NCJ Number
189942
Journal
Criminology Volume: 39 Issue: 3 Dated: August 2001 Pages: 561-590
Date Published
2001
Length
30 pages
Annotation
Using county-level data for the decennial years in the 1960-90 time period, this study re-examined the impact of conventional covariates on homicide rates and explicitly modeled spatial effects.
Abstract
Using the Land et al. specification as the baseline model, the study reassessed the robustness of the structural covariates of homicide with rigorous controls for spatial processes. In doing so, the study estimated models of spatial patterning that were consistent with potential diffusion processes. This article begins by explaining important concepts and theoretical processes relevant to structural invariance. Then, after describing data sources and methods, the article presents the results of an Exploratory Spatial Data Analysis (ESDA). ESDA is a critical first step in visualizing patterns in the data, identifying spatial clusters and spatial outliers, and diagnosing possible misspecification in analytic models. The results of the ESDA informed the multivariate analyses, wherein the researchers assessed the effects of structural variables and formally modeled spatial processes. The study found that homicide was strongly clustered in space, and this clustering could not be completely explained by common measures of the structural similarity of neighboring counties. Noteworthy regional differences were observed in the effects of structural covariates on homicide rates. Evidence was consistent with a diffusion process for homicide observed in the South throughout the 1960-90 period. The application of recently developed techniques of spatial econometrics thus offers promising opportunities for extending the understanding of the social forces that contribute to interpersonal violence. 4 tables, 4 figures, and 49 references