Are Repeatedly Extorted Businesses Different? A Multilevel Hurdle Model of Extortion Victimization

Abstract

Objectives Research consistently shows that crime concentrates on a few repeatedly victimized places and targets. In this paper we examine whether the same is true for extortion against businesses. We then test whether the factors that explain the likelihood of becoming a victim of extortion also explain the number of incidents suffered by victimized businesses. The alternative is that extortion concentration is a function of event dependence. Methods Drawing on Mexico’s commercial victimization survey, we determine whether repeat victimization occurs by chance by comparing the observed distribution to that expected under a Poisson process. Next, we utilize a multilevel negative binomial-logit hurdle model to examine whether area- and business-level predictors of victimization are also associated with the number of repeat extortions suffered by businesses. Results Findings suggest that extortion is highly concentrated, and that the predictors of repeated extortion differ from those that predict the likelihood of becoming a victim of extortion. While area-level variables showed a modest association with the likelihood of extortion victimization, they were not significant predictors of repeat incidents. Similarly, most business-level variables significantly associated with victimization risk showed insignificant (and sometimes contrary) associations with victimization concentration. Overall, unexplained differences in extortion concentration at the business-level were unaffected by predictors of extortion prevalence. Conclusions The inconsistent associations of predictors across the hurdle components suggest that extortion prevalence and concentration are fueled by two distinct processes, an interpretation congruent with theoretical expectations regarding extortion that considers that repeats are likely fueled by a process of event dependence.

Homicide rates are spatially associated with built environment and socio-economic factors: a study in the neighbourhoods of Toronto, Canada

“…In this study, homicide data have been analysed from temporal, spatial and spatio-temporal point of view. The fundamental hypothesis is that the ‘where and when’ crimes are committed are not random but follow a clustered pattern [ 65 , 66 ] concentrated on a small proportion of places [ 67 ]. Therefore, GIS provides a powerful tool to identify existing patterns of crimes and their spatio-temporal patterns (high-risk areas), something which is essential for the development of strategies for reducing crime [ 68 – 72 ].…”

Understanding How Offending Prevalence and Frequency Change with Age in the Cambridge Study in Delinquent Development Using Bayesian Statistical Models

“…Our discussion is based on Section 5.6 of Stan Development Team (2022a), which also covers the related class of zero-inflated models (Lambert, 1992). Estévez-Soto et al (2021) presented a recent use of the hurdle model in criminology. To clarify this model let us assume, for the moment, that our data are the total number of offenses y 1 , .…”

Repeat and Near Repeat Burglary Victimization in Taiwan

“…Results are presented in frequency distribution tables and the cumulative distribution is presented using Lorenz curves. The cumulative distribution is then compared with a Poisson-simulated distribution, estimated by a Monte Carlo simulation of a Poisson process with 500 replicates (see Estévez-Soto et al, 2021). Gini coefficients are also reported as an additional way to quantify burglary concentration, for which a figure of zero indicates no concentration and a figure of one indicates complete concentration.…”

“…Gini coefficients are also reported as an additional way to quantify burglary concentration, for which a figure of zero indicates no concentration and a figure of one indicates complete concentration. To examine if the observed distribution statistically differs from a reference distribution (i.e., the aforementioned Poisson distribution), we use one-sample Kolmogorov-Smirnov (KS) tests (Arnold & Emerson, 2011;Estévez-Soto et al, 2021). It is noted that since the frequency of victimization is discrete and the victimization population in the 2015 TAVS (n = 194) was greater than a suggested sample size of 30 (Dimitrova et al, 2017), the KS test is suggested to be performed using an improved R package 'KSgeneral' (Dimitrova et al, 2020).…”