Converging Evidence for Neighborhood Effects on Children’s Test Scores: An Experimental, Quasi-Experimental, and Observational Comparison

Rising income inequality has been found to be associated with rising segregation at the neighborhood level, generating concern about whether neighborhood environments themselves may influence children’s life chances, independent of other individual child and family characteristics. Because poor and minority Americans are overrepresented in our most disadvantaged neighborhoods, any “neighborhood effects” on children may contribute to persistent disparities in overall schooling outcomes across race and class lines in the United States.

A large body of nonexperimental research dating back to the Coleman Report in 1966 has produced evidence consistent with the idea of large neighborhood effects on children’s schooling outcomes. However, drawing causal inferences from these studies is complicated by the fact that the attributes of a neighborhood in which a family lives is likely correlated with characteristics of the family that predict schooling outcomes. These studies are therefore vulnerable to selection bias. The one formal randomized experiment in this literature is the five-city Moving to Opportunity (MTO) experiment, data from which suggests no statistically significant impacts, on average, on reading or math test scores for children in MTO measured four to seven years after baseline. How one should weight the findings from the MTO experiment versus the larger body of nonexperimental research remains the topic of ongoing debate within the research and policy communities.

In this chapter, we try to reconcile the experimental, quasi-experimental, and observational research literature regarding neighborhood effects on children, and we argue that the available findings are more convergent than many people believe. Drawing on a number of recent and unusually high-quality quasi-experimental and observational studies, together with a reexamination of MTO findings across the individual MTO demonstration sites, we believe that the available - 3 - evidence allows us to reject the null hypothesis that neighborhood environments never matter for children’s outcomes. Yet at the same time, the data also do not support the hypothesis that neighborhoods always matter.

In our view, the key question for research and public policy is to learn more about the conditions under which neighborhoods matter for children’s academic outcomes and why. Our ability to answer this question in the present chapter is restricted by the limited number of studies that have employed sufficiently strong research designs to support inferences about neighborhood effects on children’s outcomes, and by the fact that a disproportionate share of the studies that meet this research-design threshold have been carried out in a single city (Chicago).

With these important qualifications in mind, we believe that there is at least a suggestive case to be made that children’s test scores may be most strongly affected by community violence or may respond nonlinearly to concentrated neighborhood disadvantage or community violence. Put differently, what may matter most for children’s cognitive development is to avoid living in the most severely economically distressed or dangerous neighborhoods in the country, neighborhoods that are found in cities like Baltimore and Chicago but, surprisingly, are less prevalent even in other major urban areas such as Boston, Los Angeles, and New York. Given the limitations of the available evidence, we offer these as hypotheses to be tested further rather than as strong conclusions.