Imputing Structural Characteristics in Segments: Is the Pain Worth the Gain?

• In Event: Innovative Methods for Exploring the Spatial Distribution of Crime in Small Geographic Units

Abstract

A recent push in the criminology of place literature is to study crime at very small units, such as street segments (both sides of a street between two intersections). Although studies have successfully identified the crime patterns at street segments, a remaining challenge is studying the relationships between structural characteristics (i.e., poverty, residential instability, racial heterogeneity, etc.) of street segments and crime. Given the difficulty of obtaining such data aggregated to segments, the current study proposes using an imputation approach to apportion U.S. Census data in blocks to segments. Two imputation techniques are described that vary in their level of sophistication. A unique dataset is constructed in which block-level structural characteristics are properly apportioned to street segments utilizing the data of neighborhoods in the cities of Anaheim, Santa Ana, and Huntington Beach in Orange County, California. The results of three models are compared: two different imputation techniques for segments and a model using data aggregated to census blocks.

Author

• Young-An Kim, University of California, Irvine