Sexual Orientation,Internal Migration,and Mental Health during the Transition to Adulthood

Abstract

Previous research has suggested that sexual minorities may have higher rates of migration than heterosexuals, indicating their effort to escape stigma in the currently residing areas. However, direct evidence for the migration pattern has been lacking, and mental health implications of such coping effort have been unclear. This study seeks to fill these gaps in the literature by analyzing the Add Health data, which include longitudinal measures of residential locations, sexual orientation, and mental health. The analysis focuses on the transition to adulthood, when the rate of internal migration peaks. Among women, sexual minorities have a higher rate of migration than heterosexuals, but men do not show such a difference. Sexual minorities show better mental health when they migrate to counties with higher proportions of people living in urban areas whereas heterosexuals do not show such an association. Among sexual minority men, migration to counties with higher population density and higher proportions of college-educated residents is also linked to better mental health.

Keywords

coping internal migration mental health sexual orientation tradition to adulthood

Background

People who report same-sex or bisexual orientation (“sexual minorities”) have higher rates of mental health problems than heterosexuals, and this disparity is observed in both the adolescent population (Hatzenbuehler, McLaughlin, and Nolen-Hoeksema 2008; Marshal et al. 2011; Russell, Driscoll, and Truong 2002) and the adult population (Brewster and Tillman 2012; Grollman 2012; Meyer 2003; Ueno 2010a). Researchers have argued that this disparity results partly from stigmatization of sexual minorities in the heteronormative society, thus emphasizing structural and cultural constraints on sexual minorities’ lives. These studies have paid limited attention to the ways in which sexual minorities challenge or cope with these constraints, however. Considering sexual minorities’ internal migration as one of their coping behaviors, this study examines whether they are more likely to migrate and whether they benefit more from migration.

The study focuses on the transition to adulthood, which is an important age period for two reasons. First, the rate of internal migration peaks during the period. In the Current Population Survey, 17.9% of people between age 18 and 24 and 24.8% of people between age 25 and 29 reported that they had moved across counties in past five years (Ihrke and Faber 2012). Sexual minorities may take advantage of migration opportunities during this transition in an effort to reduce stigma exposure. Second, the extent of mental health disparities increases between heterosexuals and some sexual minorities during the transition. Specifically, an analysis of Add Health data illustrates that people who report their first same-sex experience in young adulthood show worse changes in mental health during the transition than those without such sexual experience (Ueno 2010b). Additionally, among women, those who report same-sex experience continuously from adolescence and young adulthood show worse mental health changes. The present study will examine the role of internal migration in these changing mental health disparities during the transition.

Sexual Orientation and Internal Migration

Compared to other areas, urban and politically progressive areas tend to include higher proportions of sexual minority residents, especially sexual minority men. Scholars have argued that this pattern results partly from sexual minorities’ migration to these areas in their search for social environments more accepting of sexual diversity (Black et al. 2002; Walther and Poston 2004). This interpretation is consistent with other research reporting unique challenges faced by sexual minorities living in rural, conservative areas, such as social isolation, discrimination, inability to disclose their sexual orientation, and invisibility of sexual minorities in the area (D’Augelli and Hart 1987; McCarthy 2000). A recent census study confirmed sexual minorities’ higher rate of migration by measuring changes in residential locations over a five-year period (Baumle, Compton, and Poston 2009). Specifically, the odds of interstate moves was 4% higher for women in same-sex unions than heterosexually married women, and the corresponding difference for men is 8%.

Sexual minorities may no longer head for metropolitan places, however. Using the same census information, another study demonstrated that sexual minority men tend to migrate to moderate-sized urban areas that provide rich natural amenities, whereas sexual minority women tend to move to less populous areas with large existing lesbian populations (Cooke and Rapino 2007). These findings from the census need to be interpreted with caution because the analyses were limited to people who had residing partners due to the sexual orientation measure that relied on residing partners’ gender. Nonetheless, the findings are consistent with research documenting an increasing number of gay and lesbian communities in suburban and rural areas for those who seek to break away from superficiality in the urban gay culture and to (re)integrate themselves into local communities (Annes and Redlin 2012; Kirkey and Forsyth 2001; McCarthy 2000). Further, rural communes have long existed for lesbians who escape the male domination in metropolitan areas and those who seek to entertain utopian notions of rural living and eco-feminism (Bell and Valentine 1995). The present study extends these past studies by analyzing data from a nationally representative sample and by using direct measures of sexual orientation. As we explain below, this paper also address an important question left unexplored in this literature—whether migration has different mental health consequences across sexuality groups.

Migration and Mental Health

Before addressing sexuality group differences, we discuss mental health consequences of migration more generally because past research has not adequately explored the issue. Mental health researchers have conceptualized migration and, more broadly, residential moves as stressful life events that may undermine mental health because they break daily routines and demands behavioral and psychological adjustments (e.g., Compas, Davis, and Forsythe 1985; Newcomb, Huba, and Bentler 1981). Consistent with this view, residential moves are associated with mental health problems such as depression and substance use during childhood and adolescence (Hoffmann and Johnson 1998; Tiesler et al. 2013). Further, mobile adolescents tend to display behavioral problems such as poor adjustment at school (Tucker, Marx, and Long 1998) and violent behaviors (Haynie and South 2005). These effects of relocations in early life stages may persist in adulthood (DeWit 1998; Oishi and Schimmack 2010).

The existing research in this area has three major limitations. First, to our knowledge, no study has examined the mental health consequences of residential moves during the transition to adulthood despite the high rate of moves during this period. Second, these studies paid little attention to moving distance, so the mental health consequences of migration (moves across counties), as opposed to short-distance moves, are unclear. Third, these studies only measured whether or how many times a person had moved in a given time period, and they did not consider the possibility that the mental health consequences of migration may depend on characteristics of origin and destination areas. The present study addresses these shortcomings by using direct measures of geographical distance and contextual differences of residing areas between adolescence and young adulthood. Among various contextual characteristics, we focus on urbanicity and political progressiveness (as measured by residents’ education levels), which are strongly linked to political climates for sexual diversity and the visibility of sexual minority communities (Gates and Ost 2004; Kane 2003).

Although some studies reported negative mental health consequences of residential moves, others suggested the possibility that residential moves, migration in particular, may improve mental health under certain circumstances. Specifically, people who face disadvantages and stigmatization in their currently residing areas may improve their mental health by migrating to a more protective area. Consistent with this argument, among older black people born in the South, those who moved away from the region have better functional health (Kington et al. 1998). A similar process may operate for sexual minorities who migrate in an effort to reduce exposure to sexuality stigma.

The present study conceptualizes sexual minorities’ migration as one of their coping behaviors, which have received little attention in empirical research despite the theoretical importance (Herek and Garnets 2007; Meyer 2003). Among the small number of past studied, a survey of racial-ethnic minority lesbian, gay, bisexual, and transgender youth demonstrated that they employ different coping strategies depending on the types of sexuality-related stressors (Kuper, Coleman, and Mustanski forthcoming). Other studies have examined mental health consequences of coping behaviors among lesbian women and found that avoidance, suppressive, and reactive coping behaviors are associated with increased levels of psychological distress although coping behaviors that are conceptualized as more adaptive (reflective coping) do not show any associations (Szymanski and Henrichs-Beck 2014; Szymanski and Owens 2008). To extend these studies that considered a wide range of coping behaviors within sexual minorities’ present living environments, the present study focuses on a more extreme behavior, internal migration, which involves movements across social environments. Further, unlike previous studies, our study includes a comparison group of heterosexuals and examines whether the mental health implications of migration differ between sexual minorities and heterosexuals and what implications migration has for the disparities between the sexuality groups.

Sexual Orientation, Internal Migration, and Mental Health

The existing literature provides little information about how migration impacts mental health among sexual minorities, but some cross-sectional studies have examined how their mental health varies by residing locations. For example, adolescent studies show that sexual minority adolescents living in urban areas have lower levels of depression and substance use than those living in rural areas (Poon and Saewyc 2009; Wilkinson and Pearson 2009). Contrary to this finding that indicated advantages of living in urban areas, however, Wienke and Hill’s (2013) recent analysis of three national data sets showed lower levels of happiness among sexual minorities residing in large cities relative to those in less urban areas including midsized cities, suburbs, and rural areas. Although not directly related to urbanicity, an adult study also reported that lesbians living in non-Southern states have lower rates of depression than those in Southern states (Austin and Irwin 2010), perhaps due to the more accepting climates in the non-Southern states.

These mixed findings mirror inconclusive results for the general population (Atav and Spencer 2002; Kessler et al. 2005; Levine and Coupey 2003), but they may also indicate that living in urban areas has conflicting effects on sexual minorities’ lives. For example, sexual minority men may gain access to coping resources through large sexual minority communities by migrating to urban areas, but the steep hierarchy and competition in the sexual market may increase their stress (Green 2008). Further, drug-saturated club cultures among sexual minority men in those settings may exacerbate their drug use (Green 2003; Parsons, Kelly, and Weiser 2007).

Given the conflicting results in cross-sectional associations between characteristics of residential areas and sexual minorities’ mental health, it is not surprising that the existing literature provides little information about the implications of migration for their mental health. One exception is Wienke and Hill’s (2013) aforementioned study, which measured migration experience by the population size of residential locations at the time of the interview and in adolescence. The analysis showed that gay men’s mental health did not vary by their migration history. Among women, however, those who migrated to large cities and rural areas showed higher levels of happiness than those who stayed in those areas. The study had three important limitations. First, the study analyzed sexual minorities in all stages of adulthood together and could not address the possibility that sexual minorities may show a unique pattern of migration and mental health consequences during the transition to adulthood. This is an important limitation because sexual minorities may move multiple times over the life course to different areas for different reasons (Annes and Redlin 2012). Second, the study did not include baseline measures of mental health before migration, which reduced the ability to specify the causal direction. Third, the samples only consisted of sexual minorities and thus could not address how the impact of migration may differ between sexual minorities and heterosexuals.

More recently, Everett (2014) used the Add Health data to examine how changes in neighborhood characteristics are associated with changes in depressive symptoms among sexual minorities during the transition to adulthood. The analysis showed that decreases in urbanicity and political progressiveness were associated with increases in depressive symptoms. This study also had several important limitations. First, because the study focused on neighborhood characteristics, it could not infer mental health consequences of migration or changes in social climates in broader contexts. Second, like Wienke and Hill’s (2013) study, this study was limited to sexual minorities. Third, the study only considered adulthood sexual identity as a measure of sexual orientation, which not only left unexamined the impact of adolescent same-sex sexuality but also limited the ability to specify the causal order. Fourth, the analysis combined women and men, who may have different migration patterns and experience different consequences of migration. The present study seeks to overcome these limitations of Wienke and Hill’s (2013) study and Everett’s (2014) study.

The Present Study

This study conceptualizes sexual minorities’ internal migration as one of their coping behaviors to escape sexuality stigma and examines its implications for their mental health disadvantages while focusing on the transition to adulthood as a life course context. The study makes improvements from previous research by assessing migration experience directly from the geographic distance between origins and destinations, instead of relying on respondent recalls, and by incorporating contextual characteristics of these locations. Further, the study employs a longitudinal design, which not only allows us to control for the baseline mental health states but also aids us in addressing the fluidity in sexual orientation during the life stage transition—among people who report same-sex sexuality, some people report it in both adolescence and young adulthood whereas others report it only in one life stage (Savin-Williams and Ream 2007; Ueno 2010b). The contrast between the former group and those who do not report same-sex sexuality in either life stage will show whether same-sex sexuality that is sustained from adolescence to young adulthood impacts migration patterns and mental health. Other groups may also provide important insights. For example, reporting of same-sex sexuality limited to adolescence may indicate self-exploration in an early stage of sexual development (Hewitt 1998) or effort to suppress same-sex desire as they become older (Troiden 1989). Another reporting pattern of emerging awareness of same-sex sexuality in young adulthood may reflect a high level of social control such as parents’ heteronormative expectations in adolescence or entry into a more accepting social environment such as college campus in young adulthood. Longitudinal data therefore allow us to consider the possibility that these factors associated with the dynamic reporting patterns of same-sex sexuality are also linked to migration and mental health.

We test the following two hypotheses:

Hypothesis 1: We expect that sexual minorities are more likely than heterosexuals to migrate, especially to urban places with politically progressive climates.

We expect that women may not show large sexuality group differences because their migration destinations seem to vary substantially with some of them heading to rural areas (Bell and Valentine 1995; Kazyak 2012). During the transition to adulthood, some sexual minorities migrate only temporarily to go to college, like heterosexuals do. This factor does not necessarily make the analysis less meaningful because sexual minorities may incorporate their effort to escape sexual stigma into their college choice and because migrating to a different area for college attendance often operates as a turning point in the life course (Lehmann 2014). Nonetheless, the overall high rate of migration in this life stage may reduce the extent of sexual orientation difference to be observed.

In our second hypothesis:

Hypothesis 2: We hypothesize that migration will improve mental health to a greater degree among sexual minorities than heterosexuals.

Although previous studies suggested negative consequences of relocations in early life stages for the general population, people who manage to relocate across counties may improve their living conditions especially during the transition to adulthood, when migration tends to be voluntary and motivated by attempts to seek better education, employment, and social opportunities (Geist and McManus 2008). Add Health does not allow us to discern whether migration reflects the respondent’s effort to avoid exposure to sexuality stigma. However, if sexual minorities are more likely to migrate for such reason than heterosexuals, they should receive greater mental health benefits. We further expect that this mental health benefit linked to sexual minorities’ migration will be more clearly observed for men than women because more serious stigma is attached to men’s same-sex sexuality than women’s same-sex sexuality (Herek 2002), which may amplify the extent to which migration reduces the level of exposure to sexuality stigma for men. Further, the sexuality difference in the mental health consequences may depend on destination characteristics. Considering the large, well-developed communities of sexual minority men in urban areas and the importance of political climates that strongly shifts sexual minority men’s exposure to sexuality stigma, they may improve mental health to a great extent by moving to more urban and progressive areas. In contrast, women may not show large sexual orientation differences because some sexual minority women’s affinity with country living may counteract the relative disadvantage of moving to rural areas to some extent. These hypotheses are tested in the analysis of the Add Health data.

Data and Methods

Data and Sample

Data came from the National Longitudinal Study of Adolescent Health (Add Health). In 1994, 80 high schools and 52 feeder schools (middle schools that sent graduates to those high schools) in the United States were selected for the schoolwide survey (in-school survey). Among the initial respondents, 20,745 students participated in an in-depth interview in 1995 (wave 1 in-home). About a year later, wave 2 in-home interviews were conducted with the same respondents, except those who had graduated. All wave 1 respondents, including those who did not participate in wave 2, were eligible to participate in wave 3 in-home interviews between 2001 and 2002. The wave 3 data included 15,170 respondents (73.3% of the wave 1 sample), and most of them were age between 19 and 25. Although available, wave 4 data (2008/2009) are not used in our analysis because the survey was conducted long after the transition to adulthood (the focus in the present study) and because residential location data have not been released for the wave. For this article, we treated waves 1 and 2 as adolescent data and wave 3 as young adulthood data. To avoid age overlap in the operationalization of the two life stages, the analysis focused on people under 20 years old at the time their adolescent mental health was measured in wave 1 or 2 and 20 years old or older at the time of the wave 3 interview.

Two restrictions were applied to the analysis. Among 15,170 people who participated in the wave 3 interview, 848 respondents were excluded because they were not part of the core sample and thus did not have sampling weights. An additional 1,609 respondents were dropped because they did not meet the age criteria mentioned previously. The final operational sample in the primary analysis consisted of 6,661 women and 6,052 men.

Measures

The key variables are described below. Table 1 presents means and standard deviations of each variable.

Table 1.

Descriptive Statistics by Gender.

		Women		Men
	Min, Max	Mean	SD	Mean	SD
Depressive symptoms (W3)	0, 26	4.91	4.60	4.03	3.63
Binge drinking (W3)	0, 12	1.84	2.55	3.12	3.12
Drug use (W3)	0, 120	2.16	7.32	4.47	10.20
Depressive symptoms (W2/1)	0, 56	12.05	8.63	10.12	6.80
Binge drinking (W2/1)	0, 12	1.41	2.54	1.94	2.95
Drug use (W2/1)	0, 103	1.52	5.88	2.43	7.80
Same-sex sexuality
Not reported	0, 1	.79		.88
Adolescence only	0, 1	.03		.04
Adulthood only	0, 1	.15		.06
Both stages	0, 1	.03		.02
Migrated	0, 1	.22		.22
Migrated across regions	0, 1	.07		.06
Population density (change)	−12.55, 25.84	.21	2.06	.25	2.15
Proportion urban residents (change)	−1, 1	.07	.32	.07	.30
Proportion college degree (change)	−.35, 53	−.01	.08	−.01	.07
Age W3	20, 27	22.14	1.62	22.25	1.55
Race
White	0, 1	.68		.67
Black	0, 1	.16		.15
Hispanic	0, 1	.12		.12
Other	0, 1	.05		.06
Parent education	1, 4	2.70	1.12	2.73	1.05
College graduated/attending (W3)	0, 1	.36		.29
Religiosity	1, 4	2.60	1.05	2.45	.97
Region (W2/1)
West	0, 1	.17		.16
Midwest	0, 1	.33		.31
South	0, 1	.37		.39
Northwest	0, 1	.13		.14
Population density (W1)	.00, 20.24	.46	1.52	.51	1.54
Proportion urban residents (W1)	0, 1	.60	.41	.61	.38
Proportion college degree (W1)	.08, 49	.25	.09	.25	.08
Crime rate (change)	−2,865.39, 2,111.38	−127.97	425.00	−144.13	377.87
Proportion same house (change)	−.34, .39	.00	.07	.00	.06
Median household income (change)	−32,949, 60,884	12,211.63	7,050.59	12,132.46	6,428.60
Adolescent data from W1	0,1	.24		.25

Note: N = 6,661 Women and 6,052 Men.

Sexual Orientation

The sexual orientation measure combined responses regarding attraction, dating relationships, and sexual identity in waves 1, 2, and 3. Respondents reported their attraction in all three waves. The questions in waves 1 and 3 asked, “Have you ever had a romantic attraction to a female/male?” The questions in wave 2 were worded differently and measured attraction since the wave 1 interviews. For dating relationships, wave 1 included three sets of questions. In the first set, respondents reported whether they had engaged in any “romantic” relationships in the last 18 months. To solicit information about relationships adolescents did not necessarily consider as romantic, those who reported none were asked whether they had held hands with, kissed, or told anyone (other than a family member) that they liked or loved him or her in last 18 months. When respondents reported having engaged in all three activities, Add Health called it a “liked” relationship. Respondents also reported any otherwise unreported sexual relationships (“nonromantic sexual” relationships). The gender of each partner (up to three romantic or liked partners and up to three nonromantic sexual partners) was recorded. The wave 2 interview included the same sets of questions about dating relationships occurring since wave 1. In wave 3, respondents reported romantic or sexual relationships since summer of 1995. Sexual identity was measured only in wave 3. Respondents were asked, “Please choose the description that best fits how you think about yourself? (100% heterosexual/straight, mostly heterosexual/straight, bisexual, mostly homosexual/gay, 100% homosexual/gay, or not sexually attracted to either males or females).”

We first analyzed the three measures separately because they targeted unique dimensions of sexual orientation (Badgett 2009). We were particularly interested in whether sexual identity would show a strong relationship with migration because one’s decision to migrate may presume nonheterosexual identity if such a decision is made in an effort to reduce exposure to sexuality stigma. However, none of these adolescent and young adulthood measures of sexual orientation, including sexual identity, showed significant associations with migration patterns when analyzed separately.

In our primary analysis, we combined the three sets of sexual orientation measures while paying special attention to continuities and changes between adolescence and young adulthood because the dynamic aspect of sexual orientation may shift the extents to which people experience sexuality stigma and seek to minimize it and because it is linked to mental health changes between adolescence and young adulthood as mentioned earlier. To construct the variable, we first identified whether respondents reported same-sex attraction, same-sex dating relationships, or same-sex sexual identity in each life stage (see more details about this coding technique in Ueno 2010b). For sexual identity, we considered respondents as having same-sex identity if they described their orientation as homosexuality or bisexuality (“bisexual,” “mostly homosexual but somewhat attracted to people of the opposite sex,” or “100% homosexual”). Moving “mostly heterosexual” to the same-sex identity category did not change the overall conclusions. Using these pieces of information regarding presence and absence of same-sex experience in adolescence and young adulthood, we then classified respondents into the following four groups: (1) no reporting of same-sex sexuality (5,272 women, 5,310 men), (2) adolescence only (222 women, 225 men), (3) young adulthood only (989 women, 425 men), and (4) both life stages (178 women, 92 men).¹ Due to small cell sizes, this coding did not distinguish between bisexual orientation and exclusive same-sex orientation.

Migration Experience and Changes in Residing County Characteristics

Two sets of migration measures were constructed and separately entered in multivariate models. First, a dichotomous measure of migration experience was created. In waves 1 through 3, each respondent’s home address was geocoded. To protect respondents’ confidentiality, Add Health did not disclose the addresses but provided geographical distances (straight lines) across waves (Swisher 2008). To measure migration experience between adolescence and adulthood, the study focused on the distance between wave 2 and wave 3 locations for those who participated in wave 2 (about 76% of the operational sample) and the distance between waves 1 and 3 for other respondents.² The dichotomous measure of migration distinguished between people who moved 50 miles or more (1 = migrated) and those who did not (0 = did not migrate). The cutoff of 50 miles was based on previous studies (e.g., Ham, Li, and Reagan 2011).

In addition to geographical distances, Add Health included information about the region in which respondents resided in each wave (West, Midwest, South, or Northwest). Using this information, another dichotomous measure was created to assess whether respondents moved across regions (coded ‘1’) or not (coded ‘0’). We expected that this variable would show larger sexual orientation differences in migration rates and in mental health consequences than the other dichotomous measure because cross-region migration might indicate more drastic changes in living environment. Add Health did not conduct follow-up interviews with respondents who had left the United States, so these migration measures only considered internal migration.

Second, changes in residing county characteristics were computed between waves 1 and 3. For the reasons mentioned previously, we focused on population density (number of residents per square kilometer), proportion of residents living in urban area, and proportion of residents age 25 and above holding college degrees. These measures came from the Add Health’s contextual database, which linked respondents’ wave 1 residence to the 1990 census and their wave 3 residence to the 2000 census (Billy, Wenzlow, and Grady 1998; Swisher 2008). Increases in population density and proportion of urban residents between the two life stages were assumed to indicate migration to a more urban area, and increases in proportion of college educated residents were conceptualized as a sign of migration to a politically more progressive area.³ For political progressiveness, residents’ political attitudes and political affiliations would have served as more direct measures, but such information was not available in the database. Residents’ education level served as a good proxy measure because education promotes progressive ideologies such as endorsement of social diversity (Ohlander, Batalova, and Treas 2005) and because progressive areas attract educated migrants (Florida 2002). Because residents’ education level may have reflected factors other than political progressiveness, the analysis controlled for those factors as explained below.

For these measures of county characteristics, we assumed that changes mostly reflected migration, instead of changes within the same counties. Consistent with the assumption, a majority of those who reported substantial changes (at least a half standard deviation away from the mean change in each county characteristic) were classified as “migrated” in the dichotomous measure mentioned previously. For these measures of changes in county characteristics, we examined the possibility that they were associated with mental health in a nonlinear fashion. In a supplemental analysis, however, square terms and trichotomized versions of these measures did not show nonlinear associations, so we present the results based on liner associations.

Mental Health

The analysis considered three mental health outcomes—depressive symptoms, binge drinking, and substance use. As in other studies based on the stress framework, we conceptualized depressive symptoms as an internalizing stress response and binge drinking and substance use as externalizing responses (Aneshensel, Rutter, and Lachenbruch 1991). Including externalizing symptoms was particularly important for examining the effect of migration because migration may indicate high levels of self-exploration and reductions in parental control in adulthood, which may counteract the advantages of migration in these outcomes.

Measures of young adulthood mental health were constructed from wave 3. Depressive symptoms was measured by a short version of the Center for Epidemiologic Studies Depression Scale (CES-D) (Radloff 1977). This version consisted of nine items and asked about physiological and psychological symptoms during past seven days (e.g., “you were sad,” “you could not shake off the blues”; 0 = never or rarely to 3 = most of the time or all of the time; Cronbach’s α = .79). Binge drinking summed two 7-point items, which asked on how many days they “drank five or more drinks in a row” and “gotten drunk or ‘very, very high’ on alcohol” over the past 12 months (0 = none to 6 = every day or almost every day) (r = .82). Drug use summed four items, each of which asked about how many times respondents had used a specific type of drug (marijuana, cocaine, crystal meth, and other types of illegal drugs) in the last 30 days. As control variables, the analysis also included adolescent mental health measures. These measured were constructed from corresponding items in wave 2 (or wave 1 when not available), except that adolescent depressive symptoms were based on a 19-item CES-D scale, instead of the 9-item version.

Control Variables

The analysis controlled for the following sociodemographic background variables due to its correlation with sexual orientation (Laumann et al., 1994) and with migration (Ihrke and Faber 2012) or mental health (Grollman 2012; Needham 2007). Age in wave 3 was measured in years. Race was based on wave 1 data and consisted of four dummy variables including non-Hispanic white, non-Hispanic black, Hispanic, Asian, and others. Parents’ educational background consisted of four dummy variables including less than high school, high school graduate, some college, and college graduate. Religiosity used wave 2 data (or wave 1 if not available) and was based on Adamczyk and Palmer’s (2008) scale, which summed four 4-point items (e.g., “How often did you attend religious services?” “How important is religion to you?”; α = .85). College enrollment distinguished between people who reported in wave 3 that they had earned a college degree or were currently attending college (coded ‘1’) and others (coded ‘0’). For this variable, the definition of college focused on four-year institutions with an assumption that many people who migrate to seek higher education attend four-year colleges whereas those seeking associate degrees tend to attend community colleges in the currently residing areas. The analysis also included a flag variable that identified whether the respondent’s adolescent data came from wave 1 (coded ‘1’) or wave 2 (coded ‘0’).

We also controlled for four county-level variables. Among them, three variables measured changes between waves 1 and 3 in household wealth (median household income in dollars), residential stability (proportion of people living in the same house as four years ago), and crime rate (violent crimes per 100,000 residents). Another county-level variable measured wave 1 region as a set of dummy variables. These control variables were used in the analysis because they correlated with urbanicity and political progressiveness and might also affect mental health by shifting the levels of perceived safety risks, a sense of normlessless, and availability of drugs (Aneshensel and Sucoff 1996; Wickrama, Noh, and Bryant 2005).

Analysis Plan

The first set of analyses examined whether people reporting same-sex sexuality were more likely to migrate especially to urban areas with a politically progressive climate (Hypothesis 1).⁴ Logistic regression models were used to predict migration experience and cross-region migration, and OLS models were used to predict changes in population density, proportion urban residents, and proportion college educated residents. Residing county characteristics in adolescence were entered as controls because they might affect the decision about whether and where to migrate as well as the chance of reporting same-sex sexuality.

In the second set of multivariate analysis, we used ordinary least squares (OLS) models to regress mental health in young adulthood on migration and sexual orientation while controlling for mental health in adolescence and sociodemographic background. Because adolescent mental health was entered in the model, the coefficient of each predictor indicated its association with changes in mental health between adolescence and young adulthood (Finkel 1995). The analysis first entered sexual orientation as a predictor and added migration in a subsequent model to see whether migration accounted for or suppressed sexual orientation differences in mental health changes. To test whether sexual minorities received greater benefits from migration (Hypothesis 2), multiplicative terms between sexual orientation and migration variables were entered in a subsequent model.

All analyses were conducted separately for women and men because the meaning of sexual orientation and the degree of stigma attached to same-sex orientation differ by gender (Herek 2002). Throughout the analyses, we used the “survey design” routine in Stata 12, which corrected standard errors for data dependency among those who attended the same schools in adolescence. As Chantala (2006) recommends, longitudinal weights were used along with subpopulation specification, so the results are generalizable to people who were enrolled in 7th through 12th grades in U.S. schools during the 1994–1995 school year and met the age criteria mentioned previously.

Results

Sexual Orientation Differences in Migration Patterns

We first present results from logistic and OLS regression models, which predicted migration patterns by sexual orientation. Among 10 models run for all combinations of genders and migration variables, only 2 models showed significant associations. Table 2 summarizes results from these 2 models. Model 1 shows that compared to women who did not report same-sex sexuality (the reference), those reporting same-sex sexuality only in adolescence were 73% more likely to migrate (exp[.55] = 1.73). In addition, women reporting emerging same-sex sexuality (the Adulthood Only group) were 30% more likely to migrate than the reference group (exp[.26] = 1.30). The Both Life Stage group did not significantly differ from the reference group. Model 2 presents the other significant sexuality group difference—among men, the population density in the residing county increased .74 points more for the adulthood only group than the not reported group (p < .001). Given the standard deviation of the variable (2.15), this group difference was fairly small. In short, the analysis produced mixed results for Hypothesis 1, which stated that sexual minorities are more likely than heterosexuals to migrate. An overall sexuality group difference in the migration rate was observed only for women, and greater movements toward urban counties were observed only for men’s analysis of population density. Contrary to previous research (Cooke and Rapino 2007), sexual minority women did not migrate to more rural areas than heterosexual women.

Table 2.

Regression Models Predicting Women’s Migration and Men’s Change In Population Density.

	Women’s Migration	Men’s Change in Population Density
	(logistic regression coefficients)	(ordinary least squares regression coefficients)
Same-sex sexuality (ref: not reported)
Adolescence only	.55*	.05
Adulthood only	.26*	.74*
Both stages	.11	.55
Age in wave 3	.01	.03
Race (ref: white)
Black	−.26	.91*
Hispanic	−.62**	−.24*
Other	.01	.31
Parent education	.23***	.08
Religiosity	.01	.01
College in W3	1.01***	.03
Population density (W2/1)		−.07
Adolescent data from W1	.03	−.32
Constant	−2.55**	−.47

Note: N = 6,661 Women and 6,052 Men.

p < .05. **p < .01. ***p < .001.

Two supplemental analyses were conducted to elaborate on these results. First, we examined whether the significant sexuality group differences in migration patterns could be explained by sexual minorities’ elevated stress exposure (e.g., interpersonal problems with peers, arguments with parents), lack of social support in adolescence (e.g., emotional distance from school and parents), residing region in adolescence, or college enrollment. The analysis showed, however, that these factors contributed little to the explanation of the group differences. Second, we used interaction models to test whether the sexuality group differences in migration patterns depended on sociodemographic attributes such as age, race, and parent education or by residing region in adolescence. The analysis revealed no significant interactions, however, indicating that sexual orientation was not strongly associated with migration patterns regardless of these factors.

The Association among Sexual Orientation, Migration Patterns, and Mental Health

The next analysis estimated OLS regression models to examine whether migration patterns were associated with changes in mental health and whether the association depended on sexual orientation. Table 3 presents results for women’s depressive symptoms. Model 1 entered sexual orientation and individual-level control variables as predictors. In this model, the adulthood only group and the both stage group showed significant, positive coefficients, indicating that their depressive symptom scores were 1.50 points higher for the former group (p < .001) and 1.20 points higher for the latter group (p < .01), compared to the not reported group (reference). These differences were small partly because the model controlled for depressive symptoms in adolescence. Model 2 added the dichotomous measure of migration as well as county-level control variables and revealed that migration was associated with a small reduction in depressive symptoms (b = −.28, p < .05). Model 3 added interaction terms between sexual orientation and migration to examine whether the association between migration and depressive symptoms varied across sexuality groups. None of the interaction terms was significant, however. In a similar manner, the association between cross-region migration and depressive symptoms was tested in the next two models. Model 4 showed no main effect, but Model 5 revealed significant interactions with sexual orientation. Specifically, the adulthood only group and the both stages group showed negative interaction coefficients (b = −1.38, p < .01 and b = −2.43, p < .01, respectively), indicating that cross-region migration was associated with reductions in depressive symptoms for women in these groups, consistent with Hypothesis 2. For changes in population density, Model 6 showed no main effect, and Model 7 revealed an unexpected interaction—increases in population density were positively associated with depressive symptoms for the adolescent only group and the adulthood only group (b = .18, p < .001 and b = .10, p < .05, respectively). We will return to this finding. Changes in proportion urban residents did not show main effect (Model 8), but consistent with Hypothesis 2, increases in this variable were linked to reductions in depressive symptoms for women in the adulthood only group (−2.78, p < .01, Model 9). Finally, changes in proportion of college-educated residents did not show main or interaction effects (Models 10 and 11, respectively).

Table 3.

Ordinary Least Squares Regression Models Predicting Depressive Symptoms in Young Adulthood (women).

	1	2	3	4	5	6	7	8	9	10	11
Same-sex sexuality (ref: not reported)
Adolescence only	.33	.36	.18	.22	.11	.34	.17	.33	.45	.33	.33
Adulthood only	1.50***	1.52***	1.62***	1.46***	1.58***	1.50***	1.45***	1.48***	1.66***	1.49***	1.45***
Both stages	1.20**	1.21**	1.47**	1.11**	1.35**	1.20**	1.10**	1.17**	1.22**	1.20**	1.18**
Migrated		−.28*	−.21
Interaction with:
Adolescence only			.56
Adulthood only			−.40
Both stages			−1.09
Migrated across region				−.25	.12
Interaction with:
Adolescence only					.85
Adulthood only					−1.38**
Both stages					−2.43**
Population density (change)						−.01	−.03
Population density (W2/1)						−.02	−.03
Interaction with:
Adolescence only							.18***
Adulthood only							.10*
Both stages							.19
Proportion urban residents (change)								.07	.61
Proportion urban residents (W2/1)								.57**	.56**
Interaction with:
Adolescence only									−1.67
Adulthood only									−2.78**
Both stages									−1.06
Proportion college degree (change)										−.48	−.01
Proportion college degree (W2/1)										1.22	1.20
Interaction with:
Adolescence only											1.42
Adulthood only											−3.90
Both stages											−1.04
Age in wave 3	−.11	−.11	−.11	−.11	−.11	−.11	−.11	−.11	−.11	−.11	−.11
Race (ref: white)
Black	.57**	.52**	.52**	.54**	.54**	.54**	.54**	.51**	.52**	.52**	.52**
Hispanic	.36	.30	.31	.30	.31	.34	.35	.15	.15	.27	.27
Other	.30	.27	.25	.28	.28	.28	.29	.20	.19	.25	.26
Parent education	−.08	−.07	−.06	−.08	−.08	−.08	−.08	−.10	−.09	−.09	−.09
Religiosity	−.05	−.05	−.06	−.07	−.07	−.05	−.06	−.04	−.04	−.05	−.05
College in W3	−.84***	−.81***	−.81***	−.85***	−.85***	−.86***	−.86***	−.88***	−.90***	−.87***	−.86***
Depressive symptoms (W2/1)	.16***	.16***	.16***	.16***	.16***	.16***	.17***	.17***	.17***	.16***	.16***
Region W1/2 (ref: North)
West	.13	.24	.24	.19	.18	.15	.16	.23	.21	.17	.16
Midwest	−.38	−.39	−.39	−.39	−.38	−.42	−.41	−.40	−.42	−.36	−.35
South	−.24	−.19	−.19	−.21	−.21	−.25	−.23	−.09	−.10	−.12	−.13
Crime rate (change)		−.00	−.00	−.00	−.00	−.00	−.00	−.00	−.00	−.00	−.00
Proportion same house (change)		−1.61	−1.58	−1.23	−1.03	−1.19	−1.19	−1.70	−1.47	−1.81	−1.81
Median household income (change)		−.00	−.00	−.00	−.00	−.00	−.00	−.00	−.00	−.00	−.00
Adolescent data from W1	−.13	−.13	−.13	−.13	−.12	−.13	−.13	−.12	−.11	−.14	−.14
Constant	5.77***	5.85***	5.84***	5.97***	5.97***	5.86***	5.89***	5.69***	5.52***	5.59***	5.53***
R ²	.15	.16	.16	.15	.16	.15	.16	.16	.16	.16	.16
	6,661	6,661	6,661	6,568	6,568	6,661	6,661	6,661	6,661	6,661	6,661

Note: N = 6,661.

p < .05. **p < .01. ***p < .001.

Figure 1 visually presents the result from Model 9, which showed one of the significant interactions. The x-axis (changes in proportion urban residents) is centered at the mean and extends to two standard deviations below and above the mean. The figure shows that among women who experienced large reductions in proportion urban residents (assumed to reflect migration to a more rural place), the adulthood only group had higher levels of depressive symptoms. However, their symptoms quickly declined as the change in proportion urban residents became smaller and turned positive (assumed to reflect migration to a more urban area). The slope for this group significantly differed from that of the not reported group (p < .01), which showed a positive association between proportion urban residents and depressive symptoms.

Figure 1.

Women’s Depressive Symptoms by Change in Proportion Urban Residents and Sexual Orientation.

The results for men’s depressive symptoms are summarized in Table 4. Model 1 showed that depressive symptom score was .90 point higher for the adulthood only group than the not reported group (p < .01). Models 2 and 3 revealed that the dichotomized measures of migration experience did not have main or interaction effects, and Models 4 and 5 similarly showed null results for cross-region migration. Although population density did not show a significant main effect (Model 6), it interacted with sexual orientation as shown in Model 7—increases in population density were associated with reductions in depressive symptoms only in the adulthood only group (b = –.08, p < .05). Similarly, although changes in proportion urban residents showed no main effect (Model 8), increases in the variable were associated with reductions in depressive symptoms for the both stage group (Model 9, b = −2.15, p < .01). Finally, changes in proportion college educated residents showed no main effect (Model 10), but the interaction model revealed that increases in this variable were linked to reductions in depressive symptoms for the adulthood only group (Model 11, b = −4.27, p < .05). These interactions were consistent with Hypothesis 2, which stated that migration to more urban, progressive areas improves mental health to a greater degree for sexual minorities than heterosexuals.

Table 4.

Ordinary Least Squares Regression Models Predicting Depressive Symptoms in Young Adulthood (Men).

	1	2	3	4	5	6	7	8	9	10	11
Same-sex sexuality (ref: not reported)
Adolescence only	−.10	−.10	−.03	−.07	−.05	−.10	−.07	−.12	−.13	−.09	−.10
Adulthood only	.90**	.90**	1.08**	.97**	.96**	.90**	.98**	.89**	.94**	.89**	.92**
Both stages	−.11	−.11	−.37	−.07	−.09	−.11	−.04	−.11	−.05	−.13	−.18
Migrated		.05	.09
Interaction with:
Adolescence only			−.56
Adulthood only			−.69
Both stages			1.04
Migrated across region				.19	.19
Interaction with:
Adolescence only					−.25
Adulthood only					.17
Both stages					.33
Population density (change)						.01	.02
Population density (W2/1)						−.02	−.01
Interaction with:
Adolescence only							−.07
Adulthood only							−.08*
Both stages							−.06
Proportion urban residents (change)								−.22	−.14
Proportion urban residents (W2/1)								.40	.41*
Interaction with:
Adolescence only									.67
Adulthood only									−.82
Both stages									−2.15**
Proportion college degree (change)										.46	1.02
Proportion college degree (W2/1)										1.55	1.61
Interaction with:
Adolescence only											−1.10
Adulthood only											−4.27*
Both stages											−4.52
Age in wave 3	−.09*	−.09*	−.09*	−.10*	−.10*	−.09	−.09	−.10*	−.10*	−.09*	−.09*
Race (ref: white)
Black	.56**	.55**	.55**	.54**	.54**	.56**	.55**	.56**	.57**	.53**	.52**
Hispanic	.27	.26	.25	.25	.25	.27	.27	.15	.16	.20	.19
Other	.60	.59	.60	.62	.62	.59	.59	.54	.54	.56	.56
Parent education	−.04	−.05	−.05	−.05	−.05	−.04	−.05	−.05	−.06	−.06	−.06
Religiosity	.06	.05	.05	.06	.06	.05	.05	.06	.06	.06	.06
College in W3	−.36*	−.38*	−.38*	−.37*	−.37*	−.37*	−.37*	−.39**	−.39**	−.41**	−.41**
Depressive symptoms (W2/1)	.19***	.19***	.19***	.19***	.19***	.19***	.19***	.19***	.19***	.19***	.19***
Region W1/2 (ref: North)
West	−.21	−.17	−.17	−.16	−.16	−.19	−.19	−.11	−.11	−.18	−.18
Midwest	−.39	−.39	−.39	−.38	−.38	−.41*	−.41*	−.40	−.40	−.34	−.34
South	−.47*	−.46*	−.45*	−.44*	−.44*	−.48**	−.48**	−.36	−.36	−.36	−.35
Crime rate (change)		−.00	−.00	−.00	−.00	−.00	−.00	.00	.00	−.00	−.00
Proportion same house (change)		−.66	−.65	−.64	−.66	−.87	−.88	−1.53	−1.50	−1.14	−1.07
Median household income (change)		−.00	−.00	−.00	−.00	−.00	−.00	.00	.00	−.00	−.00
Adolescent data from W1	−.20	−.20	−.20	−.21	−.20	−.20	−.20	−.20	−.20	−.20	−.20
Constant	4.46***	4.51***	4.50***	4.62***	4.63***	4.51***	4.49***	4.30***	4.30***	4.14***	4.15***
R ²	.15	.15	.15	.15	.15	.15	.15	.15	.15	.15	.15

Note: N = 6,052.

p < .05. **p < .01. ***p < .001.

Figure 2 visually presents the result from Model 11. Among men who experienced reductions in proportion of college-educated residents (assumed to reflect migration to a less progressive county), the adulthood only group had higher levels of depressive symptoms. This group’s symptom level declined quickly, however, as the change in proportion of college-educated residents became smaller and turned positive (i.e., migration to a more progressive county). The slope for this group significantly differed from the not reported group (p < .05), whose symptom level did not change much with proportion of college-educated residents. Although the both stage group also showed a negative slope, the difference from the reference group did not reach statistical significance partly due to the small group size.

Figure 2.

Men’s Depressive Symptoms by Change in Proportion of College-educated Residents and Sexual Orientation.

In both Tables 3 and 4, the coefficients for sexuality groups changed very little between the base model (Model 1) and each of the main effect models that added a different migration variable (Models 2, 4, 6, 8, and 10). This stability in the sexuality group coefficients reflected two factors: (1) sexuality groups did not differ in many of the migration variables as demonstrated previously and (2) migration was not strongly associated with mental health changes. Thus, migration patterns did not help explain nor suppress sexual minorities’ worse mental health changes during the transition to adulthood.

The multivariate analysis was repeated for binge drinking and drug use. Because these two outcome variables revealed no significant main effects of migration variables and only three significant interactions between sexual orientation and migration variables, the results are only briefly discussed here. (Detailed results are available from the authors upon request.) The analysis of binge drinking revealed a significant interaction among men—increases in population density were linked to reduced frequency in binge drinking only for the adulthood only group. The analysis of drug use showed two significant interactions—among women, cross-region migration was associated with reductions in drug use only for the both stage group, and among men, increases in population density were linked to reductions in drug use only for the adulthood only group. In short, although the analysis of binge drinking and drug use showed a fewer number of significant interactions than the analysis of depressive symptoms, those that reached significance were in the direction consistent with Hypothesis 2.

As mentioned earlier, internal migration during the transition to adulthood may indicate temporary relocations for college attendance. The main analysis previously presented controlled for college attendance, but we further examined the implications of this issue by limiting the sample to people who had not earned a college degree and those who were not attending college at the time of the wave 3 interviews. The analysis based on this subsample showed results similar to the main analysis regarding sexuality group differences in the association between migration and mental health, although these differences were smaller in the subsample, partly due to the very small cell sizes for nonheterosexual groups.

In sum, migration during the transition to young adulthood did not seem to have major mental health consequences for heterosexuals. However, several interactions appeared between sexual orientation and migration variables, and with one exception, these interactions indicated that sexual minorities benefited more from migrating especially to urban and progressive areas, thereby providing some support for Hypothesis 2. As expected, the analysis also showed that sexual minorities’ greater mental health benefits linked to migration were more consistently observed for men than women and for depressive symptoms than binge drinking and drug use.

Discussion

The analysis of the Add Health data showed mixed results for Hypothesis 1. Sexual minorities’ higher migration rate is observed only for women, and sexual minorities’ greater movements toward urban areas appear only for men’s analysis of population density. Further, for both women and men, sexual minorities do not necessarily migrate to areas with higher proportions of college-educated residents. Three factors may account for this lack of substantial differences in migration patterns across sexuality groups. First, unlike previous studies that included people in different stages of adulthood, the present analysis focused on those transitioning to adulthood. The overall high rate of migration during this transition may have diminished sexuality group differences. Second, the present study focused on a recent birth cohort, and the increasing acceptance of nonheterosexuality in society (Saad 2012) may have reduced sexual minorities’ need to migrate to escape stigma. Third, sexual minorities’ migration destinations may have been diversified and may no longer focus on urban, progressive areas (Annes and Redlin 2012; Cooke and Rapino 2007).

The analysis also showed that migration has little impact on mental health for heterosexuals as far as depressive symptoms, binge drinking, and drug use are concerned. The result may indicate that counterbalancing factors operate for this group: Although migration may cause disruptions in daily routines and social networks, new economic and social opportunities may improve life conditions. The analysis also revealed that changes in urbanicity and progressiveness of the residing counties have little mental health implications, suggesting that considering these origin and destination characteristics does not have strong bearing on the effects of migration. Perhaps large-scale contexts do not affect mental states to a great extent, and proximal contexts such as neighborhoods, workplaces, and college campuses (for students) have a more direct impact.

The analysis provided some support for the argument that migration, especially to more urban and progressive areas, improves sexual minorities’ mental health. This effect is stronger for men than women as expected. Men face more serious stigma for expressing nonheterosexual orientation (Herek 2002), especially in rural, conservative areas (D’Augelli and Hart 1987; McCarthy 2000), and migration to urban places with progressive climates may help reduce exposure to stigmatization to a greater extent for sexual minority men than sexual minority women. It is also possible that migration to these areas provides greater access to resources within sexual minority communities among men, who dominate those communities.

By considering the timing and continuity of same-sex experience, the study provided important insights into the association between migration and mental health. People who report same-sex sexuality only in adolescence do not seem to improve their mental health by migrating. Perhaps they are not exposed to sexuality stigma or do not internalize it due to their temporary sexual experience, and migration does not substantially change supportiveness of their living environment. In contrast, migration seems to promote mental health among people who start reporting same-sex sexuality in young adulthood as well as those who consistently report it in adolescence and young adulthood. Migration provides mental health benefits more consistently for the former group, and this result is interesting because people in this group do not report same-sex sexuality in the adolescent wave. The findings for this group may be interpreted in a couple of ways. First, these people’s nonheterosexual orientation is undetected in the adolescent wave perhaps because they become aware of nonheterosexual orientation in late adolescence or very early adulthood. Given that emerging sexual awareness often involves psychological shocks and network disruptions (Troiden 1989), the present results may indicate that migration helps people cope with these initial stressors by letting them quickly rebuild sexual identities and social networks. Second, the absence of adolescent same-sex experience in this group may indicate that they suppress same-sex desire due to intolerant social climates in their residing areas, and migration may help them acknowledge such desire and reduce stress resulting from the suppression. Generally, these positive mental health changes linked to sexual minorities’ migration are less clear for binge drinking and drug use than depressive symptoms. This result is consistent with the argument that migration reflects self-exploratory attitudes and socialization into drug-saturated networks, which counteract the mental health benefits of migration.

An unexpected pattern emerged in the analysis of women’s depressive symptoms: increases in population density are associated with increases in symptoms among sexual minorities. The result is particularly interesting because increases in proportion urban residents are associated with reductions in depressive symptoms among these women as expected. (Population density and proportion urban residents show a moderate positive correlation in adolescence, young adulthood, and change scores.) The data do not provide sufficient information to identify underlying mechanisms, but past research provides clues. Some sexual minority women feel excluded in metropolitan areas, which tend to emphasize women’s traditional display of femininity (Kazyak 2012), and others feel oppressed by the male dominant culture in those areas (Bell and Valentine 1995). Perhaps high population density indicates intense social interaction in a metropolitan space, which increases such cultural emphases and the pressure to conform to them. In contrast, high proportion of people living in urban area may indicate an overall progressive climate (Loftus 2001), with or without such cultural pressures.

These results need to be understood in light of three study limitations. First, Add Health did not collect any information about reasons for migration, which precluded us from directly examining the extent to which sexual minorities’ migration reflected their effort to escape sexuality stigma in the currently residing area. There may be other factors that may impact the rate and destination of internal migration in different ways for heterosexuals and sexual minorities. Second, Add Health did not provide names or other identifying information of respondents’ residing counties to protect their confidentiality, which precluded our ability to add other county-level variables to the analysis. For the same reason, we were not able to consider county- or state-level policies regarding same-sex marriage and sexual orientation discrimination at work, which may have impacted sexual minorities’ decision to migrate as well as their mental health. Third, the study was unable to address heterogeneity—people who reported nonheterosexual orientation may have differed in unobserved personal attributes and residing county characteristics, which may have impacted the chance of migration and mental health. Propensity score matching technique is a possible strategy to address heterogeneity (Rosenbaum and Rubin 1983), but Add Health did not include variables that adequately predicted sexual orientation (i.e., a requirement for this statistical technique).

Conclusion

Certain migration patterns are associated with mental health improvements among sexual minorities whereas migration does not have major mental health implications for heterosexuals. Sexual minority women benefit from migrating across regions and from migrating to counties with high proportions of urban residents, and sexual minority men benefit from migrating to counties with higher population density, higher proportions of urban residents, and higher proportions of college-educated residents. Despite the greater mental health benefits, sexual minorities are not substantially more likely to migrate than heterosexuals. This result may indicate that some sexual minorities lack sufficient resources or face unique constraints. For example, the exposure to sexuality stigma in adolescence drain their psychological resources (Boatwright et al. 1996), which may undermine their ability to plan migration. More research is necessary to investigate why some sexual minorities remain in noxious social environments. Although we conceptualize sexual minorities’ migration as a coping behavior in this article, we do not intend to blame sexual minorities for their mental health disadvantages or hold them accountable for reducing the disadvantages. Social policies should aim at directly eliminating sexuality stigma and improving living conditions for sexual minorities. These policies should target rural, politically conservative areas because sexual minorities seem to experience more serious stigma in these areas.

Footnotes

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health website (). No direct support was received from grant P01-HD31921 for this analysis.

Notes

Author Biographies

Koji Ueno is an associate professor of sociology at Florida State University. His research examines the impact of sexual orientation on mental health, friendships, and status attainment by drawing on life course and social psychological perspectives. His previous work has appeared in Social Forces, Work and Occupations, Advances in Life Course Research, and Social Psychology Quarterly.

Preeti Vaghela is a doctoral candidate in sociology at Florida State University. Her research examines the effect of discrimination on marginalized groups’ health and well-being. She also conducts research on the mental health consequences of racial and ethnic identity for second-generation Asian Americans.

Lacey Ritter is a doctoral student in sociology at Florida State University. Her current research focuses on experiences of stigma and stigma management in voluntary organizations. Her research interests include social psychology, medical sociology, and gerontology.

References

Adamczyk

Amy

Palmer

Ian

. 2008. “Religion and Initiation into Marijuana Use: The Deterring Role of Religious Friends.” Journal of Drug Issues 38(3):717–42.

Aneshensel

Carol S.

Rutter

Carolyn M.

Lachenbruch

Peter A.

1991. “Social Structure, Stress, and Mental Health: Competing Conceptual and Analytic Models.” American Sociological Review 56(2):166–78.

Aneshensel

Carol S.

Sucoff

Clea A.

1996. “The Neighborhood Context of Adolescent Mental Health.” Journal of Health and Social Behavior 37(4):293–310.

Annes

Alexis

Redlin

Meredith

. 2012. “Coming Out and Coming Back: Rural Gay Migration and the City.” Journal of Rural Studies 28(1):56–68.

Aseltine

Robert H.

Jr. Gore

Susan

. 1993. “Mental Health and Social Adaptation Following the Transition from High School.” Journal of Research on Adole-scence 3(4):247–70.

Atav

Serdar

Spencer

Gale A.

2002. “Health Risk Behaviors among Adolescents Attending Rural, Suburban, and Urban Schools: A Comparative Study.” Family & Community Health: The Journal of Health Promotion & Maintenance 25(2):53–64.

Austin

Erika L.

Irwin

Jay A.

2010. “Health Behaviors and Health Care Utilization of Southern Lesbians.” Women’s Health Issues 20(3):178–84.

Badgett

M. V. L.

2009. “Best Practices for Asking Questions about Sexual Orientation on Surveys.” The Williams Institute, UCLA School of Law. Retrieved January 10, 2012. (http://williamsinstitute.law.ucla.edu/wp-content/uploads/SMART-FINAL-Nov-2009.pdf)

Baumle

Amanda K.

Compton

D’Lane R.

Poston

Dudley L.

2009. Same-sex Partners: The Demography of Sexual Orientation. Albany, NY: SUNY Press.

10.

Bell

David

Valentine

Gill

. 1995. “Queer Country: Rural Lesbian and Gay Lives.” Journal of Rural Studies 11(2):113–22.

11.

Billy

John O. G.

Wenzlow

Audra T.

Grady

William R.

1998. National Longitudinal Study of Adolescent Health, Wave I and II Contextual Database, Part I. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill.

12.

Black

Gates

Sanders

Taylor

2002. “Why Do Gay Men Live in San Francisco?” Journal of Urban Economics 51(1):54–76.

13.

Boatwright

Karyn J.

Gilbert

Mary S.

Forrest

Linda

Ketzenberger

Kay

. 1996. “Impact of Identity Development upon Career Trajectory: Listening to the Voices of Lesbian Women.” Journal of Vocational Behavior 48(2):210–28.

14.

Brewster

Karin L.

Tillman

Kathryn H.

2012. “Sexual Orientation and Substance Use among Adolescents and Young Adults.” American Journal of Public Health 102(6):1168–76.

15.

Chantala

Kim

. 2006. “Guidelines for Analyzing Add Health Data.” Retrieved March 12, 2008. (http://www.cpc.unc.edu/projects/addhealth/data/guides/wt-guidelines.pdf).

16.

Compas

Bruce E.

Davis

Glen E.

Forsythe

Carolyn J.

1985. “Characteristics of Life Events during Adolescence.” American Journal of Community Psychology 13(6):677–91.

17.

Cooke

Thomas J.

Rapino

Melanie

. 2007. “The Migration of Partnered Gays and Lesbians between 1995 and 2000.” Professional Geographer 59(3): 285–97.

18.

D’Augelli

Anthony R.

Hart

Mary M.

1987. “Gay Women, Men and Families in Rural Settings: Toward the Development of Helping Communities.” American Journal of Community Psychology 15(1):79–93.

19.

DeWit

David J.

1998. “Frequent Childhood Geographic Relocation: Its Impact on Drug Use Initiation and the Development of Alcohol and Other Drug-Related Problems among Adolescents and Young Adults.” Addictive Behaviors 23(5):623–34.

20.

Everett

Bethany. G.

2014. “Changes in Neighborhood Characteristics and Depression among Sexual Minority Young Adults.” Journal of the American Psychiatric Nurses Association 20(1):42–52.

21.

Finkel

Steven E.

1995. Causal Analysis with Panel Data. Thousand Oaks, CA: Sage.

22.

Florida

2002. “The Economic Geography of Talent.” Annals of the Association of American Geographers 92(4):743–55.

23.

Gates

Gary J.

Ost

Jason

. 2004. The Gay and Lesbian Atlas. Washington, DC: Urban Institute Press.

24.

Geist

Claudia

McManus

Patricia A.

2008. “Geographi-cal Mobility over the Life Course: Motiva-tions and Implications.” Population Space and Place 14(4):283–303.

25.

Green

Adam I.

2003. “‘Chem Friendly’: The Institutional Basis of ‘Club Drug’ Use in a Sample of Urban Gay Men.” Deviant Behavior 24(5):427–47.

26.

Green

Adam I.

2008. “Health and Sexual Status in an Urban Gay Enclave: An Application of the Stress Process Model.” Journal of Health and Social Behavior 49(4):436–51.

27.

Grollman

Eric A.

2012. “Multiple Forms of Perceived Discrimination and Health among Adolescents and Young Adults.” Journal of Health and Social Behavior 53(2):199–214.

28.

Ham

John C.

Xianghong

Reagan

Patricia B.

2011. “Matching and Semi-parametric IV Estimation, a Distance-based Measure of Migration, and the Wages of Young Men.” Journal of Econometrics 161(2):208–27.

29.

Hatzenbuehler

Mark L.

McLaughlin

Katie A.

Nolen-Hoeksema

Susan

. 2008. “Emotion Regulation and Internalizing Symptoms in a Longitudinal Study of Sexual Minority and Heterosexual Adolescents.” Journal of Child Psychology and Psychiatry 49(12): 1270–78.

30.

Haynie

Dana L.

South

Scott J.

2005. “Residential Mobility and Adolescent Violence.” Social Forces 84(1):361–74.

31.

Herek

Gregory M.

2002. “Gender Gaps in Public Opinion about Lesbians and Gay Men.” The Public Opinion Quarterly 66(1):40–66.

32.

Herek

Gregory M.

Garnets

Linda D.

2007. “Sexual Orientation and Mental Health.” Annual Review of Clinical Psychology 3:353-75.

33.

Hewitt

Christopher

. 1998. “Homosexual Demography: Implications for the Spread of AIDS.” Journal of Sex Research 35(4):390–96.

34.

Hoffmann

John P.

Johnson

Robert A.

1998. “A National Portrait of Family Structure and Adolescent Drug Use.” Journal of Marriage and the Family 60(4):633–45.

35.

Ihrke

David K.

Faber

Carol S.

Geographical Mobility 2005 to 2010: Population Characteristics. Washington, DC: US Census Bureau.

36.

Kane

Melinda D.

2003. “Social Movement Policy Success: Decriminalizing State Sodomy Laws, 1969-1998.” Mobilization: An International Journal 8(3):313–34.

37.

Kazyak

Emily

. 2012. “Midwest or Lesbian? Gender, Rurality, and Sexuality.” Gender & Society 26:825–48.

38.

Kessler

Ronald C.

Chiu

Wai T.

Demler

Olga

Merikangas

Kathleen R.

Walters

Ellen E.

2005. “Prevalence, Severity, and Comorbidity of 12-Month DSM-IV Disorders in the National Comorbidity Survey Repli-cation.” Archives of General Psychiatry 62(6):617–27.

39.

Kington

Raynard

Carlisle

David

McCaffrey

Dan

Myers

Hector

Allen

Walter

. 1998. “Racial Differences in Functional Status among Elderly U.S. Migrants from the South.” Social Science and Medicine 47(6):831–40.

40.

Kirkey

Kenneth

Forsyth

Ann

. 2001. “Men in the Valley: Gay Male Life on the Suburban-rural Fringe.” Journal of Rural Studies 17(4):421–41.

41.

Kuper

Laura E.

Coleman

Brett R.

Mustanski

Brian S.

Forthcoming. “Coping with LGBT and Racial-ethnic-related Stressors: A Mixed-methods Study of LGBT Youth of Color.” Journal of Research on Adolescence.

42.

Laumann

Edward O.

Gagnon

John H.

Michael

Robert T.

Michaels

Stuart

. 1994. The Social Organization of Sexuality: Sexual Practices in the United States. Chicago: University of Chicago Press.

43.

Lehmann

Wolfgang

. 2014. “Habitus Transformation and Hidden Injuries: Successful Working-class University Students.” Sociology of Education 87(1):1–15.

44.

Levine

S. B.

Coupey

S. M.

2003. “Adolescent Substance Use, Sexual Behavior, and Metropolitan Status: Is ‘Urban’ a Risk Factor?” Journal of Adolescent Health 32(5):350–55.

45.

Lleras

Christy

. 2008. “Do Skills and Behaviors in High School Matter? The Contribution of Noncognitive Factors in Explaining Differences in Educational Attainment and Earnings.” Social Science Research 37:888–902.

46.

Loftus

Jeni

. 2001. “America’s Liberalization in Attitudes toward Homosexuality: 1973-1998.” American Socio-logical Review 66(5):762–82.

47.

Marshal

Michael P.

Friedman

Mark S.

Stall

Ron

King

Kevin M.

Miles

Jonathan

Gold

Melanie A.

Bukstein

Oscar G.

Morse

Jennifer Q.

2008. “Sexual Orientation and Adolescent Substance Use: A Meta-analysis and Methodological Review.” Addiction 103(4):546–56.

48.

McCarthy

Linda

. 2000. “Poppies in a Wheat Field: Exploring the Lives of Rural Lesbians.” Journal of Homosexuality 39(1):75.

49.

Meyer

Ilan H.

2003. “Prejudice, Social Stress, and Mental Health in Lesbian, Gay, and Bisexual Populations: Conceptual Issues and Research Evidence.” Psycho-logical Bulletin 129(5):674–97.

50.

Needham

Belinda L.

2007. “Gender Differences in Trajectories of Depressive Symptomatology and Sub-stance use during the Transition from Adolescence to Young Adulthood.” Social Science & Medicine 65(6):1166–79.

51.

Newcomb

Michael D.

Huba

George J.

Bentler

Peter M.

1981. “A Multidimensional Assessment of Stressful Life Events among Adolescents: Derivation and Correlates.” Journal of Health and Social Behavior 22(4):400–15.

52.

Norman

Paul

Boyle

Paul

Rees

Philip

. 2005. “Selective Migration, Health and Deprivation: A Longitudinal Analysis.” Social Science & Medicine 60(12):2755–71.

53.

Ohlander

Julianne

Batalova

Jeanne

Treas

Judith

. 2005. “Explaining Educational Influences on Atti-tudes Toward Homosexual Relations.” Social Science Research 34:781-99.

54.

Oishi

Shigehiro

Schimmack

Ulrich

. 2010. “Residential Mobility, Well-being, and Mortality.” Journal of Personality and Social Psychology 98(6):980–94.

55.

Parsons

Jeffrey T.

Kelly

Brian C.

Weiser

Jonathan D.

2007. “Initiation into Methamphetamine Use for Young Gay and Bisexual Men.” Drug and Alcohol Dependence 90(2–3):135–44.

56.

Pearlin

Leonard I.

1989. “The Sociological Study of Stress.” Journal of Health and Social Behavior 30(3):241–56.

57.

Poon

Colleen S.

Saewyc

Elizabeth M.

2009. “Out Yonder: Sexual-minority Adolescents in Rural Com-munities in British Columbia.” American Journal of Public Health 99(1):118–24.

58.

Pribesh

Downey

D. B.

1999. “Why Are Residential and School Moves Associated with Poor School Performance?” Demography 36(4):521–34.

59.

Radloff

Lenore S.

1977. “The CES-D Scale: A Self-report Depression Scale for Research in the General Population.” Applied Psychological Measurement 1(3): 385–401.

60.

Rosenbaum

Paul R.

Rubin

Donald B.

1983. “The Central Role of the Propensity Score in Observational Studies of Causal Effects.” Biometrika 70(1):41–55.

61.

Russell

Stephen T.

Driscoll

Anne K.

Truong

Nhan

. 2002. “Adolescent Same-sex Romantic Attractions and Relationships: Implications for Substance Use and Abuse.” American Journal of Public Health 92(2):198–202.

62.

Saad

Lydia

. 2012. “U.S. Acceptance of Gay/Lesbian Relations Is the New Normal.” Retrieved July 1, 2013. (http://www.gallup.com/poll/154634/acceptance-gay-lesbian-relations-new-normal.aspx).

63.

Savin-Williams

Ritch C.

2005. The New Gay Teenager. Cambridge, MA: Harvard University Press.

64.

Savin-Williams

Ritch C.

Ream

Geoffrey L.

2007. “Prevalence and Stability of Sexual Orientation Components during Adolescence and Young Adult-hood.” Archives of Sexual Behavior 36(3):385–94.

65.

Schulenberg

John

O’Malley

Patrick M.

Bachman

Jerald G.

Johnston

Lloyd D.

2000. “‘Spread Your Wings and Fly’: The Course of Well-being and Substance Use during the Transition to Young Adulthood.” Pp. 224–55 in Negotiating Adolescence in Times of Social Change, edited by Crockett

L. J.

Silbereisen

R. K.

New York: Cambridge University Press.

66.

Shilo

Guy

Savaya

Riki

. 2012. “Mental Health of Lesbian, Gay, and Bisexual Youth and Young Adults: Differential Effects of Age, Gender, Religiosity, and Sexual Orientation.” Journal of Research on Adolescence 22(2):310–25.

67.

Swisher

Raymond R.

2008. National Longitudinal Study of Adolescent Health, Wave III Contextual Database Codebook. Part I. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill.

68.

Szymanski

Dawn M.

Owens

Gina P.

2008. “Do Coping Styles Moderate or Mediate the Relationship between Internalized Heterosexism and Sexual Minority Women’s Psychological Distress?” Psychology of Women Quarterly 32(1):95–104.

69.

Szymanski

Dawn M.

Henrichs-Beck

Christy

. 2014. “Exploring Sexual Minority Women’s Experiences of External and Internalized Heterosexism and Sexism and their Links to Coping and Distress.” Sex Roles 70(1-2): 28–42.

70.

Tiesler

Carla M. T.

Birk

Matthias

Kohlböck

Gabriele

Koletzko

Sibylle

Bauer

Carl-Peter

Wichmann

H. E.

Berg

Andrea

Berdel

Dietrich

Krämer

Ursula

Schaaf

Beate

Lehmann

Irina

Herbarth

Olf

Heinrich

Joachim

. 2013. “Residential Mobility and Behavioural Problems in Children: Results from the GINIplus and LISAplus Studies.” Journal of Public Health 21(1):39–48.

71.

Troiden

Richard R.

1989. “The Formation of Homosexual Identities.” Journal of Homosexuality 17(1–2):43–73.

72.

Tucker

C. J.

Marx

Long

1998. “‘Moving on’: Residential Mobility and Children’s School Lives.” Sociology of Education 71(2):111–29.

73.

Ueno

Koji

. 2005. “Sexual Orientation and Psychological Distress in Adolescence: Examining Interpersonal Stressors and Social Support Processes.” Social Psychology Quarterly 68(3):258–77.

74.

Ueno

Koji

. 2010a. “Mental Health Differences between Young Adults with and without Same-sex Contact: A Simultaneous Examination of Underlying Mecha-nisms.” Journal of Health and Social Behavior 51(4):391–407.

75.

Ueno

Koji

. 2010b. “Same-sex Experience and Mental Health during the Transition between Adolescence and Young Adulthood.” Sociological Quarterly 51(3):484–510.

76.

Waldo

Craig R.

Hesson-McInnis

Matthew S.

D’Augelli

Anthony R.

1998. “Antecedents and Consequences of Victimization of Lesbian, Gay, and Bisexual Young People: A Structural Model Comparing Rural University and Urban Samples.” American Journal of Community Psychology 26(2):307–34.

77.

Walther

Carol S.

Poston Jr.

Dudley L.

2004. “Patterns of Gay and Lesbian Partnering in the Larger Metropolitan Areas of the United States.” Journal of Sex Research 41(2):201–14.

78.

Wickrama

A. K.

Noh

Samuel

Bryant

Chalandra M.

2005. “Racial Differences in Adolescent Distress: Differential Effects of the Family and Community for Blacks and Whites.” Journal of Community Psychology 33(3):261–82.

79.

Wienke

Chris

Hill

Gretchen J.

2013. “Does Place of Residence Matter? Rural-urban Differences and the Wellbeing of Gay Men and Lesbians.” Journal of Homosexuality 60(9):1256–79.

80.

Wilkinson

Lindsey

Pearson

Jennifer

. 2009. “School Culture and the Well-being of Same-sex-attracted Youth.” Gender & Society 23(4):542–68.