Temporary Integration, Resilient Inequality: Race and Neighborhood Change in the Transition to Adulthood

Abstract

This article focuses on neighborhood and geographic change arising with the first “selection” of an independent residential setting: the transition out of the family home. Data from two sources—the Project on Human Development in Chicago Neighborhoods, and the Panel Study of Income Dynamics—are used to provide complementary analyses of trajectories of change in geographic location and neighborhood racial and economic composition during young adulthood. Findings indicate that for young adults who originate in segregated urban areas and remain in such areas, the period of young adulthood is characterized by continuity in neighborhood conditions and persistent racial inequality from childhood to adulthood. For young adults who exit highly segregated urban areas, this period is characterized by a substantial leveling of racial inequality, with African Americans moving into less-poor, less-segregated neighborhoods. However, the trend toward racial equality in young adulthood is temporary, as the gaps between whites and blacks grow as the young adults move further into adulthood. Crucial to the reemergence of racial inequality in neighborhood environments is the process of “unselected” change, or change in neighborhood conditions that occurs around young adults after they move to a new neighborhood environment.

Introduction

The transition to adulthood represents a unique window in the life course, providing insight into how childhood advantages and disadvantages persist, fade away, or are disrupted as an individual moves into adulthood. The factors that determine the extent of change and continuity in a young adult’s developmental trajectory are diverse, but a set of life course transitions—including the transition out of the family home, or entrance into the labor force, the criminal justice system, the military, parenthood, and/or romantic partnerships or marriage—have been identified as potential “turning points” that provide individuals with the chance to chart a new course (Elder 1998; Laub and Sampson 2003; Maughan and Champion 1990; Rumbaut 2005; Settersten et al. 2005). Relatively neglected in the literature on young adulthood is a thorough consideration of change and continuity in the young adult’s neighborhood environment and geographic location during this period of the life course (Mulder 2007).

Recent research has demonstrated strong continuity in the neighborhood environments of families from one generation to the next (Sharkey 2008), suggesting that neighborhood conditions may be one central pathway by which advantages and disadvantages experienced in one generation are reproduced in the next generation. Unclear from this research, however, are the mechanisms underlying the intergenerational persistence of neighborhood advantage and disadvantage. This article draws on two data sets to shed new light on the persistence of racial inequality in neighborhood environments from childhood to adulthood by focusing attention on change in neighborhood conditions and geographic location arising with the first residential decision made by young adults during the transition to adulthood: the move out of the family home.

Race and Neighborhood Attainment

Research on neighborhood attainment has come to be viewed in terms of the resources and preferences that individuals and families bring to the housing market at different points in the life course, and the constraints that they face within that market. Spatial assimilation theory assumes that as members of minority or immigrant groups establish themselves in mainstream labor markets, they will attempt to translate economic advances into residential advantage by moving out of segregated areas and into areas occupied by members of the dominant racial/ethnic group (Alba and Logan 1993; Massey and Denton 1985; Massey and Mullan 1984). Consistent with this perspective, evidence shows that racial and ethnic gaps in neighborhood economic status and neighborhood racial composition are reduced by adjusting for group differences in social and economic status, and growing levels of black suburbanization have brought African Americans into communities that look very similar, in economic and demographic composition, to those of their white counterparts with similar social and economic status (Clark 2007; Frey 2000).

At the same time, nontrivial gaps between white and nonwhite groups remain even after considering the resources that groups bring to the housing market (Alba and Logan 1993; Alba et al. 2000; Crowder et al. 2006; Logan and Alba 1993; Sharkey 2009), suggesting that racial inequality in the nation’s neighborhoods is not driven entirely by economic inequality. This evidence reflects the constraints placed on neighborhood selection through processes of discrimination and regulation of entry, which are consistent with place stratification theory (Logan 1978; Logan and Molotch 1987; Massey and Denton 1993).

A related literature points to individuals’ preferences for the racial makeup of their neighborhoods as a primary explanation for the residential patterns that are visible in America’s cities. This research establishes that preferences for “out-group” neighbors fall along racial and ethnic lines, with whites consistently ranked as the preferred out-group, followed by Asians, Hispanics, and African Americans (Charles 2000; Clark 1992; Zubrinsky and Bobo 1996). Whereas certain segments of the African American middle-class population express a preference to remain in predominantly black communities (Clark 2007), as a whole, African Americans are the group most open to living in integrated neighborhoods, yet they are thought of as the least desirable neighbors by whites, Hispanics, and Asians.

The implications of divergent preferences for neighborhood racial/ethnic composition have been explored in several studies that simulate the process by which groups’ residential decisions interact to create aggregate patterns of racial segregation across cities (Bruch and Mare 2006; Clark 1991; Fossett 2006; Schelling 1971; Zhang 2004). One important conclusion arising from this strand of research is that relatively small divergence in preferences for neighbors’ racial background and economic status can lead to high levels of segregation in the aggregate, even in the absence of substantial discrimination in the housing market (Clark 1991; Schelling 1971; Zhang 2004). This literature thus demonstrates how sequences of interrelated decisions made by individuals responding to the change occurring around them can help explain aggregate patterns of racial and economic segregation. A more general lesson that emerges from this research is that individuals’ residential outcomes are attributable not only to their own preferences and their own resources but also to the choices being made among the neighbors that surround the family.

Home-Leaving and the Geography of Young Adulthood

While these general models of neighborhood attainment and neighborhood change may be applicable to the neighborhood trajectories of young adults, there is also a unique set of circumstances and transitions that characterize this period of the life course and that may complicate understanding of young adults’ neighborhood attainment. One basic observation about young adulthood is that residential mobility is more common during this period than at other life course stages (Long 1988; South and Deane 1993). Further, life course transitions—such as the end of schooling, entrance into the labor force, the formation of an independent family, and the continuation or dissolution of childhood peer networks—often come clustered during young adulthood, and may represent turning points with the potential to lead to changes in the young adult’s residential setting (Lee et al. 1994; McHugh et al. 1990; Long 1988). Although an extensive literature has situated the process of home-leaving alongside this set of co-occurring transitions (Goldscheider and Goldscheider 1999; Settersten et al. 2005; White 1994), only a few studies have examined changes in geography arising from the transition out of the family home (Garasky 2002; Mulder 2007; Mulder and Clark 2000).

Research on social networks and residential mobility suggests that individuals’ ties to places are closely intertwined with their ties to family and peer networks (Mulder and Cooke 2009), and multiple studies have found that residential proximity to family members and broader social networks is a central factor that helps to explain differential rates of residential mobility among different racial/ethnic groups and individuals with varying economic status (Dawkins 2006; Spilimbergo and Ubeda 2004). Alternatively, the set of transitions that occur during the transition to adulthood often leads to or else coincides with changes in residence or geographic location. Building on these ideas, the current analysis attempts to capture the different dimensions of change in neighborhood environments during the period of young adulthood by considering both change in geography as well as change in neighborhood composition. The central hypothesis guiding the analysis is that the degree of change in young adults’ neighborhood environments is closely related to the degree of change experienced in geographic location. Put differently, continuity in neighborhood characteristics during young adulthood is hypothesized to be strongest among young adults who remain within their origin city or county, and weakest among young adults who relocate to a new “place.”

This hypothesis is derived from two strands of research. First, research on neighborhoods and the life course demonstrates that changes in geographic location—particularly among youth in highly segregated cities or metropolitan areas—frequently lead to or coincide with disruptions in patterns of inequality, suggesting that when young adults relocate, they will experience the most substantial changes to multiple dimensions of their lives (Maughan and Champion 1990; Mulder 2007; Rubinowitz and Rosenbaum 2000; Sampson and Sharkey 2008). Second, research examining the transmission of neighborhood disadvantage from one generation to the next shows that the correlation of neighborhood economic status across generations is strongest for families that remain in the same place in both generations (Sharkey 2008). A related literature examining residential mobility within highly segregated metropolitan areas demonstrates that the “structure” of inequality within such places extends beyond individual neighborhoods, so that residential moves that cross the boundary of the origin neighborhood but leave the individual within the same city do not lead to pronounced changes in the character of their neighborhood environment (Sampson 2008; Sampson and Sharkey 2008; South and Deane 1993). Individual trajectories of neighborhood change are thus thought to be dependent on whether the individual remains in or exits the tightly structured landscape of segregated metropolitan areas.

Data and Methods

To assess neighborhood and geographic change during the period of young adulthood, this analysis draws on two sources of data. First, the Project on Human Development in Chicago Neighborhoods (PHDCN) is a longitudinal study of child development within Chicago neighborhoods that allows for an in-depth look at transitions occurring during young adulthood (Earls et al. 1997). This article draws on data from the PHDCN Longitudinal Cohort Study, a survey of a representative sample of children and caregivers living in Chicago as of the first interview wave, which occurred primarily in 1995. Three waves of data were collected, the second round occurring mainly in 1997/1998, and the third in 2001/2002. Because the focus is on the process of leaving home, the sample is limited to members of the 15- and 18-year-old cohorts, with particular attention placed on the 18-year-old cohort. The relevant sample for all regression models consists of 612 members of the 18 cohort and 651 members of the 15 cohort who were living in Chicago as of the first wave of data collection.

The second source of data is the Panel Study of Income Dynamics (PSID) (Hill and Morgan 1992), which allows for a broader focus on the nation as a whole and also allows for a portrait of change that covers a longer period of young adulthood. The PSID began with a nationally representative sample of roughly 5,000 families in 1968, and has followed the members of these families over time.¹ Families are matched to their census tract of residence through the PSID restricted-use geocode file, which contains census tract identifiers for sample families from 1968 through 2003.² Sample weights are used that are designed to make the sample representative of the United States’ population as of the first survey year, 1968. These weights are designed to account for sample attrition, which is extensive because of the length of the follow-up.³ The PSID sample consists of all young adults who are observed as a child in a PSID household and are then observed as a household head or the spouse of a household head. The sample size is 6,614, and the sample in high-segregation metropolitan areas is 1,148. Young adults are observed from the age of 17 through 35.

Data from the PHDCN and PSID are merged with compositional characteristics of families’ census tracts through the Neighborhood Change Database (NCDB) (GeoLytics 2003) for census years 1970, 1980, 1990, and 2000. Tract characteristics in intercensal years are imputed using linear interpolation. Online Resource 1 provides a detailed description of all variables used in the PHDCN and PSID analyses, and descriptives for each “young adult” sample are available in Table S1.

Defining Exits Out of the Parental Home

The transition out of the family home is often described as a process, rather than an event, given that young adults may move back and forth between the parental home and other supervised settings before establishing their own households (Goldscheider and DaVanzo 1989; Goldscheider and Goldscheider 1999; Goldscheider et al. 1993). As such, the transition has been defined and operationalized in multiple ways, often based on the unique features of the data source being analyzed. The PHDCN offers extensive information on the residences of respondents and the members of their households, providing the opportunity to operationalize home-leaving in a way that is consistent with theoretical criteria rather than data constraints (White 1994). For the purposes of this study, theoretical interest lies in the analysis of change and continuity in the environments of young adults as they make choices about their residential location over the period of young adulthood. The primary dimension of this transition is the residential independence of young adults from caregivers and those who may take on the role of caregivers. On the basis of these criteria, a young adult who has transitioned out of the family home is defined as one who lives independently from his own parents and from other potential caregivers, including stepparents, adoptive parents, or family members from an older generation (e.g., aunts, uncles, or grandparents).

At each of the three waves of the PHDCN survey, young adults who live in households with no potential caregivers from an older generation are identified as “home-leavers,” and those who live in households with older caregivers are identified as “stayers.” This operationalization is defined for each young adult at each wave, and thus allows for movement between the status of home-leaver and stayer between waves. Changes in status that occur between waves are not incorporated into the definition because no information on such changes is available.

Relative to the PHDCN, the PSID offers less detailed information on the precise living situations of young adults upon leaving home, and residential information for young adults during the intermediate period between living at home and starting one’s own household is not available. By necessity, therefore, in all analyses using the PSID, home-leaving is defined as a change in status from a dependent “child” in the PSID sample household to the status of “head of household” or the “spouse” of a head of household. “Home-leavers” are children who have split off from the original PSID family and who have formed their own households. Home-leaving in the PSID can thus be defined as the transition from living as a dependent in a caregiver’s household to living in an independent household.⁴

Analysis Plan

Two-level hierarchical linear models are used to estimate growth curves that describe trajectories of neighborhood conditions for individuals from different racial and ethnic groups (Bryk and Raudenbush 1992; Halaby 2003; Sampson and Sharkey 2008). This approach allows for the decomposition of multiple sources of neighborhood change, including change that arises from home-leaving and change occurring around young adults who do not leave home. Change occurring among young adults who remain at home can arise from migration into or out of a community or secular changes in social conditions that affect the circumstances of the residents in an individual’s community, such as a rise or decline in unemployment. Building on previous work decomposing change in neighborhood conditions over time (Sampson and Sharkey 2008), the analysis consists of a progression of models that allow for the decomposition of change in neighborhood conditions among three groups: stayers, home-leavers who remain within the same city or county, and home-leavers who exit their origin city/county. The dependent variables in these models are two compositional characteristics of the individual’s census tract: the poverty rate, and the percentage of African American residents.

For all analyses of the PHDCN, the unit of analysis at Level 1 is time points, t, which are nested within individuals, i, and which represent the multiple survey points at which individuals were interviewed. At Level 2, the unit of analysis is individuals. The Level 1 model is shown in Eq. (1):

$$Y_{ti}={\pi }_{0i}+{\pi }_{1i}{(\text{time})}_{ti}+{\pi }_{2i}{(\text{leaver})}_{ti}+{\pi }_{3i}{(\text{time}\times \text{leaver})}_{ti}+{\pi }_{4\text{i}}{(\text{outside}\_\text{Chicago})}_{\text{ti}}+{\pi }_{5\text{i}}{(\text{time}\times \text{outside}\_\text{Chicago})}_{ti}+e_{ti}.$$ (1)

In this specification, Y_ti is the dependent variable (e.g., neighborhood poverty) observed for all subjects i = 1 … n, at time points t = 0, 1, 2, which represent each survey wave; π_0i, the model intercept, represents the mean level of neighborhood poverty for individuals who remain at home as of Wave 1 of the survey; and time is set equal to 0 at Wave 1 of the survey, to 1 at Wave 2, and to 2 at Wave 3, and describes the average linear change in neighborhood poverty from Wave 1 to Wave 3 among young adults who remain at home. Leaver is a time-varying, dichotomous indicator for whether the individual lived outside of the parental home as of the given survey wave. The interaction of leaver with time allows for the decomposition of change in neighborhood poverty for individuals who leave home compared with those who remain at home over the course of the survey. The specification also accounts for any differences in neighborhood poverty among home-leavers and stayers that were present at Wave 1 of the survey. Outside_Chicago is an indicator for home leavers who exit Chicago, and allows for the decomposition of change in neighborhood conditions among home-leavers by distinguishing between change occurring among movers who stay in Chicago and change occurring among movers out of the city. The interaction of outside_Chicago with time estimates the change in neighborhood poverty for individuals who leave home and exit Chicago compared with those who leave home but remain within the city.

In the Level 2 between-person models, all Level 1 terms representing the initial status and change over time are allowed to vary by race and ethnicity. The Level 2 equations for the overall model intercept and for the time parameter are shown in Eq. (2), and equations for all other Level 1 terms are the same:

$${\mathrm{\Pi }}_{0i}={\beta }_{00}+{\beta }_{01}{(\text{black})}_i+{\beta }_{02}{(\text{Latino})}_i+{\beta }_{03}{(\text{other})}_i+r_{0i}$$

$${\pi }_{1i}={\beta }_{10}+{\beta }_{11}{(\text{black})}_i+{\beta }_{12}{(\text{Latino})}_i+{\beta }_{13}{(\text{other})}_i.$$ (2)

Whites serve as the reference group in these specifications, and other refers to all other individuals who are not identified as white, African American, or Latino. Interpreting the Level 1 and 2 coefficients is complex, so main results are displayed in tables and in graphical form, easing interpretation of the trajectories of change in neighborhood conditions. A detailed description of the interpretation of coefficients is available in Online Resource 1.

The primary difference between the PHDCN and the PSID is that multiple cohorts of young adults are followed in the PSID. The central focus of the analysis thus shifts from change occurring over time, as in the PHDCN models, to a focus on how the neighborhood environment changes as young adults age further into adulthood. Instead of nesting individuals within interview points, Level 1 in all PSID models represents change occurring as the respondent advances from age 17 to age 35. The Level 1 model includes two parameters that describe change as young adults age, a linear and a quadratic term:⁵

$$Y_{ti}={\pi }_{0i}+{\pi }_{1i}{(\text{age})}_{ti}+{\pi }_{2i}{({\text{age}}^2)}_{ti}+e_{ti}.$$ (3)

At Level 2, both parameters are allowed to vary by race, by a dichotomous indicator of whether the respondent moved out of their original county, and by the interaction of race and the measure of geographic mobility:

$${\pi }_{0i}={\beta }_{00}+{\beta }_{01}{(\text{black})}_i+{\beta }_{02}{(\text{changed}\text{county})}_i+{\beta }_{03}{(\text{black}\times \text{changed}\text{county})}_i+r_{0i}.$$ (4)

This model provides estimated neighborhood trajectories among whites and African Americans who remain in their county of origin and those who move on to a different county.

Results

Neighborhood Trajectories Among Young Adults in Chicago

Table 1 describes the prevalence of home-leaving and the origin and destination neighborhoods of home-leavers among sample members in the 18-year-old PHDCN cohort. About 38% of all young adults in the 18-year-old cohort lived independently at some point between Wave 1 (1995) and Wave 3 (2001) of the survey,⁶ with similar proportions of white, African American, and Latino home-leavers. The destinations of home-leavers vary by race and ethnicity. Among whites, 20% of all young adults leave home and exit Chicago compared with 13% of African Americans and only 8% of Latinos. Thus, among all groups of home-leavers, whites are the most likely to experience a significant change in geography when they leave the family home. Even among whites, though, almost three of four home-leavers remain in Illinois, suggesting that most transitions out of the family home likely do not result in a complete disruption of social ties that are formed in Chicago. This continuity in geographic location is even more pronounced among African Americans, and especially among Latinos.

Table 1.

Prevalence of home-leaving and geographic mobility of home-leavers, by race/ethnicity: PHDCN 18-year-old cohort

	Sample Size	% Who Leave Home	% Who Leave Home and Exit Chicago	% Who Leave Home and Exit Illinois
All	629	38 (n = 241)	12 (n = 77)	6 (n = 36)
White	115	41 (n = 47)	20 (n = 23)	11 (n = 13)
Black	246	41 (n = 100)	13 (n = 33)	7 (n = 18)
Latino	240	37 (n = 89)	8 (n = 19)	2 (n = 5)
Other	28	18 (n = 5)	7 (n = 2)	0 (n = 0)

Results showing trajectories of change in neighborhood poverty and neighborhood racial composition are displayed in Fig. 1; the coefficients from growth curve models that generate these figures are shown in Table 2. For the sake of clarity, the figure shows trajectories of change for home-leavers who remain in Chicago and home-leavers who exit Chicago, and excludes stayers, whose patterns of change mirror those among home-leavers who remain within the city. Trend lines are shown for whites, African Americans, and Latinos, with members of all other racial and ethnic groups not shown because of the small sample sizes and to avoid overly cluttered figures.

Fig. 1.

Trajectories of change in neighborhood poverty and percent black among home-leavers who remain in Chicago and home-leavers who exit Chicago, by race/ethnicity: PHDCN 18-year-old cohort

Table 2.

Hierarchical linear models of neighborhood poverty and percentage black, by race/ethnicity and home-leaver status: PHDCN 18-year-old cohort

	Neighborhood Poverty Rate	Neighborhood % Black
Initial Status for Stayers
Intercept (white)	0.121** (0.021)	0.038 (0.044)
Black	0.128** (0.015)	0.693** (0.034)
Latino	0.088** (0.014)	0.069* (0.033)
Other race	0.053* (0.026)	0.097† (0.053)
Initial Status for Movers Within City (relative to stayers)
Intercept (white)	0.005 (0.020)	−0.031 (0.049)
Black	−0.018 (0.026)	0.060 (0.058)
Latino	0.026 (0.025)	0.084 (0.058)
Other race	−0.077* (0.036)	−0.008 (0.135)
Initial Status for Movers Outside City (relative to movers within)
Intercept (white)	−0.001 (0.029)	0.013 (0.060)
Black	0.003 (0.036)	−0.104 (0.075)
Latino	−0.048 (0.038)	−0.009 (0.081)
“Other”	0.060 (0.048)	−0.075 (0.194)
Change for Stayers
Intercept (white)	−0.003 (0.003)	−0.001 (0.013)
Black	0.004 (0.005)	0.010 (0.016)
Latino	−0.016** (0.005)	0.001 (0.016)
“Other”	−0.013 (0.013)	−0.030 (0.028)
Change for Movers Within City (relative to stayers)
Intercept (white)	0.004 (0.012)	−0.001 (0.013)
Black	−0.003 (0.015)	0.010 (0.016)
Latino	0.001 (0.015)	0.001 (0.016)
“Other”	−0.009 (0.024)	−0.030 (0.028)
Change for Movers Outside City (relative to movers within)
Intercept (white)	0.039 (0.033)	0.020 (0.027)
Black	−0.092* (0.035)	−0.215** (0.034)
Latino	−0.069† (0.038)	−0.035 (0.038)
“Other”	−0.053 (0.038)	0.038 (0.096)

Notes: N = 629. Two-level hierarchical linear model, with individual survey years nested within individuals. Models include controls for measures of gender, age, household income, completed education, marital status, immigrant generation, and length of time at baseline address.

^†p < .10;

^*p < .05;

^**p < .01

What stands out very clearly from the top panel of Fig. 1 is the racial hierarchy in neighborhood economic status as of the first wave of the survey, with whites originating in neighborhoods with the lowest rates of poverty, followed by Latinos, and finally African Americans. Patterns of change over the course of the survey appear to disrupt this racial/ethnic hierarchy, however. Although the neighborhood poverty rate remains stable for home-leavers who remain in Chicago (with slight declines for Latinos only), the same is not true for home-leavers who exit Chicago. Among home-leavers who leave the city, there is a racial crossover in neighborhood poverty, with African Americans and Latinos experiencing sharp declines and whites experiencing sharp increases in neighborhood poverty over the course of the study. By Wave 3 of the PHDCN study, whites who leave home and depart Chicago have lost their advantaged position and end up in neighborhoods with higher poverty rates than African Americans and Latinos who leave home and leave the city.

The bottom panel of Fig. 1 examines trajectories of change in neighborhood percentage black. The racial segregation that characterizes Chicago is immediately apparent from the figure: African Americans occupy neighborhoods with extremely high percentages of black residents when compared with Latinos or whites. Racial composition remains stable for almost every group shown in the figure, with the lone exception of African Americans who leave home and exit Chicago. Among this group, there is a steep decline in neighborhood percentage black, indicating that African Americans who exit Chicago enter neighborhoods that are substantially more integrated than those from which they came. The same is not true for Latinos, who remain in neighborhoods that are between 10% and 20% black regardless of whether they remain within the city or exit Chicago. Unlike the findings for neighborhood poverty, white home-leavers who exit Chicago show minimal change in neighborhood racial composition. All groups of whites continue to live in neighborhoods with minimal black presence over the course of the study.⁷

Considered together, the two graphs indicate that racial and ethnic inequality in neighborhood conditions is maintained for young adults who remain within the highly stratified landscape of residential Chicago. However, for young adults who leave the city, there is a clear trend toward integration, with African Americans moving into racially integrated neighborhoods and an overall leveling of racial gaps in neighborhood poverty.⁸ Considering the fact that Chicago is one of the nation’s most segregated cities, this pattern might be interpreted to mean that African Americans from Chicago will inevitably end up in a more racially integrated community when they leave the city. Although this is certainly part of the explanation for the changes seen in the figures, this interpretation cannot account for the magnitude of the decline in neighborhood poverty and percentage black among African Americans, nor can it account for the racial crossover in neighborhood poverty among whites and blacks who leave Chicago. Another possible explanation for these patterns is that young adults may be moving to college campuses that are more diverse than their origin neighborhoods in Chicago. Subsequent analysis indicates that college attendance is part of the explanation for why whites experience such a sharp rise in neighborhood poverty: whites who leave Chicago and report attending a four-year college experience an increase in neighborhood poverty. However, even whites who are not attending college move to higher-poverty neighborhoods when they leave home and leave Chicago. Further, college enrollment does not explain any of the drop in neighborhood poverty or neighborhood percentage black among African American home-leavers who exit Chicago.

Whereas the previous analyses tracked changes in individuals’ residential environments over time, Table 3 shifts focus and describes changes in the neighborhoods into which young adults arrive when they leave Chicago. Specifically, the table describes how these destination neighborhoods changed from 1990 to 2000, providing a sense of whether the change in individuals’ neighborhood conditions that is brought about by leaving the city is likely to be a stable versus transitory change in young adults’ neighborhood environments.

Table 3.

Change in the destination neighborhoods of home-leavers who exited Chicago during the 1990s, by race/ethnicity: PHDCN 18-year-old cohort

	% Poor	% Black	% White
1990 Neighborhood Characteristics
White	18	6	87
Black	13	23	70
Latino	12	7	80
Average Change From 1990 to 2000
White	−0.01	+0.01	−0.05
Black	+0.03	+0.16	−0.20
Latino	+0.00	+0.03	−0.10

Table 3 provides evidence to support each interpretation. The second set of rows in the table shows that the average amount of change in the poverty rates of destination neighborhoods during the 1990s was negligible for whites and Latinos, whereas the average destination neighborhood of African Americans experienced a 3 percentage point rise in poverty from 1990 to 2000. Changes in neighborhood racial composition reveal a similar pattern. The destination neighborhoods of whites, African Americans, and Latinos all experienced some growth in the black population within the neighborhood and declines in the white population over the 1990s, although the degree of change varies markedly. The destination neighborhoods of whites experienced negligible change in percentage black and small declines in percentage white, and the destination neighborhoods of Latinos experienced slightly greater increases in percentage black and a large drop in percentage white of 10 percentage points. The destination neighborhoods of African Americans who left home and left Chicago, by contrast, experienced an increase in neighborhood percentage black of 16 percentage points and a decline in neighborhood percentage white of 20 percentage points during the 1990s.

Table 3 thus presents strong evidence to suggest that nonwhites who leave home and leave Chicago enter neighborhoods with growing concentrations of minority populations; this is particularly true for African Americans. Whereas the individual trajectories of black home-leavers who exit Chicago show steep declines in racial segregation, the results shown in Table 3 suggest a process of “resegregation,” in which the destination neighborhoods of African Americans who leave Chicago are transforming into racially and ethnically segregated neighborhoods that resemble the segregation found within Chicago.

A Wider, Longer View: Transitions Out of the Family Home in the PSID

In an effort to expand beyond Chicago, the remainder of the analysis draws on the PSID, which provides a national view of neighborhood and geographic change that extends from young adulthood through early adulthood. As an initial analysis, Table 4 shows neighborhood racial and economic change from the last year in which the young adult was classified as a child in the sample family to the first year in which the young adult was classified as a household head or the spouse of a household head. The first set of rows in the table reveals patterns of change that are much less pronounced than those found in the PHDCN. Although African Americans who move to a different county upon leaving home do end up in neighborhoods where the proportion of blacks and the poverty rate are lower than in their origin neighborhoods, the changes are nowhere near those found among young adults in Chicago.

Table 4.

Change in neighborhood racial and economic composition in the transition out of the family home, by race: PSID young adults

	White		Black
	Same County	Different County	Same County	Different County
All Young Adults
Initial status in family home
% Black	4	4	63	58
% White	91	92	32	38
% Poor	9	8	25	21
Change from family home to adult home
% Black	0.01	0.04	−0.06	−0.12
% White	−0.02	−0.06	0.05	0.10
% Poor	0.02	0.03	0.00	−0.03
n	3,227	586	2,474	327
Young Adults in High-Segregation Metro Areas
Initial status in family home
% Black	3	3	78	72
% White	93	94	17	23
% Poor	7	6	31	25
Change from family home to adult home
% Black	0.02	0.05	−0.06	−0.35
% White	−0.04	−0.07	0.06	0.36
% Poor	0.02	0.04	−0.01	−0.07
n	495	97	499	57

One possible reason for the discrepancy between the results from the PHDCN and the national results from the PSID is the severe segregation that exists in Chicago, which distinguishes the city from much of the rest of America. To test this explanation, the second set of rows examines change among young adults who originate in high-segregation metropolitan areas.⁹ The results are extremely similar to those found previously among young adults in Chicago. Whites who establish an independent household in a new county live in neighborhoods with greater representation of African Americans and higher poverty rates, while African Americans who leave segregated counties relocate to neighborhoods where the proportion of black residents and the poverty rate are substantially lower. The similarity between these patterns and those reported from the PHDCN suggests that these results may represent a general pattern of change associated with exiting the residential structure of extremely segregated urban areas. More detailed analysis of this group reveals that a large majority of young adults are following patterns of regional mobility that are reflective of large-scale trends of movement toward areas of the country that feature less-segregated metropolitan areas (Farley and Frey 1994; Logan et al. 2004).

What remains unclear is whether the dramatic change occurring during the period of young adulthood is a temporary departure from racial inequality in neighborhood environments, or whether the changes found with the initial move out of the family home extend further into adulthood. This question is pursued through an analysis of trajectories of change occurring beyond the point at which young adults form their own households and move into adulthood. Figure 2 displays trajectories of neighborhood change derived from growth curve models covering an extended period of young adulthood. Coefficients for the growth models are reported in Table 5.

Fig. 2.

Trajectories of change in neighborhood poverty and percentage black from age 17 to 35, by race: PSID young adult sample originating in high-segregation MSAs

Table 5.

Hierarchical linear models of neighborhood poverty and percentage black from age 17 to 35, by race: PSID young adult sample originating in high-segregation MSAs

	Neighborhood Poverty Rate	Neighborhood % Black
Initial Status for Stayers
Intercept (white)	0.065** (0.006)	0.022† (0.012)
Black	0.206** (0.014)	0.740** (0.031)
Initial Status for Movers Outside County (relative to stayers)
Intercept (white)	−0.006 (0.008)	−0.013† (0.007)
Black	−0.028 (0.037)	−0.033 (0.082)
Change for Stayers, Age
Intercept (white)	0.002** (0.001)	0.003** (0.001)
Black	−0.004 (0.002)	−0.020** (0.006)
Quadratic Change for Stayers, Age Squared
Intercept (white)	0.000** (0.000)	0.000* (0.000)
Black	0.000 (0.000)	0.001* (0.000)
Change for Movers to Different County, Age × Leaver
Intercept (white)	0.003 (0.002)	0.003 (0.002)
Black	−0.014* (0.006)	−0.035† (0.021)
Quadratic Change for Movers Out of County, Age Squared × Leaver
Intercept (white)	0.000 (0.000)	0.000 (0.000)
Black	0.001** (0.000)	0.002† (0.001)

Notes: N = 1,148, sample includes white and African American young adults originating in high-segregation metropolitan areas. Two-level hierarchical linear model, with individual survey years nested within individuals. Models include controls for measures of gender, total years of schooling, and time-varying measures of number of adults and children in household, marital status, household income, occupational status, and annual hours worked.

^†p < .10;

^*p < .05;

^**p < .01

A preliminary analysis of raw means for black and white young adults from the age of 17 to 35 shows that the pattern of neighborhood change over this period is curvilinear. Therefore, a quadratic specification is used to describe change occurring over young adulthood. In all figures, separate trend lines are shown for white and black young adults. Trajectories of young adults who move to a different county upon forming their own households are shown with dashed lines, while the lines for those who remain in the same county are solid. All figures display results from specifications that control for a set of covariates that capture key aspects of individuals’ economic status and life-cycle changes occurring during the period of young adulthood.

Similar to the raw differences shown in Table 4, initial analysis of the full sample of young adults reveals minimal difference in the trajectories of those who remain or exit their county of origin (results not shown). This pattern changes when the sample includes only young adults originating in highly segregated metropolitan areas (Fig. 2). As shown in the top panel of the figure, African Americans in this sample who exit their county of origin experience a pronounced drop in neighborhood poverty during early adulthood, and whites who remain or exit their counties of origin experience rising neighborhood poverty over the same period. However, these trends shift as the sample moves further into adulthood. Whereas whites experience slight declines in neighborhood poverty as they age beyond 25, the trend toward declining neighborhood poverty among African Americans who exit their county of origin flattens and reverses as they age further into adulthood. In early adulthood, there is a clear trend toward racial equality among young adults who exit highly segregated metropolitan areas, but the long-term trend suggests a reproduction of inequality in neighborhood poverty as black and white young adults move further into adulthood.

The bottom panel of Fig. 2 displays the same results, using neighborhood percentage black as the dependent variable. Results from the full sample of young adults (not shown in the figure) show that the same trends of change are present among those who remain in their county and those who exit their origin county upon forming an independent household. By contrast, the bottom panel of Fig. 2 shows substantial changes in neighborhood racial composition when black young adults from highly segregated metropolitan areas move to a different county upon forming their own households. Although whites remain in neighborhoods with minimal black presence no matter where they reside, African Americans who move to a new county enter neighborhoods that are much less segregated than those from which they came. Again, however, the longer-term trend is one of “resegregation,” as the percentage of black neighbors gradually rises as African Americans age further into adulthood. By their 30s, black adults who had moved into neighborhoods that were relatively integrated upon forming their own households find themselves back in neighborhoods that are similar to those in which they started—neighborhoods that are mostly black. Whites experience a very modest increase in neighborhood percentage black over the course of young adulthood, but they continue to live in neighborhoods with less than 10% black residents, on average, throughout this period. Thus, although the period of early adulthood shows a leveling of racial inequality in neighborhood economic status among young adults who move to a new county, there is a longer-term trend toward persistent racial inequality.¹⁰

Selected and Unselected Change: The Destination Neighborhoods of Young Adults Exiting Segregation

Although results presented to this point reveal a common pattern of strong continuity in neighborhood conditions from childhood to adulthood, young adults who exit highly segregated cities are the lone subpopulation who depart from this pattern, if only temporarily. As such, this group presents a puzzle: How is it that young adults who select out of extremely segregated areas upon establishing independent households end up returning to segregated neighborhoods further into adulthood? This section describes a final set of analyses examining change in neighborhood conditions during the first independent residential “spell,”¹¹ and change arising from the first residential move after this initial residential spell, among young adults in the PSID who exit highly segregated metropolitan areas upon establishing an independent household.

Figures 3 and 4 are based on the sample of young adults who originate in highly segregated metropolitan areas, but who leave these areas upon forming their own households. Figure 3 shows change in the destination neighborhoods of this sample of young adults over the duration of their spell in the new neighborhood. Although this first residential spell sometimes lasts a decade or longer, the vast majority of first residential spells last six years or less; therefore, change in the destination neighborhoods is plotted for the first six years of the residential spell. The figure shows that the destination neighborhoods of African Americans who exit highly segregated metropolitan areas undergo a process of demographic change during their time in the neighborhood. Whereas the racial composition in the destination neighborhoods of whites changes very little (see the top line and the bottom line in the figure), the average proportion of white residents drops steadily in African Americans’ neighborhoods, while the proportion of black residents rises. Although the average neighborhood remains racially integrated at the end of the spell of residence, there is considerable change in the racial composition of these destination neighborhoods during this time.¹²

Fig. 3.

Changes in neighborhood racial composition during young adults’ first independent residential spell: PSID young adult sample originating in high-segregation MSAs. The sample is limited to young adults who move to different county upon forming independent households

Fig. 4.

Change in neighborhood racial composition arising from young adults’ first residential move after the first independent residential spell: PSID young adult sample originating in high-segregation MSAs. The sample is limited to young adults who move to different county upon forming independent households

Figure 4 complements this analysis by plotting “selected” change in neighborhood racial composition arising from the first residential move after the initial spell of residential independence. Trends of change for African Americans look quite similar to those in Fig. 3, but a close look reveals that they are actually the opposite: residential moves lead African Americans into neighborhoods with a slightly lower percentage black and a higher percentage white than the neighborhoods from which they moved. Moves made by whites do not alter the racial composition of their neighborhood substantially, although they do lead to neighborhoods with a slightly lower percentage black and a higher percentage white.

Together, these two figures suggest that one important explanation for the persistence of racial inequality in neighborhood environments, even among young adults who have selected out of segregated environments, is a phenomenon that might be called “unselected change.” Unselected change refers to change in the neighborhood environment that occurs around individuals or families and that runs counter to the preferences of the individual as inferred by his/her decision to relocate into the neighborhood. In this example, black young adults who exit severely segregated metropolitan areas and select into racially integrated neighborhoods find themselves in neighborhoods that are undergoing a demographic shift, gradually leading toward resegregation. The pattern of unselected change suggests that the reproduction of neighborhood inequality from childhood to adulthood is not simply attributable to the decisions of white and black young adults to live in segregated neighborhoods, but rather to the decisions of those around them to exit or enter such neighborhoods. This idea is reinforced by the analysis of change arising due to the “second” move of African Americans. Like the move out of the family home, African American young adults again move into more integrated environments when they decide to relocate for a second time. In other words, “selected” change appears to lead African Americans into relatively integrated environments, whereas “unselected” change leads to increasing segregation around black young adults.

Discussion

This study identifies changes in geography and in the neighborhood environment as central transitions during young adulthood, with implications for the broader question of how childhood advantages and disadvantages persist or are disrupted as an individual moves into adulthood. More specifically, the study is motivated by research demonstrating strong continuity in individuals’ neighborhood environments from childhood to adulthood that is not explained by individual or family-level characteristics such as income or educational attainment (Sharkey 2008). This research leaves us with an unresolved question: If the reproduction of neighborhood advantage and disadvantage is not explained by human capital or the resources that individuals and families bring with them to the residential housing market, how can one account for the persistence of racial inequality in neighborhood conditions as children enter young adulthood and make their first independent residential decisions?

The pattern of findings that emerges from the analysis provides two pieces of an answer to this question. First, continuity in neighborhood conditions arises from individuals’ attachments to places, their connections to place-based social networks, and their neighborhood preferences. This conclusion is consistent with an extensive strand of research examining the role of individual preferences and attachments to places as factors that help explain racial segregation and the dynamics of neighborhood change (Altman and Low 1992; Elder et al. 1996; Fried 1982; Gerson et al. 1977; Ihlanfeldt and Scafidi 2002). This source of neighborhood continuity is exemplified by the experiences of young adults who remain within Chicago or other highly segregated cities. Regardless of whether young adults in the PHDCN sample remain living at home or leave home but remain within the city, they reproduce the racial inequality of their childhood years into the early years of adulthood. This pattern reinforces findings from previous research highlighting how individual mobility is tightly bounded within the rigid residential structure of highly segregated cities like Chicago (Sampson 2008; Sampson and Sharkey 2008). Although this study stops short of identifying the precise mechanisms that constrain the mobility decisions of individuals, the results support the consistent finding in the literature showing that mobility within the stratified landscape of cities like Chicago tends to reproduce, rather than disrupt, the overarching structure of racial and class segregation that characterizes the city’s residential neighborhoods.

By contrast, moves that cross the boundaries of segregated metropolitan areas such as Chicago lead young adults into neighborhoods that are very different than those from which they came. Even after adjusting for individual and family characteristics, there is a substantial leveling of racial inequality among young adults who exit highly segregated cities, a finding that is present among the PHDCN sample as well as the sample of young adults in the PSID who originate in the nation’s most segregated urban areas. This finding supports the central hypothesis guiding the analysis, which is that change in the racial and economic composition of young adults’ neighborhoods is strongest among those that experience geographic change. However, support for the hypothesis is qualified by the fact that this pattern is present only within the nation’s most segregated metropolitan areas.

The second source of continuity in young adults’ neighborhood conditions arises as a result of change that occurs around individuals who have selected a new environment in which to live. The impact of unselected change is particularly strong for African Americans. In the PSID sample, black young adults who exit segregated metropolitan areas and select into relatively integrated environments find themselves in neighborhoods that are undergoing change gradually leading toward resegregation. This pattern relates closely to a strand of research that considers how the preferences of different groups of individuals interact in dynamic fashion to create aggregate patterns of racial segregation (Bruch and Mare 2006; Charles 2000; Clark 2007; Schelling 1971; Zhang 2004). Central to this research is the idea that to understand neighborhood change, one must move beyond an exclusive focus on individual choices and instead consider systems of interrelated decisions made by individuals responding to the change occurring around them (e.g., see Crowder and South 2008; Quillian 1999). Although this idea has been applied to simulate aggregate patterns of residential segregation across the urban landscape, perhaps the most important implication of the present study is that processes of selected and unselected change are central to understanding neighborhood attainment over the individual life course and across generations within a family. The individual- and family-level consequences of unselected change are largely ignored in the large literature on systems of choice and aggregate levels of segregation.

The idea of unselected change thus emerges as an important piece of the explanation for why neighborhood advantages and disadvantages persist from childhood to adulthood, and fills in gaps left by other prominent approaches to analyzing neighborhood attainment. For instance, the basic tenets of spatial assimilation theory would suggest that continuity in neighborhood conditions might be explained by continuity in familial resources, educational attainment, and so forth. As noted earlier, however, the persistence of neighborhood conditions is only partially explained by these other dimensions of family background. The place stratification perspective might interpret persistent racial inequality in neighborhood conditions as evidence for pervasive barriers to residential mobility faced by African Americans. Yet, this perspective would have difficulty accounting for the fact that neighborhood conditions are reproduced even among African American young adults who select into relatively integrated neighborhood environments. Clearly, one must consider more than the resources that young adults bring with them to the residential housing market and the barriers they face within that market in order to explain continuity in neighborhood conditions over the life course. At a more general level, any theory of neighborhood attainment that focuses exclusively on the neighborhoods that individuals select is incomplete because it ignores the dynamic nature of neighborhoods and the potential for individuals’ selected neighborhoods to transform around them because of the choices of others.

The results described here suggest that the persistence of neighborhood racial inequality from childhood to adulthood is driven, in part, by young adults’ attachments to cities with high levels of segregation. However, the transmission of neighborhood advantage and disadvantage is driven also by processes that are separate from individuals’ own resources, preferences, or attachments that they bring to the residential market. Neighborhood inequality is transmitted from childhood to adulthood not only because of the choices that individuals make but also because of the choices that others make around them. This finding helps explain how it is that African Americans who select out of racially segregated environments find themselves back in increasingly segregated neighborhoods as they move further into adulthood. Although analysis of individual choice is central to understanding the reproduction of neighborhood inequality, analysis of unselected change provides a more complete picture of how the process of reproduction continues in spite of the choices that individuals make.