NCJ Number
180975
Journal
Journal of Quantitative Criminology Volume: 15 Issue: 4 Dated: December 1999 Pages: 407-422
Date Published
December 1999
Length
16 pages
Annotation
This paper explores the extent to which county murder and nonnegligent manslaughter rates in the mid-south are spatially autocorrelated.
Abstract
The data, 3-year averages of UCR murder and nonnegligent manslaughter rates from 393 mid-south counties, showed a statistically significant amount of spatial autocorrelation in the murder and nonnegligent manslaughter rates. There was no global pattern in the region, but there were some patterns of localized spatial autocorrelation. Emerging homicide research on other regions and for other time periods suggests that global spatial autocorrelation may create problems for regression analysis in other spatial and temporal contexts. The lack of spatial effects in the regression models does not negate the importance of detecting and correcting spatial autocorrelation in all statistical analysis. Spatial autocorrelation can create similar problems in descriptive and bivariate analysis. The article urges authorities who do county-level analysis to explore spatial autocorrelation with a variety of tests. Notes, figures, tables, references