|Presenters' slides and posters - International Crime and Intelligence Analysis Conference, 25-26 February, Manchester (UK)|
WHAT WORKS CLASSES
27 September 2016
7-10th November 2016
4th October 2016
24th May 2016
6th December 2016
17th May 2016
7th July 2016
12th July 2016
15th November 2016
5th-16th September 2016
ICIAC 2012 Seminar Stream 3A
Abstracts and slides
Spatial and temporal analytical techniques (I)
Mapping and correlating crime/ASB and Private Sector Housing Using the Gi* Statistic
Sophie Curtis-Ham, London Borough of Newham
Key words: Gi* statistic, geographic correlation, hotspots, crime/ASB, Private Sector Housing
Gi* (Getis-Ord statistic) analysis was conducted to inform decision-making in relation to whether and where to apply a scheme for licensing Private Rental Sector (PRS) landlords in the London Borough of Newham. A Local Authority can implement landlord licensing where PRS housing is found to be contributing to crime and disorder, in order to better control the behaviour of tenants by regulating landlords. An evidence package was compiled in support of borough-wide PRS licensing, including specific evidence of PRS housing causing crime/ASB and evidence of a correlation between the locations of PRS housing and hotspots for relevant types of crime and ASB. This paper presents the latter analysis, which determined where in Newham PRS and relevant crime/ASB co-occur, i.e. where there are overlapping concentrations of both PRS and crime/ASB, and whether these concentrations were statistically correlated.
Data included ASB records from the Council’s UNI-form database received 1/4/11-31/3/12 relating to:
- Rowdy/Inconsiderate Behaviour, Neighbour Nuisance and Verbal Abuse / Harassment /Intimidation
- Rubbish in Front Gardens
- Noise complaints
Police CRIS ‘geocoded’ crime records with ‘minor code’ Burglary-Dwelling and Police Computer Aided Dispatch (CAD) records (999 and other calls to the police) for Home Office defined ASB codes plus code 11 Drug Offences received 1/4/11-31/3/12 were also analysed. PRS housing locations were provided as a map point layer by the Council’s Geographical Information Systems Officer.
The Getis-Ord Gi* statistic was used to calculate the concentrations for each dataset, based on counts of incidents within grid cells. Cells with statistically significant Gi* z-scores at the generally accepted probability level of p<.05 (i.e. high concentrations or ‘hotspots’) were displayed on maps colour coded. Overlaying the hotspot maps for the PRS (blue) and other datasets (pink) then revealed the locations where there were significantly higher than average levels both PRS and ASB (purple). To determine whether there was a significant correlation between these concentrations of PRS and crime/ASB, correlation analysis was conducted comparing both the number of PRS properties per cell and its PRS Gi* z-score with the number of crime/ASB records per cell and its GI* z-score.
PRS was statistically significantly correlated with all datasets, meaning the more PRS in a location, the more crime/ASB (of any given type) there is and conversely that locations with fewer PRS properties also tend to see less of the crime/ASB types considered. Each crime/ASB type tended to show small hotspots scattered throughout the borough, varying by crime/ASB type. But cumulatively, overlaying all the crime/ASB types showed a much wider spread overlap between crime/ASB overall and PRS. This provided support for implementing PRS licensing on a borough-wide scale. It was emphasised in the resulting report that correlation does not imply causation – where PRS and crime/ASB were correlated, this does not mean that privately rented accommodation is a cause of ASB, only that they tend to occur in the same locations. Based on this analysis plus other, causal evidence, borough-wide PRS licensing was subsequently implemented.
Estimating offence times from victim reports
Matt Ashby and Kate Bowers, Department of Security and Crime Science, University College London
Crime analysts often attempt to draw conclusions about the temporal distribution of crime from reports of previous offences. This task is complicated in the case of many crimes against unattended property because victims of crime cannot tell police when the offence occurred.
This study used CCTV records to compare several previously-proposed methods for estimating offence times from victims’ reporting of when they left their property unattended and when they discovered the crime had occurred. These methods are all in widespread use by crime analysts in the UK and elsewhere, but they have not previously been tested because of a lack of suitable data.
This study found that two commonly-used methods, which are based on unsupportable assumptions about offender behaviour, to be inaccurate and misleading. It also found another method (aoristic analysis) to be accurate in the cases studied. This led to two conclusions: that the inaccurate methods could lead analysts astray (and so should not be used) and that aoristic analysis appears to be a suitable solution to the problem of temporal inaccuracy in this case.
The results of this study are being used by those training crime analysts and by analysts in British Transport Police to promote the use of aoristic analysis and to warn practitioners about the potential inaccuracies of some other commonly-used methods.
Page last modified on 30 jan 13 17:01