk, 2008). For example, a man may express support for the idea that men should have multiple concurrent sexual partners, but he does not feel the need to have multiple partners. In this case, the man may not have sufficient motivation to comply because he is not concerned about any possible social sanctions. The opposite example is also possible where a man thinks men should not have multiple partners, but he himself has multiple partners because of feeling pressure to comply with social norms among his referent group. This possible disconnect between attitude and behavior renders normative/ideology measures potentially insufficient at capturing the internalization of masculine gender norms (O’Neil, 2008). The individual attitude may be less important than the social norms and how an individual perceives that his social network might react. Thus, it is important to utilize a measure that can capture these aspects.

In contrast to norms/ideology measures, Gender Role Conflict or Stress Scales measure the degree to which an individual is concerned about his ability to adhere to gender norms or how he feels when acting contrary to prevailing male gender norms. On such scales, men are asked about their level of agreement with first-person statements worded to reflect conflict/stress around achieving masculine expectations or, alternately, how stressful they would find particular situations that violate traditional masculine norms. The most commonly used measures are the Gender Role Conflict Scale (O’Neil et al., 1986) and the Masculine Gender Role Stress Scale (Eisler & Skidmore, 1987). These scales draw from the theoretical concept of masculine gender role strain, which theorizes about men’s concern with achieving masculine norms, including norms of sexual prowess and sexual performance (Pleck, 1981, 1995). Pleck postulated that cultural standards for masculinity exist, socialization encourages men to attempt to live up to these norms, and pressure to conform to these norms can result in negative psychological and health outcomes for men. O’Neil (2008) described that the Gender Role Conflict Scale (GRCS) is “defined as concrete outcomes of gender role strain that can be understood and measured” (p. 364). Thus, the GRCS operationalizes gender role strain for survey research by asking men how concerned they personally are about demonstrating specific characteristics of masculinity. Since one of the hypothesized mechanisms through which norms of masculinity influence men’s sexual behaviors is related to men feeling concerned about demonstrating masculine characteristics (Courtenay, 2000), Gender Role Conflict or Stress Scales can add an important dimension to assessing the relationship between masculinity and men’s sexual risk behaviors.

While greater gender role conflict has been associated with men’s perpetration of violence (Copenhaver, Lash, & Eisler, 2000; Jakupcak, Lisak, & Roemer, 2002) and a range of other adverse mental and physical health outcomes (O’Neil, 2008, 2015), it has rarely been examined in relation to men’s sexual behaviors. To our knowledge, only two previous studies have explored the correlation between the concept of gender role conflict/stress and sexual behaviors among heterosexual men (Gottert et al., 2016; Reidy, Brookmeyer, Gentile, Berke, & Zeichner, 2016). Gottert et al. found that, among men in rural South Africa, greater gender role conflict/stress was associated with an increased odd of having multiple concurrent sexual partners. Reidy et al. created a measure of “gender role discrepancy stress” which was associated with age at first intercourse, unprotected sex, and number of partners among a convenience sample of American men interviewed on the Internet. The only other study to use gender role conflict/stress to study men’s sexual behaviors was conducted with men who have sex with men in the U.S. Malebranche, Gvetadze, Millett, and Sutton (2012) found that greater gender role conflict was associated with unprotected vaginal or anal sex with women among the behaviorally bisexual men (gender role conflict was not a significant predictor of unprotected anal sex with men). While these previous studies provide important preliminary evidence, there is a need to expand the evidence base for different cultural contexts and for men whose sexual behaviors put them at high risk of HIV and sexually transmitted infections (STI).

In this study, we examined the association between gender role conflict/stress and three different sexual risk behaviors among a sample of men in the Dominican Republic (DR). We hypothesized that higher gender role conflict/stress would be associated with increased odds of engaging in sexual risk behaviors. Since the concern a man feels about demonstrating masculine characteristics is a modifiable factor (Dworkin, Treves-Kagan, & Lippman, 2013), examining this relationship has the potential to improve upon sexual health promotion strategies for men.

Study Setting

We conducted this research as part of a feasibility study of male circumcision for HIV prevention in the DR (Brito, Caso, Balbuena, & Bailey, 2009; Brito et al., 2015; Brito, Luna, & Bailey, 2010). The parent study afforded an opportunity to examine masculine gender role conflict/stress and sexual behaviors among a relatively large sample of heterosexual men who engaged in sexual behaviors that put them at risk of HIV/STI. The study was conducted in two cities on the southeastern coast of the DR–Santo Domingo and La Romana–that both have higher HIV prevalence than the national prevalence (DIGECITSS, 2014).

De Moya (2004)—a prominent Dominican scholar focused on masculine norms in the DR—wrote about how Dominican men/boys are socialized into behaving in certain ways that are deemed culturally masculine in Dominican society. He described masculinity in the DR as a “totalitarian” regime that controls the lives of Dominican boys and young men. De Moya used participant observation and interviews with men and women to identify the “rules” associated with being a “normal” boy in the DR, including “He has to fight if he is insulted…He should not sob nor cry…He should show a vivid and visible erotic interest in all females who come close to him when he is with his peers” (de Moya, 2004, pp. 73–74). These masculine ideals are instilled in young Dominican boys and enforced by others through punishment and shaming during youth and adulthood. Thus, these rules not only stifle men, but may also cause a considerable amount of stress as they attempt to meet the rigorous standards of manhood. For more on Dominican masculinities, see de Moya (2003), Barrington (2007), Padilla (2008), Fleming, Barrington, Perez, Donastorg, and Kerrigan (2014), and Fleming et al. (2016a).

Previous research on men’s sexual risk behaviors in the DR has highlighted that their behaviors are shaped, in part, by their male peers (Barrington & Kerrigan, 2014; Barrington et al., 2009; Fleming et al., 2014). For example, Barrington et al. found that going out to sex establishments is a highly social activity and that men’s condom use with female sex workers was strongly associated with perceptions of their peers’ condom use. Fleming et al. found that male peer groups encouraged condom use, helped each other find sexual partners, and seeking sex workers was a key bonding activity for the group. Taken together, these findings provide some qualitative evidence that concern about adhering to masculine norms influences Dominican men’s sexual behaviors.

Method

Participants

The parent study used referrals and community outreach to find men who were 18–40 years old and were willing to undergo a circumcision. To reach a sample at heightened risk of HIV, female sex workers in both sites were asked to refer their sexual partners, and in La Romana, one recruiter was dedicated to recruiting men from nearby bateyes (Haitian descendent communities with a high HIV prevalence). A total of 454 men were circumcised between January 2013 and March 2014, and final follow-up occurred between July 2013 and February 2015. We conducted one survey at baseline (prior to being circumcised) and one during the men’s routine visit 6–12 months after their circumcision. For the analyses presented in this article, we used the data from the follow-up survey. Of the 454 men enrolled, 92 men were lost to follow-up and 69 were not asked about gender role conflict/stress because the scale was added after follow-up visits were initiated. As a result, our analytic sample had 293 men. Men were reimbursed for their travel to the clinic for the visit (approximately 10 USD).

Measures

Sexual Risk Behaviors

For our dependent variables, we used three sexual behaviors that have been shown to be associated with HIV/STI transmission: (1) two or more sexual partners in the last 30 days; (2) inconsistent condom use with non-steady partners in the past 6 months; and (3) drank alcohol at last sex (Anderson, 2003; Barrington et al., 2009; Kalichman, Simbayi, Kaufman, Cain, & Jooste, 2007; Mah & Halperin, 2010). For number of partners, men were asked “How many women have you had sex with in the last 30 days?” and we dichotomized their response into “two or more” and “0 or 1.” For condom use, men were asked “How frequently do you use a condom with each of your non-steady partners in the past 6 months?” and their response options were never, once, less than half the time, about half the time, more than half the time, and always. We created a dichotomous variable where “inconsistent condom use” was any response other than always. Finally, for drinking alcohol at last sex, men were asked, “The last time you had sex, were you drinking alcohol?” with the response options of yes or no.

Gender role conflict/stress

To measure gender role conflict/stress, we used a 19-item scale adapted from O’Neil et al.’s (1986) GRCS and Eisler and Skidmore’s (1987) Masculine Gender Role Stress (MGRS) Scale. These original scales were developed with US university students. Gottert et al. (2016) merged and adapted these scales for research in a rural area of South Africa; the GRCS served as the basis for the scale format and wording as well as two domains and multiple items, and the MGRS served to add new domains deemed relevant to the South African context such as “Subordination to women.” With the data from South Africa, the final 24-item Gender Role Conflict/Stress (GRC/S) Scale had a Cronbach’s α of 0.83 (Gottert et al., 2016).

We worked with a team of local Dominican researchers and two local men to adapt the GRC/S items for the Dominican context. Specifically, we removed two items deemed to be irrelevant to the Dominican context and translated items into the colloquial Spanish used by Dominican men. Because of space constraints in our survey, we also relied on recommendations from local field staff to cut several items that had the least relevance for the local context. We also created and added three additional items related to sexual prowess because they were relevant to research questions of this study. This resulted in a scale with 19 items.

Upon conducting an exploratory factor analysis (using Stata version 13.1), we found that a single-factor structure was most suitable to our data. Solutions with two or more factors proved to have low Cronbach’s alphas for sub-scales and items loaded on factors in ways that did not entirely fit with our theoretical understanding. As a result, our final unidimensional scale had 17 items, a Cronbach’s alpha of 0.75, and an overall Kaiser–Meyer–Olkin measure of sampling adequacy of 0.79.

Since there were 17 items and response options were 0 = disagree, 1 = partially agree, and 2 = strongly agree, possible scores for the GRC/S Scale ranged between 0 (lowest conflict/stress) and 34 (highest conflict/stress). We standardized the GRC/S score with a mean of 0 and SD of 1 to aid in the interpretation of results.

Control Variables

Sociodemographic control variables were assessed at baseline: age, study site, education, employment status, marital status, and monthly income. Since inconsistent condom use with casual partners is associated with number of other sexual partners (Matser et al., 2014), we also controlled for number of partners in the past 6 months when conducting analyses with our inconsistent condom use dependent variable.

Procedure

All participants provided informed written consent to participate in each component of this research study. All study procedures and protocols were approved by the Institutional Review Boards at the University of Illinois at Chicago, the University of North Carolina at Chapel Hill, and the Instituto Dominicano de Dermatología y Cirugía de la Piel Dr. Heriberto Bogaert in Santo Domingo, DR.

Statistical Analyses

We present descriptive statistics for each of the sociodemographic variables, GRC/S Scale, and dependent variables. For each of the three dependent variables, we conducted bivariate logistic regression with the standardized GRC/S score as the independent variable. We report unadjusted odds ratios (ORs) and 95 % confidence intervals for those bivariate analyses. Subsequently, we conducted multivariate logistic regression for each dependent variable with standardized GRC/S score and the full set of controls. For those analyses, we report adjusted odds ratios (AORs) and 95 % confidence intervals (CIs). All analyses reported below were conducted in SAS version 9.4.

Results

Demographic characteristics are shown in Table 1. Most men were under the age of 30 (median, 26; range, 18–41), and a majority (69 %) had at least a high school education. Seventy-three percent of men were employed (either formal or informal labor market), 11 % were unemployed, and 16 % were students. Only 8 % earned more than 25,000 Dominican Pesos (DOP) in the past month, about 625 US dollars. Nineteen percent earned less than 1000 DOP (25 USD), and the rest of the men (72 %) earned between 1000 DOP and 25,000 DOP. Half of the men reported being unmarried but had a partner, 14 % were married, and 35 % were single with no regular partner.

Table 1 Sample demographic characteristics and HIV risk behaviors
Full size table

We measured three sexual risk behaviors (Table 1). Thirty-one percent of men reported having two or more partners within the past 30 days. Of men that reported having a non-steady partner within the previous 6 months (n = 194), 43 % reported inconsistent condom use with non-steady partners in the past 6 months. When men reported on the last time they had sex, 21 % said that they were drinking alcohol.

Before standardizing, the GRC/S measure had a mean of 18.2 (range 3–34) and a SD of 5.7 (Table 2). See Table 2 for all scale items.

Table 2 Gender Role Conflict/Stress (GRC/S) Scale (n = 293)
Full size table

When testing the bivariate associations between each sexual risk behavior and men’s GRC/S scores, we found relationships in the expected direction (Table 3). Specifically, men who had a higher GRC/S score had higher odds of reporting each sexual behavior. Two of the sexual behaviors—having two or more partners in the past 30 days, and drinking alcohol at last sex—were significantly associated with having a greater GRC/S. The bivariate relationship between inconsistent use of condoms with non-steady partners and GRC/S was not significant.

Table 3 Association between men’s GRC/S score and sexual risk behaviors
Full size table

After controlling for sociodemographic characteristics (Table 3), higher GRC/S was significantly associated with increased odds of having two or more partners in the past 30 days [AOR 1.33, 95 % confidence interval (CI) 1.01–1.74], inconsistent condom use with non-steady partners (AOR 1.45, 95 % CI 1.04–2.01), and drinking alcohol at last sex (AOR 1.56, 95 % CI 1.13–2.17).

Discussion

Our study provided support for the association between GRC/S and men’s sexual risk behaviors, including number of sexual partners, inconsistent condom use with non-steady partners, and drinking alcohol at last sex.

Echoing Gottert et al. (2016) and Reidy et al. (2016), we found that men with greater concern about demonstrating masculine characteristics were more likely to engage in sexual risk behaviors. Pleck’s (1995) masculine gender role strain paradigm highlights that masculine gender norms can be conflicting and inconsistent, which causes men to worry about their ability to fulfill these norms. Given the work by de Moya (2004) in the DR emphasizing that masculinity is a “totalitarian” regime in the lives of men, men in our study felt concern about meeting these masculine expectations. One interpretation of our findings is that men may be coping with these concerns by adopting sexual risk behaviors (Glanz & Schwartz, 2008). In the DR, men’s sexual behaviors and risk-taking can be seen as a demonstration of masculinity (de Moya, 2003, 2004; Fleming et al., 2016a; Padilla, 2008). Indeed, sexual prowess (e.g., having multiple female sexual partners) and risk-taking (e.g., inconsistent condom use, alcohol use prior to sex) are key characteristics associated with masculinity in many parts of the world (Courtenay, 2000; Fleming et al., 2016b), and therefore, those behaviors represent opportunities for men to demonstrate their masculinity.

Future research on men’s sexual risk behaviors should continue to use gender role conflict/stress measures with other populations of men to verify whether this relationship holds across cultural contexts, varying age groups, and high- versus low-risk men. Given that our sample was limited to men seeking a voluntary medical male circumcision, it will be important to replicate findings in general populations of men. Additionally, longitudinal research is needed to assess whether changes in GRC/S result in changes in sexual risk behaviors. Longitudinal studies could also help to assess how GRC/S influences sexual behaviors by further exploring mechanisms driving this relationship. Finally, although psychometric analyses of our scale indicated a unidimensional factor structure–likely due to including fewer items than O’Neil’s GRCS or Gottert and colleagues’ GRC/S Scale–future research should examine which scale sub-factors most influence men’s sexual behaviors. For example, the Gottert et al. (2016) study in South Africa found that the “Subordination to Women” sub-scale was the key dimension of GRC/S that was most associated with men having multiple concurrent sexual partners. Expanding this type of research could help refine intervention strategies for reducing sexual risk behaviors and improving men’s sexual health.

Given the lack of consensus regarding how to operationalize masculinities for research on men’s sexual risk behaviors, research is also needed to critically compare measures of gender role conflict/stress to gender ideology measures to assess the relative influence of each on men’s sexual behaviors. Gottert et al. (2016) began this work, finding that conflict/stress measures are more strongly associated with men having multiple concurrent partners than a measure of gender ideology, while suggesting that it is advantageous for studies to include both constructs. Other research has just begun to explore other non-ideological measures of masculinity (e.g., gender typicality [Tate, Betergarcia, & Brent, 2015] or adherence to gender-typical behaviors [Fleming, Harris, & Halpern, 2016c]), and findings from this emerging research also need to be assessed relative to existing measures. These initial findings–including those from our study–need to be replicated and evaluated to determine what unique perspective each measure brings to the study of masculinity and men’s sexual behaviors.

Our findings also provide further support that reducing men’s sexual risk behaviors and improving their sexual health requires gender-transformative interventions (Barker, Ricardo, Nascimento, Olukoya, & Santos, 2010; Dunkle & Jewkes, 2007; Gupta, 2000). Gender-transformative interventions are focused on challenging harmful norms of masculinity and democratizing the relations between men and women (Dworkin et al., 2013; Gupta, 2000). Given that masculine norms have been shown to be influential on sexual risk behaviors, gender-transformative interventions offer a specific strategy to tackle this root cause of men’s sexual behaviors that put them at risk of HIV/STI (Dworkin, Fleming, & Colvin, 2015).

Our results also suggest that gender-transformative interventions may need to expand their approach to also directly tackle men’s concern about demonstrating masculine characteristics (i.e., gender role conflict/stress)—which relates more to men’s experience of and emotions around internalized masculine norms. Currently, these interventions have primarily relied on the empirical evidence derived from studies using measures of gender ideology, primarily the Gender Equitable Men Scale (Pulerwitz & Barker, 2008). As a result, the focus has been on changing men’s gender ideologies (Dworkin et al., 2013; Van den Berg et al., 2013) and such interventions are evaluated using measures of gender ideology (Pulerwitz & Barker, 2008; Pulerwitz, Michaelis, Verma, & Weiss, 2010; Verma et al., 2006). But, these current approaches are not addressing the internalization of masculine norms, only attitudes. Considering how to reduce men’s gender role conflict/stress could enable interventionists to develop new and innovative ideas to complement existing gender-transformative approaches for HIV/STI prevention. Since men’s gender role conflict/stress is related to concern about peers’ responses to non-masculine behavior, interventions may need to work within male peer groups to devise strategies to respond to and cope with instances where a man’s sense of masculinity is challenged.

Limitations

While our research presents some of the first empirical evidence on the association between gender role conflict/stress and men’s sexual risk behaviors, these findings should be considered in light of certain limitations. First, we used cross-sectional data and thus were unable to establish temporality between our independent variable and dependent variables. Also, our analyses did not allow us to tease apart exactly which control variables may have suppressed effects seen in the bivariate analyses. Second, we used a sample of men who were willing to undergo a voluntary medical male circumcision in settings with an HIV prevalence higher than the national average. This sample may systematically differ from the general population of men or from the population of men at risk of HIV/STI. Additionally, like most behavioral research on sex and sexuality, this analysis relied on self-reported measures of sexual behaviors. Finally, GRC/S is intended to be a multidimensional scale (Gottert et al., 2016), and thus, our findings with the unidimensional version may be obscuring important differences in the relationship between sexual behaviors and certain sub-types of gender role conflict/stress.

Conclusion

Gender is an important social determinant of health, and the field of sexual health has led the way in developing research and interventions to ameliorate the negative effects of masculinity on men’s health (Bowleg et al., 2011; Dunkle & Jewkes, 2007; Dworkin et al., 2013; Gupta, 2000; Jewkes, Sikweyiya, Morrell, & Dunkle, 2011). Despite the progress to date, our limited ability to empirically assess constructs related to masculinity and the effect they have on men’s health behaviors is stifling additional progress. Incorporating new measures, such as the GRC/S Scale, is one example of how researchers can continue to push the field forward. To improve our sexual health promotion efforts with men, we need to continue developing tools and resources to expand understanding of masculinity’s influence on sexual health.