Using National Survey Data, Administrative Data, and GIS to Determine Community Boundaries Across Cities, Suburbs, and Rural Areas

• In Event: 1-010 - Neighborhood Geocoding

Abstract

The influence of neighborhoods on the lives of children and families has long been the subject of research. However, there have been few empirical tests of the most appropriate scale for aggregating data to create valid community measures. The geographic landscape of the U.S. is complex, with people living in communities spanning densely populated large cities to resource-diffuse rural areas. The geographic scope of key community measures may differ across measure and across urbanicity. For example, rural families often travel greater distances than urban families to access resources. Additionally, some characteristics of communities (e.g., air pollution) may be most influential for child development when assessed on a small, proximal scale, while others (e.g., educational resources), may be more meaningful when aggregated at a larger scale. This research aims to validate measures of community characteristics within and across urbanicity.

This study uses survey data from the ECLS-K and a variety of administrative data sources including the Census Bureau, FBI, and EPA. The ECLS-K contains a wealth of information on the lives of children and families, which were used to validate the scope of several typically studied community characteristics, including educational resources, cultural resources, medical resources, human capital, parks, pollution, violent crime, and concentrated disadvantage. We first created measures of community characteristics at different radii (e.g. 1-mile, 5-miles, 10-miles) from children’s home zip code/census tract using Geographic Information Systems (GIS) software. Next, regression models were estimated predicting child or family measures with the community characteristics measured at different radii to assess construct validity. Within each urbanicity, we identify the optimal radii for aggregating measures of community characteristics (e.g., which radius of cultural resources is most strongly associated with parental reports of taking their child to a museum?; which radius of pollution is most strongly associated with childhood asthma?). Initial results indicate that the appropriate radii vary depending on neighborhood characteristic and urbanicity. For instance, concentrated disadvantage most strongly correlated with children’s achievement and behavior when aggregated at 1-mile from children’s census tracts consistently across all types of neighborhoods, while the number of education resources best predicted families’ use of such services at a 5-mile radius in cities and a 15-mile radius in rural areas. Results inform efforts to delineate key community factors affecting children and families.

Authors

• Portia Miller, University of Pittsburgh
  Presenting Author

• Elizabeth Votruba-Drzal, University of Pittsburgh

• Rebekah Levine Coley, Boston College

• Bryn Spielvogel, Boston College