NCJ Number
78709
Date Published
1981
Length
222 pages
Annotation
This executive summary discusses the methodology and findings of research designed to develop procedures for analyzing selected spatial and temporal characteristics of crime data.
Abstract
One of the major problems investigated was implications of aggregation in data tables for evaluating criminal justice statistics. Another problem analyzed was assessing homogeneity in cross-classified proximity data. Given an arbitrary proximity matrix that is cross-classified according to two dimensions, a nonparametric strategy generalizing Friedman's analysis-of-variance method is suggested for testing the saliences of the dimensions. The implications of unidimensional seriation for evaluating criminal justice data are also considered. The problem of validating a given unidimensional scale (ordering of a set of objects along a single dimension) is discussed in terms of a few simple properties of the data used to obtain the scale. In the examination of proximity matrix reorganization and hierarchical clustering, connections between hierarchical clustering and the seriation of objects along a continuum that depends on the patterning of entries in a proximity matrix are discussed. Several generalizations of the usual spatial autocorrelation indices are developed, based on the notion of matrix comparisons. Also introduced is an inference model for the percentage voting agreement measure which considered the composition of the parent group in evaluating the cohesion of a subgroup. Some comments on non-Euclidian mental maps are included, and possible future research is discussed.