## Abstract

Although research has long documented the relevance of gender for health, studies that simultaneously incorporate the relevance of disparate sexual orientation groups are sparse. We address these shortcomings by applying an intersectional perspective to evaluate how sexual orientation and gender intersect to pattern self-rated health status among U.S. adults. Our project aggregated probability samples from the Behavioral Risk Factor Surveillance System (BRFSS) across seven U.S. states between 2005 and 2010, resulting in an analytic sample of 10,128 sexual minority (gay, lesbian, and bisexual) and 405,145 heterosexual adults. Logistic regression models and corresponding predicted probabilities examined how poor self-rated health differed across sexual orientation–by-gender groups, before and after adjustment for established health risk factors. Results reveal distinct patterns among sexual minorities. Initially, bisexual men and women reported the highest—and gay and lesbian adults reported the lowest—rates of poor self-rated health, with heterosexuals in between. Distinct socioeconomic status profiles accounted for large portions of these differences. Furthermore, in baseline and fully adjusted regression models, only among heterosexuals did women report significantly different health from men. Importantly, the findings highlight elevated rates of poor health experienced by bisexual men and women, which are partially attributable to their heightened economic, behavioral, and social disadvantages relative to other groups.

## Introduction

The last several decades have witnessed a plethora of research documenting health disparities across a variety of status-based characteristics, including social class (Marmot 2004); racial/ethnic identity (Williams and Sternthal 2010); gender (Bird and Rieker 2008); and more recently, sexual orientation (Institute of Medicine (IOM) 2011). This work has done much to advance understanding of how characteristics that relate to social location and access to resources and power, as well as discrimination and disadvantage, shape the likelihood of living a long life in good health or a short life more burdened by disease and disability. In this study, we examine how self-reported health differs across gender and sexual identity groups. To date, studies have shown that women, on average, self-rate their health more poorly than men (e.g., Gorman and Read 2006). And recent work on self-rated health has begun establishing differences by sexual identity, finding, for example, that bisexual adults report significantly worse health than both heterosexual and gay/lesbian adults (Conron et al. 2010; Veenstra 2011).

Albeit informative, research that has examined how health outcomes vary across a single status-based category (e.g., men vs. women, heterosexuals vs. sexual minorities) is being increasingly criticized because individuals’ life chances and well-being are influenced by more than a single system of stratification (Cho et al. 2013). Further, scholars have identified several problems with empirical research that makes additive assumptions about processes of inequality and the manner in which intersectional forces are often ignored (see discussions by Bauer 2014; Bowleg 2012). Indeed, reviewing the state of the literature on sexual minority health (i.e., among persons who self-identify as lesbian, gay, bisexual, or transgender), the IOM and gender scholars have concluded that more work adopting an intersectional perspective is needed given that research rarely examines how sexual orientation operates in tandem with other status-based characteristics to shape health outcomes (Hankivsky 2012; IOM 2011; Springer et al. 2012).

In this article, we respond to these calls by explicitly considering how gender and sexual identity intersect to shape self-ratings of health status given that each is a central concept of interest in intersectional research (Hankivsky and Cormier 2009). Regarding scholarship on gender and health, this is an important expansion beyond the vast majority of past research where sexuality is not considered (Cole 2009). In addition to the intersectional focus, we expand on past scholarship investigating sexual identity and health in two key ways. First, a large proportion of past works have relied on nonprobability samples (IOM 2011), but here we draw on state-level aggregated data, collected using probability sampling that includes measures of sexual identity, gender, self-rated health status, and a range of relevant sociodemographic controls. Second, although the majority of scholarship has combined sexual minority subgroups, often to compensate for small sample sizes (Cochran et al. 2001), we follow important precedents indicating potentially distinct health profiles for gay, lesbian, and bisexual persons (Green and Feinstein 2012; McCabe et al. 2009).

Altogether, these strengths allow us to investigate whether gender and sexual identity intersect to shape self-rated health reports. In the analyses that follow, we examine both within-gender (e.g., heterosexual women vs. bisexual women) and across-gender (e.g., gay men vs. lesbians) differences in self-rated health. But more critically, and as we discuss in detail herein, we focus our attention around an evaluation of whether self-rated health reports are best among those who, on average, occupy the most-advantaged social position (i.e., heterosexual men) relative to those who occupy the least-advantaged positions (i.e., lesbians and bisexual women) as well as to those who simultaneously occupy both advantaged and disadvantaged positions (i.e., gay men, bisexual men, and heterosexual women). Further, to situate sexual minority health into the larger body of research on population health disparities, we evaluate how these relationships operate before and after adjustment for established health risk factors, including socioeconomic status (SES), health behaviors, and social networks/support and well-being.

## Gender, Sexual Orientation, and Self-rated Health

Intersectionality theory posits that gender and sexual identity interact in multiplicative ways to produce unique social contexts that shape the health and well-being of persons within each intersectional position (Choo and Ferree 2010; Collins 2000; Hankivsky and Cormier 2009). Rather than examining health disparity across a single social status, or assuming that health outcomes are the simple sum of advantages and disadvantages associated with different social statuses (e.g., woman + bisexual), an intersectional perspective emphasizes that “social identities are not independent and unidimensional but multiple and intersecting,” reflecting “interlocking systems of privilege and oppression” (e.g., sexism and heterosexism; Bowleg 2012:1267–1268). As a result, these interlocking axes of inequality cannot be disentangled, either conceptually or empirically, because all persons hold various combinations of identities that produce unique advantages and disadvantages in terms of life chances and health status (Collins 2000; Veenstra 2013).

Following McCall (2005), we adopt an intercategorical approach to assess whether and how self-rated health reports differ across the intersections of gender and sexual identity among adults. Importantly, an intercategorical approach recognizes the “complexity of relationships among multiple social groups within and across analytical categories,” and advocates for a method that is systematically comparative (McCall 2005:1786). As such, we examine self-rated health reports across all categories of intersection to fully consider and compare intersecting hierarchies of gender and sexuality (Harnois and Ifatunji 2011). And even though early work on intersectionality focused on the experiences and well-being of groups who held multiple disadvantaged statuses, it is also the case that some groups hold both disadvantaged and privileged identities (e.g., heterosexual women), indicating that the comparative nature of an intercategorical analysis also offers an important opportunity to evaluate the health status of persons who simultaneously experience social positions that are both privileged and marginalized (Bauer 2014; Bowleg 2012; Cole 2009). In our analyses that follow, we examine how gender (women, men) combines with sexual identity (heterosexual, gay/lesbian, bisexual), resulting in six distinct gender-by-sexuality groups.

Our first objective is to describe health discrepancies at the intersection of sexual orientation and gender. We begin by examining patterns of health by sexual orientation for men and women. Many early studies used convenience samples to document predominately sexual and psychiatric health disadvantages for sexual minorities. More recently, using population-based survey data from single U.S. states (Conron et al. 2010; Dilley et al. 2010) and Canada (Veenstra 2011), researchers documented poorer physical and mental health for sexual minority men and women on some indicators, and similar health on other indicators, compared with heterosexuals. Importantly, these same studies documented some gender discrepancies. For example, although gay men had higher risks than heterosexual men for various mental health issues, they were less likely to report some risks for cardiovascular disease (Conron et al. 2010). Yet, when compared with heterosexual women, lesbians were consistently in worse health (Conron et al. 2010; Dilley et al. 2010).

Because sample sizes are small and analyses will not support stratification across sexual minority groups, studies have often grouped gay/lesbian and bisexual adults under the broad umbrella of “sexual minority” (IOM 2011). This approach is problematic, however: recent studies have found worse self-reported health for bisexuals compared with gay/lesbian and heterosexual adults (Veenstra 2011), although these patterns may vary by gender (see Conron et al. 2010; Dilley et al. 2010). Some research has suggested that bisexuals may be “minorities within the minority” because they experience even greater levels of prejudice and discrimination than gays or lesbians (Herek et al. 2007). The resulting stressors and stigma experienced by bisexuals may contribute to greater health problems (Jorm et al. 2002). Thus, our first hypothesis investigates self-rated health across categories of sexual orientation for men and women:

Hypothesis 1 (H1): Within gender, heterosexuals will have the lowest odds of poor health, followed by gays and lesbians, and then bisexuals, who will have the highest odds of poor health.

Second, we examine gender patterns in health for each sexual orientation subgroup: heterosexual, gay/lesbian, and bisexual. Gender comparative studies have long documented that, on average, women report worse health than men (Gorman and Read 2006). However, whether that baseline pattern holds across sexual orientation groups is largely unknown. Conron et al. (2010) examined data from 2001–2008 in Massachusetts and found that among heterosexuals and gay/lesbian adults, women reported slightly higher rates of poor-to-fair self-rated health than men (for heterosexuals, 9.3 % among men vs. 10.1 % among women; and 8.9 % among gay men vs. 10.6 % among lesbians). These numbers also show that within each gender, self-rated health reports are very similar between heterosexual and gay/lesbian adults. However, the stand-out group was bisexuals, who reported higher rates of poor-to-fair health—especially among men—reversing the “typical” gender pattern as a higher proportion of bisexual men reported poor-to-fair health than bisexual women (19.4 % of bisexual women, and 24.7 % of bisexual men). In comparison, Dilley and colleagues (2010) examined data from Washington state from 2003—2006 and reported similar self-rated health patterns within and between heterosexual and gay/lesbian men and women. In the analyses that follow, we draw on aggregated state-level data to investigate whether women report worse self-rated health than men for each sexual identity.

Hypothesis 2 (H2): Within each sexual orientation group, the odds of poor health will be higher among women than among men.

Based on the previously cited literature, there are reasons to presume that heterosexual men would report the best self-rated health given that they represent the two categories (man and heterosexual) that are often shown to be most advantaged across a range of health-relevant characteristics. Beyond social and economic advantages, studies also show that heterosexual men experience less gender- and sexuality-based discrimination than women and sexual minority men (Meyer 2007; Pascoe 2007; Swim and Hyers 2009). In contrast, the worst health outcomes may occur among those who occupy multiple disadvantaged or subordinate positions – in this analysis, sexual minority, and specifically bisexual, women. Indeed, intersectional scholarship was proposed under the thesis that women who hold more than one disadvantaged status (including sexual minority women; see Collins 2000) experience an elevation of disadvantage across various domains of life, including health status. Thus, our third hypothesis tests for a hierarchy of health by sexual orientation and gender:

Hypothesis 3(H3): The odds of poor health will follow a sexual orientation and gender gradient and will be lowest for heterosexual men and highest for bisexual women.

Not only does focusing on health at the intersections of identity and social position improve the validity of research, but it also highlights the heterogeneity that exists in the factors that relate to and produce health disparities (Bauer 2014; Cole 2009; Connell 2012; Hankivsky 2012). As such, we aim both to describe how self-rated health varies at different intersections of gender and sexual identity and to explore how established risk factors shape health patterns across gender-by–sexual identity groups. Existing health disparities research provides a framework in which to understand group differences in health by SES (Link and Phelan 1995), health behaviors (Pampel et al. 2010), and social networks/support (Berkman and Glass 2000).

The health disparities literature has established the role of SES as a fundamental cause of disease (Link and Phelan 1995), and as such, SES is often implicated as one of the strongest contributors to health stratification. Men are socioeconomically better-off than women (e.g., Chang 2010); and although patterns have been inconsistent across studies (IOM 2011), a recent report showed that sexual minorities are at least as likely, if not more likely, to be poor than are heterosexuals (Badgett et al. 2013). In particular, the poorer self-ratings of health among bisexuals may be due largely to their lower SES. Studies have shown that bisexual men and women report lower levels of education, employment, insurance coverage, and access to a regular health care provider, as well as higher poverty rates, than heterosexual and gay/lesbian men and women (Conron et al. 2010; Dilley et al. 2010). Further, within sexual orientation groups (e.g., comparing gay men and lesbians), poverty rates are higher among women than among men (Conron et al. 2010; Dilley et al. 2010).

Health behaviors also contribute to health disparities. Men smoke and drink heavily more often than women (Read and Gorman 2010), although they also exercise more (Dilley et al. 2010). Within gender groups, health behavior patterns appear to vary by sexual orientation. Heterosexual men, compared with sexual minority men, engage in physical activity less regularly, weigh more, and have less regular contact with health care providers than sexual minority men (Brennan et al. 2010; Dilley et al. 2010). However, they also smoke less and engage in substance abuse less often (Brennan et al. 2010; Dilley et al. 2010). Sexual minority women, however, more often smoke cigarettes, drink heavily, maintain heavier body weights, and have less regular contact with health care providers than do heterosexual women (Conron et al. 2010; Dilley et al. 2010; McCabe et al. 2009; Roberts et al. 2003; Tang et al. 2004). When disaggregated, studies have also shown that bisexuals appear to have the highest health behavioral risks. Bisexual men and women have rates of smoking and heavy drinking that far exceed those of their gay, lesbian, and heterosexual counterparts (Conron et al. 2010; Dilley et al. 2010; Green and Feinstein 2012; Tang et al 2004).

For social networks and support, legal limitations have historically restricted marriage rights among sexual minorities, which may partly explain worse self-rated health among cohabiting sexual minorities relative to comparable heterosexual married persons (Denney et al. 2013; Liu et al. 2013). Among heterosexuals, men benefit more than women from the health-promoting effects of marriage (Read and Gorman 2010). At the same time, however, men generally do a poorer job of maintaining social and family ties than women (Bird and Rieker 2008), and women’s health tends to benefit from their higher level of social involvement relative to men (Umberson et al. 1996). It is less clear how social networks and support impact sexual minority health, but the health of men in same-sex relationships may be harmed by diminished essential supports typically provided by women in different-sex relationships (see Reczek and Umberson 2012). Indeed, although men appear to benefit equally from marital and different-sex cohabiting relationships (Wu et al. 2003), this benefit may not hold in the absence of a female partner.

Overall, we apply these bodies of work on the relationships among SES, health behaviors, and social networks/support, and posit two final hypotheses:

Hypothesis 4 (H4): For contrasts within gender but across sexual identity, adjusting for SES, social networks/support/well-being, and health behaviors will reduce the self-rated health disadvantage among bisexuals relative to heterosexual and gay/lesbian adults.

Hypothesis 5(H5): For contrasts within sexual identity but across gender, adjusting for SES will reduce the self-rated health gap between men and women, whereas adjusting for health behaviors and social networks/support/well-being will increase the gap.

## Data and Methods

We use Behavioral Risk Factor Surveillance System (BRFSS) data for adults aged 18 and older for seven U.S. states, which included a sexual orientation measure between 2005 and 2010 (CDC 2005–2010): California (2006–2010), Maine (2005–2010), Massachusetts (2005–2010), Montana (2010), North Dakota (2005–2010), Oregon (2005–2009), and Washington (2005–2010).1 The BRFSS uses a disproportionate stratified sample design, based on areas with a high or low density of telephone numbers, to select households with telephones in each state and then draws a random sample of one adult per household. We used two parts of the BRFSS questionnaire: the core component and optional state-added questions. The core component asks a standard set of questions to respondents in all U.S. states and territories. Although all states must include core component questions on their yearly BRFSS, the states listed here opted to include a question on self-identified sexual orientation. We constructed our sample by aggregating and merging data from these states. The final sample covers 35 state survey-years and includes 415,271 adults.

We include data from these states for the period 2005–2010 for practical reasons, and to be sensitive to the rapidly changing sociopolitical climate (Gallup 2010; Powell et al. 2010) regarding sexual minorities in the United States. Practically, we are restricted by the number of states that included a sexual orientation measure. We balanced the need for a sufficient sample size of gender-by-orientation groups with the desire to characterize relationships over a relatively short period. Most of the states included in the analysis are in the Northeast or the West, and this geographic distribution should be considered when interpreting our results.

### Measures

We examine how sexual identity and gender relate to a dichotomous version of self-rated health, where 1 = poor or fair health, and 0 = good, very good, or excellent health (hereafter referred to as “poor self-rated health” or “poor SRH”). Although this measure is a subjective (and evaluative) assessment of health status, research has established that respondents draw on many dimensions of health status when reporting self-rated health and that it is a strong and independent predictor of mortality (Idler and Benyamini 1997; Jylhä 2009; Singh-Manoux et al. 2006). Studies have also documented that beyond gender and sexual orientation, self-rated health is strongly related to the control measures (detailed later) in our study; self-rated health reports rise with socioeconomic standing and differ across racial/ethnic groups (e.g., Denney et al. 2013); are lower among adults with a higher body mass index (BMI) and among those who smoke, drink heavily, or report no regular exercise (e.g., Gorman and Read 2006); and are higher among the married and those with high levels of social integration (e.g., Gorman and Sivaganesan 2007).

Sexual orientation was measured by self-identification as heterosexual, gay, lesbian, or bisexual. Although there is some variation in the question wording across states, it was most often included as, “Do you consider yourself to be heterosexual, homosexual or gay/lesbian, bisexual, other, or don’t know?” Response categories are consistent across states and years in that every question distinguishes heterosexual, gay, lesbian, and bisexual adults; we focus on these groups. Two additional options—“other” and “asexual”—were sometimes available to respondents, but these were not included by all states. In addition to being inconsistently applied, the BRFSS provides no additional information to evaluate these individuals, so these cases are omitted from our analyses (n = 1,356). Finally, all states allowed respondents either to refuse the sexual orientation question or to respond with “don’t know.” Not all the data sets supplied by the states distinguish between the two; in some states, the two categories were combined. To maintain consistency across states, both refusals and “don’t know” responses were treated as missing data. Our resulting analytic sample includes respondents who identified as heterosexual/straight (n = 362,213), gay or lesbian (n = 5,874), or bisexual (n = 3,263), as well as cases in which sexual orientation was missing (n = 42,565). In the upcoming analyses, we provide sensitivity results that compare estimates from imputed sexual orientation and missing sexual orientation models. Although sexual orientation is a concept that is fluid over the life course and involves identity, attraction, and behavior (Herek 2006), as discussed earlier, we are restricted to a single measure of self-reported sexual identity. Consistent with our research objectives, we create a categorical measure of gender-by-orientation that includes six categories: heterosexual men, heterosexual women, gay men, lesbian women, bisexual men, and bisexual women.

All models control for state of interview and year of survey in order to partially account for regional differences in health and attitudes toward sexual minorities as well as changing attitudes toward sexual minorities (Austin and Irwin 2010; Gallup 2010; Gates and Ost 2004; Powell et al. 2010). We also include demographic controls for racial/ethnic identity (non-Latino white, non-Latino black, non-Latino Asian, Latino, and other race), and age at interview.

To address the role of established risk factors for group health differences, we consider variables available across all states and years of our sample measuring aspects of SES, health behaviors, and social networks/support and well-being. For SES, this includes completed schooling as well as employment status (currently working, unemployed, homemaker, student, retired, and unable to work); total annual household income (<$25,000,$25,000–$49,999,$50,000–$74,999, and$75,000 or more); and two medically relevant measures of SES: medical insurance (1 = insured, 0 = uninsured) and whether the respondent missed a doctor visit last year because of cost.

Health behaviors include whether the respondent has one person they think of as a regular doctor and whether the respondent visited a doctor for a routine checkup last year. We also include measures of physical exercise (where 1 = participated in any physical exercise (e.g., running) outside of work last month, and 0 = none) and smoking status (1 = currently smokes, 0 = does not smoke). For alcohol use, respondents identified whether they currently drank alcohol; if so, they were asked, on average, how many alcoholic drinks they had on the days when they drank. We define heavy drinking as drinking that exceeds more than one drink per occasion for women, and more than two drinks for men (U.S. DHHS and U.S. Department of Agriculture 2005). We also include a categorical measure of weight status (underweight, normal weight, overweight, and obese), calculating body mass index from self-reported height and weight (CDC 2012).

Last, we include three measures of social networks/support and well-being, including marital status, emotional support (which measures how often the respondent gets the support he/she needs, where 1 = never, 2 = rarely, 3 = sometimes, 4 = usually, 5 = always); and life satisfaction, which provides an assessment of psychosocial well-being (where 1 = very dissatisfied, 2 = dissatisfied, 3 = satisfied, and 4 = very satisfied).

### Analysis

Combining multiple years of data from multiple BRFSS states presents some unique challenges. We use Stata 12.1 (StataCorp 2011) to perform a variety of data management procedures and to conduct all analyses. Because BRFSS data are collected using a complex sampling design, all state data sets include stratification variables as well as sampling weights (CDC 2005–2010). Multiple survey years for the same state are included in the sample, so sampling weights are adjusted based on the size of the sample (state-year) relative to the size of the combined sample for that state; we also modify the strata variables to include state and survey year as additional levels of stratification (Korn and Graubard 1999). We adjust for the complex survey design using Stata’s svy commands, which use first-order Taylor linear approximation for this design.

Missing values come from two sources. In some cases, values are missing by design because of random assignment of respondents to different versions of the survey; these values are known to be missing completely at random. In other cases, missingness is due to item nonresponse, and evidence from the relationships of missing data with individual characteristics suggests that the data are not missing completely at random (unconditional on the observed characteristics), which makes typical approaches, including listwise deletion, inappropriate (Allison 2002). Rather than ignore the missing data mechanism, we employ multiple imputation (MI) by chained equations to generate 40 MI data sets, which we use for all reported descriptive statistics and regression models. Given the level of item nonresponse in our analytic sample, multiple imputation is a preferred technique (see Langkamp et al. 2010). Imputed values include a random component based on draws from the posterior predictive distribution of the missing data under a posited Bayesian model and, under the missing-at-random assumption (a more plausible assumption than is made by listwise deletion), provide unbiased estimates of variance (Allison 2002). The dependent variable, poor SRH, was included in the imputation model, and then cases missing on SRH (N = 1,299) were excluded from analyses following best practices (von Hippel 2007).

We use logistic regression on the multiply imputed data and report odds ratios of poor SRH. We first present baseline results and then sequentially add demographic controls and the confounders for SES, health behaviors, and social networks/support/well-being to examine the change in odds with the inclusion of each set of covariates. All logistic regression results are calculated relative to heterosexual men.

To evaluate our hypotheses examining differences within sexual orientation but across gender and within gender but across sexual orientation, we use the results of the regressions and calculate predicted probabilities and significance tests for the average difference in probabilities across the gender-by-sexual identity categories. We test average marginal effects (AME) by computing average probabilities using all observed data for hypothetical populations of all heterosexual men, all heterosexual women, and so on. We then conduct pairwise comparisons of the difference in probabilities (the average marginal effects) for Group 1 (e.g., heterosexual men) versus Group 2 (e.g., bisexual men). Using the observed data, instead of the more common approach of setting all model covariates at their means, is beneficial in that it isolates the influence of the gender-by-orientation characteristic (Bartus 2005).

## Results

### Sample Characteristics

Table 1 presents weighted mean and percentage values, stratified by gender and sexual identity, and shows that bisexual adults report the highest rates of poor health. In contrast, gay men and lesbians have the lowest proportions of poor SRH. Similar to bisexuals, gay men report poor health more often than women; only among heterosexuals do women exceed men in the proportion reporting poor SRH (15.6 % vs. 14.6 %, respectively).

Turning to demographic characteristics, despite their high proportions of poor SRH, bisexuals are considerably younger than any other group (see Conron et al. 2010 and Dilley et al. 2010 for similar findings). Heterosexual adults are the oldest, consistent with limited evidence suggesting that younger adults are more likely to report same-sex attraction and behavior (IOM 2011). Heterosexual men and women are similar in terms of race/ethnic identity, with nearly two-thirds reporting as white, one-fifth as Latino, and much smaller but roughly equivalent proportions identifying as black, Asian, or other race. Compared with heterosexuals, a higher proportion of gay men (77 %) and lesbians (71.6 %) are white, and about 12 % self-identify as Latino. A slightly greater proportion of gay men report that they are black (5.3 %) than both heterosexual men (4.8 %) and lesbians (3.9 %). However, although 4 % of gay men self-identify as Asian, this proportion is more than double among lesbians (at 8.7 %). Nationally representative data on sexual minorities is scarce, but these proportions for black and Hispanic adults are similar to recent reports (see Kastanis and Gates 2013a, b), and the proportions for Asians are larger (Kastanis and Gates 2013c). Finally, the bisexual sample shows considerable variability in racial/ethnic identity between men and women. Almost two-thirds of men and women are white, and roughly similar proportions are of Latino descent. However, three times as many bisexual women identify as black compared with bisexual men (9.2 % vs. 2.8 %), and more than three times as many bisexual men identify as Asian compared with bisexual women (15.3 % vs. 4.1 %).

For SES, Table 1 shows that gay men and lesbians are socioeconomically advantaged, as some studies have reported (Black et al. 2000), and considerably so relative to their bisexual peers. More than one-half of gay men and lesbians are college–educated, compared with about 37 % of heterosexual men and women, and even lower percentages of bisexual women (32.1 %) and men (26.5 %). Gay men and lesbians also report the highest annual household incomes, and although gay men are employed at a similar rate as heterosexual men, lesbians report substantially higher rates of employment than heterosexual women (70.3 % vs. 50.7 %). Other studies have shown that bisexual adults are the most socioeconomically disadvantaged (Badgett et al. 2013), and we find a similar pattern in our sample: bisexuals report the lowest education, income, and insurance coverage of any group; and they report the highest rate of missing a doctor visit last year because of the cost (especially bisexual women, at 30.3 %). And although bisexual women also report lower annual household income, on average, than bisexual men, bisexual men are socioeconomically more vulnerable in other ways: compared with bisexual women, they report lower average education and medical insurance coverage. Bisexual men also have a much lower employment rate than either heterosexual or gay men.

Bisexual adults also report relatively high levels of participation in unhealthy behaviors. They report the lowest rates of having a regular doctor or that they received a routine checkup last year, and the highest rates of smoking and heavy drinking. In contrast, heterosexual women report the healthiest profile: their behavior is healthier than heterosexual men on all measures except exercise and drinking (for which rates are similar across the two groups); and relative to lesbian and bisexual women, they report similar or better health behaviors (including much lower rates of smoking and heavy drinking). Gay men are similar to heterosexual men in terms of exercise and heavy drinking, but they report less obesity and higher rates of having a personal doctor and receiving a routine medical care checkup last year.

Finally, Table 1 illustrates the at-risk status of bisexual adults in terms of social networks/support and well-being. Bisexual adults (especially men) report the lowest average life satisfaction, and they report receiving less emotional support than their heterosexual and gay/lesbian peers. In contrast, heterosexual and lesbian women report equivalent rates of emotional support and life satisfaction, but heterosexual men report rates that are only slightly higher than gay men. Heterosexual adults also report the highest rates of marriage (about 60 % of adults). However, although marriage rates are much lower among sexual minorities, they are not trivial. Almost one of three self-identified bisexual women is married (29.1 %) compared with 23.8 % of bisexual men, 20.4 % of lesbians, and 13.3 % of gay men. Gay men and lesbians also report the highest rate of living as a member of an unmarried couple, followed by bisexual adults, with very low rates (about 5 %) among heterosexuals.

### Logistic Regression Models for Poor Self-rated Health

In Table 2, we present the odds ratios of poor health for each gender-by-orientation group, relative to heterosexual men. Model 1 immediately debunks the notion that the intersection of disadvantaged gender and sexual orientation statuses necessarily place adults at a higher risk for poor health: relative to heterosexual men, lesbians and gay men have significantly lower odds of poor health (OR = 0.69 and 0.78, respectively). In contrast, relative to heterosexual men, bisexual men and bisexual women have 2.05 and 1.32 times the odds of poor health, respectively. For heterosexual women, Model 1 shows that they have slightly higher odds of reporting poor SRH than do heterosexual men.

Demographic controls in Model 2 adjust away the slightly higher odds of poor health for heterosexual women and the health advantage for gay men and lesbians, making their odds of poor health statistically indistinguishable from those of heterosexual men. Driven largely by the young ages of bisexuals in our sample, their odds of poor SRH increase in Model 2 relative to heterosexual men (i.e., 2.73 times higher for bisexual men and 2.03 times higher for bisexual women).

Model 3 adds the first set of confounding factors for SES. Compared with Model 2, the results show that the pronounced socioeconomic disadvantages of bisexuals attenuates the odds of reporting poor health for men by 56 % ([2.73 – 1.75] / [2.73 – 1.00] × 100) and for women by 76 %, relative to heterosexual men. The odds for gay men and lesbians are relatively unchanged, and adjusting for SES places heterosexual women at significantly lower odds of poor health compared with heterosexual men.

Model 4 replaces the confounding SES factors with health behavior indicators and shows that these factors are far less influential in shaping gender-by-orientation relationships with poor SRH. Comparing the odds ratios from Model 2 and Model 4 shows that despite the generally poorer health behaviors of gay men, lesbians, and bisexuals (including greater proportions of current smokers), relative to heterosexual men, their odds of poor health change little.2 Model 5 replaces health behaviors with social network/support and well-being indicators and finds patterns for bisexual adults that are similar to those for SES, although less pronounced. Controlling for deficits in these factors for bisexual men and women attenuates their odds of poor health, relative to heterosexual men, by 42 % and 40 %, respectfully. For gay men and lesbians, adjusting for differences (primarily observed in marital status) leaves the two groups at lower odds of reporting poor health, relative to heterosexual men.

The fully adjusted Model 6 shows that relative to heterosexual men, heterosexual women have 12 % lower odds; gay men, lesbians, and bisexual women have similar odds; and bisexual men have 77 % higher odds of reporting poor health (although this association does not reach significance at p = .06). The findings from this model are most similar to those presented in Model 3, illustrating the prominence of SES and demographic characteristics in shaping self-rated health patterns across gender-by-orientation groups.

Finally, in Fig. 1, we present average predicted probabilities generated from the baseline and fully adjusted models to compare men and women within each sexual orientation position and to evaluate gender differences across sexual orientation. First, for heterosexuals, we replicate past research and show that gender differences in self-rated health flip so that in the fully adjusted model, the probability of poor health for men (.157) is significantly greater than that for women (.145).3 However, contrary to this established pattern of health-by-gender among heterosexuals, there are no significant differences in the baseline or fully adjusted models for gay men versus lesbians or for bisexual men versus women.

Second, to evaluate gender differences across sexual orientation, we compare heterosexual men with gay and bisexual men, and then make the same comparisons for women. For men, the baseline model shows that heterosexuals have higher predicted probabilities of poor health than gay men, but this difference is eliminated in the full model. The predicted probability of poor health is significantly higher for bisexual men (.260) than for heterosexual men (.146) and gay men (.118) in the baseline model. In fully adjusted models, those differences are smaller and remain significant only for bisexual men (.221) compared with gay men (.141).

For women, differences are more enduring across sexual orientation. In the baseline model, heterosexuals have a higher predicted probability for poor health (.156) than lesbians (.105) and a lower predicted probability than bisexuals (.185; this contrast is not significant). After full adjustments, the predicted probability of poor health for heterosexual women and lesbians is similar, and they both have significantly lower probabilities of poor health than bisexual women.

Finally, as a sensitivity test for the results presented, we reestimated our baseline and full models but included missing categories for sexual orientation for men and women. In Table 3, we compare odds ratios of these models to the estimates generated with imputed data (as was presented in Table 2). In Table 3, the baseline models that include missing data categories show that missing men and women have significantly higher odds of poor health relative to heterosexual men. But importantly, the substantive interpretation of the other categories remains the same as the odds ratios for each group shifts only slightly. That is also the case in the fully adjusted model, in which the higher odds for missing men are explained away and missing women have lower odds of poor health, although here the contrasts between lesbians and heterosexual men and between bisexual men and heterosexual men emerge as significant. Although we have no other information about the sexual orientation of these missing cases and cannot speculate about their meaning, the results are compelling enough to warrant future investigations into how respondents reply to survey questions on sexual orientation.

## Conclusion

According to the IOM (2011), health research on the sexual minority population is sparse, is often based on nonprobability samples, typically cannot differentiate across types of sexual minority populations, lacks a comparative focus relative to heterosexuals, and usually does not evaluate how sexual minority status interacts with other status-based characteristics (including gender) known to influence health status. We responded to these shortcomings by drawing on aggregated probability samples from the BRFSS and applied an intersectional framework to evaluate whether gender and sexual orientation combined to shape self-rated health status. We tested a series of descriptive hypotheses designed to investigate how the specific intersectional position that one holds relates to self-rated health, as well as evaluative hypotheses designed to assess how risk factors for poor health contribute to gender-by-sexuality patterns in self-rated health.

Our study results both confirm and challenge past findings and reveal important self-rated health disparities both within and between gender-by-orientation subgroups. The descriptive hypotheses predicted that the odds of poor self-rated health would be lowest among heterosexual men; on the contrary, our baseline models showed that self-rated health reports are lowest among lesbians and gay men. These findings demonstrated a self-reported health advantage among lesbians and gay men, which appears driven by high proportions of younger, white, and highly educated gay men and lesbians in our sample—characteristics all shown to positively impact health status across the life course (e.g., Brown et al. 2012). Other work shows that within gender, self-rated health reports among gay men and lesbians are equivalent to heterosexuals (Conron et al. 2010; Dilley et al. 2010), but these findings were drawn from single-state studies; as such, the somewhat broader geographic representation of our sample may be driving these more-positive health assessments. Black et al. (2000) reported similar demographic results using the General Social Survey (GSS), but they cautioned that their results could be driven by well-educated persons being more willing to identify as gay or lesbian. We concur and think that, particularly for a survey conducted over the telephone (such as the BRFSS), highly educated gays and lesbians may be more apt to identify as such, raising questions about how to best collect information on sexual identity. The addition of sexual orientation on face-to-face surveys, such as the National Health Interview Survey (NHIS), will allow future research to assess the accuracy of different data collection methods, and we encourage this comparative research as probability-based data that include information on sexual orientation become available. Overall, we contribute to recent laudable attempts to understand the intersections of status for health (see Bowleg 2012; Veenstra 2011, 2013) and highlight how health differences may be misrepresented if based on only gender and not sexual orientation.

We found strong support for disproportionate levels of poor health among bisexuals: results showed that bisexual men and women report poorer health than their heterosexual and gay and lesbian counterparts. This finding is similar to recent work based on eight years of BRFSS data from Massachusetts (Conron et al. 2010); however, it differs from findings by Dilley and colleagues (2010), who used four years of BRFSS data from Washington state and found that reports of poor self-rated health were elevated among bisexual women, but not bisexual men. These results along with our own suggest that sociopolitical and sociocultural environments may be important factors to consider in future work.

Furthermore, our results confirm past research and provide new evidence for gender differences in health. As others have (Ross and Bird 1994), we found that heterosexual women report poor SRH at a higher rate than heterosexual men at baseline and that SES adjustments switch those higher odds to lower odds. However, adjusting for social networks/support and well-being did little to influence the gender gap for heterosexuals, but adjusting for health behaviors reduced the gap to nonsignificance. By contrast, for sexual minorities, we found that gay and bisexual men, relative to lesbians and bisexual women, have higher probabilities of poor SRH, although the differences were not significant. These patterns are consistent with research describing more gender equality among sexual minorities (e.g., Reczek and Umberson 2012) and suggests that gender may play less of a role in distinguishing health status among sexual minority adults than it does among heterosexual adults. And although not displayed in Fig. 1, once again we found that SES adjustments proved to have the greatest effect on the probabilities of poor SRH across groups, showing suppression of associations with poor health among gay men and lesbians and attenuation for bisexual men and women.

Altogether, our findings suggest that the major mechanisms contributing to health differences for sexual minorities are similar to the mechanisms that are contributing to health disparities more generally. In particular, bisexual men and women are disproportionately disadvantaged on important social, economic, and behavioral factors that are strongly associated with health and well-being. Although the impact of these disadvantages has been described, their origins among bisexuals remains largely unknown. If bisexuals are minorities within the minority and experience unique and extreme forms of discrimination, that may contribute to disparities in, for example, earnings, educational attainment, the propensity to smoke cigarettes, and ultimately in overall assessments of well-being (IOM 2011). Although beyond the ability of our study to evaluate, determining how much disadvantage in physical health among sexual minority men and women, and particularly bisexuals, is carried in processes of stress, stigma, and discrimination will be the task of much future population-level work. Indeed, that our models were more successful in reducing the self-rated health gap for bisexual men than women may be attributable to our inability to adjust for stress-related factors, such as depression and anxiety, which women experience at higher rates than men (Read and Gorman 2010).

The study of sexual minority health at a population level is in its infancy, especially when examined with an intersectional perspective (IOM 2011). Our study makes important contributions to this small but growing field of research, but our use of cross-sectional data limits our ability to establish causal pathways (including teasing out the confounding vs. mediational role of SES, health behaviors, and social networks/support for gender-by-orientation disparities in health status). And although we were able to aggregate a large probability-drawn sample of more than 10,000 sexual minority adults, the limited geographic scope of our seven-state study overrepresents states with larger sexual minority populations, some with more favorable laws regarding sexual minority rights (Gates 2011; Gates and Ost 2004). As such, our findings might underrepresent health disparities across sexual orientation groups and need replication with national-level data. We also recognize limitations in our identification of sexual minorities via a single measure of self-identified sexual orientation given that sexual orientation is a multifaceted concept that includes components of identity, attraction, and behavior (McCabe et al. 2009). For example, a woman may be involved romantically with another woman yet not identify herself as a lesbian (Laumann et al. 1994). In the face of current data restrictions, our work provides important insights and a solid base to build upon. However, moving forward, calls to include multiple measures of sexual orientation in national health surveys should be implemented if we hope to achieve a more complete understanding of sexual minority health in population perspective.

In closing, we drew on intersectional theory to examine how gender and sexual identity intersect to pattern self-rated health reports—an important strength of our article as scholars have noted that conceptualizations of sexual identity as a status characteristic has little meaning when considered independently of gender (Cole 2009). We see this focus as a meaningful contribution because research on intersectionality has developed separately from dominant paradigms in health disparities research, including fundamental cause theory and social ecology (Krieger 2001; Link and Phelan 1995). Yet, as others have noted (Weber and Parra-Medina 2003), the application of an intersectional framework has much potential to generate new knowledge that can be used to more effectively guide policy actions aimed at reducing health disparities. By focusing on the intersections of gender and sexual identity, we highlighted that it is overly simplistic and inaccurate to conclude that women report poorer health than men (a statement that applied to only heterosexuals in our sample). Further, examining “sexual minorities” conflates the health status of gay/lesbian and bisexual adults, which is a substantial problem given the poorer health status of bisexual adults and the manner in which they experience a significant elevation in risk factors associated with poor health. Going forward, research would do well to collect new, probability-based data on sexual orientation and health, in addition to using existing data sources in creative ways. This would encourage better understanding of intersections not only between sexual orientation and gender, but also with age, race/ethnicity, and other status-based characteristics shown to influence health outcomes.

## Acknowledgments

A previous version of this article was presented at the 2012 annual meeting of the Southern Demographic Association. We thank Erin Cech for her helpful comments during the writing of this article. The second author acknowledges support from the Health Disparities Scholar Program, National Institute on Minority Health and Health Disparities, National Institutes of Health. The opinions and conclusions expressed herein are solely those of the authors and should not be construed as representing the opinions or policies of the National Institutes of Health.

## Notes

1

We include only state survey-years in which at least some respondents were asked all the questions used to form the variables in our regression models.

2

In supplementary analyses, we entered the health behaviors variables one at a time and found that accounting for smoking reduced the odds of poor health for bisexual men and women by 11 % (from 2.73 to 2.53) and 8 % (from 2.03 to 1.94), respectively. No other substantial changes were observed in these supplementary analyses.

3

Each significance test evaluates the pairwise comparison for the average marginal effect (AME) of being a member of Group 2 versus being a member of Group 1 (in this example, heterosexual woman vs. heterosexual man). AME uses all observed data for all covariates in the full model (rather than means on all covariates) to predict average probabilities for a hypothetical population of all heterosexual men (Group 1) and a hypothetical population of all heterosexual women (Group 2). In this probability framework, the AME for Group 2 versus Group 1 is .145 – .157 = –.012; SE = .003; t = –3.06; p = .002.

## References

Allison, P. D. (
2002
).
Missing data
.
London, UK
:
Sage Publications
.
Austin, E., & Irwin, J. (
2010
).
Health behaviors and health care utilization of Southern lesbians
.
Women's Health Issues
,
20
,
178
184
. 10.1016/j.whi.2010.01.002
Badgett, M. V. L., Durso, L. E., & Schneebaum, A. (
2013
).
New patterns of poverty in the lesbian, gay, and bisexual community
.
Los Angeles, CA
:
The Williams Institute
.
Bartus, T. (
2005
).
Estimation of marginal effects using margeff
.
Stata Journal
,
5
,
309
329
.
Bauer, G. R. (
2014
).
Incorporating intersectionality theory into population health research methodology: Challenges and the potential to advance health equity
.
Social Science & Medicine
,
110
,
10
17
. 10.1016/j.socscimed.2014.03.022
Berkman, L. F., & Glass, T. (
2000
).
Social integration, social networks, social support, and health
. In Berkman, L. F., & Kawachi, I. (Eds.),
Social epidemiology
(pp.
137
173
).
Oxford, UK
:
Oxford University Press
.
Bird, C. E., & Rieker, P. P. (
2008
).
Gender and health: The effects of constrained choices and social policies
.
New York, NY
:
Cambridge University Press
.
Black, D., Gates, G., Sanders, S., & Taylor, L. (
2000
).
Demographics of the gay and lesbian population in the United States: Evidence from available systematic data sources
.
Demography
,
37
,
139
154
. 10.2307/2648117
Bowleg, L. (
2012
).
The problem with the phrase women and minorities: Intersectionality—An important theoretical framework for public health
.
American Journal of Public Health
,
102
,
1267
1273
. 10.2105/AJPH.2012.300750
Brennan, D. J., Ross, L., Dobinson, C., Veldhuizen, S., & Steele, L. (
2010
).
Men’s sexual orientation and health in Canada
.
Canadian Journal of Public Health
,
101
,
255
258
.
Brown, D., Hayward, M. D., Hummer, R. M., & Montez, J. K. (
2012
).
The significance of education for mortality compression in the United States
.
Demography
,
49
,
819
840
. 10.1007/s13524-012-0104-1
Centers for Disease Control and Prevention (CDC)
. (
2005–2010
).
The BRFSS data user guide.
Centers for Disease Control and Prevention (CDC)
. (
2012
).
Chang, M. L. (
2010
).
Shortchanged: Why women have less wealth and what can be done about it
.
Oxford, UK
:
University Press
.
Cho, S., Crenshaw, K. W., & McCall, L. (
2013
).
Toward a field of intersectionality studies: Theories, applications, and praxis
.
Signs
,
38
,
785
810
. 10.1086/669608
Choo, H. Y., & Ferree, M. M. (
2010
).
Practicing intersectionality in sociological research: A critical analysis of inclusions, interactions, and institutions in the study of inequalities
.
Sociological Theory
,
28
,
129
149
. 10.1111/j.1467-9558.2010.01370.x
Cochran, S. D., Mays, V., Bowen, D., Gage, S., Bybee, D., Roberts, S. . . . White, J. (
2001
).
Cancer-related risk indicators and preventive screening behaviors among lesbian and bisexual women
.
American Journal of Public Health
,
91
,
591
597
.
Cole, E. R. (
2009
).
Intersectionality and research in psychology
.
American Psychologist
,
64
,
170
180
. 10.1037/a0014564
Collins, P. H. (
2000
).
Black feminist thought: Knowledge, consciousness, and the politics of empowerment
.
New York, NY
:
Routledge
.
Connell, R. (
2012
).
Gender, health, and theory: Conceptualizing the issue, in local and world perspective
.
Social Science & Medicine
,
74
,
1675
1683
. 10.1016/j.socscimed.2011.06.006
Conron, K. J., Mimiaga, M., & Landers, S. (
2010
).
A population-based study of sexual orientation identity and gender differences in adult health
.
American Journal of Public Health
,
100
,
1953
1960
. 10.2105/AJPH.2009.174169
Denney, J. T., Gorman, B. K., & Barrera, C. B. (
2013
).
Families, resources, and adult health: Where do sexual minorities fit?
.
Journal of Health and Social Behavior
,
54
,
46
63
. 10.1177/0022146512469629
Dilley, J. A., Simmons, K., Boysun, M., Pizacani, B., & Stark, M. (
2010
).
Demonstrating the importance and feasibility of including sexual orientation in public health surveys: Health disparities in the Pacific Northwest
.
American Journal of Public Health
,
100
,
460
467
. 10.2105/AJPH.2007.130336
Gallup
. (
2010
.)
Gay and lesbian rights
. Retrived from http://www.gallup.com/poll/1651/Gay-Lesbian-Rights.aspx
Gates, G. J. (
2011
).
Census snapshot: 2010 methodology. Adjustment procedures for same-sex couple data
.
Los Angeles, CA
:
The Williams Institute
Gates, G. J., & Ost, J. (
2004
).
The gay and lesbian atlas
.
Washington, DC
:
The Urban Institute Press
.
Gorman, B. K., & Read, J. G. (
2006
).
Gender disparities in adult health: An examination of three measures of morbidity
.
Journal of Health and Social Behavior
,
47
,
95
110
. 10.1177/002214650604700201
Gorman, B. K., & Sivaganesan, A. (
2007
).
The role of social support and integration for understanding socioeconomic disparities in self-rated health and hypertension
.
Social Science & Medicine
,
65
,
958
975
. 10.1016/j.socscimed.2007.04.017
Green, K. E., & Feinstein, B. A. (
2012
).
Substance use in lesbian, gay, and bisexual populations: An update on empirical research and implications for treatment
.
Psychology of Addictive Behaviors
,
26
,
265
278
. 10.1037/a0025424
Hankivsky, O. (
2012
).
Women’s health, men’s health, and gender and health: Implications of intersectionality
.
Social Science & Medicine
,
74
,
1712
1720
. 10.1016/j.socscimed.2011.11.029
Hankivsky, O., & Cormier, R. (
2009
).
Intersectionality: Moving women’s health research and policy forward
.
:
Women’s Health Research Network
.
Harnois, C. E., & Ifatunji, M. (
2011
).
Gendered measures, gendered models: Toward and intersectional analysis of interpersonal racial discrimination
.
Ethnic and Racial Studies
,
34
,
1006
1028
. 10.1080/01419870.2010.516836
Herek, G. M. (
2006
).
Legal recognition of same-sex relationships in the United States: A social science perspective
.
American Psychologist
,
61
,
607
621
. 10.1037/0003-066X.61.6.607
Herek, G. M., Chopp, R., & Strohl, D. (
2007
).
Sexual stigma: Putting sexual minority health issues in context
. In Meyer, I. H., & Northridge, M. E. (Eds.),
The health of sexual minorities: Public health perspectives on lesbian, gay, bisexual, and transgender populations
(pp.
171
208
).
New York, NY
:
Springer
.
Idler, E. L., & Benyamini, Y. (
1997
).
Self-rated health and mortality: A review of twenty-seven community studies
.
Journal of Health and Social Behavior
,
38
,
21
37
. 10.2307/2955359
Institute of Medicine (IOM)
. (
2011
).
The health of lesbian, gay, bisexual, and transgender people: Building a foundation for better understanding
.
Washington, DC
:
National Institutes of Health
.
Jorm, A. F., Korten, A. E., Rodgers, B., Jacomb, P. A., & Christensen, H. (
2002
).
Sexual orientation and mental health: Results from a community survey of young and middle-aged adults
.
British Journal of Psychiatry
,
180
,
423
427
. 10.1192/bjp.180.5.423
Jylhä, M. (
2009
).
What is self-rated health and why does it predict mortality? Towards a unified conceptual model
.
Social Science & Medicine
,
69
,
307
316
. 10.1016/j.socscimed.2009.05.013
Kastanis, A., & Gates, G. J. (
2013a
).
LGBT African-Americans and African-American same-sex couples
.
Los Angeles, CA
:
The Williams Institute
Kastanis, A., & Gates, G. J. (
2013b
).
LGBT Latino/a individuals and Latino/a same-sex couples.
Los Angeles, CA
:
The Williams Institute
Kastanis, A., & Gates, G. J. (
2013c
).
LGBT Asian and Pacific Islander individuals and same-sex couples
.
Los Angeles, CA
:
The Williams Institute
Korn, E., & Graubard, B. (
1999
).
Analysis of health surveys
.
New York, NY
:
John Wiley & Sons, Inc.
.
Krieger, N. (
2001
).
Theories for social epidemiology in the 21st century: An ecosocial perspective
.
International Journal of Epidemiology
,
30
,
668
677
. 10.1093/ije/30.4.668
Langkamp, D. L., Lehman, A., & Lemeshow, S. (
2010
).
Techniques for handling missing data in secondary analyses of large surveys
.
,
10
,
205
210
. 10.1016/j.acap.2010.01.005
Laumann, E. O., Gagnon, J. H., Michael, R. T., & Michaels, S. (
1994
).
The social organization of sexuality: Sexual practices in the United States
.
Chicago, IL
:
University of Chicago Press
.
Link, B., & Phelan, J. (
1995
).
Social conditions as fundamental causes of disease
.
Journal of Health and Social Behavior
,
35
(Extra issue),
80
94
.
Liu, H., Reczek, C., & Brown, D. (
2013
).
Same-sex cohabitors and health the role of race-ethnicity, gender, and socioeconomic status
.
Journal of Health and Social Behavior
,
54
,
25
45
. 10.1177/0022146512468280
Marmot, M. G. (
2004
).
The status syndrome: How social standing affects our health and longevity
.
New York, NY
:
Henry Holt and Company, LLC.
.
McCabe, S. E., Hughes, T. L., Bostwick, W. B., West, B. T., & Boyd, C. J. (
2009
).
Sexual orientation, substance use behaviors and substance dependence in the United States
.
,
104
,
1333
1345
. 10.1111/j.1360-0443.2009.02596.x
McCall, L. (
2005
).
The complexity of intersectionality
.
Signs
,
30
,
1771
1800
. 10.1086/426800
Meyer, I. H. (
2007
).
Prejudice and discrimination as social stressors
. In Meyer, I. H., & Northridge, M. E. (Eds.),
The health of sexual minorities: Public health perspectives on lesbian, gay, bisexual, and transgender populations
(pp.
242
267
).
New York, NY
:
Springer
.
Pampel, F. C., Krueger, P. M., & Denney, J. T. (
2010
).
Socioeconomic disparities in health behaviors
.
Annual Review of Sociology
,
36
,
349
370
. 10.1146/annurev.soc.012809.102529
Pascoe, C. J. (
2007
).
Dude, you’re a fag: Masculinity and sexuality in high school
.
Berkeley, CA
:
University of California Press
.
Powell, B., Bolzendahl, C., Geist, C., & Steelman, L. (
2010
).
Counted out: Same-sex relations and Americans’ definitions of family
.
New York, NY
:
Russell Sage Foundation
.
Read, J. G., & Gorman, B. (
2010
).
Gender and health inequality
.
Annual Review of Sociology
,
36
,
371
386
. 10.1146/annurev.soc.012809.102535
Reczek, C., & Umberson, D. (
2012
).
Gender, health behavior, and intimate relationships: Lesbian, gay, and straight contexts
.
Social Science & Medicine
,
74
,
1783
1790
. 10.1016/j.socscimed.2011.11.011
Roberts, S. A., Dibble, S., Nussey, B., & Casey, K. (
2003
).
Cardiovascular disease risk in lesbian women
.
Women's Health Issues
,
13
,
167
174
. 10.1016/S1049-3867(03)00041-0
Ross, C., & Bird, C. (
1994
).
Sex stratification and health lifestyle: Consequences for men’s and women’s perceived health
.
Journal of Health and Social Behavior
,
35
,
161
178
. 10.2307/2137363
Singh-Manoux, A., Martikainen, P., Ferrie, J., Zins, M., Marmot, M., & Goldberg, M. (
2006
).
What does self-rated health measure? Results from the British Whitehall II and French Gazel cohort studies
.
Journal of Epidemiology and Community Health
,
60
,
364
372
. 10.1136/jech.2005.039883
Springer, K. W., Hankivsky, O., & Bates, L. M. (
2012
).
Gender and health: Relational, intersectional, and biosocial approaches
.
Social Science & Medicine
,
74
,
1661
1666
. 10.1016/j.socscimed.2012.03.001
StataCorp.
(
2011
).
Stata statistical software: Release 12
.
College Station, TX
:
StataCorp LP
.
Swim, J. K., & Hyers, L. I. (
2009
).
Sexism
. In Nelson, T. D. (Ed.),
Handbook of prejudice, stereotyping, and discrimination
(pp.
407
430
).
New York, NY
:
Psychology Press
.
Tang, H., Greenwood, G. L., Cowling, D. W., Lloyd, J. C., Roeseler, A. G., & Bal, D. G. (
2004
).
Cigarette smoking among lesbians, gays, and bisexuals: How serious a problem? (United States)
.
Cancer Causes and Control
,
15
,
797
803
. 10.1023/B:CACO.0000043430.32410.69
Umberson, D., Chen, M. D., House, J. S., Hopkins, K., & Slaten, E. (
1996
).
The effect of social relationships on psychological well-being: Are men and women really so different?
.
American Sociological Review
,
61
,
837
857
. 10.2307/2096456
U.S. Department of Health and Human Services (U.S. DHHS) and U.S. Department of Agriculture
. (
2005
).
Dietary guidelines for Americans, 2005
. (6th ed.).
Washington, DC
:
U.S. Government Printing Office
.
Veenstra, G. (
2011
).
Race, gender, class, and sexual orientation: Intersecting axes of inequality and self-rated health in Canada
.
International Journal for Equity in Health
,
10
(
3
). doi:10.1186/1475-9276-10-3
Veenstra, G. (
2013
).
Race, gender, class, sexuality (RGCS) and hypertension
.
Social Science & Medicine
,
89
,
16
24
. 10.1016/j.socscimed.2013.04.014
Von Hippel, P. T. (
2007
).
Regression with missing Ys: An improved strategy for analyzing multiply imputed data
.
Sociological Methodology
,
37
,
83
117
. 10.1111/j.1467-9531.2007.00180.x
Weber, L., & Parra-Medina, D. (
2003
).
Intersectionality and women’s health: Charting a path to eliminating health disparities
.
Advances in Gender Research
,
7
,
181
230
. 10.1016/S1529-2126(03)07006-1
Williams, D. R., & Sternthal, M. (
2010
).
Understanding racial-ethnic disparities in health: Sociological contributions
.
Journal of Health and Social Behavior
,
51
,
S15
S27
. 10.1177/0022146510383838
Wu, Z., Penning, M. J., Pollard, M. S., & Hart, R. (
2003
).
“In sickness and in health”: Does cohabitation count?
.
Journal of Family Issues
,
24
,
811
838
. 10.1177/0192513X03254519