Argentina’s robbery rates are among the highest in the world, and for decades the nation has endured consistently high inflation and at times hyperinflation. There is little research on the association between inflation and acquisitive crime in Argentina, but given the widespread negative effects of inflation on other outcomes in Argentina, it is important to know if it is associated with crime. We examined this association. Our unit of analysis was the Argentina-year from 1973 to 2022 (N = 50). Outcome variables were total acquisitive crime (sum of official robbery and theft categories) and homicide rates. We obtained acquisitive crime data from (Sistema, 2023) and inflation data from the World Bank (2023). We employed dynamic time series modeling. Results indicated that inflation is associated with both outcomes across time. Drawing from rational choice and routine activity theories, inflation may be associated with both the motivation and opportunity for crime by expanding illicit markets, especially for stolen durable goods.

ABSTRACT Geographic masking is a key technique to protect individual privacy when sharing spatial point data. Among probabilistic approaches, donut masking is widely used because of its conceptual simplicity and ease of implementation. Whereas most previous studies have focused on average trends linking masking parameters to privacy and analytical utility, this study highlights the often-overlooked role of execution-level variability: differences in outcomes from repeated masking executions under the same parameter settings. Using synthetic crime data and census-based household distributions from Suginami Ward in Tokyo, we applied donut masking across a range of displacement radii and assessed its effects on both spatial k-anonymity and the preservation of spatial analytical results derived from K-function analysis. To quantify analytical preservation, we introduced two novel metrics based on p-value profiles: the Euclidean distance and consistency rate. Our results showed that while a larger displacement increased privacy, it also introduced substantial variability in the analytical outcomes. However, some executions, even under strong masking, exhibited high fidelity in the analysis results. Comparisons with other masking methods suggest that combining execution-level selection with probabilistic masking can help balance privacy protection and analytical preservation. Based on these findings, we proposed two practical strategies for selecting favorable executions: fixed-count selection and threshold-based iterations. Processing time assessments confirmed the feasibility of this approach in standard computing environments. Rather than viewing geographic masking as a fixed-parameter process, this study reframes it as a flexible selection problem that enables more effective anonymization with minimal analytical compromise.

The proliferation of big, complex, and spatial data presents opportunities to further advance criminological theories by measuring the physical environments in which crime occurs. Drawing on high-resolution Sentinel-2 satellite imagery and administrative crime, demographic, and land-use data for Chicago, this study develops micro-scale indices of physical disorder, environmental instability, and vegetation intensity. We integrate these novel environmental measures into multivariate negative binomial models of violent, property, drug, and other crime counts, including community-area fixed effects to isolate within-neighborhood variation. Satellite-derived physical disorder consistently predicts elevated crime counts across categories, even within community areas, while vegetation and environmental instability display nuanced associations that vary by offense type. These findings extend two frameworks within the environmental criminology field, routine activity, and social disorganization theories, by revealing how fine-grained physical context shapes crime opportunity and guardianship beyond traditional sociodemographic measures. This paper illustrates how novel data infrastructures can refine and expand classical criminological theories in the era of digital urban analytics.

Individuals with developmental language disorder (DLD) are disproportionately represented in the criminal justice system. The prospective associations between DLD and offending, and the educational and criminal justice pathways through which DLD might increase the risk of offending and reoffending, remain unclear. We analysed existing data from the Avon Longitudinal Study of Parents and Children (maximum N = 6,800; 51% female; 9% with DLD) with linked school data (national pupil database) and crime records (Avon and Somerset police records for offences committed between ages 13 and 29 years in the region). DLD was determined when the individuals were aged 7-9 years using direct assessments and parent reports. Regression and mediation models were fitted to the data. Individuals with DLD were more likely (odds ratio 1.74, 95% confidence intervals 1.25, 2.44) to have a recorded offence (i.e. charged or cautioned by the police) compared to those without DLD. School suspension was a significant mediator of the relationship between DLD and recorded offending. However, SEN identification was not associated with recorded offending for those with DLD. There was also no difference in the odds of being given an out of court disposal or reoffending for individuals with DLD compared to those without DLD. Individuals who have DLD are more likely to be cautioned or convicted for an offence by the police than those without DLD, and this may in part be because they are more likely than those without DLD to be suspended from school.

ABSTRACT This study presents an overview of official elder abuse data for US states and is the first to assess county-level predictors of elder abuse and neglect through the lens of social disorganization theory. Five years of official elder abuse and neglect, violent crime, and property crime data from three midwestern states, Illinois, Kansas, and Missouri, were aggregated and studied to determine whether the same factors that predict violent and property crime are useful for understanding variation in elder abuse. Results indicate concentrated disadvantage is a consistent explanatory variable, and variables measuring characteristics of the entire county population are better for modeling elder abuse counts than variables measuring characteristics of the elderly population. These findings indicate partial support for social disorganization theory and suggest policies developed to reduce crime by lessening concentrated disadvantage may have benefits for reducing elder abuse prevalence.

Crime rates often change abruptly across the borders of adjacent communities, yet the mechanisms producing these localized disparities remain poorly understood. This study develops a boundary centered analytical framework to examine how roadway network permeability and boundary land use context jointly shape crime rate differences between neighboring areas. Using incident level crime data from six US cities, we apply crime informed community detection to identify approximately 2,000 adjacency boundaries and analyze crime disparities using nonlinear descriptive models, regressions, and SHAP assisted machine learning diagnostics. Results reveal a nonlinear relationship between permeability and crime disparities in which gaps are largest under low connectivity, narrow rapidly at moderate connectivity, and weaken or diminish at high connectivity. Infrastructure barriers and vacant or industrial land create boundary vacuums associated with persistent and volatile violent crime disparities, whereas commercial corridors amplify property crime disparities through increased opportunity exposure. Overall, violent crime patterns reflect insulation produced by physical and land use barriers, while property crime patterns reflect exposure to target rich environments. These findings position urban boundaries as a critical ecological unit for understanding micro scale crime divergence and offer actionable insights for place-based security interventions, resource allocation, and applied models linking to localized crime dynamics.

This research examines how individual-level power distance shapes deviant tourist behavioral intentions and how multi-layered social influences condition this relationship. Integrating Social Cognitive Theory, Moral Disengagement Theory, and the Theory of Planned Behavior, we propose that power distance increases deviant intentions by heightening uncertainty about the social costs of punishment and fostering favorable attitudes toward norm violations. Social influence is conceptualized as operating at situational, contextual, and structural levels, which amplify these cognitive processes. Evidence from four complementary studies (N = 1123) supports the proposed framework. Study 1 identifies power distance as a robust cultural antecedent of deviant tourist intentions. Studies 2A and 2B show that witnessing other tourists’ deviance elevates intentions, with stronger effects among high–power distance individuals and variation across deviant domains. Study 3 confirms a sequential mediation pathway via punishment uncertainty and attitudinal approval, and demonstrates that subjective norms condition early risk appraisals. Study 4 incorporates county-level crime data, revealing that long-term exposure to permissive normative environments strengthens key cognitive pathways to deviance. Together, these findings advance a multilevel explanation of deviant tourist behavior and offer culturally informed insights for managing misconduct and promoting sustainable visitation.

This paper proposes a time series model with Bernoulli counting series dependent high-order random coefficient INAR and investigates its stationarity and ergodicity. The unknown parameters are estimated by the conditional least squares method and the modified quasi-likelihood estimator, with the asymptotic properties of these estimators rigorously established. Simulation studies are conducted to evaluate the performance of the proposed estimation approach, demonstrating that the modified quasi-likelihood estimator outperforms the conditional least squares estimator in certain regions of the parameter space. Furthermore, the practical utility of the model is illustrated through an empirical analysis of criminal intent crime data.

Research suggests that the longer firearms remain in circulation following their first known use in a crime, the more likely they are to be used in multiple offenses. However, little is known about how repeat-use firearms journey in time and space from one crime to another. This study uses spatially and temporally referenced crime data from Houston, Texas, to address four questions: 1) How far do repeat-use firearms travel between crime incidents? 2) How many days elapse before the same firearm is used again in another crime? 3) Is there a relationship between time and distance for repeat-use firearms? 4) What factors are associated with the distance and time between repeat-use firearm incidents? Results from spatial analysis indicate that the median straight-line distance between matched gun crime cases is 4.33 mi, with a median traveling time of 59 days. Results show a significant relationship between time and distance for repeat-use firearms, even after accounting for initial offense type, including homicide, assault, deadly conduct, robbery, and criminal mischief. Exploratory findings reveal that guns journey faster from one crime to another in more densely populated neighborhoods and when the initial offense is robbery, and that guns first used in robberies tend to travel greater distances between crimes relative to other offense types. Overall, these findings suggest that repeat criminal gun use occurs within a relatively short time frame, that firearms move only short distances between incidents, and that both offense type and local population context are associated with the spatiotemporal movement of repeat-use firearms.

Crime prediction has become an important application of artificial intelligence because public safety agencies need faster and more reliable ways to identify crime trends. This project presents CrimeCast, a web-based crime prediction and analysis system developed with Python and Flask, trained on crime data from Tamil Nadu, India. The system uses a Random Forest Classifier to predict the most likely crime type from inputs such as state, city, latitude, longitude, year, and domestic status. The model is designed to classify six major crime categories: Assault, Burglary, Cyber Crime, Domestic Violence, Robbery, and Theft. The application combines machine learning with a secure, responsive web interface built with Bootstrap and an SQLite database for storing user accounts and prediction history. It also includes analytics dashboards that show crime distribution, yearly trends, city-wise frequency, and domestic versus non-domestic comparisons. An interactive heatmap built with Leaflet.js provides a geographic visualization of crime density across Tamil Nadu. With its prediction engine, history tracking, and visual

This study examines how decisions to report crime vary across population groups, testing the hypothesis that individuals possess “windows of surprise” – learned thresholds for institutional engagement – shaped by social, economic, and psychological factors. A cross-sectional online survey (n = 1948) presented participants with crime vignettes of increasing severity. K-means clustering identified three latent reporting profiles, with entropy values used to assess response variability. Cluster 1 (lower income, mixed employment) displayed the broadest reporting threshold and highest entropy, especially for low-severity offences. Cluster 0 (moderate income, stable employment) showed narrow, consistent thresholds. Cluster 2 (higher income, higher education) revealed context-sensitive patterns. Income, gender, education, religion, and household structure significantly differentiated clusters. Results indicate that crime reporting is shaped by structured cognitive filters rather than uniform offence recognition. While serious crimes prompt consistent reporting, lower-harm offences are unevenly interpreted, particularly in disadvantaged groups, raising concerns about representational equity in crime data and the need for policing strategies that acknowledge these disparities.

Rapid urbanization and population growth in Indian smart cities have increased the complexity of crime prevention and public safety management. Traditional crime analysis methods often on previous records and manual interpretation, which are limited in handling dynamic spatial and temporal crime patterns. This research paper proposes an Enhanced AI-Spatio Crime Prediction and Hotspot Visualization System for Indian Smart Cities that integrates artificial intelligence, geospatial analytics, and data visualization to improve crime forecasting and decision-making. The proposed system uses machine learning algorithms to analyze historical crime data, location-based factors, demographic patterns, and time-series trends to predict potential crime occurrences. Advanced clustering techniques are applied to identify crime hotspots, while interactive visualization dashboards provide real-time maps, heatmaps, and analytical insights for law enforcement agencies and city administrators. The model is designed specifically for the Indian urban environment by considering city-specific challenges such as population density, traffic flow, socio-economic diversity, and rapidly changing infrastructure. Experimental results indicate that the proposed approach enhances prediction accuracy, enables proactive policing, optimizes resource allocation, and supports safer urban planning. The study demonstrates how AI-driven crime intelligence systems can contribute significantly to the development of secure, efficient, and sustainable smart cities in India.

Can mass protests for racial justice influence police patrol tactics? Which racial and spatial mechanisms underlie these changes? Previous literature suggests that high-profile events can lead individual agents to alter their enforcement activities by ‘de-policing.’ However, literature is scarce (and conflicting) about the relationship of de-policing with crime, and effects across racial groups are underexplored. In this article, I argue that this apparent de-policing is, in reality, a positive reform of police patrol tactics induced by public pressure, which forced law enforcement to reconsider its approach to police-civilian interactions and focus on adopting a conservative stopping strategy, more targeted against higher-risk individuals, and more sensitive regarding the unequal treatment of minorities. To do so, I analyze this issue combining pedestrian stops and crime data from Chicago, asking whether the 2020 BLM protests led the Chicago Police Department to change its patrol strategy. When protests erupted, crime and policing got progressively decoupled: when criminality rose, stops did not. While overall crime simply returned to its pre-BLM (and pre-COVID-19) trends and levels, policing changed radically: stops, searches and arrests dropped and became stationary, while hit rates rose sharply. Stops also decreased differently across racial groups: almost exclusively for Black civilians, and mostly in minority districts. This decline in minority policing is evident across all officer groups, further indicating that the change reflects a broad shift in policing tactics rather than individual-level shirking.

The Crime Rate Prediction System is designed to help authorities and individuals understand and anticipate crime patterns in different regions. It analyzes historical crime data, user inputs, and environmental factors to predict future crime rates. By studying past incidents and trends, the system can identify high-risk areas and provide useful insights for crime prevention and safety planning. At the core of this system is a machine learning-based model that uses multiple factors such as location, time, type of crime, and past records. It not only considers historical data but also analyzes patterns and relationships within the data to make accurate predictions. The system can identify whether crime is likely to increase or decrease in a particular area, helping law enforcement agencies take preventive measures. The Crime Rate Prediction System also allows users to visualize crime data through graphs and reports. It helps in tracking crime trends over time and provides alerts for potential high-crime zones. The system is designed to be user-friendly, making it easy for both officials and the general public to access and understand crime-related information. This system can be considered a smart decision-support tool that improves public safety by providing reliable crime predictions. It continuously learns from new data, improving its accuracy over time and helping build safer communities. Keywords: Crime Prediction, Machine Learning, Crime Analysis, Data Visualization, Public Safety, Predictive Analytics.

BackgroundTraditional policing interventions associated with the standard model of policing, such as programmes designed to arrest and prosecute repeat offenders, have not been effective in controlling crime. In contrast, a growing number of rigorous programme evaluations find focused deterrence strategies, designed to change offender behaviour through a blended law enforcement, social service and opportunity provision, and community-based action approach, are effective in controlling crime. While evidence is growing, there are few randomised controlled trials evaluating this type of intervention, particularly assessing the effect on repeat serious violence offenders.MethodsThe police force will identify individuals involved in 3 or more serious violence offenses, with the most recent serious violence offense occurring within the last 24 months in the recorded crime data. Eligible individuals are placed within one of six strata based on gender and age group. Within each stratum, 50% of individuals are randomly allocated to the treatment condition and the rest to the control condition. All participants will receive usual care, but those randomly assigned to the experimental condition will receive a focused deterrence intervention visit from police providing a scripted empathetic talk and a list of local resources. The police will collect crime data for all individuals for the 12 months following the date of randomisation. The primary outcome is the total crime harm, measured by the Cambridge Crime Harm Index, perpetrated by the individual across all crimes committed in the year following randomisation (across England and Wales). The secondary outcomes include the number of arrests for violent crime, the total number of arrests, the number of non-violent arrests, and the time to the first arrest within the same timeframe.DiscussionCurrent RCTs investigating FD interventions focus on crime counts, but this assumes parity amongst crimes. Our study will assess the impact on both crime harm and crime counts to provide a comprehensive review of the impact of this intervention. If the hypotheses are supported, this single-contact FD intervention would likely have significant operational appeal for police and communities to prevent and reduce crime and harm with light-touch engagement.Trial registrationThe study is now listed on the ISRCTN registry with study registration number ISRCTN35233331. Registered on 24 July 2024.

Under rapid urbanization, street crime poses growing challenges to urban safety. Existing studies often treat gray and green spaces as independent variables, limiting the understanding of nonlinear crime patterns and spatiotemporal heterogeneity. Using day–night street crime data from Shanghai between 2010 and 2020, this study applies an interpretable machine learning framework combining CatBoost and SHAP to examine how the coupling of gray–green spatial structures influences street crime. Gray–green spatial morphology is quantified using both MSPA- and Fragstats-based indicators, which are integrated into composite coupling indices. The results indicate that gray–green structural coupling exhibits significant nonlinear and threshold-dependent effects on street crime. Compared with conventional Fragstats metrics, MSPA-based structural indicators demonstrate stronger explanatory power. Theft-specific analysis further indicates that gray-space core–edge structures exhibit higher crime risk at night, with this effect becoming more pronounced in the later period. Across both study periods and day–night contexts, green branch areas (G_BRANCH) consistently show stable inhibitory effects, with the strongest suppression occurring when G_BRANCH values range between 0 and 1.6 and interact with gray core–edge structures (B_CORE and B_EDGE). These findings provide quantitative evidence that gray–green spatial structures function through coupled, nonlinear interactions and offer targeted spatial planning implications for crime prevention in high-density cities.

The 2021 census for the Cayman Islands asked two new questions that did not feature in any previous census. Section 11 of the questionnaire asked whether any respondent had been a victim of crime (excluding traffic) during the past 12 months and, if so, whether they had reported it. Because there is no crime survey undertaken in the Islands, this is a valuable addition to the census, and the results can be viewed in a number of ways in combination with the remainder of the demographic data from the questionnaire. Furthermore, the experience of crime and the claimed levels of reporting of crime can be compared to the police statistics for actual recorded crime to provide a clearer picture of crime reporting. While these new census questions provide valuable insights, the study also highlights limitations in the current approach, including the lack of data on crime types, frequency and impact. We recommend implementing a more comprehensive annual crime survey to address these gaps and enhance understanding of crime patterns. This research contributes to the broader discussion on crime data collection methodologies in small island nations and their implications for policy development and resource allocation in law enforcement.

NCRB’s annual report, ‘Crime in India,’ includes crime data for states, districts, and metropolitan cities. The official crime data of metropolitan cities in India often overlook the complexities associated with the jurisdiction of police administration and ULBs, as well as their impact on the crime rate. The crime rates of many cities, such as Jabalpur, are often inaccurate due to jurisdiction-related complexities. These complexities may lead to overcounting of crime incidents and under-counting of crime rates. Crime researchers mostly ignore these complexities when quoting NCRB data to explore the crime in a city or comparing crime between metropolitan cities.

Operating on granular, publicly available crime data from 2017–2022, we utilize negative binomial regressions to investigate the spatial relationship between gamedays and crime in Atlanta, comparing effects of football and soccer. Our models suggest that football increases crime near the stadium in Atlanta substantially relative to the effects of soccer. Football games are correlated with a 113% increase in crime—concentrated within 0.5 miles of the stadium and robust across models. We support our findings using a temporal analysis within gamedays, which illustrates that the distribution of crimes is centered around the hours surrounding the start of the game. Our results support the existence of a substitution effect of crime towards the stadium on game days and the existence of demographic differences between football and soccer fans in the United States. Mercedes-Benz stadium should implement additional security measures, including free lockers within the stadium, to shield their patrons from theft.

Economic and demographic shifts can shape property crime patterns in rural communities, challenging assumptions that these areas are inherently safe. Prior research has linked factors such as population size, housing turnover, employment, and resource production to crime, yet few studies examine their combined effects in rural settings. This study analyzes the relationship between economic indicators and property crime in rural Natrona County, Wyoming, guided by Routine Activity Theory, Strain Theory, and Social Disorganization Theory. The study hypothesizes that decreases in oil and gas production, heightened drought levels, a shrinking labor force, rising unemployment and residential availability, and lower high school graduation rates will be associated with increases in property crime. Additionally, it is expected that Natrona County will experience lower levels of property crime during periods of lockdown mandates. An Ordinary Least Squares (OLS) regression model with the independent variables, COVID-19, gas and oil production, total population, labor force size, and homes for sale, was applied to monthly property crime data from January 2021 through December 2023. The model explained 91% of the variance in crime. Findings show COVID-19 negatively predicted property crime, while population totals and homes for sale were positively associated. Gas production was significant without a clear trend, and oil production and labor force size were not significant. Results highlight the critical role of housing and population dynamics in rural crime patterns. Future research should expand economic measures and improve data reliability to better capture rural crime dynamics.