Crime Data Research Articles

Towards spatio-temporal crime events prediction

The importance of early prediction in reducing the impact of crime cannot be overstated. Machine learning algorithms have proven to be effective in this regard, but their inability to capture key features automatically can be a hindrance. To overcome this challenge, we propose a deep neural network model that is capable of extracting salient features automatically for predicting crime categories using real-world crime data sourced from the Chicago open data portal. To ensure the robustness of our proposed model, we carried out an extensive exploratory data analysis to determine the impact of socioeconomic indicators on crime occurrences. Additionally, we implemented a data upsampling technique to handle class imbalance issues, and we leveraged hyperparameter optimization algorithms to fine-tune the model. The results of our study were impressive. Our proposed model outperformed the baseline model and other algorithms, with an average improvement of 6% in macro F1 score. This suggests that our model is highly effective, if not superior, in predicting crime categories accurately. Overall, our study provides a solid framework for using deep neural network models in crime prediction, while highlighting the importance of automatic feature extraction in enhancing the accuracy of predictions. By reducing the impact of crime through early prediction, we can help to create a safer and more secure society.

Crime risk prediction incorporating geographical spatiotemporal dependency into machine learning models

The spatiotemporal distribution of crime is closely related to the environment, exhibiting a typical characteristic of “spatiotemporal autocorrelation”. However, most of the existing machine learning-based crime prediction methods have difficulty in simulate the spatiotemporal dependence of crime. In this study, we mitigate the spatiotemporal dependence embedded in crime data by introducing a spatiotemporal lag variable. To verify the feasibility of the proposed methods, four machine learning methods were used to determine whether considering spatiotemporal dependency could improve model prediction accuracy and explore the impact of various factors (i.e., environmental factors and demographical factors) on crime risk intensity in different locations using crime data collected from June 2014 to May 2018 in Dallas. The results indicated the following: (1) incorporating spatiotemporal lag variables can effectively improve the prediction accuracy of machine learning models; (2) variables predicting crime are highly nonlinear over time and space, and tree-based nonlinear models greatly outperform linear models in predicting crime; and (3) interpretable machine learning models can reveal the unique contribution of each variable to researchers and practitioners. These findings contribute to our understanding of the mechanism of crime occurrence and may guide the development of crime prevention strategies.

Exploring the impact of measurement error in police recorded crime rates through sensitivity analysis

It is well known that police recorded crime data is susceptible to substantial measurement error. However, despite its limitations, police data is widely used in regression models exploring the causes and effects of crime, which can lead to different types of bias. Here, we introduce a new R package (‘rcme’: Recounting Crime with Measurement Error) that can be used to facilitate sensitivity assessments of the impact of measurement error in analyses using police recorded crime rates across a wide range of settings. To demonstrate the potential of such sensitivity analysis, we explore the robustness of the effect of collective efficacy on criminal damage across Greater London’s neighbourhoods. We show how the crime reduction effect attributed to collective efficacy appears robust, even when most criminal damage incidents are not recorded by the police, and if we accept that under-recording rates are moderately affected by collective efficacy.

The Effect of Land Cover Protection on Preventing Crime: The Case Study of the State of New York, USA

The settlements’ traditional character and architectural texture will strengthen people’s sense of belonging and prevent migration, while local urban identity will be the most significant impediment to crime. The study investigates the relationship between the environmental and criminal conditions in New York, the world’s most diverse demographic. In New York, not every county has the same settlement or vegetation homogeneity level. To avoid the immigrant factor, the location where the offenders committed the crime was considered; crime data were statistically analyzed to assess the vegetation-crime relationship. With an interdisciplinary approach, the research also draws attention to the mean-world syndrome regarding crime and urbanization in the current cultural environment in the context of the media-violence relationship. Under the crime index category, the crimes committed were divided into major groups, and statistical analyses were performed on all the groups. The study also touches on proactive policing used in the United States. According to our findings, the keyword of sustainable development, equity, has been successfully realized in the United States. As a result, the classic scenario of income distribution disparities in the state driving citizens to crime does not apply to New York.

Modelling underreported spatio-temporal crime events.

Crime observations are one of the principal inputs used by governments for designing citizens' security strategies. However, crime measurements are obscured by underreporting biases, resulting in the so-called "dark figure of crime". This work studies the possibility of recovering "true" crime and underreported incident rates over time using sequentially available daily data. For this, a novel underreporting model of spatiotemporal events based on the combinatorial multi-armed bandit framework was proposed. Through extensive simulations, the proposed methodology was validated for identifying the fundamental parameters of the proposed model: the "true" rates of incidence and underreporting of events. Once the proposed model was validated, crime data from a large city, Bogotá (Colombia), was used to estimate the "true" crime and underreporting rates. Our results suggest that this methodology could be used to rapidly estimate the underreporting rates of spatiotemporal events, which is a critical problem in public policy design.

MACHINE LEARNING FOR THE STUDY AND PREDICTION OF CRIMINAL DATA

The goal of the Crime Data Analysis Project is to analyze and interpret enormous amounts of crime-related data in order to find insightful patterns and trends. In order to improve crime prevention methods, maximize resource allocation, and foster safer settings, this project aims to offer law enforcement agencies, lawmakers, and communities relevant information. Gather pertinent crime data from a variety of sources, including law enforcement organizations, public records, and emergency services, while assuring data accuracy and consistency. Keywords: Criminology, Crime Analysis, Crime Prediction.

Perceptions on jurisdiction risk: a cross-country analysis

PurposeThe purpose of this article is to make a contribution to the existing knowledge by using the unique cross-jurisdiction data drawn from the FCA’s REP-CRIM submissions to explore dynamics behind firms’ perceptions on financial crime. Capturing firm’s sentiment is notoriously challenging, and any relevant regulatory data is usually not available in the public domain. A recent exception is the UK Financial Conduct Authority’s (FCA’s) financial crime data return (REP-CRIM) submissions which include the cross-country regulatory data on the UK financial institutions’ perceptions of jurisdiction risk. Despite a broad literature with respect to financial crime, there exists an important gap in the existing knowledge with respect to factors that are associated with the perceptions of firms with respect to jurisdiction risk, which this article aims to close.Design/methodology/approachUsing cross-country regulatory data on the UK financial institutions’ perceptions of jurisdiction risk, this study empirically determines that perceptions of jurisdiction risk is significantly and positively associated with anti-money laundering and countering the financing of terrorism (AML/CFT) framework, as well as with tax burden on business and institutional and legal risk in the case of 165 jurisdictions.FindingsThe findings lend support to the proposition that unsystematic efforts and too much publicity may ascertain the high-risk image of a jurisdiction, deterring cross-border business. Policy implications that emerge from the study also add to the case for strengthening institutional and legal frameworks, as well as relieving the tax burden on doing business.Research limitations/implicationsFindings of the present study should be interpreted with caution, as the dependent variable used in the present study reflects UK firms’ perceptions of jurisdiction risk, which may depend on various factors such as different risk appetites and the countries in which firms carry out business, and not necessarily the actual level of risks based on financial crime statistics. For example, a jurisdiction which may indeed be considered high risk, would not necessarily be ranking high on the FCA’s list of UK firms’ jurisdiction risk perceptions due to few firms operating in that particular country. As a result, the list could differ from the Financial Action Task Force’s black and grey lists. Findings based on the regulatory data on the UK financial institutions’ perceptions of jurisdiction risk should be considered preliminary in nature, given that they are based on a single year cross sectional data. As global and country-level AML/CFT efforts continue to intensify and as more regulatory data becomes publicly available, it would be imperative to bring further empirical evidence to bear on the question of whether financial crime perceptions are likely to be more pronounced for jurisdictions where AML/CFT efforts are more intensified. Likewise, from a policy standpoint, it would be equally important to explore further the role that institutional and legal risk, as well as tax burden on businesses, play in shaping firms’ perceptions of jurisdiction risk.Practical implicationsFindings lend support to the proposition that unsystematic efforts and too much publicity may ascertain the high-risk image of a jurisdiction, deterring cross-border business. Therefore, rather than waiting for more data to be made available by other financial regulators, which could lead to a more conclusive evidence in the future, on balance, the findings of this study add to the case for carefully designing and systematically implementing AML/CFT measures in a less publicized manner. Findings lend support to the theoretical postulation that disorderly efforts and undue publicity regarding AML/CFT efforts serve to ascertain the high-risk image of a jurisdiction, which could deter cross-border business and could be detrimental to how firms undertake due diligence. They also suggest that disorderly implementation of AML/CFT measures may hinder access to formal financial service and jeopardize authorities’ ability to trace the movement of funds, which may also add to negative perceptions of jurisdiction risk.Social implicationsFindings are in line with the theoretical expectations that perceptions of jurisdiction risk would be expected to be higher in countries with inadequate disclosure rules, lax regulation and opacity jurisdiction. Likewise, results are aligned with the expectations that tax burden on business would be expected to be in a positive relationship with jurisdiction risk, as it would increase the likelihood of tax evasion, which incentivizes financial crime. Therefore, policy implications that emerge from the study also add to the case for strengthening institutional and legal frameworks and relieving the tax burden on doing business as part of efforts to improve the international image of jurisdictions with respect to financial crime risks.Originality/valueUsing the cross-country regulatory data on the UK financial institutions’ perceptions of jurisdiction risk, this study has empirically determined that perceptions of jurisdiction risk is significantly and positively associated with AML/CFT framework, as well as with tax burden on business and institutional and legal risk. These findings have implications from a policy standpoint.

The (mis)measure of race and ethnicity in crime data

The measurement of crime in the United States is one that has constantly evolved since national efforts began in 1930. However, the measurements of victim and offender characteristics, specifically race and ethnicity, have not developed at the same pace, nor as rapidly for crime data as it has for other fields. This paper examines two primary criminal justice data sources, the Uniform Crime Reports (UCR) and the National Incident-Based Reporting System (NIBRS) and compares them for context to the United States Census. Analysis shows that the Census has continually expanded racial and ethnic categories while the crime data instruments have mostly refined them only by name. The paper concludes with suggestions for revision of the crime data collection instruments, specifically NIBRS, and a discussion of research and policy implications.

Information fusion in crime event analysis: A decade survey on data, features and models

Crime event analysis (CEA) has become increasingly important in assisting humans in preventing future crimes. A fundamental challenge in the research community lies in the dynamics of criminal intents. Offenders’ criminal intents may evolve due to their surroundings, making it difficult for machine learning models to capture them from limited existing information, leading to model uncertainty. As a result, there has been a surge of works exploiting various information related to the evolution of criminal intents, thus enhancing prediction accuracy. This work conducts a comprehensive survey of the past decade (2013–2023) of CEA methods from the perspective of information fusion. We first investigate the categories of crime data and briefly introduce existing CEA tasks as well as evaluation metrics. Then, fusion is systematically reviewed from the bases of multi-modal data, features and machine learning models in terms of different categories of crime data. Finally, we conclude by highlighting some limitations and identifying several future research directions.

Municipal bankruptcies and crime

While the importance of municipal bonds for the provision of public services is well-documented, the consequences of municipal bankruptcies remain understudied. We contribute to this literature by studying the effects of municipal bankruptcies on crime. Using the staggered difference-in-differences approach and agency-level crime data, we find that violent and property crime rates rise after successful (approved by the bankruptcy courts) Chapter 9 bankruptcy filings. The association between successful municipal bankruptcy and crime remains robust to alternative estimation methods, including a novel empirical technique that allows for the treatment effect heterogeneity and dynamics, and several robustness checks and falsification exercises. Our exploratory evidence suggests that the rise in crime rates is due to lower public safety expenditures. Our findings highlight the importance of efficient financial management for local government entities and the need for particular attention to law and order in the local jurisdiction going through bankruptcy.

The Use of K-Nearest Neighbour Approach for Crime Rate Prediction in Public Universities: Case Study Adekunle Ajasin University, Akungba–Akoko (AAUA), Ondo State, Nigeria

The prediction of crime rates has been a challenging problem for law enforcement agencies and policymakers. The main challenge in developing a predictive model for crime rate is the unavailability and quality of data. Crime data is often incomplete, inconsistent, and biased, which has always affected the accuracy of the scholars’ designed model. For example, the use of traditional methods such as manual crime analysis and statistical analysis are often limited by the volume and complexity of the data. The need for a more accurate and efficient predictive model that can handle large and complex datasets is increasing. This research developed a predictive model using the KNN algorithm that is into accurate prediction of crime rates and helps policymakers to make better decisions. The KNN algorithm was implemented using Python programming language with scikit-learn libraries. The primary dataset was collected from security units of Adekunle Ajasin University, Akungba Akoko and Akungba police station. The Algorithm was implemented on a Laboratory Local Area Network at the Information and Communication Technology Application Centre (ICTAC), Adekunle Ajasin University, Akungba Akoko. The performance evaluation was based on precision, recall, and F1 score. The results show that the KNN algorithm can accurately predict crime rates with an average accuracy of 88%. The model demonstrates that the KNN algorithm can be a useful tool for predicting crime rates. In addition, the model is hoped to enhance crime prevention strategies, reduce crime rates, and ultimately improve the quality of life in university communities. Keywords: Crime, Crime Rate, Prediction, K-Nearest Neighbour, Public Universities. Akinwumi, D.A. (2023): The Use of K-Nearest Neighbour Approach for Crime Rate Prediction in Public Universities: Case Study Adekunle Ajasin University, Akungba–Akoko (AAUA), Ondo State, Nigeria. Journal of Advances in Mathematical & Computational Science. Vol. 11, No. 2. Pp 11-24. dx.doi.org/10.22624/AIMS/MATHS/V11N2P1. Available online at www.isteams.net/mathematics-computationaljournal.

WEATHER CHANGES INFLUENCE THE CRIME ACTIVITY TOWARDS THE QUALITY OF LIFE

The controversy over weather changes will lead to more stress, less social control, reduced social support, criminal tendencies, more possibilities for crime, and societal conflict. The factors driving criminal activity have received much attention, and voluminous literature has related the incidence of changing weather changes to criminal activity. However, weather variables can affect the environment and human behaviour differently. For example, past scholars found that homicidal crime has a connection with the weather change. However, although studies look at the relationship between crime and weather change, the source of the data is debatable. In addition, empirical studies showing the relationship between property crime and weather change are challenging to obtain because none can be found. Therefore, this paper aims to identify the pattern of criminal activity based on weather changes to address this situation. In this paper, the research methodology uses quantitative data. The crime data were gathered from the Royal Malaysia Police (RMP), and the weather data was gathered from Meteorological Department for Selangor, Kuala Lumpur, and Johor from 2011 to 2020. The data were analysed using SPSS statistical and illustrated in the bar graph method to identify the pattern and relationship between crime and weather. This study found that non-aggressive and property crime is less influenced by the high temperature but influenced by the ambient temperature and humidity. Future research on the relationship between crime and weather towards the quality of life of people must be theoretically grounded and use measures of crime and weather that are justified by the quality-of-life index theory.

K-Means and AHC Methods for Classifying Crime Victims by Indonesian Provinces: A Comparative Analysis

Crime is a common phenomenon that often occurs in society and has a negative impact both individually and collectively. Gaining a deeper understanding of crime can help us tackle the problem more efficiently. In an era that is increasingly complex and globally connected as it is now, crime has undergone significant developments and changes. Crime remains a serious threat to our security, integrity, and well-being. Some common types of crime include theft, robbery, fraud, physical abuse, and murder. Crime can happen anytime and anywhere. To tackle crime, data mining techniques can be used to analyze the surrounding situation and gain new knowledge. One approach is to classify provinces based on crime data from previous years so that crime-prone areas can be identified and security measures can be increased. In this study, two grouping methods were used, namely K-Means and AHC using the complete linkage mode. There are 34 provinces in Indonesia which are grouped based on the number of victims of crime from 2019 to 2021. The grouping results using the K-Means method yield two groups with 17 provinces each. However, using the AHC complete linkage method, there is a difference in the number of provinces between cluster 0 and cluster 1 compared to the K-Means results. In addition, there are differences in the location of the province in the cluster between the two methods. In the K-Means method, provincial data is located in cluster 0, while in the AHC method, the province's data is in cluster 1. Thus, this study provides insight into crime in Indonesia and provides information about the grouping of provinces based on crime rates using the K-Means method. Means and AHC

Cross-jurisdictional financial crime risks: what can we learn from the UK regulatory data?

PurposeFinancial crime presents a serious threat to the stability and integrity of the global financial system. To combat illicit financial activities, regulatory bodies worldwide have implemented various measures, including the requirement for financial institutions to assess the financial crime risks they are exposed to in the jurisdictions they operate in. These risks include inadequate anti-money laundering and countering the financing of terrorism frameworks and other financial crime risks that have significant strategic implications for firms’ geographical footprints and customer risk classifications. This paper aims to make a contribution to the literature by undertaking a cross-country analysis of 158 countries to shed light on what drives perceived jurisdiction risk of the UK financial services firms.Design/methodology/approachCapturing firms’ perceptions of financial crime risk requires significant data collection efforts, including surveys and interviews with key personnel. This can be highly resource-intensive and may require access to sensitive information that firms may be reluctant to share. Furthermore, the dynamic nature of financial crime risks means that perceptions can change rapidly in response to changes in the regulatory and geopolitical landscape. As a result, capturing and monitoring firms’ perceptions of financial crime risks requires ongoing monitoring and analysis. Capturing firms’ perceptions of financial crime risks at a cross-jurisdictional level is a particularly complex and challenging task that requires careful consideration of a range of factors. As a result of data limitations, empirical investigation of the factors underlying the firms’ perceptions of jurisdiction risk is in its infancy. This paper uses regulatory financial crime data from the UK in a multivariate regression analysis, following a general-to-specific approach where any redundant variables were removed from the general model sequentially.FindingsResults suggest that perceived jurisdiction risk is significantly and positively associated with evasion of tax and regulations, while it is significantly and negatively associated with political stability and regulatory stringency. These have important implications for home and host supervisors with respect to the factors that drive perceived jurisdiction risks and the evaluation of the nature of inherent financial crime risks within regulated firms. The findings confirm the critical role of the shadow economy, political stability and regulatory rigor in shaping jurisdiction risk perceptions. From a policy standpoint, the findings support the case for taking prompt policy action to identify, prioritize and implement specific and targeted measures with respect to the shadow economy, political stability and rigor of regulations to improve international firms’ perceptions of jurisdiction risk.Originality/valueWhile there exists different measures of financial crime risk, it is notoriously challenging to capture firms’ perceptions of it, particularly at a cross-jurisdiction level. This is because financial crime risks can vary significantly across different jurisdictions due to differences in legal and regulatory frameworks, cultural norms and levels of economic development. This makes it difficult for firms to compare and evaluate the financial crime risks they face in different jurisdictions. Besides, firms’ perceptions of financial crime risks can be influenced by a range of subjective factors, including personal experiences, media coverage and hearsay. These perceptions may not always align with objective risk assessments, which are based on more systematic and empirical methods of risk measurement. This paper contributes to the existing literature by undertaking a cross-country analysis drawing on a unique set of UK regulatory financial crime data, which is based on a total of 1,900 annual financial crime data regulatory return (REP-CRIM) submissions to the UK’s Financial Conduct Authority.

Measuring the impact of the state of emergency on crime trends in Japan: a panel data analysis

Purpose/BackgroundCity-specific temporal analysis has been commonly used to investigate the impact of COVID-19-related behavioural regulation policies on crime. However, these previous studies fail to consider differences in the intensity of intervention among cities and the impact of these behavioural regulation policies on crime trends nationwide. This study performs panel data analyses to examine how the declaration of a state of emergency (SoE) affected ambient population and crime in Japan, taking advantage of the fact that the SoE was implemented at different times in different prefectures.MethodsThe current study uses two sets of panel data of 47 prefectures for 22 weeks from February to July 2020: (1) the data on ambient population in five types of locations provided by the Google Mobility Reports, and (2) official crime data of six types of crime: residential burglary, commercial burglary, theft of/from vehicle, bicycle theft, sexual assault, and violence and injury. Firstly, an ordinary least squares regression analysis was performed to examine the impact of the SoE on the ambient population. Then a negative binomial model with fixed effects was adopted to examine the effect of the ambient population on the crime trends.FindingsThe SoE declaration was found to increase the ambient population in ‘residential’, and decrease that in other settings including ‘workplaces’, ‘transit stations’, and ‘retail and recreation’ in targeted prefectures. Spill-over effects of the SoE were observed on the ambient population of non-SoE prefectures. The ambient population have impacted five out of the six types of crime examined, except for sexual assault. After controlling for the ambient population, we observed an increase in commercial burglary and theft of/from the vehicle in all prefectures during the SoE weeks, compared to the weeks when the SoE was not declared.ConclusionsThe declaration of the SoE during the COVID-19 pandemic changed the ambient population in the SoE-prefectures, resulting the changes in crime levels as well. In addition, the implementation of the SoE in specific prefectures was found to have a contextual impact on national-level crime trends. Furthermore, the implementation of the SoE caused changes in some crime types that could not be explained by the changes in the ambient population, suggesting that the implementation of the SoE affected offenders’ decision-making. It is also worth noting that the changes in ambient population and crime trends during the pandemic were observed in Japan where the behavioural regulation policy without law enforcement was introduced.

The crime and the place: Robbery in the night‐time economy

AbstractThe night‐time economy (NTE) provides many opportunities for crime as there is an abundance of potential victims who are often intoxicated and clustered in a small geographical area. Previous research on NTE violence has primarily focused on assault. However, other offences are also common, such as robbery. This study focused on NTE‐related robbery using police recorded crime data relating to 1624 personal robberies (including attempts) from West Midlands Police, UK. The data was binary coded to identify and compare offence characteristics. Robbery offences in the NTE showed unique characteristics compared to robberies unrelated to this context. In particular, there were differences in alcohol, use of violence, injuries, approach style and crime locations. The findings of the current research align with theoretical frameworks from environmental criminology (e.g. crime generators and attractors), have implications for crime prevention and investigations and can feed into developing policing strategies that take into account the background context for offending.

Crime in the era of COVID‐19: Evidence from England

AbstractThis paper examines the effects of COVID‐19‐induced lockdowns on recorded crime in England. The enforcement of lockdowns at both the national and local levels allows unveiling the impact on criminal activities by type of shutdown policy. Using official crime data across the universe of local authorities, we find that unlike local lockdowns national lockdowns significantly change the shape of recorded criminal activity, with the first nationally‐imposed lockdown having the strongest impact. Findings also reveal that police operations play a prominent role in explaining changes in reported crimes. Back‐of‐the‐envelope calculations suggest that lockdowns reduced the economic costs of crime by £4.2 billion.

Experimental Analysis of the Machine Learning Algorithms for Crime Web Page Classification

The evolution of web 3.0 results in storing data in plenty in the web which are huge as that of hard to search and its specific classification and this sort of issue attracts research attention. This task specifically addresses the problem of classifying web documents covering the crime data in it. Manual classification of crime related documents from the available ones is unimaginable and also observed that the existing models have not produced considerable results. It is obvious that machine learning approaches performs well in handling large amount of data. Hence, the problem of crime related web document classification is handled with an automated web page classification technique based on word similarity. The baseline is to identify the best performing feature selection approaches and combine them with machine learning models for isolating the best performing combination. The feature selection methods used are Jaccard Similarity, Information Gain, Gain Ratio and Chi Square Analysis and each of them is individually combined with four machine learning classifiers such as Support Vector Machine, K-Nearest Neighbor, J48 and Naïve Bayes. The models are tested with the data set containing Web documents pertaining to Crime, Politics, Sports and Terrorism collected from different sources. The models have proved better performance when added with feature selection approaches than without them. When it comes for comparing different combinations of feature selection approaches and machine learning models, Support vector machine combined with Jaccard Similarity is found to perform well than the other approaches with an accuracy of 98.74%.

Blockchain-Based Approach for Tracking Global Criminals

Blockchain is the latest technology, which is now widely used in various fields due to its benefits such as higher tamper-proof security. Each transaction is stored in an immutable distributed ledger on each blockchain node. Criminal records are confidential and should be kept safe from hackers and intruders. Law enforcement agencies or investigative agencies can use these security recordings to analyze and detect criminal activity. Due to the lack of a global online system, Investigative agencies are having difficulty locating and researching the past of anonymous foreign criminal who is committing crimes abroad. This problem will only be solved if investigative agencies have a common global criminal database and applications to extract information. Investigation agencies, Copes, and other organizations can access global crime data using this technology on the distributed node of the blockchain in their nation. Authorities can instantly and efficiently access global criminal histories from local blockchain databases and identify the names and details of anonymous individuals in response to any foreigners suspected of engaging in unsocial activities in their country of origin. This will reduce the time and process of gathering information from the country where the suspect is located. The system would eliminate the possibility of falsification and tamper with criminal records. Key Words: Blockchain, light, Global Criminal, Criminal information

The effect of various social determinants of health on overall survival in the SCLC population.

8588 Background: Small cell lung cancer (SCLC) comprises 10-15% of all lung cancers yet little is known about the impact of social determinants of health (SDOH) on outcomes in patients with SCLC. This study aims to assess the impact of various SDOH on overall survival (OS) in patients diagnosed with SCLC at a single institution. Methods: We have established a large clinical/pathologic/genomic database of all patients diagnosed and treated for SCLC between 1998-2022 at our institution for retrospective study. Demographic data including sex, race, age, stage, address, and smoking status were collected. Poverty index was assessed by percent of school-aged children in poverty at the school district level based off 2021 census using the SAIPE tool and stratified as low (poverty index <=15.1) and high (poverty index >15.1). Water/Air carcinogens were obtained by zip code through EPA TRI Explorer database at year of SCLC diagnosis. Food desert designation was acquired through USDA low income and low access atlas at 0.5 miles for urban residences and 10 miles for rural residences respectively. Violent crime rates were obtained from the FBI Crime Data Explorer by year of diagnosis and police department jurisdiction. OS was computed from date of diagnosis to death or last known follow up date, and univariate Cox proportional hazards (PH) regression analysis was performed. Factors that were significant at significance level of 0.25 in univariate analysis were further evaluated using multivariable Cox PH model. Results: A total of 982 patients with SCLC were included. Univariate PH analysis demonstrated statistically significant improvement in OS with younger age (HR = 1.03, 95% CI 1.02-1.03, p <0.0001), female sex (HR = 0.8, 95% CI 0.7-0.92, p 0.001), low poverty index (HR = 1.12, 95% CI 0.98-1.12, p = 0.097), limited stage disease (HR = 3.12, 95% CI 2.66-3.66, p = <0.0001), and lack of air pollution exposure (HR = 1.09, 95% CI 0.95-1.24, p = 0.227). Multivariable PH analysis showed statistically significant OS improvement for those from lower poverty areas (HR = 1.22, 95% CI: 1.07- 1.4, p = 0.004) as well as in those without air pollution exposure (HR = 1.21, 95% CI: 1.05- 1.39, p = 0.008), controlling for age, sex, stage and smoking. Exposure to water pollution, violent crime rate, food desert designation and pack years smoked were not found to be independent predictors of OS. Conclusions: In this large detailed cohort of 982 patients with SCLC, there was a significantly lower OS for patients living in high poverty areas and for patients exposed to air pollutants. This effect was durable even after for controlling for age, sex, stage, and smoking status, demonstrating that these are independent risk factors associated with poorer OS. Our next research steps will be to examine the biologic basis for these disparities through genomic and transcriptomic analysis.