Neighborhood attributes as determinants of children's outcomes - How robust are the relationships?

Estimates of neighborhood effects on children's outcomes vary widely among the studies that seek to identify their existence and magnitude, reflecting substantial variation in data and model specification. Here, we re view that literature, and ask if the disparity in estimates of neighborhood effects may reflect the differences among studies in the specification of family characteristics, and hence omitted variables bias. We report a systematic set of robustness results for three youth outcomes thigh school graduation, the number of years of completed schooling, and teen nonmarital childbearing) using data on about 2,600 children from the Panel Study of Income Dynamics. We observe these children over a period of at least 21 years and have included an extensive set of neighborhood variables for these individuals measured over the entire school-age period. We measure the relationship of these neighborhood variables to the three outcomes, moving from basic models containing no individual and family characteristic variables to models containing an extensive set of individual and family statistical controls. We conclude that the reliability of estimates of these impacts may be an artifact of the degree to which family background is characterized in model specification. Confidence that reported neighborhood effects reveal true relationships requires statistical controls for the full range of family and individual background that may also influence children's attainments; not all variables with coefficients showing asterisks have significant effects.