|
Activity Number:
|
66
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 6, 2006 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #305899 |
|
Title:
|
Treatment of Spatial Autocorrelation in Geocoded Crime Data
|
|
Author(s):
|
Krista Collins*+ and Colin Babyak
|
|
Companies:
|
Statistics Canada and Statistics Canada
|
|
Address:
|
16G R.H. Coats Building, Ottawa, ON, K1A0T6, Canada
|
|
Keywords:
|
spatial autocorrelation ; spatial lag model ; crime data
|
|
Abstract:
|
In examining the spatial distribution of crime in city neighborhoods, the measurements at one location may be influenced by the measurements at neighboring or nearby locations. This can result in spatial autocorrelation, which violates the assumption of independence among observations. We examine the basic theory and methods used to analyze spatial lattice data (data aggregated to regions as opposed to discrete points). Our focus is on spatial autoregressive models for lattice data and how to account for the spatial structure of the data. In a study funded by the National Crime Prevention Center at Public Safety Canada, we implement the spatial lag model in a study of crime rates in the neighborhoods on the Island of Montreal, Canada. We examine the differences in the models obtained with and without modeling the spatial effects.
|
