|
Abstract
|
Purpose - This paper aims to quantify characteristics of areas that experienced rioting across the UK in August 2011. Through exploring the areas where riots occurred, and those that did not have any problems, variables and factors that may have contributed to a geographical area experiencing rioting can begin to be identified. Through doing so, it is hoped that local authorities can be better prepared to deal with potential similar situations in the future.
Design/methodology/approach - National indicator data were collated for all local authority areas across England. After a literature review, and preliminary analysis, a set of indicators that the authors thought could be potential factors were identified. Using these indicators as input variables logistic regression was completed, with the target variable defined as whether or not the local authority had experienced any rioting (target variable was binary categorical - Yes/No).
Findings - A logistic regression equation was produced that gave a risk score to each local authority area. Using an ROC Curve an optimum cut off point was created. Anything over this cut off point was deemed to be vulnerable to rioting. The overall accuracy of the model was 88.4 per cent. The positive predictive value was 97.2 per cent and the negative predictive value was 42.9 per cent. Predictor variables included in the model were: existing acquisitive crime rates; unemployment deprivation; and education/level 4 attainment.
Research limitations/implications - Due to the short turnaround time of producing this insight, only limited data are available to build a model on. The paper focuses on characteristics of the geographical areas where rioting occurred, rather than the traits of the culprits themselves. Since completing the paper, more information has become available.
Practical implications - With key predictor variables identified, it is possible to look at areas that are a potential risk. The false positives and false negatives (local authority areas that did not behave as the model suggested) also pose interesting questions. Why did rioting not occur in certain areas that showed the same characteristics as those that experienced rioting?
Social implications - Where a riot risk is noted as high, local public services may consider any actions that could be taken to tackle the risk and can use this information to justify continued resources in that area.
Originality/value - Many discussions were had in the months that followed the August riots. These discussions tended to focus on the perpetrators and not the areas that are analysed in this paper. |