Beyond Blocks and Tracts: Evaluating a New Method of Neighborhood Aggregation

• In Event: Rethinking Place and Spatial Dynamics

Abstract

While neighborhood effects scholars typically aggregate to census units for analyses, a growing body of evidence suggests that more attention should be paid to what residents consider to be a part of their neighborhood. Using data from the Los Angeles Family and Neighborhood Survey, US Census, Reference USA, and TIGER/Line GIS Shapefiles, this paper develops a new method of aggregation based on resident perceptions of the spatial distribution of their neighborhood. Data are then aggregated to this neighborhood unit, census blocks, and a static buffer approach to allow for model comparison. Results indicate that accounting for resident perceptions when aggregating data to the neighborhood level has the ability to greatly increase the explanatory power of these models. In examining social cohesion, simply allowing boundaries to fluctuate based on resident perceptions led to roughly 2X more explanatory power than those using census blocks. Additionally, variation was also found among key correlates between models. These results suggest that prior research may in fact suffer from misspecification issues, further highlighting the importance of proper selection of areal units to neighborhood effects research.

Author

• Ben Forthun, University of Northern Colorado