George Galster (Wayne State University) and Anna Santiago (Michigan State University)
Abstract found at: http://usj.sagepub.com/content/early/2015/07/28/0042098015598067.fullInvestigating the adverse consequences of residential segregation for ethnic minority people has a long and distinguished social scientific history in the U.S. Rigorous studies have identified substantial socioeconomic and health costs that segregation imposes on Latinos and African Americans, though often with contingencies based on ethnicity and gender.
What these quantitative studies do not reveal, however, are the complex underlying pathways through which segregation exacts its toll. Do the primary mechanism(s) operate through confining minorities in inferior neighborhoods that constrain opportunities because they offer inadequate public services, more intense pollution, poorer access to employment, higher rates of violence and/or concentrated poverty and joblessness? Or, does segregation operate independently of these ethnically correlated aspects of geographic context and instead work through ethnic composition of the neighborhood per se? Might concentrations of ethnic minorities impose costs (or benefits) regardless of the particulars of the places where they are concentrated, due to socialization processes within minority communities and/or stereotyping and stigmatizing processes by majority populations external to them? Stated differently, the issue needing empirical investigation is the degree to which ethnic residential segregation affects the life chances of minorities irrespective of the particular features of the places where they are concentrated beyond disproportionate ethnic clustering.
In this paper we focus on a variety of outcomes for individual children, youth and young adults (“children” hereafter) from the groups commonly thought to suffer the greatest harms from residential segregation in the U.S.: low-income Latino and African American families. Our original approach bridges two traditionally distinct literatures: “consequences of segregation” and “neighborhood effects.” The forces of segregation operating at the regional scale inevitably generate one direct outcome at the neighborhood scale: increasing exposures of minority children to neighbors from other minority groups. They generate another outcome indirectly following from the concentration of minorities: shaping the geographic flows of resources (financial, institutional, human, environmental) and thereby the opportunities and quality of life available to minority children in those neighborhoods. Potentially both direct and indirect effects of segregation affect child developmental outcomes by altering their neighborhood contexts. Distinguishing these two effects constitutes the core empirical test of our paper.
Specifically, our research questions for low-income Latino and African American children are:
To what extent are the odds of ever experiencing a variety of childhood outcomes in the domains of health, exposure to and use of violence, education and fertility related to the percentages of Latino and African American residents in their census tracts? [i.e., the direct effects of segregation]
If so, to what degree do these relationships persist when controlling for neighborhood indicators of socioeconomic status, safety, housing stock, institutional resources and environmental conditions ?[ i.e., the indirect effects of segregation]We answer these questions through a stratified, logistic regression analysis of data on minority children that we collected through a retrospective survey of Latino and African American families whose children resided in Denver (CO) Housing Authority (DHA) dwellings during at least part of their childhood. This program represents an unusual natural experiment holding great potential for overcoming geographic selection bias because assignment of households on the DHA waiting list to public housing mimics a random process, thereby permitting causal inferences about neighborhood indicators and child outcomes. Because DHA dwellings are located in a wide variety of neighborhoods throughout Denver County, it also offers an unusual opportunity of observing how low-income minority children fare in places that differ on a host of other neighborhood characteristics beside ethnic composition that we measure comprehensively.
Regarding our first research question, results show that low-income, Latino and (especially) African American children growing up amid greater concentrations of African American neighbors in Denver pay penalties across several outcome domains, controlling for their own, households’ and caregivers’ characteristics. By contrast, growing up among greater concentrations of Latino neighbors is associated with both positive and negative outcomes for African American and (especially) Latino children.
Specifically, low-income African American children living in a neighborhood with a standard deviation-higher percentage of African American residents are predicted to have greater odds of: (1) an asthma diagnosis by 52 percent; (2) witnessing violence in the neighborhood by 62 percent; (3) being victimized in the neighborhood by 78 percent; (4) using violence themselves by 39 percent; and (5) having a child outside of marriage by 51 percent. These associations with harmful outcomes are more limited for low-income Latino children. Latino children living in a neighborhood with a standard deviation-higher percentage of African American residents are predicted to have greater odds of: (1) being victims of neighborhood crime by 82 percent; and (2) repeating a grade in school by three percent.
By contrast, concentrations of Latino neighbors are associated with positive and negative outcomes for children of both ethnicities. Latino children living in a neighborhood with a standard deviation-higher percentage of Latino residents are predicted to have 56 percent greater odds of witnessing violence, but lower odds of: (1) using violence by 35 percent; (2) dropping out before a diploma by two percent; and (3) having children outside of marriage by 29 percent. Of interest, the positive dropping out result also applied to African American children in our sample. However, if they were to live amid a standard deviation-higher percentage of Latino residents they are predicted to have 51 percent greater odds of being victimized in the neighborhood.
Our second research question asks the extent to which the aforementioned associations persist when a host of non-ethnic neighborhood indicators are added to the models. Results reveal that all the aforementioned associations between African American neighborhood composition and adverse outcomes for African American children disappear when other aspects of residential context are controlled. The same is true for Latino children, with one exception (odds of repeating a grade). Similarly, the few aforementioned associations between African American neighborhood composition and adverse outcomes for African American children disappear when other aspects of residential context are controlled. Indeed, with the one exception of repeating a grade, all five statistically significant parameters for neighborhood ethnic composition indicate more favorable minority child outcomes. Controlling for non-ethnic aspects of residential context, low-income African American children living in a neighborhood with a standard deviation-higher percentage of African American residents are predicted to have 84 percent lower odds of an asthma diagnosis. Were they to live in a neighborhood with a standard deviation-higher percentage of Latino residents, they are predicted to have lower odds of: (1) witnessing violence by 78 percent; (2) using violence by 71 percent; and (3) having a nonmarital birth by 89 percent. Ethnic effects on low-income Latino children appear more limited. When they live in a neighborhood with a standard deviation-higher percentage of Latino residents they are predicted to have eight percent lower odds of dropping out of school.
So if ethnic composition per se is not responsible for the negative associations between segregation and negative outcomes, what is? Two aspects of the residential environment stand out as consistent predictors of negative outcomes: higher property crime rates and lower occupational prestige among resident workers. For both Latino and African American children, higher property crime rates are associated with substantially greater odds of: asthma diagnosis, witnessing and being victimized by violence, and repeating a grade; for Latinos the same associations apply to engaging in violence, dropping out, and having children outside of marriage. For African American children, lower occupational prestige is associated with substantially greater odds of: asthma diagnosis, witnessing violence, dropping out and having children outside of marriage. For Latino children it is associated with substantially greater odds of: being victimized by and engaging in violence, dropping out and having children outside of marriage.
Several other aspects of neighborhood context prove predictive for certain outcomes and ethnicities. Places with negative peer influences as assessed by caregivers are associated with greater odds that African American children will witness, be victimized by, and engage in violence. Places with more social problems (an index focusing in violent and criminal behaviors) as assessed by caregivers are associated with greater odds that Latino children will have these same exposure to violence outcomes, plus have children outside of marriage.
In conclusion, growing up amid concentrations of African American residents was associated with a variety of adverse outcomes for low-income Latino and (especially) African American children, though outcomes associated with concentrations of Latino residents were more mixed. Virtually all of the negative associations disappeared, however, when other aspects of the residential context were controlled, and several positive ones persisted. The adverse developmental consequences of ethnic segregation appeared to be generated primarily by concentrating minority children in neighborhoods with higher rates of property crime and lower occupational prestige.
We see our study as bridging two substantial scholarly literatures: consequences of ethnic segregation and neighborhood effects. We have linked them by positing that the forces of segregation operating at the metropolitan scale directly shape children’s intra-neighborhood exposure to various ethnic groups. Moreover, because ethnic composition of neighborhoods affects the flows of various resources into them, segregation indirectly affects children’s exposure to non-ethnic aspects of their local context. Our empirical findings indicate that it is the indirect effects of segregation that create by far the more pernicious neighborhood effects for low-income Latino and African American children.
Our results clearly suggest that policymakers should be cognizant of neighborhood ethnic segregation as an important metropolitan force that continues to shape the developmental context of low-income minority children. It is not neighborhood ethnic composition per se, but rather the correlated aspects of safety and social status that appear crucial. Our finding is thus fully consistent with the position that desegregation efforts should not be motivated by an implicit valorization of “whiteness” or stigmatization of “non-whiteness.”
The daunting policy challenge is providing neighborhood environments that are developmentally friendly to low-income, minority children. This can be approached in principle by either by improving the neighborhoods where they live through community development strategies and/or by improving their access to developmentally superior neighborhoods through well-designed assisted housing programs backed by vigorous fair housing enforcement. We acknowledge the political difficulties involved in doing so in practice, but urge continued efforts in these regards nevertheless.
 We acknowledge that this is an ambiguous finding insofar as it may indicate a lower probability of caregivers seeking medical diagnosis given a set of symptoms, instead of a lower probability of symptoms.