Robust Spatial Analysis of Rare Crimes

1999

Crime maps are becoming significant tools in crime and justice. Advances in the areas of information technology, computing and Geographic Information Systems (GIS) have opened new opportunities for the use of (computerized) mapping in crime control and prevention programs. Crime maps are also valuable for the study of the ecology and the locational aspects of crime. Maps enable areas of unusually high or low concentration of crime to be visually identified. Maps are however only pictorial representations of the results of more or less complex spatial data analyses. This paper discusses the methodological problems with the spatial analysis of crime rates. The Poisson distribution is a natural candidate to model this type of data, but it has the problem that the variance is a function of the mean. In addition, rates in areas with small populations are more variable than in areas where populations are large. Moreover, data may exhibit spatial correlation, so a simple Poisson model fails to account for the whole variability in the underlying crime rates. As a consequence, maps based on crude crime rates and probability maps are misleading. A hierarchical model dealing with all these problems is proposed and applied to the regional analysis of firearm homicides in the Eastern Australian states. The analysis addresses the question of whether rural and non-rural areas differ in their associated risks of firearm-related homicide. It is shown that this risk is not associated with the rural character of an area, but with factors relating to social and economic disadvantage, which affect both rural and non-rural regions equally.

Behavior and Social Issues, 2006

Violent crime is often studied with individual level variables, using population characteristics as predictors. This study attempts to predict an additional amount of the variability in violent crime using an environmental variable-population density-in a single U.S. city. Data aggregated to the census block group level are used to test a model that compares the urban center of the city with the entire county and the non-urban parts of the county. Drawing on Jane Jacobs' (1961) theories of urbanism and the occurrence of crime, it was hypothesized that population density at the census block level would negatively predict violent crime in the urban areas. Based on evidence of a non-linear relationship between crime and density (Regoeczi, 2002), it was conversely hypothesized that density would have a positive predictive effect on violent crime in the suburban areas, due to differences in urban and suburban/rural crime. The analyses support the hypotheses for the urban areas, but fail to support the hypotheses for the suburban areas, providing insight into an elusive relationship-and the effects of environments on behavior patterns.

2004

The author(s) shown below used Federal funds provided by the U.S. Department of Justice and prepared the following final report: Document Title:

The analysis and understanding of spatial crime patterns is crucial for law enforcements to improve strategic and tactical decision-making. In this context, generalized linear models, such as count regressions, are commonly applied. These non-spatial models are challenged by spatial autocorrelation effects, contradicting fundamental model assumptions. Therefore, the purpose of this research is to present a spatially explicit approach, which combines a negative binomial model and spatial filtering to explain the spatial distribution of nonviolent offences in Houston, TX, for the year 2010. The results provide evidence that the non-spatial negative binomial model is biased while the supplementary consideration of a spatial filter is capable to absorb these undesirable spatial effects and results in a wellspecified regression model. Moreover, besides the significant importance of space in the explanation of the non-violent crime patterns, only the percentage of renter-occupied housing units and the percentage of Asian population are significantly related to the crime. The former covariate has a stimulating effect while the latter has an inhibiting effect. Jekel, T., Car, A., Strobl, J. & Griesebner, G. (Eds.) (2013): GI_Forum 2013. Creating the GISociety. © Herbert Wichmann Verlag, VDE VERLAG GMBH, Berlin/Offenbach. ISBN 978-3-87907-532-4. © ÖAW Verlag, Wien. eISBN 978-3-7001-7438-7,

Social Science Computer Review, 2007

The present research examines a structural model of violent crime in Portland, Oregon, exploring spatial patterns of both crime and its covariates. Using standard structural measures drawn from an opportunity framework, the study provides results from a global ordinary least squares model, assumed to fit for all locations within the study area. Geographically weighted regression (GWR) is then introduced as an alternative to such traditional approaches to modeling crime. The GWR procedure estimates a local model, producing a set of mappable parameter estimates and t-values of significance that vary over space. Several structural measures are found to have relationships with crime that vary significantly with location. Results indicate that a mixed model— with both spatially varying and fixed parameters—may provide the most accurate model of crime. The present study demonstrates the utility of GWR for exploring local processes that drive crime levels and examining misspecification of ...

Proceedings of the ICA

Crime has an inherent geographical quality and when a crime occurs, it happens within a particular space making spatiality essential component in crime studies. To prevent and respond to crimes, it is first essential to identify the factors that trigger crimes and then design policy and strategy based on each factor. This project investigates the spatial dimension of violent crime rates in the city of Detroit for 2019. Crime data were obtained from the City of Detroit Data Portal and demographic data relating to social disorganization theory were obtained from the Census Bureau. In the presence of spatial spill over and spatial dependence, the assumptions of classical statistics are violated, and Ordinary Least Squares estimations are inefficient in explaining spatial dimensions of crime. This paper uses explanatory variables relating to the social disorganization theory of crime and spatial autoregressive models to determine the predictors of violent crime in the City for the period. Using GeoDa 1.18 and ArcGIS Desktop 10.7.1 software package, Spatial Lag Models (SLM) and Spatial Error Models were carried out to determine which model has high performance in identifying predictors of violent crime.

2013

The main techniques used for quantitative analyses of urban crime can generally be divided into three categories: descriptive studies of crime dispersion over a specific urban area without any substantial statistical modeling, traditional statistical spatial models whose normality assumptions do not hold and count models which do not take into account the spatial configuration of the urban layouts. In this work we discuss how configurational components can be introduced in the count data modeling of crime illustrating our point with a case study centered on a highly populated area of the City of Genoa on three crime typologies. The statistical modeling of crime at street level is performed using count models which include the usual economic and socio-demographic variables, complemented with a set of configurational variables, built using the techniques of Space Syntax Analysis, in order to include, among the regressors, the graph complexity of the urban structure. The configurational variables included in this model are statistically significant, consistently with the criminological theories stating that the urban layout has a role in crime dispersion over a city and their use among the set of regressors, substantially improves the overall goodness of fit of the models. The configurational variables herein introduced add an implicit spatial correlation structure of crime to the models and give new and useful information to Municipalities to interpret how crime patterns relate to the urban layout and how to intervene through the means of urban planning to reduce or prevent crime.

2000

The new century brings with it growing interest in crime places. This interest spans theory from the perspective of understanding the etiol- ogy of crime, and practice from the perspective of developing effec- tive criminal justice interventions to reduce crime. We do not attempt a comprehensive treatment of the substantial body of theoretical and empirical research on place and crime

2012

Crime rates-numbers of crimes divided by the population living in an area-have problems when used for small areas. Some small areas include substantial nonresidential areas that contribute to the risk of crime, can be the location of crimes, but that have no populations. Negative binomial models to predict counts of the numbers of crimes in small areas are used to incorporate multiple measures of the risk or exposure to crime that cannot be accomplished using crime rates. Population, several measures of employment, and numbers of students in small areas from a transportation planning dataset all contribute to exposure and the prediction of crime in Indianapolis. Because these data are specific to Indianapolis, models using generally available data from the Census Transportation Planning Products and only data from the census of population are evaluated as alternatives. As the initial exposure data were available for the entire metropolitan area, alternative crime rates using these data are estimated and compared with the traditional population-based crime rates for 14 municipalities in the metropolitan area.

Journal of Quantitative Criminology, 2020

Objectives Crime counts are sensitive to granularity choice. There is an increasing interest in analyzing crime at very fine granularities, such as street segments, with one of the reasons being that coarse granularities mask hot spots of crime. However, if granularities are too fine, counts may become unstable and unrepresentative. In this paper, we develop a method for determining a granularity that provides a compromise between these two criteria. Methods Our method starts by estimating internal uniformity and robustness to error for different granularities, then deciding on the granularity offering the best balance between the two. Internal uniformity is measured as the proportion of areal units that pass a test of complete spatial randomness for their internal crime distribution. Robustness to error is measured based on the average of the estimated coefficient of variation for each crime count. Results Our method was tested for burglaries, robberies and homicides in the city of Belo Horizonte, Brazil. Estimated "optimal" granularities were coarser than street segments but finer than neighborhoods. The proportion of units concentrating 50% of all crime was between 11% and 23%. Conclusions By balancing internal uniformity and robustness to error, our method is capable of producing more reliable crime maps. Our methodology shows that finer is not necessarily better in the micro-analysis of crime, and that units coarser than street segments might be better for this type of study. Finally, the observed crime clustering in our study was less intense than the expected from the law of crime concentration.

Stirring up a hornests’ nest: Geographical distribution of crime. Journal of Economic Behavior and Organization, Volume 152, 2018, pages 17-35 (with. I. Lopez Cruz and G. Torrens)., 2018

How to make police deployment strategies more e¢ cient is becoming the crucial research agenda for the economics of crime and law enforcement. We contribute to this agenda developing the …rst general equilibrium model designed to study how the geographic distribution of police protection a¤ects the decision to pursue illegal activities, the intensity and location of crime, residential choices, housing prices, and the welfare of di¤erent socioeconomic groups. The target is to explore the positive and normative long-run e¤ects of di¤erent ways of spatially allocating police forces in an urban area. We …nd that, when the police protect some neighborhoods (concentrated protection), the city becomes segregated, while when the police are evenly deployed across the city (dispersed protection), an integrated city emerges. Unequal societies face a di¢ cult dilemma in that concentrated protection maximizes aggregate welfare but exacerbates social disparities. Taxes and subsidies can be employed to o¤set the disadvantages of police concentration. Private security makes an integrated city less likely to occur in equilibrium. Even under dispersed public protection, rich agents may use private security to endogenously isolate themselves in closed neighborhoods. JEL classi…cation codes: K42, R12

Journal of Criminal Justice, 2015

Purpose: Using community structure and the racial-spatial divide as a framework, this study examines whether geographic sub-regions of violent crime exist in a large metropolitan area, and if the systemic model of crime can predict them. In addition, surrounding social structure measures are included to determine whether they demonstrate the same violent crime links seen in recent work on concentration impacts. Methods: A LISA analysis is used to identify violent crime clusters for 355 jurisdictions in the Philadelphia (PA)-Camden (NJ) primary metropolitan area over a 9-year period. Multinomial logit hierarchical/mixed effects models are used to predict cluster classification using focal and lagged structural covariates. Results: Models confirmed links of focal jurisdiction socioeconomic status and residential stability with subregion classification. Models with spatially lagged predictors show powerful impacts of spatially lagged racial composition.

The aim of this paper is to address two critical but largely neglected issues in the spatial analysis of urban crime which are spatial spillover effects of crime penetrating neighborhood boundaries and non-stationarity regarding the relationships between contextual factors and neighborhood crime. We use a GIS-based spatial approach to normalize the estimate of burglary crime at block group level and use the geographically weighted regression (GWR) to investigate the correlates of neighborhood crime. Results suggest that the use of normalized measure of neighborhood crime helps better reveal the spatial patterns of burglary crime and the use of GWR accounts for the spatial variations of relationships between contextual factors and crime. In particular, the normalized measure of crime has implications for improving the measurement accuracy of the risk of crime across urban neighborhoods and can be applied to the spatial analysis of other socioeconomic issues such as housing foreclosures and environmental hazards which are also plagued by the spatial spillover issue when geographically contiguous data are analyzed.