Upward Mobility Context and Health Outcomes and Behaviors during Transition to Adulthood: The Intersectionality of Race and Sex

Abstract

This study investigates how upward mobility context affects health during transition to adulthood and its variations by race and sex. Using county-level upward mobility measures and data from the Panel Study of Income Dynamics, we apply Propensity Score Weighting techniques to examine these relationships. Results show that low upward mobility context increases the likelihood of poor self-rated health, obesity, and cigarette use, but decreases alcohol consumption probability. Conversely, high upward mobility context raises the likelihood of distress, chronic conditions, and alcohol use, but reduces cigarette use likelihood. In low-opportunity settings, Black individuals have lower risks of chronic conditions and cigarette use than white men. In high-opportunity settings, Black women are more likely to experience depression and chronic conditions, while Black men are likelier to smoke than white men. Our findings emphasize the complex link between upward mobility context and health for different racial and sex groups.

The ages 18 to 29 mark a critical transition to adulthood, pivotal for acquiring human capital (Berlin, Furstenberg Jr., and Waters 2010; Waters, Kefalas, and Holdaway 2011). Health during this period significantly affects long-term economic and social well-being, potentially leading to lasting inequalities (Wood et al. 2018). Amid newfound independence and exploration, young adults face substantial risks, adopting risky behaviors that contribute to poor health outcomes (Park et al. 2006).

The local upward mobility context in which young people find themselves may play a crucial role in their transition to adulthood. Scholars have recently shown increased interest in the impacts of place and local context on individual health outcomes (Alvarado 2019; Ludwig et al. 2013). In particular, a growing body of literature has demonstrated that county-level intergenerational upward mobility levels, referred to as “upward mobility context” henceforth, have a positive association with adult health (Venkataramani et al. 2016a, 2016b, 2020). This concept, defined as the average county-level expected percentile rank in the national income distribution at age 29 for individuals from low-income families, directly reflects the likelihood of achieving the “American Dream” (Chetty and Hendren 2018). It is influenced by various county-level characteristics, such as the quality of education, employment opportunities, physical environment, and public service availability (Chetty and Hendren 2018; Venkataramani et al. 2020). Growing up in a county with low upward mobility can affect health by reducing individuals’ chances of achieving upward mobility and high socioeconomic status (SES). Additionally, it may also increase stress due to limited opportunities and expose individuals to unsafe and unhealthy social environments, even when they manage to achieve upward mobility in such low-opportunity environments.

The relationship between upward mobility context and health likely varies by race and sex (Assari et al. 2017; Bishop et al. 2020), especially during the transition to adulthood. This goes beyond the mere difference in exposures to upward mobility. For example, even Black and white youths growing up in similar upward mobility contexts might react differently to those circumstances. The unique socialization of Black women, for instance, could make them more resilient to low upward mobility contexts, attenuating the negative impacts compared to the white population (Rosenfield 2012). Conversely, racial and sex discrimination faced by Black women, even in high mobility environments, could suppress health benefits relative to white people (Ciciurkaite 2021; Hargrove 2018). Little research has studied racial and sex disparities in how upward mobility context affects health in transitioning to adulthood. Overlooking intersectionality might obscure the true association between upward mobility context and health for young adults with multiple marginalized identities.

This study bridges gaps in literature, investigating the link between county-level upward mobility and individual health outcomes and behaviors during the transition to adulthood, and how this varies by race and sex intersections. Utilizing data from the Panel Study of Income Dynamics (PSID) and its Transition into Adulthood Supplement (TAS), we employ linear regressions to examine these relationships. To address potential selection bias, considering different SES families’ self-selection into counties with diverse upward mobility levels, we implement Propensity Score Weighting (PSW) techniques.

This study contributes to the existing literature in three ways. Firstly, it is the first to examine how upward mobility context affects health during the transition to adulthood and how this relationship varies by race and sex. Specifically, we utilize race- and sex-specific mobility measures, distinguishing our study from previous ones that employed race- and sex- combined mobility measures (O’Brien, Venkataramani, and Tsai 2017; Venkataramani et al. 2016a, 2016b, 2020). Secondly, we investigate a broader range of health outcomes and behaviors than previous studies. This allows us to differentiate the effects of upward mobility context on physical and mental health, as well as health behaviors. Thirdly, our study accounts for selection bias. While our approach still relies on assumptions, it provides estimates closer to causal ones compared to previous research.

BACKGROUND

Place and the Transition to Adulthood

During the transition to adulthood, many young adults leave their childhood homes for the first time for work, education, or other experiences. Gaining independence and escaping supervision allows exploration (Berlin et al. 2010; Waters et al. 2011). While fostering growth, these opportunities can lead to risky health behaviors, impacting later health (Hope, Hoggard, and Thomas 2015; Liu et al. 2012). Additionally, poor health outcomes may directly manifest during this period. Prior research has noted increased mental health risks during this time (Mossakowski 2008; Snedker and Hooven 2013).

Transitioning to adulthood yields varied experiences and paths, rendering it among the most diverse life stages (Berlin et al. 2010; Wood et al. 2018). Social and environmental interactions, heightened by increased independence and less oversight, are pivotal. Concurrently, significant time and effort are dedicated to contemplating future prospects. Therefore, social context’s impact on present and future health and behavior amplifies (Snedker and Hooven 2013).

Several scholars have recognized the importance of understanding the effect of local upward mobility context on individual health and well-being (Chetty, Hendren, and Katz 2016). While some studies have begun to address this topic (Daza and Palloni 2022; O’Brien et al. 2017, 2020; Venkataramani et al. 2016a, 2016b, 2020), few have specifically focused on the health of young adults during the transition to adulthood. Upward mobility context can be influenced by various county characteristics, such as school quality, physical environment, and the availability of services. The lack of upward mobility in a county may reflect limited access to material benefits, services, and medical resources, which are crucial for the health and health behaviors of its residents.

Upward mobility context may influence health and health behaviors during the transition to adulthood through various pathways. One evident pathway is that being raised in counties with low levels of upward mobility reduces the chances of achieving higher SES in life. Low SES is known to be associated with poorer health outcomes (Kivimäki et al. 2020; Venkataramani et al. 2016a). However, for young adults who typically have not yet established a stable SES, this mechanism might be less significant.

Relatedly, being raised in a low-opportunity environment may reduce the chances of individual upward mobility, and individual levels of social mobility can independently influence health, irrespective of original or attained SES. Sociological literature has extensively explored the impact of individual social mobility on health and well-being, but the findings are mixed (see Zang, Sobel, and Luo [2023] for a review). Earlier studies in the 1950s suggested that social mobility, regardless of upward or downward, disrupts social ties and leads to social isolation and adaptation, resulting in worse health (Durkheim 1952; Hollingshead, Ellis, and Kirby 1954; Sorokin 1959). However, more recent studies show that individual upward mobility either has no effect on health or has positive effects on health, net of childhood and current SES (Chen, Brody, and Miller 2022; Zang et al. 2023).

Growing up in a county with low levels of upward mobility may have negative psychological impacts, separate from an individual’s actual mobility outcomes (Leventhal and Brooks-Gunn 2003; Venkataramani et al. 2016b). For instance, experiencing challenges toward upward mobility can increase young adults’ stress and lead to negative perceptions of future outcomes. Chronic stress faced by young adults has been shown to have severe negative impacts on health (Borrell, Dallo, and Nguyen 2010). Individuals with low expectations for their future well-being may have weaker incentives to prioritize their health (Venkataramani et al. 2016a, 2020). Emerging research indicates that Black people, who face both a low-opportunity environment and structural racism, may experience adverse health effects when striving for upward mobility (Subramanyam et al. 2013). This could be attributed to their prolonged and intense exposure to psychosocial stressors, which trigger the utilization of coping mechanisms requiring heightened effort (James 1994). This phenomenon, often referred to as “John Henryism,” can lead to early-onset cardiovascular disease and other negative health outcomes, despite individual upward mobility (James, Hartnett, and Kalsbeek 1983; Subramanyam et al. 2013). Additionally, socioeconomic disadvantage plays a significant role in contributing to this effect, which is not limited to Black people. Notably, studies on tokenism and health reveal that individuals from lower SES backgrounds in prestigious universities, who achieve upward mobility, often grapple with mental health challenges due to the exclusive nature of such institutions (Aries and Seider 2005; Billings 2021). Corresponding findings emerge for women in esteemed corporations and fields characterized by substantial gender inequality (McGee and Bentley 2017; Roth 2004).

While growing up in a high-opportunity environment may offer numerous benefits in terms of upward mobility prospects, it often comes with competitiveness and constant pressure to compare oneself with others, leading to significant stress. This stress can affect individuals’ health behaviors and outcomes in such settings. The literature on relative income indicates that people’s subjective perceptions of their economic position compared to others play a crucial role in their health (Miller and Paxson 2006; Subramanyam et al. 2009). In high-opportunity environments, where upward mobility opportunities are abundant, many individuals strive to improve their socioeconomic status. Consequently, they are more likely to engage in social comparisons, particularly regarding economic achievements (Subramanyam et al. 2009). The pressure to succeed and the fear of falling behind in the race for upward mobility can contribute to increased stress levels. Indeed, research has shown that in emerging markets with high levels of economic opportunities, despite experiencing higher levels of upward mobility compared to countries with fewer opportunities, people’s subjective well-being tends to be worse (Brockmann et al. 2009; Graham and Pettinato 2002). This stress, particularly among young adults, can contribute to the adoption of unhealthy behaviors (Nelson et al. 2008). For example, well-to-do neighborhoods often boast a higher concentration of restaurants and bars, potentially amplifying the likelihood of escalated alcohol consumption as a coping mechanism for stress among young adults (Pedersen, Bakken, and von Soest 2018; Slutske, Deutsch, and Piasecki 2016).

Finally, being raised in an area with low upward mobility may not only lead to poor health behaviors, but also increase a young person’s likelihood of interacting with others who exhibit these health behaviors. For instance, Christakis and Fowler (2007) found evidence of network effects in the spread of obesity, showing that individuals who have friends or family members with obesity are more likely to develop obesity themselves. Similarly, Papachristos and Wildeman (2014) found that homicides follow a networked structure, affecting the health and safety of others in the same county.

Drawing from the aforementioned evidence, growing up in a county characterized by either low or high levels of upward mobility may lead to heightened probabilities of engaging in unhealthy behaviors and experiencing adverse health consequences (H1). Nevertheless, while both scenarios may elevate the vulnerability to mental health issues attributable to stress, it is noteworthy that inhabiting a low-opportunity environment could accentuate the risk of encountering physical health complications as well, stemming from a dearth of essential resources and underdeveloped healthcare infrastructure. In these contrasting environments, the inclination towards certain unhealthy behaviors could be magnified, contingent upon the specific attributes unique to each context. For instance, in regions with robust opportunities for upward mobility, young adults might exhibit a pronounced tendency to resort to alcohol consumption as a coping mechanism in response to stress, fueled by the presence of a plethora of restaurants and bars that often characterize such areas. Conversely, individuals growing up in locales where prospects for upward mobility are limited may be more prone to manifest other unhealthy behaviors, potentially stemming from the lack of accessible recreational facilities or avenues for constructive stress relief. This underscores the intricate interplay between environmental factors and individual behaviors, underscoring the need for a nuanced understanding of the mechanisms that drive unhealthy choices and health outcomes across diverse settings.

There are two major challenges when examining the impact of upward mobility context on health. First, the relationship may differ depending on the specific health outcome or behavior under consideration. Levels of upward mobility are likely to be associated with family and community structures, as well as access to amenities, which can influence individual health indicators differently. Thus, while higher mobility context may generally improve health outcomes, it could have negative impacts on specific health behaviors. For example, Hudson et al. (2012) found that despite the generally protective influence of higher SES on health, it can also be associated with higher rates of depression for certain populations. These results indicate that to fully understand the impact of upward mobility on health, we need to examine a range of health outcomes and behaviors.

Second, it is important to note that the relationship can be influenced by selection bias. Socioeconomically advantaged young people, who tend to be healthier than others, may choose to live in places with high levels of upward mobility, rather than upward mobility being the sole reason for their better health. For instance, immigrants living in counties with high levels of upward mobility tend to be healthier than non-immigrants (Riosmena, Kuhn, and Jochem 2017). Additionally, residents who are more socioeconomically advantaged in counties with low levels of upward mobility are more likely to afford moving to a county with greater upward mobility. However, previous studies have rarely addressed these selection issues when examining the relationship between upward mobility context and individual health. An exception is Daza and Palloni (2022), but this paper does not focus on outcomes during the transition to adulthood or racial and sex disparities.

Racial and Sex Disparities in How Upward Mobility Context Affects Health

Substantial racial and sex disparities in health outcomes and behaviors during the transition to adulthood are well-documented. Black females have higher prevalence of obesity and chronic conditions than whites (Hargrove 2018). The compounding impact of racial and sex discrimination faced by Black women in education and work settings may elevate stress levels (Berdahl and Moore 2006), subsequently increasing the likelihood of chronic health conditions and obesity (Salleh 2008). Additionally, Black mothers encounter heightened stress due to being primary earners and caretakers, while Black fathers’ economic challenges and incarceration risks contribute to this dynamic (Rosenfield 2012). Rates of depression are generally higher among white females than other groups (Mulye et al. 2009), and substance use is especially high among white boys (Mulye et al. 2009; Park et al. 2006). Traditional gender roles in white communities lead to unique mental health challenges for women. Men often dominate economic and relational aspects while women are relegated to domestic tasks, potentially contributing to higher depression rates due to work–family stress (De Coster and Heimer 2017; Rosenfield 2012). High substance use rates among white males may be attributed to factors like toxic masculinity (Tamás et al. 2019) and a positive correlation between income and substance use (Kar et al. 2018), as white men often possess greater economic resources.

These findings imply that neither race nor sex can fully explain poor health. Instead, they must be considered together. The lens of intersectionality theory aids our analysis. Introduced by Kimberlé Crenshaw (1989) in the context of Black feminism, intersectionality theory suggests that individuals with multiple marginalized identities face intricate discrimination experiences that transcend the sum of individual identities. Crenshaw’s work highlighted how the experiences of a Black woman differ from those of Black individuals and females combined, warranting separate analysis (Crenshaw 1989). Later scholars like Stephanie Nixon (2019) stressed analyzing both disadvantaged and advantaged groups within intersectionality theory. This approach unveils disparities rooted in social structures and inequality systems. This study employs this expanded intersectionality theory to scrutinize health disparities in race- and sex-specific groups.

The theoretical interaction between upward mobility context, race, and sex in shaping the health outcomes and behaviors of transitioning adults presents inherent ambiguity. On the one hand, the possibility of alleviating the adverse effects of low mobility contexts for Black adults, as well as specific health outcomes and behaviors among white women, is evident (H2a). This prospect finds support in the concept of racial socialization, a process whereby individuals from a specific racial group are primed to anticipate potential challenges or realities arising from living within racially discriminatory societal frameworks (Iruka, Curenton, and Eke 2014). This priming can effectively alleviate the adverse health effects stemming from unfavorable environments (Pamplin and Bates 2021). For instance, the upbringing of young Black women often involves instilling values of independence, resilience, and resourcefulness, acknowledging the distinct obstacles they confront (Rosenfield 2012). These challenges encompass not only workplace discrimination but also heightened parenting responsibilities attributed to the economic disadvantages and the disproportionate risk of incarceration faced by Black men. Furthermore, research indicates that communities of color tend to possess robust familial and non-familial support systems, a factor shown to diminish health risks amidst adverse circumstances (Ursache, Barajas-Gonzalez, and Dawson-McClure 2022). For certain health outcomes, white women might experience less impact from a low-opportunity environment compared to white men. This could be attributed to their potential proficiency in handling stress within such environments, stemming from their often-heightened experience of gender discrimination. Traditional American gender roles frequently assign the economic responsibility of a household to men. Consequently, women, on average, might encounter lower stress levels attributed to low-opportunity environments.

On the other hand, the advantageous effects of high mobility environments on Black adults, particularly Black women, may be comparatively limited compared to their white counterparts (H2b). This hypothesis finds support in various prior studies that have noted how beneficial social contexts do not yield equivalent protective advantages against poor health for Black adults in comparison to white adults (Assari 2018; Assari et al. 2017). This discrepancy could potentially stem from the persisting discrimination that Black adults encounter even within favorable environments (Ciciurkaite 2021; Hargrove 2018). Simultaneously, the coping strategies that Black communities deploy to address adversity and safeguard health in low mobility contexts might be less applicable in high mobility environments. Black women notably face elevated instances of workplace harassment, surpassing both white women and Black men in frequency and severity (Berdahl and Moore 2006). This adversity could be amplified in high-opportunity environments, often termed “White Spaces,” where racial discrimination might be more pronounced (Hudson et al. 2016). Navigating this intricate landscape might require amplified efforts and more resilient coping mechanisms, potentially diminishing the advantages for Black women compared to their white counterparts (Farmer, Wray, and Haas 2021). The presence of gender inequality in such environments could further erode the health benefits experienced by Black women compared to Black men.

DATA AND METHODS

Data and Measures

We utilized data from the PSID and its TAS, which provide insights into health, socioeconomic conditions, and life experiences (Insolera et al. 2019). The PSID Child Development Supplement (CDS) follows a subset aged 0 to 12 in 1997 over three waves. Starting in 2005, the TAS follows participants biennially who were either (1) at some point members of the CDS before 2017 or (2) between the ages of 18 and 28 starting from 2017. Additional information about PSID and TAS can be found on their website (https://simba.isr.umich.edu/default.aspx). The TAS offers comprehensive information on health behaviors and outcomes during the transition to adulthood, rendering it a valuable data source for our study.

Our main dependent variables included self-rated poor health, mental health variables, physical health variables, and health behaviors for youths aged 18 to 29. All variables were available for each year of the TAS from 2005 to 2017. Mental health variables encompassed depression and non-specific distress. Physical health variables comprised the presence of chronic conditions and obesity. Health behaviors encompassed drug use, cigarette use, and alcohol use. Among all the information available in our dataset, these health outcomes and behaviors have been shown to hold relative importance during the transition to adulthood (Johnston et al. 2015).

Self-rated poor health in this study was assessed using the indicator of “fair” or “poor” health. Depression was determined based on clinical diagnosis by a medical professional. Non-specific distress was measured using the K-6 Non-Specific Psychological Distress Scale, a composite scale ranging from 0 to 24 that evaluates mental health. Participants were asked if they felt nervous, hopeless, restless; had trouble concentrating; felt that everything was an effort; experienced sadness; or felt worthless in the past 30 days. A score of 13 or higher on this scale has been identified as the optimal cut-off for clinically positive cases of serious mental illness (Kessler et al. 2003). This scale has been validated by several studies since its inception (Peiper et al. 2016). The presence of chronic conditions was determined by the diagnosis of any chronic condition received from a healthcare professional. Consistent with prior research (Alvarado 2019), obesity was defined as having a body mass index (BMI) equal to or exceeding 30. Cigarette use was indicated by having ever smoked cigarettes. Alcohol use was indicated by having ever consumed alcohol. Similarly, following previous studies (Heflin 2019), drug use was indicated by any illicit use of four recreational drugs in the past 12 months: amphetamines, cannabis, cocaine, and barbiturates.

Our main independent variable is county-level race- and sex-specific intergenerational upward mobility during childhood. This variable was computed as the average of county-level intergenerational upward mobility levels across ages 0 to 17 for each individual. County-level intergenerational upward mobility measures were obtained from the Opportunity Atlas project (https://www.opportunityatlas.org/), created by Chetty and colleagues (2018) using administrative tax records of American children born between 1978 and 1983 and their parents. Consistent with the definition of “absolute upward mobility” in prior research (Chetty et al. 2014), the upward mobility context is established as the expected percentile rank in the national income distribution at age 29 for individuals from low-income families. Low-income status was designated as the 10th percentile of the national income distribution when the individual was approximately 16 years old (Chetty et al. 2014). Robustness checks using the 25th and 1st percentiles to define low-income yielded results consistent with our primary findings (detailed results available upon request).

Compared to upward mobility measures that track mobility from age 16 to ages significantly beyond 29, the measure employed in this study captures a “legacy effect of advantaged or disadvantaged origins” (Zang and Kim 2021:pg 2), which could hold particular relevance for transitioning to adulthood outcomes. The birth cohorts included in the TAS were born between 1983 and 1998. Building upon prior research (O’Brien et al. 2017; Venkataramani et al. 2016a, 2016b, 2020), we assumed that upward mobility levels within a county remained relatively stable over a two-decade period. Additionally, the TAS cohorts observed the 1978 to 1983 cohorts experiencing upward mobility during their young adulthood, a factor that could notably influence their perceptions of upward mobility prospects.

We opted for a binary measure of upward mobility context instead of a continuous one to mitigate sensitivity to model misspecification and outliers in our PSW analysis (Naimi et al. 2014). Furthermore, binary measures are better equipped to capture potential nonlinear relationships between upward mobility context and health, as opposed to continuous measures. Specifically, we categorized individuals based on whether they grew up in counties with upward mobility levels below the 30th percentile or above it. Additionally, we compared individuals from counties with upward mobility levels below the 70th percentile to those from counties with levels at or above the 70th percentile. As part of robustness checks, we employed cutoffs at the 20th and 80th percentiles, and the results demonstrated substantial consistency (detailed results available upon request).

Race encompassed white and Black categories. Due to inconsistent reporting of Hispanic origin across waves, our focus was solely on the reported race. Sex comprised male and female categories. For model simplicity, we further condensed these categories into four race and sex combinations. We accounted for various individual and family characteristics from participants’ childhood (ages 0 to 17 years) that hold theoretical predictive value for health outcomes and behaviors. These encompassed household income, household wealth, whether the child’s family lived in mobile homes, homeownership status, employment status of the household head, retirement status of the household head, food stamp usage in the household the year before the survey, age of the household head, highest educational attainment of the household head in years, whether the household head was a single mother, marital status of the participants’ parents, and the number of children in the household. To capture the average exposure during childhood, we calculated time-constant measures for time-varying variables across ages 0 to 17. Additionally, we included controls for the quadratic form of age and survey year indicators in the model. Importantly, we refrained from controlling for any post-childhood county-level SES measures, such as household income, as they could potentially act as mediators. Including these variables would reduce the specific impacts attributed to childhood mobility contexts (Angrist and Pischke 2009).

The original TAS sample had 4,776 unique individuals. After narrowing to ages 18 to 29, we had 4,146 individuals. Focusing on Black and white respondents left 3,663 individuals, with 19 missing covariate details. We linked TAS individuals with county-level mobility data using restricted-use identifiers, yielding 3,644 individuals. Unfortunately, 2 lacked linkage due to missing county identifiers. Plus, 4 lacked upward mobility data due to small population sizes, making reliable estimates challenging. Our final sample contained 10,208 to 10,235 person-year observations across 3,638 individuals, based on specific outcomes. Table 1 presents summary statistics. Notably, observations show that Black respondents, especially women, tended to grow up in counties with lower upward mobility than whites.

Table 1:

Descriptive Statistics

	Number of Individuals	Person- Years	Mean/Percentage	SD
Upward Mobility Context (Bottom 30 Percentiles)
White Men	906	2,534	.22
White Women	977	2,950	.27
Black Men	866	2,319	.35
Black Women	889	2,432	.35
Upward Mobility Context (Top 30 Percentiles)
White Men	906	2,534	.46
White Women	977	2,950	.41
Black Men	866	2,319	.29
Black Women	889	2,432	.27
Race/Ethnicity and Sex
White Men	906	10,246	24.73
White Women	977	10,246	28.79
Black Men	869	10,246	22.69
Black Women	890	10,246	23.78
Control Variables
Family Lives in Mobile Home	3,642	10,246	.07	.20
Family Owns Home	3,642	10,246	.58	.39
Household Head Is Retired	3,642	10,246	.02	.09
Any Family Member Uses Food Stamps	3,642	10,246	.19	.29
Age of Household Head	3,642	10,246	37.81	7.10
Educational Attainment of Household Head	3,642	10,246	2.63	.97
Household Head Is a Single Mother	3,642	10,246	.30	.37
Parents of Child Are Married	3,642	10,246	.67	.38
Household Income	3,642	10,246	51354.66	48162.44
Household Wealth	3,642	10,246	46.26	24.38
Number of Children in Household	3,642	10,246	2.31	.90
Household Head Is Employed	3,642	10,246	.82	.26
Outcome Variables
Self-Rated Poor Health	3,641	10,234	9.51
Depression	3,642	10,239	6.30
Distress	3,640	10,227	4.76
Obesity	3,624	10,138	21.60
Chronic Conditions	3,641	10,228	34.07
Alcoholic Use	3,642	10,246	64.91
Drug Use	3,642	10,219	54.46
Cigarette Use	3,641	10,234	32.88

Note: Data are from the Panel Study of Income Dynamics.

Methods

We began by conducting linear probability models to explore the relationship between upward mobility context and transition to adulthood outcomes. We employed linear probability models instead of logistic regressions because the coefficients from the former are more intuitively interpretable. Nevertheless, the findings from logistic regressions aligned with our primary results (see online Appendix S1). Our regression models controlled solely for race and sex, along with the quadratic form of age and survey year indicators. It is important to acknowledge that individuals from varied demographic and family backgrounds might be sorted into different counties of residence, each characterized by distinct levels of upward mobility. To address this potential selection bias, we employed the PSW technique, which enhances causal inference by offering a stronger foundation. In comparison to alternative propensity score-based methods like propensity score matching, PSW offers several unique benefits. Firstly, it eliminates the need to exclude numerous observations that could not be matched, thereby preserving the sample size. Secondly, unlike propensity score matching that may exacerbate covariate imbalance, as shown by King and Nielsen (2019), PSW does not encounter this issue (Desai and Franklin 2019).

We utilized inverse probability weights (IPW) to balance individual- and family-level covariates between children who grew up in counties with high versus low upward mobility contexts. These weights were obtained by generating propensity scores through probit regressions based on individual and childhood family characteristics. The distribution of propensity scores, segregated by treatment status, exhibited substantial overlaps (see online Appendix Figures A1 and A2). Our balancing assessments demonstrated satisfactory balancing properties for the majority of covariates (see online Appendix Tables A1 and A2, along with online Appendix Figures A3 and A4). Leveraging these weights, we conducted doubly robust estimation, enabling the effect estimator to withstand misspecifications in either the treatment or outcome models. This approach also accounted for a few covariates that were not fully balanced in the treatment model (e.g., cases where the household head is retired).

The specification of the outcome model is below:

$$y_{it}={\alpha }_0+{\alpha }_{1}UpMo{b}_i+{\alpha }_{2}RaceSe{x}_i+{\alpha }_{3}UpMo{b}_i\times RaceSe{x}_i+{\alpha }_4{X}_{it}+{\epsilon }_{it}$$ (1)

where $y_{it}$ is the health outcome or behavior of individual $i$ in year $t$, $UpMo{b}_i$ is the upward mobility context of the county individual $i$ lives in, $RaceSe{x}_i$ is the combination of race and sex for individual $i$, $X_{it}$ includes time-constant and time-varying covariates, and ${\epsilon }_{it}$ is an error term. We separately estimated the model with and without the interaction term. When the interaction term is included, ${\alpha }_3$ captures the heterogeneous effects by race and sex.

RESULTS

Upward Mobility Context and Health Outcomes and Behaviors during the Transition to Adulthood

Figure 1 presents the outcomes from the linear probability models that elucidate the connection between upward mobility context and individual health outcomes and behaviors during the transition to adulthood. Comprehensive results can be found in online Appendix Table A3. As illustrated in Panel A, being raised in a county within the bottom 30 percentiles for upward mobility amplifies the likelihood of reporting poor health by 1.7 percentage points (p < 0.1). Given that around 9.5% of young adults in our sample indicated poor self-rated health (see Table 1), the 1.7 percentage points translate to 17.5% of this group, indicating a moderate effect size. Notably, Panel B does not indicate significant associations with mental health. Panel C reveals that growing up in a county within the bottom 30 percentiles for upward mobility heightens the likelihood of obesity by 3.4 percentage points (p < 0.1). Considering that approximately 22.4% of young adults in our sample had obesity (see Table 1), the 3.4 percentage points correspond to roughly 15% of this cohort, suggesting a moderate effect size.

Figure 1: Upward Mobility Context and Health during Transition to Adulthood, Without Controlling for Childhood Family Characteristics.

Note: Y-axis represents predicted probabilities. Results are derived from the linear probability models adjusting for the quadratic form of age and survey year dummies. Whiskers represent 95% confidence intervals. Data are from the Panel Study of Income Dynamics, restricted use data. The sample size is 10,223 (Self-Rated Poor Health), 10,228 (Depression), 10,216 (Distress), 10,127 (Obesity), 10,217 (Chronic Conditions), 10,235 (Alcohol Use), 10,208 (Drug Use), 10,223 (Cigarette Use).

Regarding health behaviors, Panel D indicates that growing up in a county within the bottom 30 percentiles for upward mobility elevates the likelihood of cigarette use by 6.0 percentage points, while simultaneously reducing the probability of alcohol use by 5.1 percentage points. Around 32.9% of young adults have smoked cigarettes, while 64.9% have consumed alcohol. These effect sizes represent 18.3% and 7.8% of these individuals, respectively. In summary, growing up in a low-opportunity county appears to yield moderate negative impacts on health and health behaviors, coupled with relatively minor positive effects on alcohol consumption. Intriguingly, consistent patterns emerge—being raised in a county within the top 30 percentiles for upward mobility lessens the likelihood of cigarette use by 5.5 percentage points, while augmenting the probability of alcohol use by 7.5 percentage points. No significant links were established for drug use.

Figure 2 displays the outcomes obtained using the PSW technique, which addresses selection into counties with varying levels of upward mobility. Detailed results are available in online Appendix Table A4. In alignment with the findings in Figure 1, albeit with smaller magnitudes, growing up in a county within the bottom 30 percentiles for upward mobility is linked to heightened probabilities of poor self-rated health (p < 0.1), obesity, and cigarette use. Simultaneously, it’s associated with diminished probabilities of alcohol use. Similarly, being raised in a county within the top 30 percentiles for upward mobility context is associated with elevated probabilities of distress, chronic conditions (p < 0.1), and alcohol use, while displaying reduced probabilities of cigarette use.

Figure 2: Upward Mobility Context and Health during Transition to Adulthood, Propensity Score Weighting.

Note: Y-axis represents predicted probabilities. Whiskers represent 95% confidence intervals. Data are from the Panel Study of Income Dynamics, restricted use data. The sample size is 10,223 (Self-Rated Poor Health), 10,228 (Depression), 10,216 (Distress), 10,127 (Obesity), 10,217 (Chronic Conditions), 10,235 (Alcohol Use), 10,208 (Drug Use), 10,223 (Cigarette Use).

These outcomes further reinforce the conclusions drawn from Figure 1, where growing up in a county characterized by high upward mobility levels was linked to a lower risk of smoking and an increased risk of drinking. Notably, distinct from the outcomes in Figure 1, being raised in a county within the top 30 percentiles for upward mobility context amplifies the probability of distress by 1.7 percentage points and raises the probability of having chronic conditions by 2.0 percentage points. Given that 4.8% of young adults in our sample reported distress and 34.1% reported having any chronic conditions, these effect sizes encompass 35.1% and 5.8% of these individuals, respectively. Notably, the effect size for distress is substantial compared to other outcomes, whereas the effect size for chronic conditions is relatively modest. These outcomes, for the most part, align with H1: Both very low- and very high-opportunity environments can result in detrimental health outcomes and behaviors. Moreover, high-opportunity environments are distinctly linked to heightened alcohol consumption.

Given that PSW methods enable the mitigation of selection bias by accounting for self-selection into counties with varying mobility contexts, without relying on functional form assumptions, these findings suggest that the influence of mobility on health might actually extend more broadly—impacting a greater range of health indicators—than initially suggested by the linear regressions in Figure 1.

Heterogeneous Effects by Race and Sex

Figures 3 and 4 presents the findings concerning the heterogeneous effects during the transition to adulthood based on the intersectionality of race and sex. Detailed results can be found in online Appendix Table A5. Among white men, being raised in a county situated within the bottom 30 percentiles for upward mobility context, as opposed to the remaining range, is linked to an elevated probability of chronic conditions by 4.7 percentage points. Interestingly, we observe that growing up in a county situated within the bottom 30 percentiles for upward mobility context, while leading to higher risks of being diagnosed with chronic conditions for white men, does not necessarily lead to higher risks of being diagnosed with chronic conditions for Black young adults, especially Black women. For Black men, the corresponding number is −3.0, which is 7.7 percentage points lower compared to white men. Similarly, for Black women, the corresponding number is −5.3, which is 10.0 percentage points lower compared to white men.

Figure 3: Upward Mobility Context in the Bottom 30 Percentiles and Health during Transition to Adulthood, by Race and Sex.

Note: Y-axis represents predicted probabilities. Whiskers represent 95% confidence intervals. Data are from the Panel Study of Income Dynamics, restricted use data. The sample size is 10,223 (Self-Rated Poor Health), 10,228 (Depression), 10,216 (Distress), 10,127 (Obesity), 10,217 (Chronic Conditions), 10,235 (Alcohol Use), 10,208 (Drug Use), 10,223 (Cigarette Use).

Figure 4: Upward Mobility Context in the Top 30 Percentiles and Health during Transition to Adulthood, by Race and Sex.

Note: Y-axis represents predicted probabilities. Whiskers represent 95% confidence intervals. Data are from the Panel Study of Income Dynamics, restricted use data. The sample size is 10,223 (Self-Rated Poor Health), 10,228 (Depression), 10,216 (Distress), 10,127 (Obesity), 10,217 (Chronic Conditions), 10,235 (Alcohol Use), 10,208 (Drug Use), 10,223 (Cigarette Use).

Among white men, being raised in a county situated within the bottom 30 percentiles for upward mobility context, as opposed to the remaining range, is linked to an elevated probability of cigarette use by 11.5 percentage points. The corresponding number for cigarette use is 10.6 percentage points lower among white women than among white men. Additionally, the likelihood of cigarette use is 8.1 percentage points lower among Black men and 9.1 percentage points lower among Black women compared to white men. These findings lend support to H2a, suggesting that the adverse impact of low mobility contexts can be mitigated among Black adults, and white women specifically in terms of cigarette use.

Furthermore, as shown in Figure 4, we find that Black women who grew up in a county situated in the top 30 percentiles for upward mobility context actually had higher probabilities of experiencing chronic conditions or depression compared to their counterparts in other mobility contexts. Among white men raised in a county located within the top 30 percentiles for upward mobility context, in contrast to the remaining range, the probability of depression is reduced by 1.1 percentage points and the probability of chronic conditions is increased by 0.7 percentage point. By contrast, the corresponding figures for Black women are 2.9 and 5.3 percentage points higher, respectively. These findings provide substantial support for H2b and suggest that the health benefits gained by Black women from high upward mobility contexts are comparatively diminished when compared to the advantages experienced by white men. These conclusions align with other studies that report diminished benefits of higher childhood SES for Black women. For example, Assari (2018) finds that higher parental education attainment is associated with a smaller boost in upward mobility for Black women compared to white women and men. In relation to alcohol and cigarette use, the probabilities for white men stand at 6.5 and −3.8 percentage points, respectively. Notably, the probability of alcohol use sees a decrease of 6.1 percentage points among Black men, while the likelihood of cigarette use increases by 9.1 percentage points in comparison to white men. This implies that specific health behaviors of Black men derive greater benefits from favorable opportunity environments compared to white men, whereas in other aspects, the advantages are comparatively less pronounced.

DISCUSSION

In this study, we investigate the impact of upward mobility context on health outcomes and behaviors during the pivotal transition to adulthood. Moreover, we explore how this relationship is influenced by the intersectionality of race and sex. By employing race- and sex-specific mobility measures and analyzing a diverse range of outcomes, our PSW findings reveal that low levels of upward mobility context correspond to moderately heightened risks of self-rated poor health, obesity, and smoking, yet slightly diminished risks of alcohol consumption. Additionally, growing up in a county with high mobility is linked to substantially increased risks of distress, and slightly raised risks of chronic conditions and alcohol consumption. Conversely, it is associated with moderately reduced risks of smoking. These outcomes largely lend support to H1.

Our findings highlight the intricate interplay of race and sex in shaping the relationship between upward mobility context and health outcomes and behaviors during the transition to adulthood. In situations of limited opportunity, Black respondents experienced relatively fewer negative impacts in terms of the probabilities of developing chronic conditions and engaging in cigarette use (also true for white women) as compared to white men. Conversely, in environments characterized by elevated opportunities in contrast to other mobility contexts, Black women derived lesser advantages towards risks of depression and chronic conditions, while Black men experienced diminished benefits concerning smoking risks as compared to white men. These outcomes lend support to hypotheses H2a and H2b, underscoring the intricate nature of these interactions between race, sex, upward mobility context, and health-related factors.

Our findings regarding the adverse impact of high mobility contexts on distress, chronic conditions, and alcohol consumption might appear counterintuitive at first glance. However, our findings on alcohol use are indeed congruent with prior studies indicating that growing up in socioeconomically advantaged neighborhoods, which often boast a higher concentration of bars and restaurants, predicts increased alcohol consumption during the transition to adulthood (Pedersen et al. 2018; Slutske et al. 2016). Regarding chronic conditions, it is plausible that environments characterized by ample opportunities tend to offer better medical resources, thereby potentially reducing the likelihood of underdiagnosing—a common occurrence in regions with fewer opportunities. In any case, among the three outcomes, the effects on chronic conditions and alcohol consumption are relatively modest, whereas the impact on distress is large, indicating that the negative effects primarily pertain to mental health. As hypothesized, high opportunity environments, particularly during young adulthood, can be exceptionally competitive, leading to heightened social comparison and subsequently elevated stress levels (Salleh 2008). Additionally, high opportunity environments might increase the odds of individual upward mobility, potentially resulting in poorer mental health due to the erosion of social ties from one’s social origin and the instability of social networks within an individual’s community (Sorokin 1959).

Overall, our findings underscore our argument that the influence of upward mobility context varies according to the specific health indicator under consideration and is intricately shaped by the intersection of race and sex. Generally, the adverse impact of low mobility environments on physical health, particularly in terms of chronic conditions, appears to be mitigated among Black adults. Simultaneously, their health gains from high mobility environments are comparatively lesser than those experienced by white men. Our application of intersectionality theory has allowed us to illuminate distinct pathways through which upward mobility context relates to health outcomes and behaviors within each racially and gender-specific group. The distinct process of racial socialization within the United States, alongside a heightened cultural emphasis on community and familial support, may have equipped Black individuals with better coping mechanisms to navigate challenges in adverse environments in contrast to their white counterparts (Rosenfield 2012; Ursache et al. 2022). However, high-opportunity environments in the United States often exhibit a dominance of white influence, potentially subjecting Black people to heightened psychosocial stressors. In managing these challenges, there could be associated costs to both physical and mental health among Black adults (Ciciurkaite 2021; Hargrove 2018; James 1994).

Our findings make contributions to the field of medical sociology in several ways. Primarily, existing research on the relationship of social mobility and health has historically centered on individual mobility, often neglecting the consideration of county-level intergenerational mobility. Such studies frequently fail to accommodate the diverse contextual landscapes in which individuals reside. Notably, even when folks achieve similar degrees of upward mobility on an individual level, the broader upward mobility context can differ greatly. This variation significantly shapes their experiences, which in turn holds critical implications for their health outcomes. For instance, attaining upward mobility within a low-opportunity environment might entail significantly more hurdles, potentially leading to heightened stress levels compared to achieving the same mobility in a high-opportunity setting. Our findings suggest that this nuanced perspective may help elucidate the disparate and sometimes conflicting findings observed in this line of research (Präg and Gugushvili 2020; Zang and Dirk de Graaf 2016).

Secondly, a substantial body of literature has advocated for the strategy of relocating racial minorities to more favorable environments as a means to alleviate racial disparities in socioeconomic outcomes (Cashin 2001). However, aligning with the emerging discourse on racial and ethnic disparities in health (Ellickson 2010; Massey and Tannen 2018), our findings present a compelling argument against the effectiveness of adopting a tokenistic approach that involves simply transferring select racial minorities to high-opportunity environments. Notably, these environments often exhibit a predominant white influence. It becomes evident that relying solely on such relocation tactics will fall short in effectively addressing the broader issue of narrowing racial disparities in health outcomes and behaviors. While it is true that relocating racial minorities to improved environments may yield enhanced SES outcomes, as indicated by certain studies (Leventhal and Brooks-Gunn 2003), it is vital to recognize that such progress might come at the expense of health-related outcomes. Instead, a more comprehensive approach is warranted. Our research underscores the necessity for policies and interventions that target the fundamental sources of structural racism, along with its cascading effects on SES, healthcare infrastructure, and subconscious racial biases prevalent in the environments where racial minorities are situated.

There are several limitations to this study. Firstly, our upward mobility context measures were ultimately derived from regression estimates, introducing a degree of uncertainty (Chetty et al. 2018). However, this uncertainty primarily affects smaller counties. Our examination reveals that the majority of counties in the TAS have sizable populations (i.e., 80% of individuals in our sample reside in counties with at least 65,000 people). Consequently, we anticipate reduced uncertainty around these central estimates.

Secondly, due to data constraints, we did not examine the mechanisms through which upward mobility context affects young adults’ health and behaviors, especially in relation to race and sex. Our investigation suggests that racial trends are partly tied to SES disparities (see online Appendix S2). Relatedly, our study uses broad indicators for upward mobility. Future research could use more refined measures like county-level SES, health facilities, and environment for deeper insights. Moreover, our emphasis on upward mobility at the county level is rooted in the county’s role as the most granular administrative unit for government policymaking. Future work could investigate even smaller scales like schools and neighborhoods, revealing more about upward mobility dynamics.

Thirdly, while our PSW analysis helps address selection biases, there might still be unobserved factors influencing the results. Exploring causal links between context and health remains a challenge for future research. Fourthly, our assumption of stable county-level upward mobility over two decades underpins the link between context and respondents. Aligning cohorts with updated mobility context calculations could enhance future studies. Fifthly, our analysis lacked data on Hispanic, Asian, Native American, and non-binary individuals due to sample size limitations. Lastly, due to data constraints, some health outcomes and behaviors were broad. Despite the presence of missing observations, checks with relatively refined measures showed consistent findings (see online Appendix S3).

Despite these limitations, our study has important policy implications. Our results emphasize that individuals’ health is often a result of their histories: particularly the environments in which they spend their formative years. It is critical to invest in community programs for promoting economic opportunities and health among young and emergent adults in disadvantaged communities. Specifically, our results suggest the possible benefits of strengthening obesity- and smoking-prevention programs in areas with low upward mobility, with the aim of mitigating later life impacts. A large literature exists to guide the design of these programs (King et al. 2011; Kornet-van der Aa et al. 2017). In addition, our results illustrate the importance of policies that respond to the complicated interaction between race, sex, place, and health. The interventions required by one demographic in a locality may differ from those required by others. For instance, high mobility contexts are associated with lower rates of cigarette use in general. However, there may nonetheless be a benefit in providing targeted support for Black men in these areas who appear not to reap the rewards of high opportunity environments as much as white men. Finally, policy attention should also be paid to the higher levels of distress, chronic conditions, and alcohol consumption among young adults growing up in high opportunity environments.

Ultimately, our results suggest that the focal point of efforts should shift towards enhancing the quality of the environments in which racial minorities reside. Addressing the root causes of disparities requires a multi-faceted strategy that not only uplifts socioeconomic prospects, but also actively dismantles the barriers perpetuated by structural racism, thus creating a more equitable foundation for health and well-being.

Biographies

Emma Zang, assistant professor in the Department of Sociology at Yale University. Dr. Zang’s research interests intersect health and aging, family demography, and inequality, with a particular focus on examining these dynamics in both the United States and China. Her work has appeared in journals such as the American Journal of Sociology, Demography, Social Science & Medicine, JAMA Internal Medicine, and Nature Human Behaviour. She is a Butler-Williams Scholar, and an Early Career Faculty Scholarship recipient from the National Institute on Aging.

Melissa Tian, senior undergraduate student in the Department of Sociology at Yale University. Melissa’s research interests include economic mobility and health, social determinants of health, and effectiveness of healthcare policy. She is currently working on a senior thesis project that looks to examine the perceptions of healthcare workers on the effectiveness of social determinants of health screening.