Open research questions in Crime Patterns and Interventions

What the literature leaves open

• There are several limitations to the use of clustering approaches, which are important to note. First, clustering analytics often assume that all clusters have the same radius; when this assumption is violated, the resulting clusters may behave in an unusual way (see Raykov et al., 2016 for examples). To offset this potential limitation, the strength of both the mean silhouette coefficients, and silhouette coefficients for each cluster were reviewed, to ensure developed clusters behaved in a logical manner. This was also a function of anchor point selections, which hinge on research questions and researcher application (Times Square and hospitals, in this case). Second, metrics employed to analyze clustering techniques are typically predi- cated on Euclidean distance because Euclidean distance functions in linear space. This means outliers may contribute to dramatic changes to evaluation metrics (Raykov et al., 2016). This is particularly a limitation of k-means analyses [see Tech- nical Appendix], which we attempted to mitigate by implementing the k-medoids algorithm to develop clusters. K-medoids is less vulnerable to outliers influencing evaluation metrics. Additionally, the number of clusters was optimized through use of the silhouette coefficient, to avoid within cluster outliers. Third, as the Technical Appedix explains, the k-medoids method assumes that the number of clusters in the data is known prior to analysis and this information is gen- erally used to evaluate accuracy (Raykov et al., 2016; Adepeju et al., 2021). Given the fact that this research sought to identify latent clusters, and interpret their distribu- tion, this was not the case. This was not necessarily a limitation in this research, as the intention in this research was to understand an opaque phenomenon. Identifying the optimal number of clusters prior to analysis helped mitigate this limitation. However, it bears note that employing a supervised method may improve understanding of model accuracy, where the data supports such an approach. Finally, there was an important practical limitation to implementing the DoD meth- odology. The computational time may be marginally longer than more simple cluster- ing analytics, primarily due to complexity of applying the algorithm to determine the cluster ensemble among larger datasets (Dsouza et al., 2017). A solution may be to par- allelize the analysis to increase computational speed (Subhashree & Prakash, 2015), employ a cloud-based platform for analysis, or alternatively to run these analytics on a computer with appropriate specifications. Despite this limitation, accuracy of the modelling was not impacted by the speed of at which findings were computed.

  Measuring macro and micro geospatial changes in New York City’s motor vehicle thefts · 2026 · DOI

• Although this is the first study to examine the public’s pretrial decision-making preferences, it is not without limitations. First, the 23 factorial design of the experiment partitioned the sample into eight sub-groups, and with an overall sample size of 1092 participants, this resulted in approximately 136 participants per group. While we would have liked to explore the additional factors Scurich and Krauss (2020) suggested as important to pretrial decision-making, including more crime-types, adding even one manipulation would reduce the sub-group sample size to about 68 participants each. To avoid sub-grouping the sample too much (Auspurg & Hinz, 2014; Wallander, 2009), we elected to maintain the 23 factorial design. While our focus on similarly situated violent and non-violent crimes is consistent with prior sentencing vignette studies (Lehmann et al., 2019; Miller et al., 1991; Wozniak et Making sense of mixed signals: case factors & pretrial decision-making1 3 al., 2022) and criminological research more generally (Gottschalk, 2016; Petersen & St. Louis, 2024; Seeds, 2017; St. Louis & Petersen, 2024), future research should explore other crime types. In addition, we recommend that future work distinguish weapon involvement and victim injury from current charges and consider factors such as pending charges. Besides these aggravating factors, researchers should also consider mitigating factors, such as the ability to pay, family support, childcare, housing, or employment. These practical constraints are not peripheral, as over 80% of judges report these factors as somewhat or very relevant to their decision-making (Ulmer et al., 2023). Yet, incorporating them into judicial discretion can widen disparities, as pretrial decisions may reflect unequal access to resources rather than actual risk. It is for this same reason that some pretrial risk assessments opt to exclude personal/social characteristics from their instruments (Desmarais et al., 2021), as they may be correlated with demographic characteristics and undermine the race-neutrality of risk assessments. Accordingly, future research may include manipulations related to defendant demographics (e.g., race, gender) to assess how they influence pretrial decision-making. We held these factors constant by explicitly excluding defendant demographics from our vignettes and experimentally manipulating legally relevant factors instead (e.g., offense type, prior record, flight risk). However, given the racial typification of crime (Chiricos et al., 2004; Russell-Brown, 2009) and the relationship between racial resentment and pretrial justice attitudes (St. Louis & Petersen, 2024), incorporating a race and/or gender manipulation is another next step.

  Making sense of mixed signals: an experimental analysis of case factors and pretrial decision-making · 2026 · DOI

• This underscores the need for additional research on the nature and influences of DWI CFS in this area as our findings demonstrate that game conditions influence the concentration of DWI CFS in particular areas but these conditions do not significantly influence the number of DWI CFS per day per grid.

  Spurring crime? The geospatial clustering of DWIs in San Antonio, Texas on NBA gamedays · 2026 · DOI

• Some limitations need to be addressed. To provide a neighbourhood that appears as realistic as possible and immerses the participants, multiple elements were pres- ent that might have impacted the participants’ overall experience. Amongst those things were active sprinklers on two of the properties and the camera being present in every condition. When looking at a real-life neighbourhood, then naturally sprin- klers would be active, and a camera would be displayed to everybody who is walking around. However, in the context of this study it might have interfered with the experi- ence of the participants. Another difficulty was the possible interference of the slow loading of the scene with the self-switching lights and possibly self-closing blinds condition. There is a possibility that participants did not see these dynamic guardians just because of a technical delay and were therefore not deterred by it. It might also have been for self-switching lights and self-closing blinds that there was inattentional blindness. Especially in high cognitive load settings such as VR-environments this likelihood increases (Hyman et al., 2010). Self-switching lights and self-closing blinds are also more subtle cues than a smart-home camera with an added voice, especially because as humans we are attentive to spoken language (Kühne et al., 2020). When cues are subtle, they are most likely ignored in high cognitive load settings, in which in VR- settings overlooking cues is more likely (Gajjar et al., 2017).

  Dynamic defence: dynamic guardianship for enhanced neighbourhood security · 2026 · DOI

• Burglar perspective: activity level Burglar perspective: target hardening Burglar perspective: chances of getting caught Burglar perspective: social cohesion Burglar perspective: house occupation Residents’ perspective Residents’ perspective: reduced scale Residents’ perspective: freed resources Residents’ perspective: perceived level of protection Residents’ perspective: ability to protect Residents’ perspective: convenience…

  Dynamic defence: dynamic guardianship for enhanced neighbourhood security · 2026 · DOI

• This study centers on the epicenter of the 2020 “Minneapolis reckoning” (Phelps 2024) within the context of a 30 percent national homicide increase (Gramlich 2021; Council on Crime and Justice 2023). Although similar patterns may appear in other metropolitan areas that experienced high-profile police killings, generalizations beyond Hennepin and Ramsey Counties should be approached cautiously. The interrupted-time-series design assumes that no unobserved time-varying factors coincide precisely with the treatment; while inclusion of COVID-19 caseloads, mobility, policy timing, weather, and temporal autoregressive terms support the tenability of this assumption, residual confounding cannot be ruled out. Time-varying heterogeneity in the movement of the pandemic and related behavioral responses not captured by our time-varying covariates (e.g., changes in routine activities not captured by mobility and policy controls; alterations in informal social controls temporally orthogonal to these measures, between- tract heterogeneity in pandemic response in our panel models) could confound our estimates. Therefore, the validity of our estimates of the effect of Mr. Floyd’s murder rests upon these statistical adjustments, which may not completely capture the significant homicide-related impacts of the pandemic. Further, it is beyond the scope of our current data and analysis to separate the effect of the murder from the broader context of the pandemic. Given we do not have any counterfactual observation of the murder outside of the time of a global pandemic, we are not able to speak with certainty to the effect of the murder in non-pandemic contexts (i.e., a potential moderating relationship). Further, our causal-mediation test evaluates only one mechanism—police stops—within Minneapolis (Imai et al.2010; Tingley et al. 2014). Relationships in suburban or adjacent tracts may differ, and we cannot directly measure changes in informal social control, legal cynicism, or mental health that likely co-evolved with policing and homicide. Finally, as national data show an increasing proportion of “unknown” circumstances in homicide records (Council on Crime and Justice 2023), even hand-coded classifications remain subject to measurement uncertainty. Overall, these caveats suggest that our estimates should be viewed as upper-bound effects in a strong-case city rather than precise forecasts for all jurisdictions. Nonetheless, the patterns we document, namely an enduring, compositionally distinctive, and spatially concentrated spike in homicide following Floyd’s murder, partially mediated by depolicing, offer a clear signal about how acute institutional ruptures can interact with chronic inequality to reshape lethal violence. For criminologists, the Twin Cities experience demonstrates the value of integrating event-level data, neighborhood context, and institutional behavior to understand how periods of crisis alter both the level and character of crime, and it points toward the development of dynamic, multi-level models capable of capturing similar patterns in other places and times.

  The Changing Nature of Homicide in the Twin Cities: An Interrupted Time Series Analysis of the Police Murder of George Floyd · 2026 · DOI

• Several limitations bear on the interpretation of these findings. First, the analysis is cross- sectional, with crime data drawn from the 2024–25 period and perception data from the 2023–24 CHA survey wave. While the one-year offset is modest relative to the structural stability of neighborhood-level conditions, it introduces some temporal misalignment into the gap measure. Second, the community area is a relatively coarse spatial unit; patterns that are invisible at this scale, like block-level environmental conditions and the spatial distribution of crime within community areas, may drive meaningful variation in the per- ception gap that the analysis cannot detect. The HDBSCAN point-pattern analysis, while not included as a focus of this study, suggested that violent crime concentrates at spatial scales far smaller than community areas, and this scale mismatch may itself contribute to the gap in high-crime areas where a few blocks account for disproportionate shares of neighborhood-level crime totals. Third, with n = 77 community areas, the study has limited statistical power to detect small effects in regression models, and the non-normal residuals driven by Fuller Park and a small number of other outliers suggest that robust or quantile regression methods might better characterize the conditional distribution of perception gaps across neighborhood types in future work.

  Fear vs. Reality: Exploring the Crime Perception Gap in Chicago Neighborhoods · 2026 · DOI

• Based on the findings the recommendations are given below: 1. Although very weak and statistically insignificant relationships were observed for public awareness, community involvement, police trust and frequency of interaction; regular meetings with the support of local communities and additional measures such as crime-prevention campaigns or neighborhood watch activities must strengthen police-community linkages. 2. Responses indicated that law enforcement strength, degree of policy implementation and perception of the availability of legal support and conviction rates are perceived in a positive manner by respondents; therefore, existing legislative frameworks could be more consistently applied and implemented to contain urban social crimes4. 3. These results indicate the four components resource availability, police training, public cooperation, and agency coordination are performing relatively adequate. Hence, it is imperative that law enforcement agencies get better trained, more logistical support and enhanced coordination with local government bodies, legal services authorities and social welfare organizations. 4. Because the highest mode values from the results were also received for using technology, urban policing should provide CCTV monitoring, digital complaint systems with multiple applications such as trip booking and nearby area alerts, emergency response apps, crime mapping, information on how to minimize or refrain from crime propagation through information portals and websites, online reporting into dedicated desks. 5. But when it came to communication channels, engagement programmers and police transparency, the scores were only average. To gain the public's trust and cooperation police need to keep residents up to date on what is going on with crime-prevention efforts and treat them fairly. 8. LIMITATIONS This study has several limitations. While the survey had 400 respondents, this sample may not be representative for all urban residents because crime patterns and policing conditions vary by area. Due to the study being based on perception-type responses, personal experience, fear, or trust in police may influence the outcome as well. Moreover, weak and nonsignificant correlations of public awareness with community participation, trust, and interaction frequency restrict the conclusiveness of strong statements on community involvement. It is important to note that the study did not analyze official crime data or court records. Because it is cross-sectional, this can never fully allow explanations of longer-term shifts in how effective policing is or legal strategies over long periods of time to be very clear. 9. CONCLUSION This study finds that while community policing and legal strategies have been shown to decrease social crimes in urban areas, they are not as effective alone.

  Community Policing and Legal Strategies for Reducing Social Crimes in Urban Areas · 2026 · DOI

• This study has several limitations. While the survey had 400 respondents, this sample may not be representative for all urban residents because crime patterns and policing conditions vary by area. Due to the study being based on perception-type responses, personal experience, fear, or trust in police may influence the outcome as well. Moreover, weak and nonsignificant correlations of public awareness with community participation, trust, and interaction frequency restrict the conclusiveness of strong statements on community involvement. It is important to note that the study did not analyze official crime data or court records. Because it is cross-sectional, this can never fully allow explanations of longer-term shifts in how effective policing is or legal strategies over long periods of time to be very clear. 9. CONCLUSION This study finds that while community policing and legal strategies have been shown to decrease social crimes in urban areas, they are not as effective alone. It illustrates that public awareness, community involvement and interaction frequency and trust in police are weakly associated and not statistically significantly related with crime prevention at the present moment, hence the parameters of community participation do not seem to work together well for crime prevention. The positive perception of law enforcement strength, policy implementation, legal accessibility, and conviction rate implies that formal institutional mechanisms matter in regulating urban crimes. Findings regarding the subsystems showed that resource availability, police training, public cooperation, and agency coordination are at a moderate level (indicating some institutional capacity but needing development). Use of technology was seen as a potential tool for enhancing policing while need is more on communication, engagement programs and police transparency. In conclusion, the study concludes that integrated efforts are required to curb urban crime through enhanced criminal justice system capacity enhancement, community participation, specialized training for police personnel and collaboration among policing agencies as well as technological adoption. Nor can sustainable crime control rely solely on police action but public acceptance and implementation of the law must be accompanied by an ongoing partnership between citizens, policing efforts and local actors. REFERENCES 1) Abdillah, Widianingsih, I., Buchari, R. A., & Nurasa, H. (2024). From urban crime areas to urban resilience: Lessons learned from Bandung City, Indonesia. Cogent Social Sciences, 10(1), 2369171.

  Community Policing and Legal Strategies for Reducing Social Crimes in Urban Areas · 2026 · DOI

• The current study addresses this gap in research by comparing racial/ethnic portraits of crime (cross-sectionally and over time) across four of the FBI's “public facing” official crime data systems—(1) Crime in the United States (CIUS) reports, (2) NIBRS Offender data, (3) NIBRS Arrestee data, and (4) NIBRS Arrestee National Estimates.

  The New Dawn of NIBRS: Race/Ethnicity and Crime Patterns and the NIBRS Transition · 2025 · DOI

• Abstract Purpose Domestic violence and abuse (DVA) cases remain under-reported and under-prosecuted in the criminal justice system (CJS), with researchers frequently having limited access to raw police data.

  ‘Understanding Domestic Violence and Abuse: Victim, Suspect and Crime Predictors of Police Outcomes’ · 2025 · DOI

• Even though federal law requires state child welfare agencies to conduct annual credit checks for adolescent youth in foster care, little is known about the factors contributing to risk of identity crime victimization among foster care-involved youth, either before or after their initial entry into care.

  Predictors of Identity Crime Victimization of Adolescent Youth in Foster Care · 2024 · DOI

• Although most people have heard the terms 'souvenirs', 'trophies', and 'mementos', discussed in books and movies on the true crimes of sexual murderers, limited research has delved into the phenomenon of theft in sexual homicide (SH).

  Trophy, souvenir, or simple theft? Taking items from the victim in sexual homicide · 2024 · DOI

• Our findings support the limited evidence on NFS in England and Wales, and Australia in finding that most victims are female and most suspects are male, typically victims and suspects are young-middle-aged adults, and most NFS is domestic in nature.

  Non-Fatal Strangulation: An Empirical Review of the New Offence in England and Wales · 2024 · DOI

• Few studies have considered the impact of restricting access to pubs or bars on crime, and the present study aims to provide more insight into this by using the restrictions to combat the COVID-19 pandemic as a natural experiment.

  COVID-19 restrictions, pub closures, and crime in Oslo, Norway · 2022 · DOI

• This suggests aneed for further study on contextual differences of different locations for CCTV, to ascertain whether the effect on violence in deprived neighbourhoods can be replicated.

  CCTV in deprived neighbourhoods – a short-time follow-up of effects on crime and crime clearance · 2020 · DOI

• Abstract An increasing number of police departments use information technology (IT) to optimize patrolling strategies, yet little is known about its effectiveness in preventing crime.

  Crime is Terribly Revealing: Information Technology and Police Productivity · 2020 · DOI

• Exposure to information conveying this stereotype has been found to reinforce the shooter bias as measured by the FPST, even though the cognitive mechanisms involved in this bias are not well understood.

  Modeling Effects of Newspaper Articles on Stereotype Accessibility in the Shooter Task · 2019 · DOI

• The federal response to the problem of sex offending has resulted in an exponential increase in the number of sex offenders on federal post‐conviction supervision; however, relatively few studies have explored whether and how well the actuarial risk instrument currently used by federal probation officers—the federal Post Conviction Risk Assessment instrument (PCRA)—accurately predicts reoffending behavior among the federal sex offender population.

  Predicting Sex Offender Recidivism: Using the Federal Post Conviction Risk Assessment Instrument to Assess the Likelihood of Recidivism Among Federal Sex Offenders · 2018 · DOI

• Then 12 experts in survey-based measures of crime-related issues were involved in a Delphi panel with the objective of enhancing a participatory design of a new questionnaire addressing individual and space-based determinants of the perception of insecurity, which has been poorly explored to date.

  Participatory design of a thematic questionnaire in the field of victimization studies · 2018 · DOI

• In this study, we attempt to address the inconsistent results in previous studies and shed light on the relationship between rural crimes and shale oil development at the Bakken using county‐level data for Montana and North Dakota from 2000 through 2014.

  Does Shale Energy Development Mean More Crime? The Case of the Bakken Oil Boom · 2018 · DOI

• This shifting street dynamic is underexplored in prior research, thus, drawing on qualitative data from Scotland and Bourdieu’s theory of social field, the current study presents three cases of gangs at different stages of evolution and examines how levels of gang organization affect spatial relationships.

  Putting the ‘Street’ in Gang: Place and Space in the Organization of Scotland’s Drug-Selling Gangs · 2018 · DOI

• As in the United States, there is no consensus across countries on the exact definition of the approach, but the fundamental ideas of community engagement and partnerships, proactive approaches to crime prevention, and organizational transformation/decentralization are typically consistent (albeit to varying degrees; Donnelly, 2013).

  PROTOCOL: Community‐oriented policing to reduce crime, disorder, and fear and improve legitimacy and satisfaction with police: a systematic review · 2017 · DOI

• Thus, the current study helps to address a gap in the research by determining the exact near repeat space-time clustering of three crime types (burglary, theft from a motor vehicle [TFMV], and common assault) across three Canadian cities (Edmonton, Alberta; Moose Jaw, Saskatchewan; and Saint John, New Brunswick).

  Near Repeat Space-Time Patterns of Canadian Crime · 2017 · DOI

• These findings imply that (a) changes in weather patterns modify peoples’ routine activities and, in turn, influence when crime is committed; (b) universal crime seasonality patterns should not be assumed across all cities; and (c) crime seasonality should be studied at a disaggregate or crime-specific level.

  Crime Seasonality across Multiple Jurisdictions in British Columbia, Canada · 2017 · DOI

Most-cited papers in Crime Patterns and Interventions

• Structural Covariates of Homicide Rates: Are There Any Invariances Across Time and Social Space? · American Journal of Sociology · 1990 · 857 citations
• Criminal Deterrence Research at the Outset of the Twenty-First Century · Crime and Justice · 1998 · 562 citations
• Work as a Turning Point in the Life Course of Criminals: A Duration Model of Age, Employment, and Recidivism · American Sociological Review · 2000 · 475 citations
• PROBLEM‐ORIENTED POLICING IN VIOLENT CRIME PLACES: A RANDOMIZED CONTROLLED EXPERIMENT* · Criminology · 1999 · 419 citations
• Crime prevention through environmental design (CPTED): a review and modern bibliography · Property Management · 2005 · 415 citations
• Social Osmosis and Patterns of Crime · Journal of Political Economy · 1991 · 407 citations
• Impact of social distancing during COVID-19 pandemic on crime in Los Angeles and Indianapolis · Journal of Criminal Justice · 2020 · 344 citations
• THE PHILADELPHIA FOOT PATROL EXPERIMENT: A RANDOMIZED CONTROLLED TRIAL OF POLICE PATROL EFFECTIVENESS IN VIOLENT CRIME HOTSPOTS* · Criminology · 2011 · 320 citations
• Fear of Crime and Constrained Behavior Specifying and Estimating a Reciprocal Effects Model · Social Forces · 1988 · 310 citations
• Childhood Risk Factors for Adolescent Gang Membership: Results from the Seattle Social Development Project · Journal of Research in Crime and Delinquency · 1999 · 271 citations

Most recent work

• Focused deterrence can reduce crime: A systematic review of randomized controlled trials and quasi‐experiments · Criminology & Public Policy · 2026
• Exploring the Impact of Socio-Economic Disparities on Crime Rates in Urban India · Trends in Finance and Economics · 2026
• The Changing Nature of Homicide in the Twin Cities: An Interrupted Time Series Analysis of the Police Murder of George Floyd · Journal of Quantitative Criminology · 2026
• An empirical examination of the group hazard hypothesis: The role of formal and informal detection in group offending · European Journal of Criminology · 2026
• Criminal network analysis and the study of organized crime* · Global Crime · 2026
• Employer attitudes towards hiring people with a criminal record: A review of empirical methodologies · International journal of law, crime and justice · 2026
• The role of urban environment maintenance in crime prevention: a machine learning analysis of street crime in Manhattan · Open House International · 2026
• Crime Rate Prediction Using Machine Learning · International Journal of Creative and Open Research in Engineering and Management · 2026
• Smart Policing with Reinforcement Learning: A Hybrid Framework for Crime Forecasting and Resource Optimization · INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT · 2026
• Crime Prediction and Analysis using Machine Learning · International Journal of Science, Strategic Management and Technology · 2026

Related topics in Social Sciences