SPATIAL STATISTICS AND CRIME

A meeting organised by the General Applications Section of the
                  Royal Statistical Society.

           Tuesday 13th May, 5.00pm (Tea 4.30pm)

     The Royal Statistical Society, Errol Street, London
   (Directions: http://www.rss.org.uk/about/direction.html)

                 Free meeting, all welcome

"A Statistical Look at Crime Pattern Analysis"
CHRIS BRUNSDON (University of Newcastle)

Over recent years there has been a growing interest in the statistical
analysis of crime patterns. This is in part due to advances in the
collection and collation of crime data,  and in part due to advances in
the statistical techniques used to analyse this data. Although the subject
is not exclusively geographical, a great deal of emphasis has been placed
on the spatial distribution of crime patterns and how these may evolve
over time. In this talks, we will give some insights into the kinds of
methods that may be used, together with some practical examples.

Discussion will focus on a number of approaches: techniques  for
identifying 'hot spots' of crime using kernel density estimation will be
considered; and methods for identifying space-time clustering of crime,
such as the Knox test. In addition to this,  modelling of  daily, weekly,
and yearly levels of crime will be presented. Finally, the future of crime
pattern analysis will be considered.

One important aspect of crime pattern analysis is that it is a visual as
well as a statistical tool - and that much of the output can be readily
interpreted by non-statisticians with responsibility for police resource
management and crime detection.  For this reason, emphasis will also be
placed on graphical output from many of the techniques.


"The Use of Crime Data and Crime Pattern Analysis in Policy Evaluation"
SHANE JOHNSON* (University of Liverpool)

This aim of this presentation will be to explore how the analysis of
aggregate and disaggregate crime data can be used to attribute outcomes
(e.g. burglary reduction) to specific policy interventions (e.g. crime
prevention activities).

Particular attention will be paid to:

·  the impact of the timing and intensity of policy inputs, (e.g.
expenditure), outputs (e.g. publicity brochures, bolts, window locks,
etc) on crime rates using multi level models;
·  the longitudinal tacking of protected properties using
disaggregate data within a GIS
·  the measurement of crime displacement  and the development of a
new method for its quantification - the weighted displacement quotient;
·  the identification of diffusion of benefit (i.e. where significant
reductions in crime net of local trends occurs in areas not directly
targeted by crime prevention measures by in close proximity to them);
· the measurement of crime switch (e.g. from burglary to theft from
vehicles);
·	the identification of the number of burglaries prevented, value of
burglaries saved and  associated cost benefit ratios.

The development and use of these techniques will be discussed in relation
to a an  evaluation of the Reducing Burglary Initiative (RBI) recently
completed for the Home Office and an analysis of the impact of installing
gates to secure alleys at the rear of terraced properties (the Merseyside
'alleygating' initiative).


(* note that this is a change of speaker to that advertised on the RSS
Meetings card)