Abstract

Though research suggests that sexual minorities (e.g., nonheterosexual individuals) are more geographically mobile in the transition to adulthood than their heterosexual counterparts, quantitative estimates are rare and previously used data sources have significant limitations. Using data from the National Longitudinal Study of Adolescent to Adult Health (N = 11,705) that directly measure sexualities across dimensions (i.e., identity, behavior, and attraction), I examine variation in geographic mobility between childhood (ages 11–17) and adulthood (ages 26–34) across various sexualities (e.g., gay/lesbian and bisexual). Three findings emerge. First, mobility varies across sexualities. Individuals with gay/lesbian identity, same-sex behavior, and same-sex attraction are more geographically mobile than individuals with heterosexual identity, different-sex behavior, and different-sex attraction, respectively. By contrast, individuals with bisexual identity, both-sex behavior, and both-sex attraction tend to be statistically indistinct from individuals with heterosexual identity, different-sex behavior, and different-sex attraction, respectively. Second, mobility differences are largest and most prevalent when sexualities are operationalized according to identity. Third, evidence suggests that the effects of gay/lesbian identity, same-sex behavior, and same-sex attraction on mobility are larger for men than for women. In providing the first quantitative estimates of geographic mobility differences across broader sexual minority and heterosexual populations, this study expands inquiry related to sexualities and mobility.

Introduction

Young people move for a variety of reasons: for example, to go to college, to establish independence, to seek employment, or to start a new household (Horowitz and Entwisle 2021; Mulder 2009). Missing from most of these accounts, however, is the consideration of sexual orientation.1 Despite this lack of attention, there are several reasons to expect that sexual minorities (e.g., nonheterosexual individuals) are more geographically mobile than their heterosexual counterparts between childhood and adulthood. Sexual minorities may be more likely to move away from their communities of origin to avoid rejection or stigmatization (Gorman-Murray 2009; Lewis 2014), evade surveillance from family (Gorman-Murray 2009; Lewis 2014; Luo 2022; Rosenfeld 2007; Rosenfeld and Kim 2005; Wimark 2016b), or increase their chances of finding a partner (Carillo 2017; Gorman-Murray 2009). They may possess weaker ties to their place of origin (Hatzenbuehler et al. 2012; Ueno 2005) or may be especially mobile to live among others who share their sexual orientation and life experiences (Compton and Baumle 2012; Ghaziani 2014; Gorman-Murray 2009).

While the geographic distribution of sexualities in the United States has received extensive scholarly attention (see Brown-Saracino 2019, Ghaziani 2019, and Kazyak 2020 for recent reviews), variation in geographic mobility across sexualities has received comparatively little scrutiny. Quantitative analyses are especially rare. Relying on U.S. Census data, these analyses find that individuals in same-sex couples are more geographically mobile between birth and young adulthood than their different-sex coupled counterparts (Rosenfeld 2007; Rosenfeld and Kim 2005). However, census data have major limitations. Most importantly, the census does not measure sexual orientation directly. Instead, we can only surmise someone's sexual orientation from whether they coreside with a same- or different-sex partner. Differences in geographic mobility between the broader young adult sexual minority and heterosexual populations—regardless of relationship status—are therefore unclear. Moreover, variation in these differences across dimensions of sexual orientation (i.e., identity, behavior, and attraction), specific sexualities (e.g., bisexual and gay/lesbian individuals), and other characteristics (e.g., gender) remains unexplored.

Several questions emerge. First, how does geographic mobility between childhood and adulthood vary across sexualities? Second, what individual, household, parent, and geographic characteristics best explain these differences? Third, how do mobility gaps vary across dimensions of sexual orientation, sexualities, and other characteristics? For example, are mobility differences between heterosexuals and gay/lesbian individuals similar to those between heterosexuals and bisexuals?

To answer these questions, I examine the relationship between sexual orientation and geographic mobility between childhood and adulthood, as well as the factors that mediate or moderate this relationship. Analyses use nationally representative longitudinal data from the National Longitudinal Study of Adolescent to Adult Health (Add Health). Representing detailed information about respondents, their sexualities, and their residences over time, these data enable robust examination of geographic mobility across various operationalizations of sexual orientation.

Results reveal substantial variation in geographic mobility in the transition to adulthood across sexualities. Individuals with gay/lesbian identity, same-sex behavior, and same-sex attraction live farther away from their communities of origin and are more likely to move counties between childhood (ages 11–17) and adulthood (ages 26–34) than their heterosexual, different-sex-behaving, and different-sex-attracted counterparts, respectively. By contrast, geographic mobility differences between individuals with bisexual and heterosexual identity, both- and different-sex behavior, and both- and different-sex attraction tend to be small and statistically indistinct. These findings are robust across various model specifications. Moreover, I find mobility differences vary across operationalizations of sexualities and gender. Specifically, mobility gaps are largest according to sexual identity, rather than sexual behavior or attraction, and evidence suggests the effects of gay/lesbian identity, same-sex behavior, and same-sex attraction on mobility are larger for men than women

To my knowledge, this is the first study to quantitatively estimate geographic mobility differences between the broader sexual minority and heterosexual populations. Adopting a life course perspective, this study highlights mobility differences specifically between childhood and adulthood, advancing knowledge about the ways in which sexualities shape mobility in the transition to adulthood.

Literature Review

The Role of Sexual Orientation in Shaping Geographic Mobility

The transition to adulthood is marked by several life changes, including moving away from one's community of origin (Shanahan 2000). Investigations of geographic mobility during this life stage, however, often neglect the role of sexual orientation (Horowitz and Entwisle 2021; Mulder 2009).

Past research offers several reasons to expect sexual minorities to experience greater geographic mobility between childhood and adulthood than their heterosexual counterparts. For one, sexual minorities might move at a higher rate to avoid stigmatization, violence, or discrimination in their communities of origin. Despite growth in support for gay rights and acceptance of sexual minorities, prejudice remains. In 2020, about a quarter of Americans believed gay and lesbian relations should be illegal (Gallup Organization 2020). Perhaps unsurprisingly, sexual minorities experience comparatively high rates of school bullying, early-life adversity (e.g., child abuse, expulsion from the home), and hiring and housing discrimination (Friedman et al. 2013; McLaughlin et al. 2012; Mishel 2016; Mittleman 2019; Tilcsik 2011). Importantly, these attitudes and experiences vary geographically, potentially inducing the migration of sexual minorities from more to less stigmatizing places. Qualitative evidence supports this idea, finding that desire for comfort and safety often motivate moves of young adult sexual minorities (Gorman-Murray 2009; Lewis 2014; Wimark 2016a).

Concerns of stigmatization may implicate sexual minorities' families of origin as well. On one hand, sexual minorities may be more likely than heterosexuals to experience disapproval of their sexual orientation from family members (Katz-Wise and Hyde 2012; Rosario et al. 2014). As a result, sexual minorities may be especially mobile to avoid scrutiny from their families and reduce their families' control over their romantic lives (Gorman-Murray 2009; Lewis 2014; Luo 2022; Rosenfeld 2007; Rosenfeld and Kim 2005; Wimark 2016b). On the other hand, sexual minorities might move to protect their families from the stigma of having a sexual minority child (Carillo 2017).

Compared with their heterosexual counterparts, sexual minorities tend to be less socially integrated during adolescence and early adulthood. Adolescent sexual minorities experience greater social isolation and weaker attachments to school than their heterosexual peers (Hatzenbuehler et al. 2012; Ueno 2005), and they report relatively weaker ties and worse relationships with parents through young adulthood (Needham and Austin 2010; Pearson and Wilkinson 2013). Sexual minorities may, thus, possess weaker ties to their communities of origin, opening the door for greater and farther migration.

Educational attainment is one avenue sexual minorities may gravitate toward to minimize stigmatization and maximize social acceptance. On one hand, sexual minorities may exert greater effort in school than heterosexuals to boost their chances of escaping stigmatization in their place of origin. On the other hand, they may be more likely to pursue higher education to counterbalance anticipated discrimination in the labor market or because they believe college environments are relatively accepting (Hewitt 1995; Mollborn and Everett 2015). These behaviors, in turn, may increase the residential mobility (e.g., to college) and horizons of sexual minorities.

Finally, sexual minorities may migrate at comparatively higher rates because of homophily, the tendency of individuals with similarities to associate with one another (McPherson et al. 2001). Adult sexual minorities are highly geographically concentrated (Newport and Gates 2015). As a result, adolescent sexual minorities might move away from their communities of origin at relatively high rates to live among a critical mass of others who share their sexual orientation, interests, and life experiences. Indeed, sexual minorities continue to feel drawn to areas with sexual minority institutions and high sexual minority concentrations (Compton and Baumle 2012; Ghaziani 2014; Gorman-Murray 2009).

Each of these threads leads to the same main hypothesis:

  • Hypothesis 1: Sexual minorities are more likely than heterosexuals to experience geographic mobility between childhood and adulthood.

Estimates of Geographic Mobility Differences Across Sexualities

Despite these foundations, we know little about migration differences between sexual minorities and heterosexuals. Prior work related to sexualities and geographic mobility is largely qualitative. Study samples are limited to sexual minorities—typically, gay men—and research questions tend to concern transnational migration (e.g., Cantú 2009; Carillo 2017; Stone 2007).

Few studies examine sexualities and geographic mobility quantitatively. Cooke (2005) found no relationship between mobility and the working behaviors of same-sex couples, while Cooke and Rapino (2007) uncovered gender differences in the mobility patterns of same-sex partnered men and women. By restricting analyses to same-sex couples, however, they did not estimate mobility differences across sexualities. Baumle et al. (2009) found greater interstate mobility between 1995 and 2000 among same-sex partnered men and women than different-sex married men and women. However, they did not analyze geographic mobility specifically between childhood and adulthood.

Only Rosenfeld and Kim (2005) estimated sexual orientation differences in mobility between childhood and adulthood. In their examination of the independent life stage and its consequences for union formation (later reproduced in Rosenfeld 2007), they found that young same-sex couples (aged 20–29) were more likely than their different-sex coupled counterparts to live in a state different than their birth state. According to 2000 census data, the odds of interstate mobility were about 1.23 times as high among same-race, same-sex couples than among married, same-race, different-sex couples. Young interracial, same-sex couples were the most mobile: the odds of interstate mobility among such couples were about 2.05 times as high among their married, same-race, different-sex coupled peers.

Quantitative studies of sexualities and geographic mobility rely on census data. Despite very large sample sizes, such data have important limitations. First, census measurement of sexualities is crude. Census data can only identify individuals in same-sex coresidential couples, thereby concealing an unknown fraction—possibly more than half (Jones 2017)—of the full sexual minority population and introducing potential selection bias. Using census data also makes it impossible to distinguish among dimensions of sexual orientation (i.e., identity, behavior, and attraction) and specific sexualities (e.g., gay/lesbian and bisexual). For example, census data cannot capture the nuance of belonging to a same-sex couple but identifying as bisexual. Second, same-sex couple measures derived from 1990 and 2000 census data are prone to measurement error. Survey question language and tabulation procedures in these years likely overestimated the number of same-sex couples; up to 40% of same-sex couples identified in the 2000 census were likely misclassified different-sex couples (Black et al. 2000, 2007; O'Connell and Gooding 2007). Finally, census measures of geographic mobility are narrow, limiting analysis of moves between specific life stages or geographic units, such as counties. For instance, interstate mobility between birth and young adulthood—as used in Rosenfeld and Kim (2005) and Rosenfeld (2007)—does not capture county moves within a state and could include moves between birth and adolescence instead of adolescence and adulthood.

Given these limitations, prior estimates are likely distorted representations of mobility differences between the broader young adult sexual minority and heterosexual populations. Moreover, prior work does not attend to the wide range of sexualities and operationalizations of sexual orientation, obscuring potential variation and meaningful facets of the relationship between sexualities and mobility.

Uncovering Variation in the Relationship Between Sexualities and Geographic Mobility

Sexual orientation is typically operationalized according to three dimensions: identity, behavior, and attraction. Sexual identity refers to the labels individuals use to describe themselves, sexual behavior refers to sexual activity, and sexual attraction refers to sexual interest. These dimensions do not always align (Laumann et al. Michaels 1994; Silva 2019; Ward 2015). In fact, recent estimates indicate that only about 62% of gay men and lesbian women report same-sex-only attraction, and about 7% of heterosexual-identified men and women report engaging in same-sex behavior in their lifetimes (Mishel 2019).

Why might sexual orientation differences vary by dimension? For one, individuals who share a social identity tend to share attitudes and behaviors (Stets and Burke 2006). Compared with sexual behavior or attraction, sexual identity may shape the life course more strongly, particularly if moving is a common behavior (or perceived behavior) among individuals who identify as sexual minorities. Further, sexual identity may be constituted through place or migration (Brown-Saracino 2017; Gorman-Murray 2009; Kazyak 2011; Knopp 2004; Weston 1995). For example, Brown-Saracino (2017) found that sexual minority women articulate distinct sexual identities depending on the environment in which they live, while Gorman-Murray (2009) found that gay men and lesbian women often move to “come out of the closet” and identify as gay/lesbian. Putting aside claims about causal order, the evidence suggests relatively stronger associations between sexual identity and geographic mobility.

  • Hypothesis 2: Geographic mobility differences between sexual minorities and heterosexuals are larger when sexualities are operationalized according to sexual identity than to sexual behavior or sexual attraction.

Sexual orientation differences in migration may vary by subgroup as well. About 50% of gay men and lesbian women believe their sexual orientation makes a difference in their lives, compared with only 30% of bisexuals (Pew Research Center 2013a). Hence, sexual orientation may guide the behavior of gay men and lesbian women to a greater degree than for bisexuals. Relatedly, compared with gay men and lesbian women, bisexuals tend to report fewer negative stigma-related experiences, such as violence, discrimination, and verbal abuse (Herek 2009), potentially reducing their motivations to move. Lastly, bisexual men and women tend to have worse educational outcomes (Mollborn and Everett 2015) and experience greater wage penalties (Mize 2016) than do gay men and lesbian women. Even if bisexuals have the same desire and motivations to move as gay men and lesbian women, they may not have the resources to do so at the same rate.

  • Hypothesis 3: Geographic mobility differences between gay/lesbian-identified, same-sex-behaving, and same-sex-attracted individuals and heterosexual-identified, different-sex-behaving, and different-sex-attracted individuals are larger than those between bisexual-identified, both-sex-behaving, and both-sex-attracted individuals and heterosexual-identified, different-sex-behaving, and different-sex-attracted individuals.

Because of asymmetry in attitudes about men and women's sexuality (England 2016; Mize and Manago 2018a; Pascoe 2007), sexual orientation gaps may be larger among men than women. Because sexual minority status carries greater stigma for men than for women, on average (Herek 2009; Mize and Manago 2018a, 2018b; Pew Research Center 2013b), it is possible that sexual minority men experience stronger motivation to move to avoid stigmatization. Moreover, the relationship between sexualities and migration may vary by gender because of gender differences in sexualities themselves. Women are more likely than men to be bisexual or have both-sex behavior or attraction, while men are more likely than women to be gay or have same-sex behavior or attraction (Mishel 2019). Women also appear to be more sexually “fluid” than men, oscillating to a greater degree between sexual identities, behaviors, and attractions (Diamond 2008; Kaestle 2019; Mock and Eibach 2012; Savin-Williams et al. 2012). If the “bar” for sexual minority identity, behavior, or attraction is different for women and men, sexual minority women and men might differ in other ways associated with mobility.

  • Hypothesis 4: Sexual orientation differences in geographic mobility between childhood and adulthood are larger among men than women.

In addition to producing estimates of geographic mobility differences between the broader sexual minority and heterosexual populations, then, this study aims to examine variation in the relationship between sexualities and mobility.

Data

This study uses data from Add Health, a longitudinal study of a nationally representative sample of adolescents in the United States. Respondents were recruited from high schools and followed for over two decades. Respondents were between the ages of 11 and 21 during Wave I (1994–1995) and between 24 and 34 during Wave IV (2008). Questionnaires gathered various information about respondents' health and well-being. Survey administration occurred in respondents' homes through the use of audio computer-assisted self-interview (ACASI) software, which might reduce reporting bias on sensitive questions (Brown et al. 2013). Response rates for Waves I and IV are 79% and 80.3%, respectively. Attrition across survey waves is small and incurs negligible nonresponse bias (Harris 2013).

Parent and geographic context data are also available. A parent—typically, the resident mother or father of each respondent—was asked to complete an in-home questionnaire during Wave I. This questionnaire collected information about respondents' childhood households and parents. Merged data from such external sources as the U.S. Census and American Community Survey (ACS) provide additional information about the contexts in which respondents live. Main analyses use respondent, parent, and geographic context data from Wave I and respondent data from Wave IV.

Add Health offers several advantages over data sources used in prior studies of sexualities and geographic mobility. To start, the longitudinal design of Add Health enables estimation of geographic mobility over more precise periods of time—for example, between childhood and adulthood. Second, in measuring sexualities directly, Add Health largely avoids measurement error related to same-sex couple studies using census and ACS data. Third, in identifying sexual minorities regardless of relationship status, selection bias related to coupling inherent in census and ACS measures is mitigated. Fourth, more detailed measures of sexualities enable examination of heterogeneity across sexualities and dimensions of sexual orientation.

Because Add Health data rely on self-administered questionnaires, however, data for certain measures collected during specific waves—namely, early waves when respondents are young adolescents—might be unreliable. For example, the sexual attraction measure from Wave I likely overstates the proportion of same-sex attraction among Wave I respondents because of misunderstanding of the survey question or “mischievous” responses among youth (i.e., dishonest responses intended as a joke) (Katz-Wise et al. 2015; Robinson-Cimpian 2014; Savin-Williams and Joyner 2014). To mitigate potential bias, I define sexual identity, behavior, and attraction according to Wave IV measures collected during adulthood (described in detail in the following section). In addition, all outcome measures are collected by Add Health researchers rather than self-administered questionnaires.

Sample and Measures

Analyses restrict the sample to individuals with nonmissing weights who are ages 11–17 during Wave I and ages 26–34 during Wave IV and who are neither in prison, homeless, or living in group quarters during Wave IV (N = 11,705). I exclude individuals who are in prison, homeless, or living in group quarters as these circumstances incur residential changes and, therefore, bias estimates of geographic mobility. Age restrictions help isolate geographic mobility between childhood and adulthood.

Geographic Mobility

The main outcome of interest is geographic mobility, defined as moves between Wave I and Wave IV.2 I use three separate measures—each derived from geocoded address information—to capture geographic mobility: distance from childhood address, intercounty mobility, and interstate mobility. I use these measures, rather than neighborhood mobility or number of moves, because possible reasons for mobility differences across sexualities (e.g., avoidance of stigmatization, evasion of family control, quests for belonging) presume moves over meaningful distances. I construct distance from childhood address using information about the actual physical distance between respondents' Wave I and Wave IV addresses. I convert the original measure from meters to miles and log-transform the result to account for skewness.3 To construct dichotomous measures of intercounty mobility and interstate mobility, I use identification codes characterizing the state and county respondents reside within at each wave.4 I conduct parallel analyses for each outcome measure. Notably, moves occur across various context: for example, moves can occur between urban and nonurban, only nonurban, or only urban areas.

Sexualities

Primary independent variables are sexual identity, sexual behavior, and sexual attraction, depending on analysis. Sexual identity is measured according to the following six categories: 100% heterosexual, mostly heterosexual, bisexual, mostly homosexual, 100% homosexual, and no sexual identity. I recode respondents who answer 100% or mostly heterosexual as heterosexual, those who answer mostly or 100% homosexual as gay/lesbian, and those who do not think of themselves as any of these categories as having no sexual identity. Sexual behavior is measured according to the proportion of male and female sexual partners respondents report having in adulthood (i.e., individuals that respondents “had sex” with at age 18 and older). I code respondents with only different-sex partners as having different-sex partners, those with more than 0% to 75% same-sex partners as having both-sex partners, and those with no adult sexual activity as having no sexual behavior. Because same-sex-behaving adults often have sexual experiences with members of a different sex before settling into consistent same-sex behavior (Kaestle 2019), I code respondents with 75–100% same-sex partners as having same-sex partners.5Sexual attraction is defined according to two complementary yes/no questions asked during Wave IV: “Are you romantically attracted to females?” and “Are you romantically attracted to males?” I recode respondents with different-sex-only attraction as different-sex attracted, those who report attraction to both males and females as both-sex attracted, those with same-sex-only attraction as same-sex attracted, and those attracted to neither males nor females as having no attraction. All sexual orientation measures are collected during Wave IV.

Confounders

Certain models adjust for various geographic context, household, parent, and individual-level confounders. Unless otherwise indicated, confounders are measured during Wave I. Except for household number of children and parent occupation, all household and parent measures are reported through the parent questionnaire.

Geographic context confounders include childhood county unemployment rate and logged childhood county median household income. These covariates account for common migration “push” factors (Molloy et al. 2011). Household confounders include logged household income, number of children (residents under age 18), and English language usage (1 = English is primary language, 0 = English is not primary language) of the respondent's childhood household.

Parent confounders include age (in years), gender (man or woman), educational attainment (no high school diploma, high school diploma or GED, some college, and college degree or more), immigrant status (1 = born in the United States, 0 = not born in the United States), marital status (1 = married, 0 = not married), employment status (1 = employed, 0 = not employed), full-time status (1 = works ≥35 hours per week, 0 = works <35 hours per week or does not work), and two employed parents (1 = two employed parents, 0 = one or no parent employed). Occupation refers to the kind of work a respondent's resident mother and father do. If respondents provide occupations for both resident parents, I use the occupation of the resident father, as men are more likely to be employed than women, and children's status depends more on the occupational status of fathers than mothers (England et al. 2020; Hout 2018). Parental college expectations measures how disappointed a parent would be if their child did not graduate from college (1 = very, 2 = somewhat, 3 = not at all).6

Respondent confounders include age (in years), gender, and race/ethnicity (White, Black, Latino/Hispanic, Asian, and other).7 Add Health offers only two possible response categories when asking about sex: male or female.8 I define gender according to this measure. Several measures capture respondents' social integration and ties to place during adolescence. Residential duration measures the proportion of respondents' lives (up to Wave I) in which they lived at their Wave I address. Neighborhood satisfaction is a five-point categorical measure ranging from not at all satisfied to very much satisfied. I use living with parent (1 = lives with parent, 0 = does not live with parent), measured during Wave IV, to control for moves that occur with parents in tow.9

Mediators

Additional models include potential mediators of the relationship between sexualities and geographic mobility; unless otherwise indicated, mediators are measured during Wave I. Mediators are organized into two sets: the first captures social integration, while the second captures educational aspirations and attainment. Social integration measures include social acceptance, family understanding, closeness to parent, and desire to move. Social acceptance measures how much respondents agree with the statement “You are socially accepted”; response categories range from strongly agree (1) to strongly disagree (5). Desire to move captures how much respondents want to leave their childhood home, and family understanding captures how much respondents feel their family members understand them; categories for these measures range from not at all (1) to very much (5). Closeness to parent is a binary measure of whether respondents feel quite close or very close to at least one parent (1 = quite or very close, 0 = not quite or very close). Educational attainment measures include GPA (continuous, ranging from 1 to 4), college expectations (five categories, ranging from very unlikely to very likely to go to college), and educational attainment (no high school diploma, high school diploma or GED, some college, and college degree or more, measured during Wave IV).10

Table 1 displays descriptive statistics of the analytic sample, including the mean, standard deviation, and proportion of missing observations for selected measures. Although missingness among main independent and dependent variables is low, the amount of missing data for certain covariates is nontrivial. Most missingness results from the relatively low response rate of the parent questionnaire (Harris 2013). To account for potential bias, I use multiple imputation by chained equations to impute missing values for each covariate with missing information. Multiple imputation is a commonly used model-based approach that uses observed values of some variables to predict missing values of other variables (Enders 2010; van Buuren 2018). I perform multiple imputation with Stata's mi command, creating 40 multiply imputed data sets with which analyses are run. Details of the imputation procedure are included in the online appendix.

Methods

To estimate differences in geographic mobility across sexualities, I run a series of ordinary least-squares (OLS) and logistic regressions depending on outcome measure (i.e., OLS for continuous outcomes and logistic for dichotomous outcomes). Specifically, I regress geographic mobility on sexual orientation:

γ=β0+β1Χ1+ε,

where γ is a measure of geographic mobility between Wave I and Wave IV (i.e., either log miles from childhood address, an indicator of intercounty mobility, or an indicator of interstate mobility) and Χ1 is a categorical measure of either sexual identity, behavior, or attraction. I run separate analyses for each dimension of sexual orientation.

Additional models include sets of confounders and mediators that are characteristics of respondents, or their household, parents, or geographic context:

γ=β0+β1Χ1+βkXk+...+ε,

where γ is a measure of geographic mobility between Wave I and Wave IV; Χ1 is a categorical measure of sexual identity, behavior, or attraction (depending on analysis); and Xk is a vector of household, parent, individual, and geographic context confounders and mediators.11 Models that adjust for mediators explore the effect of potential mechanisms on the relationship between sexualities and geographic mobility. Because results controlling for confounders are very similar to those controlling for confounders and mediators, while tables show both, the text generally discusses confounder-adjusted results (i.e., the total effect of sexualities on mobility).

Finally, to explore gender variation in the association between sexual orientation and mobility, I run models that interact gender with sexual orientation and all other covariates. This is statistically equivalent to running gender-specific models (e.g., separate models for men and women) but is more computationally efficient (Long and Mustillo 2021).

All analyses are run with 40 multiply imputed data sets, account for Add Health survey design, and use appropriate cross-sectional weights (Chen and Chantala 2014). To aid interpretation of results, tables present average marginal effects, calculated with mimrgns in Stata 17 (Klein 2014). Because of small sample sizes of sexual minorities in nationally representative samples, I report p < .10 two-tailed tests as marginally significant. I interpret these results with caution and only when there are clear patterns across models.

Results

Table 2 displays descriptive statistics of the analytic sample for selected measures organized by sexual orientation dimension and subgroup. Regardless of dimension, mean differences between sexual minorities and heterosexuals tend to be small. For example, sexual minorities are about as likely as heterosexuals to live with their parents during Wave IV, and differences in childhood county log median incomes range from 0% to 2%, depending on dimension. That said, parents of heterosexuals are slightly more likely than parents of sexual minorities to have a college degree and be married. Moreover, individuals with gay/lesbian identity, same-sex behavior, or same-sex attraction are less likely to be White, and individuals with bisexual identity, both-sex behavior, or both-sex attraction feel less socially accepted, have a stronger desire to move, and have worse educational outcomes (e.g., GPA and college attainment), on average.

Most striking are mean differences in gender and geographic mobility. Depending on dimension, 81–85% of individuals with bisexual identity, both-sex behavior, or both-sex attraction are women, while 57–67% of individuals with gay/lesbian identity, same-sex behavior, or same-sex attraction are men. Individuals with heterosexual identity, different-sex partners, or different-sex attraction, by contrast, are split evenly according to gender. Regarding mobility, individuals with gay/lesbian identity, same-sex behavior, or same-sex attraction tend to be the most mobile, while individuals with heterosexual identity, different-sex behavior, or different-sex attraction tend to be the least mobile. For example, gay- and lesbian-identified individuals live an average of 324 miles away from their childhood address by Wave IV (about the distance from San Francisco to Los Angeles), while heterosexual-identified individuals live 183 miles away (about the distance from New York to Boston).

Table 3 presents average marginal effects calculated from regressions predicting various geographic mobility outcomes, organized by dimension of sexual orientation. Results represent sexual orientation differences in predicted outcomes (e.g., predicted distance from childhood address, predicted probabilities of intercounty or interstate mobility). I back-transform log mile differences into mile differences to aid interpretation of results. Model specifications are nested, meaning each set of covariates is added to the model in the prior column.

Across dimensions and models, sexualities tend to shape distance from childhood address and county moves between childhood and adulthood. Mobility differences appear almost exclusively between gay/lesbian and heterosexual individuals, same- and different-sex-behaving individuals, and same- and different-sex-attracted individuals. According to models that adjust for all confounders, individuals with gay/lesbian identity, same-sex behavior, and same-sex attraction live 115–171 miles farther from their childhood address than do individuals with heterosexual identity, different-sex behavior, and different sex attraction, respectively. The probability of moving counties between childhood and adulthood is 8–16 percentage points higher among individuals with gay/lesbian identity, same-sex behavior, or same-sex attraction than among individuals with heterosexual identity, different-sex behavior, or different-sex attraction. These results are statistically significant at the .05 level. While interstate mobility differences between these groups tell a similar story, they are either marginally significant at the .10 level or nonsignificant.

Differences between gay/lesbian and heterosexual individuals, same- and different-sex-behaving individuals, and same- and different-sex-attracted individuals change little after accounting for potential mediators. Adjusting for childhood social integration reduces differences in miles from childhood address by 3–4% and county migration by 0–0.2 percentage points. Accounting for educational attainment shrinks differences in miles from childhood address by an additional 10–16% and intercounty mobility by an additional 1.1–1.5 percentage points.

Variation Across Detailed Sexualities and Dimensions of Sexual Orientation

Mobility differences between individuals with bisexual identity, both-sex behavior, and both-sex attraction and individuals with heterosexual identity, different-sex behavior, and different sex attraction tend to be small and nonsignificant. After accounting for confounders, only interstate mobility differences between both- and different-sex-behaving individuals are statistically significant at the .05 level: individuals with both-sex behavior are 3.6 percentage points less likely to move states between childhood and adulthood than their counterparts with different-sex behavior.

Looking across dimensions, sexual orientation differences in mobility are largest and most prevalent when defined according to identity, but smallest and least prevalent when defined according to behavior. For example, after adjusting for confounders, while individuals with same-sex behavior are eight percentage points more likely to move counties than individuals with different-sex behavior, individuals with gay/lesbian identity are 15.6 percentage points more likely to move counties than individuals with heterosexual identity.

Interestingly, in confounder-adjusted analyses, differences between gay/lesbian- and bisexual-identified individuals are almost always smaller than those between same- and both-sex-behaving individuals and same- and both-sex-attracted individuals.12 Moreover, differences between same- and both-sex-behaving individuals and same- and both-sex-attracted individuals are always statistically significant at the .05 level, while with the exception of intercounty mobility differences, differences between gay/lesbian- and bisexual-identified individuals are marginally significant or nonsignificant.13 Sexual orientation gaps in geographic mobility, therefore, vary by dimension.

Variation by Gender

Given that men and women are unevenly represented across sexualities (Mishel 2019), results in Table 3 may obscure gender-specific sexual orientation gaps in geographic mobility. For example, because 81–85% of individuals with bisexual identity, both-sex behavior, or both-sex attraction are women (as evident in Table 2), mobility gaps between bisexual and heterosexual women, both- and different-sex-behaving women, and both- and different-sex-attracted women may override mobility gaps between analogous groups of men. Small and statistically insignificant coefficients for individuals with bisexual identity, both-sex behavior, and both-sex attraction in Table 3 could result despite large mobility differences between men with these characteristics and men with heterosexual identity, different-sex behavior, or different-sex attraction.

Table 4 displays results of analyses conducted to address these concerns and explore gender variation in the association between sexual orientation and geographic mobility. Models in panel A adjust for all confounders; models in panel B further adjust for all mediators. In each panel, the first column displays sexual orientation differences among men, the second column displays sexual orientation differences among women, and the third column displays differences in these differences (i.e., “second differences”). Sexual orientation differences are derived from regressions that interact gender with all model covariates.

There is some evidence that sexual orientation gaps in mobility vary by gender. Across nearly all analyses in panel A, sexual orientation gaps tend to be larger among men than women. The effect of both-sex attraction appears to operate in a different direction for men and women as well, with both-sex-attracted men being more mobile than different-sex-attracted men, but both-sex-attracted women being less mobile than different-sex-attracted women. However, only one second difference is statistically significant at the .05 level—that for interstate mobility differences between same- and different-sex-attracted individuals: the interstate mobility gap between same- and different-sex-attracted men is 17.6 percentage points higher than that between same- and different-sex-attracted women.

Supplemental and Sensitivity Analyses

I conduct several supplemental and sensitivity analyses. First, I repeat analyses in Table 3 with additional Wave IV covariates (e.g., log earnings, occupational status, marital status, parenthood, and the proportion of residents in respondents' neighborhoods in same-sex couples). Results from these analyses are displayed in Table A1 in the online appendix. Although some differences become nonsignificant, particularly miles from childhood address differences between same- and different-sex-behaving individuals and same- and different-sex-attracted individuals (p values ranging from .11 to .16), results are substantively similar. Accounting for additional adulthood characteristics does not change interpretations of the main results.

Table A2 in the online appendix presents results of analyses using alternative measures of sexual identity and behavior (i.e., coding “mostly heterosexual” and “mostly homosexual” as bisexual and respondents with 0–99% same-sex sexual activity as having both-sex partners). Results are consistent with those in Table 3. Because sexualities measured at Wave IV may be endogenous to geographic mobility, I also repeat the main analyses with earlier measures of sexual identity and attraction from Wave III. Table A3 shows these results. Despite small differences in the size and statistical significance of estimates, which may be expected owing to changes in the sample, survey question language, and individuals' sexualities across waves, results are consistent with those in Table 3.

Finally, in analyses that are available upon request, I explore additional sources of variation in the relationship between sexualities and geographic mobility. Specifically, I explore variation across childhood county size (1 = ≥500,000 residents, 0 = <500,000 residents), childhood county political behavior (proportion of voters who voted for the 1992 Republican presidential candidate), and respondents' parents' knowledge of their sexual identity (1 = ≥1 parent knows about respondents' sexual identity, 0 = neither parent knows about respondents' sexual identity; all heterosexual-identified individuals are coded as 1). Results from childhood county size and political behavior analyses are in the anticipated direction. Sexual orientation differences in geographic mobility appear to be smaller among individuals who grow up in large counties than those who grow up in small counties, and larger among individuals who grow up in counties with greater shares of Republican voters. However, almost all second differences (i.e., differences in sexual orientation differences across these county-level characteristics) are nonsignificant. Estimates from analyses that interact sexual orientation with an indicator of whether the respondent has at least one parent who knows their sexual identity are nonsignificant and display no clear patterns.14

Discussion and Conclusion

Sexual orientation shapes geographic mobility between childhood and adulthood, but for whom and according to which operationalizations? Individuals with gay/lesbian identity, same-sex behavior, and same-sex attraction tend to be much more geographically mobile in the transition to adulthood than their heterosexual-identified, different-sex-behaving, and different-sex-attracted peers, respectively. After accounting for various confounders, gay/lesbian-identified individuals live about 170 miles farther from their childhood addresses than their heterosexual peers. By contrast, individuals with bisexual identity, both-sex behavior, and both-sex attraction tend to be statistically indistinct from individuals with heterosexual identity, different-sex behavior, and different-sex attraction, respectively. These findings hold across various outcome measures and model specifications.

Sexual minorities and heterosexual individuals—particularly gay/lesbian and heterosexual individuals—are similar across a range of characteristics: where they grow up, the families they come from, race and ethnicity, academic performance, and educational attainment. Perhaps unsurprisingly, then, common explanations of geographic mobility—local labor market conditions, residential satisfaction, socioeconomic characteristics, life course milestones (e.g., marriage, parenthood)—fail to fully explain variation across sexualities. Even after accounting for various geographic context, household, parent, and individual-level confounders and mediators, substantial differences between gay/lesbian and heterosexual individuals, same- and different-sex-behaving individuals, and same- and different-sex-attracted individuals remain. Adjusting for all confounders and mediators decreases differences between these groups in distance from childhood address by 7% at most and in intercounty mobility by 1.1–1.5 percentage points.

While this study identifies clear mobility differences between sexualities, its ability to explain these differences is limited. Though Add Health measures are numerous and span various levels of analysis, they are not exhaustive. Remaining gaps may be explained by disproportionate stigmatization across sexualities, but only studies that directly measure stigmatization (or victimization, perceptions of discomfort, etc.) can provide sufficient support for this hypothesis. Moreover, data sources that measure other theoretically relevant characteristics, such as the proportion of sexual minorities living in one's community of origin, may facilitate broader exploration of potential mediators and moderators.

Similarly, reasons for mobility differences among sexual minorities (e.g., greater mobility among individuals with same-sex attraction than among individuals with both-sex attraction) are not readily apparent. Given that individuals with a bisexual identity, both-sex behavior, or both-sex attraction report the least social acceptance and greatest desire to move in childhood (as shown in Table 2), it seems unlikely they have stronger ties to their place of origin or less motivation to move than individuals with gay/lesbian identity, same-sex behavior, or same-sex attraction. It is possible that individuals with bisexual identity, both-sex behavior, or both-sex attraction are relatively more motivated to move but have fewer resources to do so (Mize 2016; Mollborn and Everett 2015). However, accounting for adulthood resources does not change the interpretation of main results. More simply, bisexual identity, both-sex behavior, and both-sex attraction may be particularly noisy categories (Rust 2000), especially as individuals with these traits demonstrate relatively high sexual “fluidity” across adolescence and adulthood (Kaestle 2019). Subsequent work, quantitative and qualitative alike, could attend to these discrepancies, exploring variation in migration motivations among sexual minorities, as well as the ways in which meanings individuals attach to the sexualities they embody shape life course milestones.

Importantly, differences in geographic mobility depend on the operationalization of sexualities. Gaps tend to be largest and most prevalent according to sexual identity, suggesting that sexual identity shapes mobility more than sexual behavior or attraction. Differences between bisexual- and gay/lesbian-identified individuals also tend to be smaller than those between both- and same-sex-behaving individuals and between both- and same-sex-attracted individuals, suggesting relatively greater shared experiences and outlooks among sexual minority-identified individuals than among individuals with sexual minority behavior or attraction. Results are, therefore, consistent with theories about social identity (Stets and Burke 2006) and sexual identity formation (Gorman-Murray 2007; Knopp 2004). On one hand, sexual minority-identified individuals may attempt to emulate others in their identity group who are (or are perceived to be) highly geographically mobile. On the other hand, geographic mobility (and the destination of such mobility) may constitute or provide sufficient conditions for sexual minority identity itself (Brown-Saracino 2017; Gorman-Murray 2007, 2009; Kazyak 2011; Lewis 2012, 2014). For instance, moving away from one's community of origin may provide sufficient physical and psychic distance to develop a sexual minority identity without fear or judgment (Gorman-Murray 2009; Lewis 2014). Although this study cannot fully discern between these accounts, and both may occur concurrently, future research could attend to disentangling these processes. In addition to theoretical implications, these findings have important methodological implications. Because sexualities are multidimensional, measurement decisions can have real empirical consequences. Researchers studying sexualities should address the multidimensionality of the construct and take care in choosing measures that are theoretically appropriate to the research question at hand.

Complementing prior work that finds gender differences in the migration patterns of same-sex couples (Cooke and Rapino 2007), I uncover evidence that suggests sexual orientation differences in mobility vary by gender. Specifically, the effect of same-sex attraction on interstate mobility is greater for men than for women. The fact that same-sex attraction has a larger effect on interstate mobility, rather than intercounty mobility, among men is perhaps telling. Sexual minority status tends to confer greater stigma on men than on women (Herek 2009; Mize and Manago 2018a, 2018b); same-sex-attracted men may be more likely to make moves across meaningful distances (i.e., across state lines) to mitigate this stigmatization. While within-gender variation may exist, sample size limitations prevent this level of analysis. Additional research could further investigate how stigma may induce mobility among men or women who may be relatively more stigmatized as a result of particular intersections of gender and, for example, race, ethnicity, class, or gender presentation. Still, this study illustrates one way in which theoretical links between gender and sexual orientation established in prior work (England 2016; Pascoe 2007) may play out in the transition to adulthood.

Because this study focuses on mobility between childhood and adulthood, it is unclear whether mobility differences persist throughout the life course. Qualitative evidence suggests that the mobility of sexual minorities and heterosexuals may converge in midlife (Lewis 2014), but this has yet to be tested empirically. Nonetheless, focusing on the transition to adulthood opens new avenues of inquiry. Given that geographic mobility between childhood and adulthood is associated with movement across neighborhood socioeconomic contexts (Leibbrand and Crowder 2018; Sharkey 2008), my findings lay the groundwork for inquiry regarding sexualities and residential choice, particularly as research shows that same-sex couples tend to live in relatively advantaged areas (Anacker and Morrow-Jones 2005; Gates and Ost 2004). In addition, my results raise questions about the ways in which sexualities shape other life course milestones in the transition to adulthood, such as going away to college, cohabitation, marriage, and childbearing.

In providing the first quantitative estimates of geographic mobility differences between the broader sexual minority and heterosexual populations, this study reveals substantial mobility differences across sexualities and uncovers multiple sources of variation in the relationship between sexualities and mobility. In doing so, the study highlights important considerations for the measurement of sexualities and lays the groundwork for future research related to sexualities, mobility, and the life course.

Acknowledgments

I am grateful to Siwei Cheng, Paul DiMaggio, Paula England, Mike Hout, and Eric Klinenberg for their helpful input and encouragement. I also thank each anonymous reviewer and the Demography editorial team for constructive feedback and guidance. This research uses data from Add Health. Add Health is directed by Robert A. Hummer at the University of North Carolina at Chapel Hill and funded by the National Institute on Aging cooperative agreements U01 AG071448 (Hummer) and U01AG071450 (Aiello and Hummer). Waves I–V data are from the Add Health Program Project, funded by grant P01 HD31921 (Harris) from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Add Health was designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill.

Notes

1

I use “sexual orientation” as an umbrella concept to encompass various sexual and romantic identities, practices, feelings, and dispositions. When possible, I use “sexualities” to acknowledge and refer to the range of subgroups defined by sexual orientation. I use “sexual minority” to refer to individuals with nonheterosexual identities, behaviors, or attractions. I recognize that sexualities are socially constructed and vary across time and place. Categories attempt to capture meaningful patterns.

2

Most respondents who attended college completed their program several years before the administration of Wave IV. For most respondents, then, it is unlikely that Wave IV addresses resulted from moves specifically to attend college.

3

Specifically, I take the log of miles from childhood address plus one (i.e., ln(miles +1)) to retain individuals who live at the same address.

4

Add Health assigns identification codes to maintain confidentiality. Codes are harmonized according to 2010 census boundaries. Actual names of counties and states are not available, curtailing investigation of mobility between specific places.

5

Supplemental analyses use two alternative coding schemes for sexual identity and sexual behavior, recoding “mostly heterosexual” and “mostly homosexual” as bisexual and respondents with 0% to 99% same-sex sexual activity as having both-sex partners. Results are nearly the same as those of the main analyses.

6

I do not include a measure of parent race in regression models because of multicollinearity (e.g., variance inflation factor above 10).

7

I define race/ethnicity according to Add Health’s constructed race variable, which is based on respondents’ and parents’ responses about their race, and respondents’ answers about their Latino or Hispanic background. I assign respondents who report Latino or Hispanic origin as Latino/Hispanic and those who do not as either White, Black, Asian, or other race.

8

Questionnaires preload the respondent’s sex from prior survey waves. Sex is originally collected through in-school questionnaires that ask respondents about their sex (“What sex are you?”). Interviewers are instructed to confirm the respondent’s sex during each wave. Nine respondents in the main analytic sample (0.08%) have a different sex recorded in Wave I and Wave IV. Reasons for these discrepancies are unclear. Discrepancies could result from errors in preloading or data entry or actual changes in the respondent’s sex. The percentage of cases that report different sexes in Waves I and IV is very small, however, and is unlikely to significantly bias results.

9

I treat living with parent as a confounder rather than a mediator given the ambiguous causal order of living with a parent and geographic mobility. Wanting to live away from a parent could cause geographic mobility, or geographic mobility could cause living away from a parent. Models that exclude living with parent, or treat living with parent as a mediator, produce results substantively similar to those of the main analyses.

10

Because of potential endogeneity, I exclude additional Wave IV covariates in the main analyses. Supplemental analyses with additional Wave IV covariates produce substantively similar results.

11

While analyses of distance from childhood address and intercounty mobility include county unemployment rate and county median household income measures, analyses of interstate mobility do not because complementary state-level measures are unavailable.

12

The only exception is that intercounty mobility differences between gay/lesbian- and bisexual-identified individuals are larger than intercounty mobility differences between same- and both-sex-behaving individuals.

13

These analyses are available upon request.

14

Ambiguous results may reflect the limitations of the available measure: (1) parents’ knowledge of one’s sexual identity is an imperfect measure of being “out”; (2) this measure is collected only during Wave III, and sexual identity and disclosure may change substantially between Waves III and IV (Kaestle 2019); (3) this measure is only asked of respondents who identify as bisexual, gay, or lesbian; (4) there is no clear way to operationalize being “out” among heterosexuals; (5) being “out” may not be relevant to sexual behavior or attraction; and (6) no sexual identity, behavior, and attraction is recoded as missing in order for models using imputed data to run. For these reasons, I do not include this measure in the main analyses. When included, results are substantively similar.

References

Anacker, K. B., & Morrow-Jones, H. A. (
2005
).
Neighborhood factors associated with same-sex households in U.S. cities
.
Urban Geography
,
26
,
385
409
.
Baumle, A. K., Compton, D. R., & Poston, D. L.Jr. (
2009
).
Same-sex partners: The social demography of sexual orientation
.
Albany
:
SUNY 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
.
Black, D., Gates, G., Sanders, S., & Taylor, L. (
2007
).
The measurement of same-sex unmarried partner couples in the 2000 U.S. Census
(CCPR Working Paper, No. CCPR-023-07).
Los Angeles
:
California Center for Population Research, University of California, Los Angeles
.
Brown, J. L., Swartzendruber, A., & Diclemente, R. J. (
2013
).
Application of audio computer-assisted self-interviews to collect self-reported health data: An overview
.
Caries Research
,
47
(
Suppl. 1
),
40
45
.
Brown-Saracino, J. (
2017
).
How places make us: Novel LBQ identities in four small cities
.
Chicago, IL
:
University of Chicago Press
.
Brown-Saracino, J. (
2019
).
Aligning our maps: A call to reconcile distinct visions of literatures on sexualities, space, and place
.
City & Community
,
18
,
37
43
.
Cantú, L.Jr. (
2009
).
The sexuality of migration: Border crossings and Mexican immigrant men
(Naples, N. A. & Vidal-Ortiz, S., Eds.).
New York
:
NYU Press
.
Carillo, H. (
2017
).
Pathways of desire: The sexual migration of Mexican gay men
.
Chicago, IL
:
University of Chicago Press
.
Chen, P., & Chantala, K. (
2014
).
Guidelines for analyzing Add Health data
(CPC report).
Chapel Hill
:
Carolina Population Center, University of North Carolina at Chapel Hill
. https://doi.org/10.17615/C6BW8W
Compton, D. R., & Baumle, A. K. (
2012
).
Beyond the Castro: The role of demographics in the selection of gay and lesbian enclaves
.
Journal of Homosexuality
,
59
,
1327
1355
.
Cooke, T. J. (
2005
).
Migration of same-sex couples
.
Population, Space and Place
,
11
,
401
409
.
Cooke, T. J., & Rapino, M. (
2007
).
The migration of partnered gays and lesbians between 1995 and 2000
.
Professional Geographer
,
59
,
285
297
.
Diamond, L. M. (
2008
).
Sexual fluidity: Understanding women's love and desire
.
Cambridge, MA
:
Harvard University Press
.
Enders, C. K. (
2010
).
Applied missing data analysis
.
New York, NY
:
Guilford Press
.
England, P. (
2016
).
Sometimes the social becomes personal: Gender, class, and sexualities
.
American Sociological Review
,
81
,
4
28
.
England, P., Levine, A., & Mishel, E. (
2020
).
Progress toward gender equality in the United States has slowed or stalled
.
Proceedings of the National Academy of Sciences
,
117
,
6990
6997
.
Friedman, S., Reynolds, A., Scovill, S., Brassier, F., Campbell, R., & Ballou, M. (
2013
).
An estimate of housing discrimination against same-sex couples
(Report).
Washington, DC
:
U.S. Department of Housing and Urban Development
.
Gallup Organization
. (
2020
).
Gay and lesbian rights
.
Gallup
. Retrieved from https://news.gallup.com/poll/1651/gay-lesbian-rights.aspx
Gates, G. J., & Ost, J. (
2004
).
The gay and lesbian atlas
.
Washington, DC
:
Urban Institute Press
.
Ghaziani, A. (
2014
).
There goes the gayborhood?
Princeton, NJ
:
Princeton University Press
.
Ghaziani, A. (
2019
).
Cultural archipelagos: New directions in the study of sexuality and space
.
City & Community
,
18
,
4
22
.
Gorman-Murray, A. (
2007
).
Rethinking queer migration through the body
.
Social and Cultural Geography
,
8
,
105
121
.
Gorman-Murray, A. (
2009
).
Intimate mobilities: Emotional embodiment and queer migration
.
Social and Cultural Geography
,
10
,
441
460
.
Harris, K. M. (
2013
).
The Add Health study: Design and accomplishments
(CPC report).
Chapel Hill
:
Carolina Population Center, University of North Carolina at Chapel Hill
. Retrieved from https://addhealth.cpc.unc.edu/wp-content/uploads/docs/user_guides/DesignPaperWave_I-IV.pdf
Hatzenbuehler, M. L., McLaughlin, K. A., & Xuan, Z. (
2012
).
Social networks and risk for depressive symptoms in a national sample of sexual minority youth
.
Social Science & Medicine
,
75
,
1184
1191
.
Herek, G. M. (
2009
).
Hate crimes and stigma-related experiences among sexual minority adults in the United States
.
Journal of Interpersonal Violence
,
24
,
54
74
.
Hewitt, C. (
1995
).
The socioeconomic position of gay men: A review of the evidence
.
American Journal of Economics and Sociology
,
54
,
461
479
.
Horowitz, J., & Entwisle, B. (
2021
).
Life course events and migration in the transition to adulthood
.
Social Forces
,
100
,
29
55
.
Hout, M. (
2018
).
Americans' occupational status reflects the status of both of their parents
.
Proceedings of the National Academy of Sciences
,
115
,
9527
9532
.
Jones, J. M. (
2017
,
June
22
).
In U.S., 10.2% of LGBT adults now married to same-sex spouse
.
Gallup
. Retrieved from https://news.gallup.com/poll/212702/lgbt-adults-married-sex-spouse.aspx
Kaestle, C. E. (
2019
).
Sexual orientation trajectories based on sexual attractions, partners, and identity: A longitudinal investigation from adolescence through young adulthood using a U.S. representative sample
.
Journal of Sex Research
,
56
,
811
826
.
Katz-Wise, S. L., Calzo, J. P., Li, G., & Pollitt, A. (
2015
).
Same data, different perspectives: What is at stake? Response to Savin-Williams and Joyner (2014a)
.
Archives of Sexual Behavior
,
44
,
15
19
.
Katz-Wise, S. L., & Hyde, J. S. (
2012
).
Victimization experiences of lesbian, gay, and bisexual individuals: A meta-analysis
.
Journal of Sex Research
,
49
,
142
167
.
Kazyak, E. (
2011
).
Disrupting cultural selves: Constructing gay and lesbian identities in rural locales
.
Qualitative Sociology
,
34
,
561
581
.
Kazyak, E. (
2020
).
Introduction to special issue “Geographies of sexualities
.”
Journal of Lesbian Studies
,
24
,
173
185
.
Klein, D. (
2014
).
mimrgns: Stata module to run margins after mi estimate
[Computer software]. Retrieved from https://econpapers.repec.org/RePEc:boc:bocode:s457795
Knopp, L. (
2004
).
Ontologies of place, placelessness, and movement: Queer quests for identity and their impacts on contemporary geographic thought
.
Gender, Place & Culture
,
11
,
121
134
.
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
.
Leibbrand, C., & Crowder, K. (
2018
).
Migration, mobility, and neighborhood attainment: Using the PSID to understand the processes of racial stratification
.
Annals of the American Academy of Political and Social Science
,
680
,
172
192
.
Lewis, N. M. (
2012
).
Remapping disclosure: Gay men's segmented journeys of moving out and coming out
.
Social and Cultural Geography
,
13
,
211
231
.
Lewis, N. M. (
2014
).
Moving “out,” moving on: Gay men's migrations through the life course
.
Annals of the Association of American Geographers
,
104
,
225
233
.
Long, J. S., & Mustillo, S. A. (
2021
).
Using predictions and marginal effects to compare groups in regression models for binary outcomes
.
Sociological Methods & Research
,
50
,
1284
1320
.
Luo, M. (
2022
).
Sexuality, migration and family: Understanding Jia and its impact on Chinese young gay men's migration motives from a temporal perspective
.
Journal of Ethnic and Migration Studies
,
48
,
578
593
.
McLaughlin, K. A., Hatzenbuehler, M. L., Xuan, Z., & Conron, K. J. (
2012
).
Disproportionate exposure to early-life adversity and sexual orientation disparities in psychiatric morbidity
.
Child Abuse & Neglect
,
36
,
645
655
.
McPherson, M., Smith-Lovin, L., & Cook, J. M. (
2001
).
Birds of a feather: Homophily in social networks
.
Annual Review of Sociology
,
27
,
415
444
.
Mishel, E. (
2016
).
Discrimination against queer women in the U.S. workforce: A resume audit study
.
Socius
,
2
. https://doi.org/10.1177/2378023115621316
Mishel, E. (
2019
).
Intersections between sexual identity, sexual attraction, and sexual behavior among a nationally representative sample of American men and women
.
Journal of Official Statistics
,
35
,
859
884
.
Mittleman, J. (
2019
).
Sexual minority bullying and mental health from early childhood through adolescence
.
Journal of Adolescent Health
,
64
,
172
178
.
Mize, T. D. (
2016
).
Sexual orientation in the labor market
.
American Sociological Review
,
81
,
1132
1160
.
Mize, T. D., & Manago, B. (
2018a
).
Precarious sexuality: How men and women are differentially categorized for similar sexual behavior
.
American Sociological Review
,
83
,
305
330
.
Mize, T. D., & Manago, B. (
2018b
).
The stereotype content of sexual orientation
.
Social Currents
,
5
,
458
478
.
Mock, S. E., & Eibach, R. P. (
2012
).
Stability and change in sexual orientation identity over a 10-year period in adulthood
.
Archives of Sexual Behavior
,
41
,
641
648
.
Mollborn, S., & Everett, B. (
2015
).
Understanding the educational attainment of sexual minority women and men
.
Research in Social Stratification and Mobility
,
41
,
40
55
.
Molloy, R., Smith, C. L., & Wozniak, A. (
2011
).
Internal migration in the United States
.
Journal of Economic Perspectives
,
25
(
3
),
173
196
.
Mulder, C. H. (
2009
).
Leaving the parental home in young adulthood
. In Furlong, A. (Ed.),
Handbook of youth and young adulthood: New perspectives and agendas
(pp.
203
210
).
London, UK
:
Routledge
. https://doi.org/10.4324/9780203881965-34
Needham, B. L., & Austin, E. L. (
2010
).
Sexual orientation, parental support, and health during the transition to young adulthood
.
Journal of Youth and Adolescence
,
39
,
1189
1198
.
Newport, F., & Gates, G. J. (
2015
,
March
20
).
San Francisco metro area ranks highest in LGBT percentage
.
Gallup
. Retrieved from https://news.gallup.com/poll/182051/san-francisco-metro-area-ranks-highest-lgbt-percentage.aspx
O'Connell, M., & Gooding, G. (
2007
).
Editing unmarried couples in Census Bureau data
(Working paper).
Washington, DC
:
U.S. Census Bureau
. Retrieved from https://www.census.gov/content/dam/Census/library/working-papers/2007/demo/twps07.pdf
Pascoe, C. J. (
2007
).
Dude, you're a fag: Masculinity and sexuality in high school
.
Berkeley
:
University of California Press
.
Pearson, J., & Wilkinson, L. (
2013
).
Family relationships and adolescent well-being: Are families equally protective for same-sex attracted youth?
Journal of Youth and Adolescence
,
42
,
376
393
.
Pew Research Center
. (
2013a
).
A survey of LGBT Americans: Attitudes, experiences and values in changing times
(Report).
Washington, DC
:
Pew Research Center
.
Pew Research Center
. (
2013b
).
In gay marriage debate, both supporters and opponents see legal recognition as ‘inevitable’
(Report).
Washington, DC
:
Pew Reasearch Center
. Retrieved from https://www.pewresearch.org/wp-content/uploads/sites/4/legacy-pdf/06-06-13-LGBT-general-public-release.pdf
Robinson-Cimpian, J. P. (
2014
).
Inaccurate estimation of disparities due to mischievous responders: Several suggestions to assess conclusions
.
Educational Researcher
,
43
,
171
185
.
Rosario, M., Reisner, S. L., Corliss, H. L., Wypij, D., Frazier, A. L., & Austin, S. B. (
2014
).
Disparities in depressive distress by sexual orientation in emerging adults: The roles of attachment and stress paradigms
.
Archives of Sexual Behavior
,
43
,
901
916
.
Rosenfeld, M. J. (
2007
).
The age of independence: Interracial unions, same-sex unions, and the changing American family
.
Cambridge, MA
:
Harvard University Press
.
Rosenfeld, M. J., & Kim, B. (
2005
).
The independence of young adults and the rise of interracial and same-sex unions
.
American Sociological Review
,
70
,
541
562
.
Rust, P. C. R. (
2000
).
Review of statistical findings about bisexual behavior, feelings, and identities
. In Rust, P. C. R. (Ed.),
Bisexuality in the United States: A social science reader
(pp.
129
184
).
New York, NY
:
Columbia University Press
.
Savin-Williams, R. C., & Joyner, K. (
2014
).
The politicization of gay youth health: Response to Li, Katz-Wise, and Calzo (2014)
.
Archives of Sexual Behavior
,
43
,
1027
1030
.
Savin-Williams, R. C., Joyner, K., & Rieger, G. (
2012
).
Prevalence and stability of self-reported sexual orientation identity during young adulthood
.
Archives of Sexual Behavior
,
41
,
103
110
.
Shanahan, M. J. (
2000
).
Pathways to adulthood in changing societies: Variability and mechanisms in life course perspective
.
Annual Review of Sociology
,
26
,
667
692
.
Sharkey, P. (
2008
).
The intergenerational transmission of context
.
American Journal of Sociology
,
113
,
931
969
.
Silva, T. J. (
2019
).
Straight identity and same-sex desire: Conservatism, homophobia, and straight culture
.
Social Forces
,
97
,
1067
1094
.
Stets, J. E., & Burke, P. J. (
2006
).
Identity theory and social identity theory
.
Social Psychology Quarterly
,
63
,
224
237
.
Stone, A. L. (
2007
).
Queer disaporic (non) identity: Japanese lesbians return home
. In Asgharzadeh, A., Lawson, E., Oka, K. U., & Wahab, A. (Eds.),
Diasporic ruptures: Globality, migrancy, and expressions of identity
(Vol.
II
, pp.
39
53
).
Rotterdam, the Netherlands
:
Sense Publishers
.
Tilcsik, A. (
2011
).
Pride and prejudice: Employment discrimination against openly gay men in the United States
.
American Journal of Sociology
,
117
,
586
626
.
Ueno, K. (
2005
).
Sexual orientation and psychological distress in adolescence: Examining interpersonal stressors and social support processes
.
Social Psychology Quarterly
,
68
,
258
277
.
van Buuren, S. (
2018
).
Flexible imputation of missing data
(2nd ed.).
Boca Raton, FL
:
CRC Press
.
Ward, J. (
2015
).
Not gay: Sex between straight White men
.
New York
:
NYU Press
.
Weston, K. (
1995
).
Get thee to a big city: Sexual imaginary and the great gay migration
.
GLQ
,
2
,
253
277
.
Wimark, T. (
2016a
).
Migration motives of gay men in the new acceptance era: A cohort study from Malmö, Sweden
.
Social and Cultural Geography
,
17
,
605
622
.
Wimark, T. (
2016b
).
The impact of family ties on the mobility decisions of gay men and lesbians
.
Gender, Place and Culture
,
23
,
659
676
.
This is an open access article distributed under the terms of a Creative Commons license (CC BY-NC-ND 4.0).

Supplementary data