Introduction

While they are often examined separately, attitudes about gender roles are often intertwined with attitudes about sexuality (Lefkowitz et al. 2014). At the core of this attitudinal nexus is heteronormativity. Heteronormativity sets standards and norms for sex and gender roles in families and work, both in and out of the home, that dictate which intimate pairings are acceptable, privileging other-sexed couples. While heteronormativity has clear consequences for attitudes regarding sexuality, it also has implications for gender role attitudes (Habarth 2008; Kitzinger 2005).

There is substantial research on how gender attitudes influence heterosexuals’ attitudes towards sexual minority individuals (Davies 2004; Whitley and Ægisdóttir 2000; Henry and Wetherell 2017; Kerns and Fine 1994; Kite and Whitley 1996), but comparatively little research has considered the gender attitudes of sexual minority individuals. In short, how do gay or lesbian and bisexual individuals differ from heterosexual individuals in their attitudes about gender roles, and how might these differences be moderated by gender?

We advance the hypothesis that because gay or lesbian and bisexual individuals are more likely to reject heteronormativity, these individuals will also tend to reject so-called traditional attitudes and norms concerning gender that are intertwined with heteronormativity. This is particularly the case for gender norms surrounding marriage and family (i.e., male breadwinner, female homemaker), which largely do not resonate with or apply to same sex couples. However, experiences and attitudes often differ by gender, so the interaction of gender and sexual minority status is important to consider. Since traditional gender norms tend to benefit men and disadvantage women, we hypothesize that gender will moderate the effect of sexual identity, with gay and bisexual men being less likely than lesbian and bisexual women to reject traditional gender attitudes. We argue this will be particularly the case for gender roles in the public sphere, for which male sexual minorities might still benefit from gender inequality.

While these hypotheses are extrapolated from the findings of past research, much of that research has utilized samples that are geographically-limited and\or lack a probability-based sampling design (e.g., Habarth 2008; Warriner et al. 2013). To advance this area of research beyond these limitations, this study utilizes the General Social Survey, a survey of US adults with a probability-based sampling design, to examine how an individual’s sexual identity is associated with his or her gender attitudes. Through cross-tabulations and logistic regression analyses, we find that gender and sexuality both influence gender role attitudes. According to our findings, gay men and lesbian women are both more likely than their heterosexual peers to reject traditional gender role attitudes related to the family and household. However, for public sphere gender role attitudes, such as the suitability of women for political offices, lesbian women are more likely than heterosexual women to reject traditional gender attitudes but gay men do not significantly differ from heterosexual men. Given the increased discussion of gender and sexuality, further examination of these patterns of beliefs and how they affect behaviors need to be conducted.

Literature Review

Relevance of Sexuality to Gender Role Attitudes

Heteronormativity refers to the social, legal, cultural, institutional, and interpersonal practices and beliefs that are produced by and reproduce a set of assumptions about what is acceptable in regards to sex and gender (Kitzinger 2005). Heteronormativity gives privilege to heterosexual individuals, assuming that being attracted to the opposite sex is “normal” and “natural” (Habarth 2008; Herek 2004; Kitzinger 2005; Torkelson 2012). As Habarth (2008: 2) put this, heteronormativity, “affects individuals regardless of sexual orientation, proscribing and requiring different kinds of actions and experiences based on gender, and creating categories of acceptable and unacceptable groups of people.” This becomes clear in cases when individuals violate these requirements and norms. For example, Lupton (2000) found that male nurses were perceived as highly feminine or homosexual due to their non-traditionally masculine occupational choice thus undermining their capability and masculinity.

A number of studies have demonstrated an association between the endorsement of traditional gender role attitudes and hostility or opposition towards gay, lesbian, and other sexual minority individuals (Brown and Henriquez 2008, 2011; Kerns and Fine 1994; Kite and Whitley 1996; Willoughby et al. 2010). For example, Nagoshi et al. (2008) found that the closer individuals adhered to traditional norms about gender roles the more likely they were to score highly on a homophobia scale. Similarly, Brown and Henriquez (2011) found in their survey of heterosexual college students that individuals who held less traditional gender role attitudes expressed greater support for homosexuals’ rights. Kerns and Fine (1994) found that heterosexual individuals who supported traditional gender roles were more likely to have negative attitudes towards gay men and lesbian women. Similarly, Whitley and Ægisdóttir (2000) found that endorsing traditional gender role attitudes had a stronger effect on heterosexuals’ attitudes about gay and lesbian individuals than simply holding a generic conservative belief system. Henry and Wetherell (2017) extended this dynamic in a cross-national analysis, showing that nations with more gender equality have more positive aggregate attitudes towards sexual minority individuals. The link between gender role attitudes and sexuality attitudes has motivated studies examining how heterosexual individuals’ gender attitudes shape their acceptance of and attitudes towards sexual minorities (Davies 2004; Kite and Whitley 1996) but there is a dearth of studies that examine the gender role attitudes of sexual minority individuals.

Sexual minority individuals challenge the heteronormative structure that creates and maintains traditional gender roles simply by being attracted to individuals of the same sex. Interestingly, heteronormative thinking about gender role attitudes is reflected in some of the survey questions often utilized in studies assessing individuals’ attitudes towards gender roles. For instance, a question appearing in the General Social Survey that has been used in many studies asks respondents to indicate how much they agree or disagree that “it is much better for everyone involved if the man is the achiever outside the home and the women takes care of the home and family” (Carter et al. 2009; Liao and Cai 2016; Powers et al. 2003; Smith et al. 2017). Such survey questions not only reflect how traditional gender roles are closely tied to heteronormative assumptions but also how individuals who are not heterosexual are unlikely to see themselves fitting into such traditional gender roles.

Even when a particular survey question might have less explicit heteronormative wording, we expect sexual minority individuals to reject such traditional gender attitudes as the attitudes themselves can be perceived to be part of a larger system of attitudes and norms that exclude individuals who are not heterosexual. Moreover, sexual minority individuals may perceive such attitudes and norms as representing a latent belief system that is actively hostile towards them. This perception would not be unfounded.

Given the association between traditional gender role attitudes and homophobic attitudes and the heteronormative nature of many traditional gender role attitudes, it would seem likely that sexual minority individuals would tend to reject traditional gender roles attitudes. Indeed, Habarth (2008) found that sexual minority individuals were more likely than their heterosexual peers to reject heteronormative and politically conservative ideologies. Similarly, Shechory and Ziv’s (2007) comparison of heterosexual, gay, and lesbian couples found more liberal gender attitudes among the gay and lesbian couples. Findings such as these lead to the following hypothesis:

H1

Sexual minority individuals will be less likely to endorse traditional gender role attitudes as compared to heterosexual individuals.

Diverging Interests of Sexual Minority Women and Men

Perhaps not surprisingly, an individual’s gender is often pointed to as a factor in his or her gender attitudes (Emmers-Sommer 2014). Fan and Marini (2000) found that young women tend to have more egalitarian gender role attitudes than young men. Young women also tend to place greater value on non-market, home-based work than young men thus indicating that men see the traditionally masculine role of earning outside of the home as more valuable than the traditionally feminine role of home-making. Other studies (e.g., Carter et al. 2009; Kane 1992; Powers et al. 2003; Emmers-Sommer 2014) have also shown that gender is a major predictor of a person’s endorsement or rejection of traditional gender roles with women being less likely to support traditional gender roles than men. This may be due to the fact that many gender role attitudes about women tend to be negative and therefore women are less likely to ascribe to traditional gender role attitudes.

Such findings naturally raise the question of how gender might interact with sexuality in shaping gender role attitudes. Although sexual minority men and women share an interest in the deconstruction of heteronormatively-influenced stereotypes and gender role attitudes, the salience of this interest may not be equal between the two groups. That is, men and women are likely affected by and perceive traditional gender roles differently regardless of their sexual identity.

Indeed, past research has suggested that the sexual minority effect on gender role attitudes might depend on the gender of the sexual minority individual. For instance, some studies have shown that gay and bisexual men have greater levels of internalized homophobia than lesbian women, which has been linked to having more traditional gender role attitudes (Herek et al. 1998; Kahn 1991; Szymanski and Carr 2008). Furthermore, Warriner et al. (2013: 1310) found that gay men are similar to heterosexual men in expressing more benevolent and hostile sexism than lesbian women. In explaining this finding, they note that “being outside the social norm on sexual orientation does not change how the mechanism of sexism occurs in men. Lesbians are still a threat to gay men’s dominant social role, despite the gay man’s sexual orientation status, perhaps taking him out of the heterosexual male dominant status.”

In sum, both lesbian women and gay men may be more likely to reject traditional gender role attitudes due to the heteronormative nature of the attitudes. However, this sexual minority effect might be weaker for men, as they are still potentially influenced by their privileged position in the gender hierarchy. This leads to the second hypothesis of the present study.

H2

Gender will moderate the association between being a sexual minority and rejecting traditional gender role attitude, with sexual minority women rejecting such attitudes more strongly than sexual minority men.

Methods

The data utilized to examine the above hypotheses come from the 2008 through 2016 General Social Survey (GSS) (Smith et al. 2017). Primarily sponsored by the National Science Foundation, the GSS has been conducted annually or biennially since 1972. However, the GSS did not begin asking about respondents’ sexual identity until 2008, hence our focus on the 2008–2016 years. The GSS uses a probability-based sampling design that results in data that are representative of the non-institutionalized US adult population. The large majority of interviews are conducted in person and last about 90 min. Response rates have been around 70% in the past twenty years. We utilize Stata 15.1 to conduct our statistical and include data weights that account for the sample structure of the GSS and patterns of nonresponse. The name of this weight in the GSS codebook is WTSSNR.

Outcome: Gender Role Stereotypes

The GSS has been used extensively to examine attitudes about gender roles (Brooks and Bolzendahl 2004; Carter et al. 2009; Cotter et al. 2011; Kenneavy 2012; Powers et al. 2003; Rice and Coates 1995). We utilize three items that have often been used in this past research as our outcome measures of traditional gender role attitudes. Two of these items representing more of the private or domestic sphere, while one represents more of the public sphere.

Regarding the private sphere, the first items asks respondents to agree or disagree with the statement, “It is much better for everyone involved if the man is the achiever outside the home and the woman takes care of the home and family.” The second item asks respondents to agree or disagree with the statement, “A preschool child is likely to suffer if his or her mother works.” We note that both of these items are grounded in a heteronormative framework, as both explicitly or implicitly describe an other-sexed couple. The heteronormativity of current survey measures of gender role attitudes is a limitation of this study and of the study of gender role attitudes more broadly. We acknowledge that the measures of gender role attitudes that we include in this study are heteronormative and therefore may not apply to the LGB community and we urge future research to use more sexuality inclusive measures of gender role attitudes. We expect that both male and female sexual minorities will freely reject such traditional norms, both because of their heteronormative nature and because neither has much to directly gain from enforcing such norms. In other words, while we except Hypothesis 1 to be supported with these two items, we are less confident that Hypothesis 2 will be supported.

Our measure of traditional gender norms in the public sphere asks respondents whether they agree or disagree with the statement that “Most men are better suited emotionally for politics than are most women.” This question has the advantage that, at least on the surface, the wording of this statement does not appear to be inherently confounded with respondents’ sexuality. That is, it would seem that a gay or lesbian respondent would find this question just as relevant and meaningful as a heterosexual respondent. We still expect that sexual minority individuals will be more likely than heterosexual individuals to reject this statement given its associations with heteronormativity. However, male sexual minorities might have more to gain from enforcing such traditional gender norms given the nature of this statement. Given this possibility, we expect that gender might moderate the sexuality association with endorsing this statement. In other words, we expect both Hypothesis 1 and Hypothesis 2 to be supported with this outcome.

Possible responses for all three items were either agree or disagree, which we reverse coded so that 0 = disagree and 1 = agree. Around 4–5% of cases in each year do not provide either an agree or disagree response and are instead coded as being “not sure.” As with past research (e.g., Carter et al. 2009), we exclude these cases.

Predictors of Interest: Sexual Identity and Gender

As stated in our hypotheses, our primary interest is in how sexuality is associated with gender role stereotypes and how this association is moderated by gender. The GSS began to include a question about respondents’ sexual identity in 2008. The question asks, “Which of the following best describes you?” This question is asked during a self-administered portion of the GSS interview. Possible responses are (1) gay, lesbian, (2) bisexual, and (3) heterosexual or straight. We exclude individuals that said that they “don’t know” or did not provide an answer to this question. The heterosexual category serves as the reference group in the analysis. Of the 5391 individuals included in the analysis, 96.25% (N = 5166) identified as heterosexual, 1.81% (N = 105) identified as gay or lesbian, and 1.94% (N = 120) identified as bisexual.

Respondents’ gender is coded as either man or woman. The man category serves as the reference group in the analysis. It is worth noting that this measure is coded by the interviewer and is not asked explicitly, which means that it is possible that some respondents are coded differently than their actual gender identity or their biological sex.

Controls

Although our hypotheses are focused on the sexuality and gender predictors, we must take into account other influences on respondents’ gender role stereotypes to come to an accurate conclusion about those predictors. Given this, we include a number of controls highlighted by past theory and research.

Researchers (Kane 1992, 2000; Powers et al. 2003; Carter et al. 2009) have shown that when comparing the acceptance and endorsement of traditional gender role attitudes of African Americans and white Americans, race and gender both have an effect and have a combined effect. Minority men, minority women, and white women are less likely to support traditional gender role attitudes than white men who benefit the most from the traditional heteronormative structure in the United States (Powers et al. 2003; Carter et al. 2009). To account for any potential racial effect, we include indicators representing the respondent’s race. This measure includes three categories: (1) white, (2) black, and (3) other. The white category serves as the reference group in the analysis. We chose white as the reference group in part because it is the largest group and in part because research has found that white individuals tend to have more conservative attitudes on some gender-related attitudes (Kane 1992, 2000).

Many researchers have shown that younger cohorts tend to be more egalitarian in regards to gender role attitudes than older cohorts were (Brooks and Bolzendahl 2004; Cotter et al. 2011; Fan and Marini 2000). Therefore, we control for the respondent’s age that is measured continuously from 18 to 89, with the 89 value representing respondents who are age 89 and above. Part of the reason as to why younger cohorts tend to be more liberal than older cohorts is that younger cohorts tend to be more educated and higher levels of education has been linked to more liberal views (Brooks and Bolzendahl 2004; Buchmann and DiPrete 2016; Cotter et al. 2011; Habarth 2008). The more educated a person is, the less likely they are to have traditional gender role attitudes. We also include a measure of the respondent’s education, which is based on the respondent’s highest educational degree. Responses were (0) less than a high school degree, (1) high school degree, (2) junior college\associate’s degree, (3) bachelor’s degree, or (4) graduate degree.

The more religious a person is, the more likely they are to have traditional gender role attitudes (Bolzendahl and Brooks 2005; Siordia 2016). Siordia (2016) found results that indicate a direct relationship between gender role ideology and religious ideology. We use a measure of an individual’s view of the bible to measure their religious beliefs. The question asks, “Which of these statements comes closest to describing your feelings about the Bible?” Possible responses were (1) The Bible is the actual word of God and is to be taken literally, word for word, (2) The Bible is the inspired word of God but not everything in it should be taken literally, word for word, (3) The Bible is an ancient book of fables, legends, history, and moral precepts recorded by men, or (4) Other. The first response serves as the reference category in the analysis, and we expect that the other responses will be associated with more liberal gender attitudes. We also include a measure representing the respondent’s political ideology. The question asks, “We hear a lot of talk these days about liberals and conservatives. I’m going to show you a seven-point scale on which the political views that people might hold are arranged from extremely liberal–point 1–to extremely conservative–point 7. Where would you place yourself on this scale?”

Additionally, given the questions used to measure gender role attitudes were concerned with work and family roles, we included measures of respondents’ occupational and marital status. Work status is measured with indicators representing either (1) working full-time, (2) working part-time, (3) unemployed or temporarily not working, (4) retired, (5) in school, (6) keeping house, and (7) other. Marital status is measured with five indicators representing (1) currently married, (2) widowed, (3) divorced, (4) separated, and (5) never married. A measure is also included representing the number of children the respondent has ever had. This ranges from 0 to 8, with the last value representing 8 or more children.

We also include controls representing respondents’ region and context of residence. There is evidence to suggest that there are regional differences in gender role attitudes with the people from the United States south being more likely to support traditional gender role attitudes than people from other regions (Powers et al. 2003; Carter et al. 2009). Region is measured with four indicators corresponding to the Census-defined regions of Northeast, Midwest, South, and West. The Northeast region serves as the reference category in the analysis. Additionally, researchers have found differences in gender role attitudes depending on the context of a person’s place of residence with those living in more urban places being less likely to support traditional gender role attitudes (Carter et al. 2009). Context of residence is measured with indicators representing whether the respondent resides in a standard metropolitan statistical area (SMSA) or not, the size of that SMSA, if applicable, and where in the SMSA the respondent resides. The precise categories are: (1) Central city of the 12 largest SMSAs, (2) Central city of the remaining 13-100 SMSAs, (3) Suburb of the 12 largest SMSAs, (4) Suburb of the remaining 13-100 SMSAs, (5) Non-SMSA county with town of 10,000 or more, and (6) Non-SMSA county without a town of 10,000 or more. Finally, we include a measure representing the year of the GSS.

Results

After excluding cases that were missing on the measures described above, the final analytic sample consisted of 5324 individuals. Descriptive statistics for all the measures can be found in Table 1. As seen in this table, 29.04% of sampled U.S. adults agreed that men are better suited for politics than women, 39.04% agreed that it is better for a man to work while the woman takes care of the home, and 43.47% agree that preschool kids suffer if the mother works. Overall, then, traditional gender roles are somewhat more likely to be endorsed in the domestic or private sphere than in the public sphere.

Table 1 Descriptive statistics
Full size table

Looking at the sexual identity measure, we find that 95.81% of the sample identified as heterosexual, while 1.95% identified as gay or lesbian and 2.24% identified as bisexual. As would be expected, the percentage of self-identified sexual minority individuals in the GSS data mirrors other national probability sample surveys (Newport 2015).

Table 2 examines the responses to the three outcome measures across the sexual identity groups. We see in this table that gay and lesbian individuals (6.02%) are less likely than heterosexual individuals (20.35%) to agree that men are better suited for politics. The gap, however, is smaller when examining bisexual individuals (16.54%). We see similar patterns for the two private sphere gender role items. About 30% of heterosexual individuals agree that a traditional male breadwinner-female homemaker household is better. This compares to 12.36% of gay and lesbian individuals and 28.99% of bisexual individuals. Just over 31% of heterosexual individuals agree that preschool children suffer if their mother works. This compares to 13.45% of gay and lesbian individuals and 25.30% of bisexual individuals.

Table 2 Cross-tabulations of gender role attitudes by sexual identity
Full size table

In sum, at least for gay and lesbian individuals, we find some initial support for Hypothesis 1. We find less support for Hypothesis 1 when focusing on bisexual individuals. In a moment we will examine the results of an analysis that accounts for other differences between the sexual identity groups.

Hypothesis 2 states that the sexual identity differences in traditional gender attitudes will be moderated by gender. Table 3 presents a first look at this hypothesis. This table breaks out responses not only by sexual identity but also by gender. For example, we see that gay men appear to be less likely to agree with the statement that men are better suited for politics relative to heterosexual men (11.12% to 21.04%), but bisexual men are actually more likely to agree (32.04%). Overall, though, the differences across the sexual identity groups are not statistically significant for men.

Table 3 Cross-tabulations of gender role attitudes by sexual identity and by sex
Full size table

On the other hand, we find larger and more consistent differences among women. Compared to 19.73% of heterosexual women, almost no lesbian women agree (.78%) with the idea that men are better suited for politics. Bisexual women are also less likely than heterosexual women to disagree with this statement (9.20%). These findings in Table 3 offer some initial support for Hypothesis 2.

Looking at the second outcome in Table 3 also shows some possible, although weaker, moderation of sexual identity differences by gender. Looking at the male columns shows that the overall differences across sexual identity do not quite reach the p < .05 level of statistical significance, while the female columns do reach this level. On the other hand, we do not find gender differences in the sexual identity gaps when examining the preschool child outcome.

While the patterns seen in Tables 2 and 3 provide some support for our expectations, these patterns could be driven by other factors associated with gender role attitudes that differ across the sexual identity groups. That is, what would the results look like if we compared heterosexual, gay and lesbian, and bisexual individuals who were identical on education, age, political ideology, and other factors that might also be associated with gender role attitudes? To address this question, we conducted logistic regression analyses for the outcomes where 0 = respondent disagrees with traditional gender role statement and 1 = respondent agrees with traditional gender role statement. The results of this analysis, in the form of odds ratios, are shown in Table 4. Values above 1 indicate that a measure is associated with an increase in the odds of a respondent agreeing with the traditional gender role statement, while values below 1 represent a decrease in the odds of a respondent agreeing with that statement.

Table 4 Logistic regression models predicting agreement with gender role statements (odds ratios presented)
Full size table

Models 1 through 3 examine the “men are better suited emotionally for politics” outcome. Model 1 examines only the association between sexual identity and agreement with this statement. We find that gay and lesbian individuals have significantly reduced odds, relative to heterosexual individuals, of saying that men are better suited for politics. This generally corresponds to what was seen in Table 2 and supports Hypothesis 1, although this model shows that the percentages for bisexual individuals in that table are not significantly different from heterosexual individuals.

Model 2 in Table 4 introduces the control measures. After introducing these controls, the previously significant difference in the odds of saying that men are better suited for politics between gay and lesbian individuals and heterosexual individuals becomes non-significant. However, as seen earlier in Table 3, this overall difference might be hiding differences between male and female gay and lesbian individuals. In a moment we will consider whether gender moderates the sexual identity differences.

Looking first at the control measures, though, we see that female individuals have significantly lower odds compared to male individuals of saying that men are better suited for politics. Compared to white individuals, those of an other race have significantly higher odds of saying that men are better suited for politics. We do not find any significant differences across marital statuses or number of children. We do find, though, that retired individuals have higher odds of saying that men are better suited for politics as compared to those who are currently working full-time.

Education is negatively associated with the odds of saying that men are better suited for politics, while those saying that the bible is the actual word of God have higher odds of expressing the traditional gender role attitude as compared to those saying the bible is the inspired word of God, book of fables, or those expressing an other position on the bible. Political conservativism is also associated with higher odds of endorsing the traditional gender attitude. We do not find any significant differences across region or residential size and context net of the other measures in the model. Finally, we find that the year of the survey is negatively associated with the odds of expressing the traditional gender role attitude.

Model 3 examines the potential moderating effect of gender on the association between sexual identity the “men are better suited emotionally for politics” outcome. This moderating effect was proposed in Hypothesis 2 and provided some initial support in Table 3. Because of the inclusion of the interaction terms, the sexual identity odds ratios represent the differences among males, while the gender differences represent the differences among heterosexual individuals.

Looking at the sexual identity measures, we see that among men there are no significant differences between the heterosexual category and the gay and bisexual categories in the odds of expressing the traditional gender role attitude that men are better suited for politics than women. Examining the interaction terms, however, we find that among women gay, lesbian, and bisexual individuals have significantly reduced odds of expressing the traditional gender role attitude as compared to heterosexual women. In other words, being a sexual minority is significantly associated with rejecting the idea that men are better suited for politics among women, but not among men. This finding provides support for Hypothesis 2.

Models 4 through 6 are similar to the first three models, but these examine the “much better…if the man is the achiever” outcome. In Model 4 we see that, relative to heterosexual individuals, gay and lesbian individuals have significantly reduced odds of agreeing with this statement. Model 5 introduces the control measures. In this model we find that gay and lesbian individuals do not significantly differ from heterosexual individuals in their odds of agreeing with this traditional gender role sentiment, but bisexual individuals actually show significantly higher odds relative of agreeing with the statement relative to heterosexual individuals.

Looking at the control measures in Model 5, we find fairly similar patterns to what were seen in Model 2, although we find more consistent work status differences relative to individuals working full-time (with those not working full time having higher odds of agreeing with the traditional gender statement) and regional differences relative to those residing in the Northeast (with those in the South and West agreeing with the traditional gender statement). Model 6 examines whether the sexual identity differences might depend on or be moderated by gender. Unlike with Model 3, though, we do not find evidence of such moderation. In sum, both gay men and lesbian women are less likely to agree with the idea of a male breadwinner-female homemaker compared to heterosexual individuals. This sexual identity difference itself seems to be largely a function of demographic and ideological differences between the gay and lesbian individuals and heterosexual individuals, as this gap disappears in Model 5.

Models 7 through 9 examine the “preschool child is likely to suffer if his or her mother works” outcome. Model 7 again shows that gay and lesbian individuals have lower odds of agreeing with this statement relative to heterosexual individuals. Bisexual individuals do not significantly differ, though. These effects remain in Model 8 after the demographic and ideological controls are included. Model 9 introduces the interaction terms to consider whether the sexual identity effects differ by gender. However, these interaction terms are not significant, which indicates that both gay men and lesbian women are less likely to agree with this statement relative to heterosexual men and women.

Looking at Table 4 overall suggests that gay men and lesbian women do tend to reject traditional gender roles as compared to heterosexual individuals as suggested by Hypothesis 1. although in some cases this partly a function of demographic and ideological differences between the sexual identity groups. For private sphere or domestic gender roles, there is no difference between gay men and lesbian women. For public sphere gender roles, however, lesbian women standout in rejecting such traditional gender role attitudes, while gay men do not significantly differ from their heterosexual counterparts. It is also noteworthy that across the models bisexual individuals are not more likely than heterosexual individuals to reject traditional gender roles. This suggests that bisexual individuals are not quite as opposed to such heteronormative and traditional gender attitudes as gay and lesbian individuals. We discuss these patterns further below.

Discussion

Previous research examining the connection between sexuality and gender role attitudes has been limited by the data employed. Some studies have focused on how gender role attitudes shape attitudes towards sexuality among heterosexuals (Brown and Henriquez 2011; Kerns and Fine 1994; Whitley and Ægisdóttir 2000), but this obviously does not tell us much about the attitudes of sexual minority individuals and how they might compare to heterosexual individuals. Studies that have looked at these questions using data that includes sexual minority individuals have been limited by samples that are not representative (Brown and Henriquez 2011; Habarth 2008; Warriner et al. 2013).

The study presented here utilized a probability sample of US adults to examine how sexual minority individuals differ from heterosexual individuals in their attitudes about traditional gender roles and how any sexual minority effect might be moderated by gender. Interestingly, our findings largely support those of past research. Our analysis of the General Social Survey found that lesbian women are significantly less likely to endorse traditional gender role attitudes in the private and public spheres as compared to their heterosexual counterparts. On the other hand, gay men do not significantly differ from their heterosexual counterparts in rejecting traditional gender attitudes in the public sphere, although they are less likely to support traditional gender role attitudes in the private sphere than their heterosexual counterparts. In this way, Hypothesis 1 is partially supported due to the increased likelihood of gay men and lesbian women to be more likely to reject traditional gender role attitudes in the private sphere than their heterosexual counterparts. Moreover, with the divergence of support between gay men and lesbian women for traditional gender role attitudes in the public sphere, Hypothesis 2 is supported because gender is a moderating factor endorsement of traditional gender role attitudes. As others have argued (Warriner et al. 2013), we suggest that this is a function of traditional gender attitudes largely benefiting men, even sexual minority men. This mutes the potential effect of being a sexual minority in rejecting these heteronormatively-driven gender attitudes for sexual minority men.

However, due to the lack of differences between the support for traditional gender role attitudes between bisexual and heterosexual individuals, hypothesis one is not fully supported. Previous research suggests that this may be due to bisexuals’ purposeful separation of ideologies from that of gays and lesbians (Worthen 2013). Similarly to our study, previous research has found that not only does sexuality matter in determining a person’s attitudes and beliefs but that gender of the person is also a key factor to consider (Worthen 2013). It may be the case that much like how gay men still benefit from the heteronormative patriarchal structure, bisexual individuals too can benefit from being able to be in other-sexed relationships that sets them apart from their gay and lesbian counterparts. More research needs to be done in order to determine the causal mechanisms behind why bisexual individuals may be more similar in gender role attitudes to their heterosexual rather than their gay/lesbian counterparts.

Limitations

Of course, this study has its own limitations. One of these limitations is that the analysis was limited to using three measures of traditional gender role attitudes. However, the reason for only using these three measures was due to the fact that other questions utilized in past research about traditional gender roles have an explicit heterosexual framing. Even in this study, two of our three measures of traditional gender role attitudes had explicitly heterosexual contexts. In order to look at gender role attitudes amongst sexual minority individuals in other contexts such as the home, there needs to be more inclusive questions asked on the GSS or other surveys so that a variety of partnerships are included beyond heterosexual partnerships. The use of more inclusive measures of gender role attitudes would allow for a more robust analysis of the gender role attitudes of sexual minority individuals and would make for more inclusive surveys.

While a central contribution of our analysis comes from its use of a comparatively large probability sample of U.S. adults, the total number of sexual minority individuals in these data is still small. This obviously makes it more difficult to make conclusions regarding the nature of smaller differences observed between the sexual identity groups. For example, as seen in the percentages in Table 3 and the odds ratio in Model 3 of Table 4, gay men do appear superficially more likely than heterosexual men to reject the traditional gender norm of men being better suited for politics. Additionally, our evidence suggests that bisexual individuals do not differ from heterosexual individuals in their endorsement of traditional gender role attitudes. We cannot say with confidence, though, whether gay men differ from heterosexual men or whether bisexual individuals differ from heterosexual individuals.

Conclusion

Despite these limitations, our results reinforce what others have found on the importance of considering how gender, sexuality and other characteristics interact with each other (e.g., Swank and Fahs 2013; Civettini 2016). More research should be done to further understand how differences within particular categories (male versus female, gay versus lesbian, bisexual versus gay/lesbian, bisexual vs. heterosexual) affects peoples’ attitudes towards traditional gender roles along with other types of beliefs and attitudes. Understanding the relationship between gender, sexuality and gender role attitudes is more important now than ever before due to the increasing acceptance and prevalence of individuals who do not fit within traditional gender and sexuality norms.