October 2020
Volume 62, No. 4 | Go to abstracts
Articles
Page 1
Are Conditional Sentence Orders Used Differently for Indigenous Offenders? A Comparison of Sentences and Outcomes in Canada
Leticia Gutierrez, Nick Chadwick
Page 30
Spatial Patterns of Immigration and Property Crime in Vancouver: A Spatial Point Pattern Test
Olivia K. Ha, Martin A. Andresen
Page 52
Designing an Explainable Predictive Policing Model to Forecast Police Workforce Distribution in Cities
Mark Parent, Aurélien Roy, Claudele Gagnon, Noémie Lemaire, Nadine Deslauriers-Varin, Tiago H. Falk, Sébastien Tremblay
Page 77
Citizen Characteristics, Neighbourhood Conditions, and Prior Contacts with the Police: A Comparative Study of Public Satisfaction with the Police
Mengyan Dai, Xiaochen Hu, Feng Gu
Abstracts
Are Conditional Sentence Orders Used Differently for Indigenous Offenders? A Comparison of Sentences and Outcomes in Canada
Leticia Gutierrez, Nick Chadwick
Conditional sentence orders (CSOs) were introduced in Canada in 1996, largely as a mechanism to address the overreliance on incarceration. This sentencing option is particularly relevant for Indigenous individuals who have been vastly overrepresented in custodial settings. Despite being in place for over twenty years, little is known about the use and effectiveness of CSOs. The current study examined the use and outcomes of CSOs for Indigenous (n = 749) and non-Indigenous offenders (n = 1,625) in one Canadian province. Specifically, the length of CSO, frequency and type of optional conditions, number of breaches, and rates of reoffending were compared between the two groups. Results from a logistic regression, controlling for risk relevant co-variates, indicated that Indigenous individuals tended to receive shorter CSOs compared to Caucasian individuals. However, Indigenous individuals were 35% more likely than Caucasian individuals to be convicted of a breach and more likely to incur multiple breaches while on a CSO. Despite the differences in the rates of breaches, the likelihood of reoffending over a two-year period was equivalent across the two groups. Although the reasons for breach were not available for the current study, future research should investigate this further to determine whether increased breach rates for Indigenous individuals are the result of more rule-violating behaviour, an inequity in the fairness of the conditions applied to Indigenous versus Caucasian individuals, or whether it is possible that breaches are over-detected and convicted at a higher rate for Indigenous individuals. Greater awareness of the underlying mechanisms related to increased breach rates affords the opportunity to ensure that CSOs are consistently implemented, something which will contribute to achieving the objective of successful diversion from imprisonment.
_____________________________________________
Spatial Patterns of Immigration and Property Crime in Vancouver: A Spatial Point Pattern Test
Olivia K. Ha, Martin A. Andresen
_____________________________________________
Designing an Explainable Predictive Policing Model to Forecast Police Workforce Distribution in Cities
Mark Parent, Aurélien Roy, Claudele Gagnon, Noémie Lemaire, Nadine Deslauriers-Varin, Tiago H. Falk, Sébastien Tremblay
Despite extensive research, measurable benefits of predictive policing are scarce. We argue that powerful models might not always help the work of officers. Furthermore, developed models are often unexplainable, leading to trust issues between police intuition and machine-made prediction. We use a joint approach, mixing criminology and data science knowledge, to design an explainable predictive policing model. The proposed model (a set of explainable decision trees) can predict police resource requirement across the city and explain this prediction based on human-understandable cues (i.e., past event information, weather, and socio-demographic information). The explainable decision tree is then compared to a non-explainable model (i.e., a neural network) to compare performance. Analyzing the decision tree behaviour revealed multiple relations with established criminology knowledge. Weather and recent event distribution were found to be the most useful predictors of police workforce resource. Despite wide research showing relationships between socio-demographic information and police activity, socio-demographic information did not contribute much to the model’s performance. Though there is a lack of research on measurable effects of predictive policing applications, we argue that combining human instinct with machine prediction reduces risks of human knowledge loss, machine bias, and lack of confidence in the system.
_____________________________________________
Citizen Characteristics, Neighbourhood Conditions, and Prior Contacts with the Police: A Comparative Study of Public Satisfaction with the Police
Mengyan Dai, Xiaochen Hu, Feng Gu
This study takes a comparative approach to examine public satisfaction with the police, focusing on three theoretical models: the demographic model, the neighbourhood conditions model, and the prior contacts with the police model. Using survey data collected from two mid-sized communities in the U.S. and Canada, this study analyzes the similarities and differences in the factors affecting satisfaction with the police with both statistical methods and random forests analysis. The statistical results suggest a great amount of similarity in the effects of theoretically relevant factors across the two samples. The random forests analysis further points to the consistent importance of age, quality of life, and education in predicting public satisfaction. In addition, both analyses find that the effects of age and quality of life are stronger for the sample in the U.S. than those for the sample in Canada. This study suggests that police departments in these jurisdictions could effectively improve satisfaction with the police by addressing quality of life issues in their communities and improving their relationship with younger citizens and citizens with lower levels of education through better interactions.