Near-repeat mapping
Dr Andy Evans
Repeat crimes
~31% to 76% criminals return to commit further crimes
Some rapidly return to the same victim (7-15% burglaries)
Some rapidly return to the same area (10-40% burglaries)
Some return to same area but over a longer time (hotspots)
Some crimes are 'slippery' - criminals repeat crimes slowly moving along streets
Townsley et al. (2003)
Near-repeats
Crimes clustered in space and time
Criminals return to an area over a few weeks until risk rises or reward drops
Might be thought of as an "optimal forager" strategy
If we can predict their return pattern we can prevent the crimes
Johnson & Bowers (2004); Bernasco (2008)
Analysis
Utilise Knox test
Work out distance between all event pairs in space and time
Put event pairs in a table where columns are time bins and rows space bins
| 1 week | 2 weeks | 3 weeks | 4 weeks | 5 weeks | |
|---|---|---|---|---|---|
| 100m | 40 | 30 | 10 | 0 | 1 |
| 200m | 35 | 20 | 5 | 0 | 2 |
| 300m | 10 | 10 | 5 | 5 | 0 |
| 400m | 10 | 5 | 0 | 2 | 0 |
| 500m | 0 | 5 | 2 | 0 | 1 |
Johnson et al. (2007)
Prediction : Trafford method
Simple 400m buffers around crimes
Colour the buffers by weeks-elapsed for three weeks
Separate graph plots of crime densities for each area by time of day
Fielding & Jones (2012)
Prediction : Trafford method
Super-cocooning: target hardening and awareness raising in concentric forms around crimes
Trafford burglary down by up to 26%
Application
Determine critical scales
Prospective or Trafford methodology
Picks up repeat victimisation as well, but not general hotspots without time clustering
These are picked up by long term risk surfaces
Works best in high risk target- and environment-homogeneous areas