{"paper_id":"0dbb8bc9-8346-4375-a3c0-99f413123824","body_text":"The Effects of Emergency Homeless Shelters on Crime during the COVID-19 Pandemic | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article The Effects of Emergency Homeless Shelters on Crime during the COVID-19 Pandemic Brianna Camero This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7151989/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract This study examines the associations between emergency homeless shelters and crime in New York City (NYC) during the COVID-19 pandemic, when public health imperatives required rapid housing solutions for individuals experiencing homelessness. In response, the city leased commercial hotels to serve as temporary emergency shelters, raising public concerns about potential increases in neighborhood crime. Using data from the NYC Department of Homeless Services and publicly available crime data, this study assesses the effects of both traditional congregate shelters and hotel-based emergency shelters on crime across 59 community districts at six time points between 2020 and 2022. Crimes were categorized as total, violent, property, or disorder. Quasi-Poisson regression models with community and time fixed effects were used to estimate associations. The findings indicate that traditional shelters have no significant relationship with crime. Emergency hotel shelters, however, were associated with modest reductions in total and property crime, although these findings approached but did not reach conventional levels of statistical significance with p-values around 0.051. These results challenge common public perceptions linking homelessness interventions to increased crime and underscore the importance of shelter design and implementation context. While limitations include constraints in traditional shelter data variation and shelter size information, this study contributes to the evidence base informing public health and housing policy. This study suggests that hotel-based shelters may not only support vulnerable populations during public health crises but may also have neutral or even protective effects on neighborhood safety. Homelessness Shelters Crime New York City COVID-19 Introduction Homelessness has long been a pressing social issue in the United States (US), with approximately 400,000 people living in shelters and an estimated 200,000 individuals sleeping on the streets on any given night, although the latter number is particularly difficult to assess with certainty (Meyer et al., 2023 ). In New York City (NYC), homelessness had already reached record highs prior to the COVID-19 pandemic, but the onset of this public health crisis exacerbated the situation. Owing to a lack of access to private spaces, individuals experiencing homelessness were at significantly greater risk of COVID-19 transmission, hospitalization, and death (Coalition for the Homeless, 2020 ). In April 2020 alone, COVID-19-related deaths among individuals experiencing homelessness were 157% higher than the city’s average monthly homeless death toll for all causes in 2019, and the mortality rate among individuals in homeless shelters was 62% higher than that of NYC’s overall population (Coalition for the Homeless, 2020 ). In response, NYC implemented emergency measures, including the leasing of commercial hotels to serve as temporary homeless shelters. These hotel-based shelters allowed for greater physical distancing and offered private rooms without the requirements, providing a degree of autonomy and safety not typically available in traditional congregate shelter settings. Unlike conventional shelters, hotel-based shelters often have fewer behavioral requirements, less oversight, and greater privacy (Padgett et al., 2023 ). While these emergency shelters aimed to reduce viral transmission and support the well-being of vulnerable populations, their presence in residential and commercial neighborhoods also raised public concerns, particularly around crime. Public perceptions frequently link homelessness to crime, even though this association is often rooted more in stigma than in empirical evidence. Past national surveys have shown that many people perceive people experiencing homelessness as dangerous or unpredictable (Link et al., 1995 ; Tsai et al., 2017 ), reinforcing social stigma and shaping public discourse and policy. These perceptions are further reinforced by criminological theories such as broken windows theory (Wilson & Kelling, 1982 ), which suggests that visible signs of disorder, such as loitering, public intoxication, or vandalism, signal weakened informal social control, which can lead to further deviance. Under this framework, the presence of homeless shelters might be viewed as a form of social disorder that contributes to neighborhood decline and increased crime. Relatedly, routine activity theory (Cohen & Felson, 1979 ) provides another lens through which to understand possible links between homelessness and crime. According to this theory, crime is likely to occur when a motivated offender encounters a suitable target in the absence of a capable guardian. The rapid introduction of emergency shelters may disrupt local routines, increase foot traffic, or draw individuals into new public spaces, thereby altering the opportunity structure for certain types of crime. Moreover, because many individuals experiencing homelessness are at heightened risk of victimization, particularly from assault, robbery, and theft (Ellsworth, 2019 ), the increase in visible homelessness in certain areas may also reflect vulnerability rather than criminality. Empirical findings on the relationship between homelessness and crime, however, are mixed and context dependent. In Vancouver, Faraji and colleagues ( 2018 ) reported that the introduction of temporary winter shelters was associated with a 56% increase in property crime in the immediate area, primarily driven by theft from vehicles, vandalism, and other thefts. However, the rates of breaking and entering commercial properties actually decreased by 34% in the immediate proximity of these shelters, suggesting a complex and uneven effect. Similarly, a study in Los Angeles reported that higher levels of homelessness were significantly associated with increased property crime in affected neighborhoods (Ee & Zhang, 2022). Moreover, a longitudinal analysis in Finland linked homelessness to increased rates of violent crime, property crime, and public intoxication (Markowtiz, 2024). These findings underscore that the presence of shelters or unhoused individuals can influence crime patterns, but the mechanisms driving those patterns remain poorly understood and are likely moderated by broader social, environmental, and policy factors. At the individual level, however, there is evidence to suggest that certain characteristics associated with homelessness, particularly mental illness, substance use disorders, and survival behaviors, can increase the likelihood of involvement in criminal activity. Approximately one-third of individuals experiencing homelessness suffer from major mental illnesses, excluding substance-use disorders (Mechanic et al., 2014 ). These conditions can hinder access to stable housing and employment and, when left untreated, may contribute to behaviors that bring individuals into contact with the criminal justice system (Calsyn & Winter, 2002 ). Research shows that individuals suffering from homelessness in conjunction with serious mental illness are more likely to be involved in aggressive or violent incidents (Link et al., 1999 ; Silver, 2002 ). Additionally, survival-related offenses, such as shoplifting, trespassing, loitering, and public urination—sometimes referred to as “quality of life” crimes or disorder offenses—are more commonly committed by unhoused individuals (Fischer et al., 2008 ; McGuire & Rosenbeck, 2004). These actions, while often nonviolent, contribute to the broader perception of individuals experiencing homelessness as criminal actors, although these behaviors often arise from necessity rather than intent to harm. Individuals experiencing homelessness are also disproportionately subject to arrest and law enforcement scrutiny. Snow and colleagues ( 1989 ) documented elevated arrest rates among individuals who are homeless, especially those with nonviolent offenses. Similarly, Ferguson and colleagues ( 2012 ) reported that compared with their housed counterparts, young adults who were homeless were significantly more likely to engage in property crime and drug usage. However, the high rates of criminal justice involvement among unhoused individuals do not necessarily indicate a direct causal link between homelessness and increased criminal behavior. Rather, they highlight the criminalization of homelessness and broader systemic failure in providing adequate housing, mental health care, and social services. Importantly, the literature offers limited insight into the specific effects of emergency homeless shelters, particularly hotel-based models, on neighborhood-level crime. Faraji and colleagues’ ( 2018 ) study remains one of the only peer-reviewed analyses to directly examine the short-term impacts of temporary shelters on crime, but it focused on congregate-style winter shelters and not the more private, decentralized hotel-based model implemented during the COVID-19 pandemic in NYC. Moreover, hotel-based shelters offered distinct benefits compared with traditional shelters: residents experienced improvements in physical and mental health, hygiene, safety, nutrition, and overall well-being (Padgette et al., 2023). These enhanced living conditions may serve as protective factors against both victimization and criminal involvement. The current study contributes to the growing but still nascent literature by examining how emergency hotel shelters, implemented as public health interventions during the COVID-19 pandemic, may have affected crime at the community district level in NYC. In doing so, it provides a unique opportunity to differentiate the impacts of congregate shelters and hotel-based shelters and to evaluate whether the design and conditions of temporary housing influence crime outcomes. By shedding light on this relationship, this study aims to inform both homelessness policy and public safety strategies in post-pandemic urban environments. Methods Measures To conduct this study, I use data obtained from the Department of Homeless Services in New York City through a freedom of information law request. These data are broken down at the community district level (n = 59) according to the number of standard homeless shelters and the number of commercial hotels used as temporary emergency homeless shelters (referred to as emergency hotels in the results tables) to combat the spread of COVID-19 at six different time points: September 2020, January 2021, March 2021, April 2021, October 2021, and February 2022. With 59 community districts and six time points, the total sample size was 354. To assess the impact of these shelters on crime, I obtained data from the New York City open data website, which was geocoded into community districts using the geographic coordinates included in the crime data and a community district shapefile retrieved from New York City’s Department of City Planning website. This geocoding process was completed via the “sf” package (Pebesma & Bivand, 2023; Pebsema, 2018), whereas the analysis was completed via the “fixest” package (Berge, 2018 ), both of which are in RStudio (R Core Team, 2022 ). I then aggregated the crime data by month into four categories: total crime, violent crime, property crime, and disorder crime. Total crime is a broad measure of all crime incidents that occur within each community district, including violent, property, and disorder crime. Violent crimes are crime incidents where a victim is harmed by or threatened with violence, such as robbery, aggravated assault, homicide, and rape. Property crimes are crime incidents where a victim’s property is stolen, damaged, or destroyed without physical harm to the victim; these include burglary, larceny, and vandalism. Finally, disorder crimes are relatively minor physical or social crime incidents where they may have caused a disruption to the community, such as noise disturbances, prostitution, loitering, and solicitation. Table 1 presents the correlation matrix for the variables. This was largely done to ensure that emergency hotels and standard shelters were not highly related to each other and that similar phenomena were measured. This was not the case, with the correlation being approximately − 0.01, indicating that they are independent measures. Furthermore, standard shelters are moderately correlated with each of the crime types, whereas emergency hotels are weakly correlated, suggesting a different relationship for each type of shelter with crime. This weak correlation for emergency hotels, however, could be an artifact of low variability within the emergency hotel variable. Table 2 thus shows the descriptive statistics of each variable, with emergency hotels having a mean of 0.75 and a standard deviation of 1.73, suggesting fairly high variability. The standard shelters variable was notably very different from the emergency hotels variable, with a mean of 4.918 and a standard deviation of 5.482. Table 1 Correlation Matrix Emergency Hotels Standard Shelters Emergency Hotels 1.0000 -0.0132 Standard Shelters -0.0132 1.0000 Total Crime 0.1490 0.4946 Violent Crime 0.0165 0.6695 Property Crime 0.2707 0.1064 Disorder Crime 0.067 0.5374 Table 2 Descriptive Statistics Table Emergency Hotels Standard Shelters Total Crime Violent Crime Property Crime Disorder Crime Mean 0.7542 4.9180 599.5000 121.0600 234.7000 191.9000 Median 0.0000 3.0000 552.5000 107.5000 205.0000 176.5000 Standard Deviation 1.7309 5.4815 230.5848 63.6986 107.8252 73.3715 Minimum 0.0000 0.0000 35.0000 5.0000 26.0000 3.0000 Maximum 11.0000 22.0000 1378.0000 323.0000 786.0000 418.0000 Analytic Strategy Because the dependent variables of crime are count variables, I initially set out to utilize a Poisson or negative binomial regression model. However, for both of these model types, the data did not pass the Kolmogorov-Smirnov test. With p-values of less than 0.001 for each crime type, the crime data significantly differ from the Poisson and negative binomial distributions. Because of this, I opted for a quasi-Poisson model after ensuring that the assumption of the variance being a linear function of the mean was met. The standard errors of these models are also clustered at the community district level due to the aggregation of the data. Furthermore, I use time and place fixed effects in the quasi-Poisson model to control for the differences between community districts and months that cannot be controlled for due to a lack of data availability, such as the amount of physical disorder in a place, collective efficacy within community districts, or the amount of foot traffic in a community district in a given month. Thus, by adding month and community district fixed effects, I can ensure a more accurate estimate of the actual relationship between homeless shelter types and crime. Results Table 3 presents the results of all four regressions of crime type on emergency hotels and standard shelters. However, while I present McFadden’s R 2 , this number is essentially uninterpretable because the fixed effects are included within the model as well. Thus, the amount of variance in crime accounted for by only the homeless shelters is unknown. Within each of the crime models, while standard shelters have consistently positive coefficients, they are insignificant in each model, with p-values ranging from 0.21 to 0.73. Thus, standard shelters appear to have no significant effect on crime. However, this may be due to the inability to isolate the effects of standard shelters, since it is unlikely that there would be any major change in the number of standard shelters in a community district between time points. For emergency hotels, however, the coefficients for total crime and property crime approach significance at p = 0.0506 and 0.051, respectively. The effect of emergency hotels is also consistently negative, suggesting that the presence of more emergency hotels may result in a crime reduction effect, with a percentage decrease of approximately 2.7% for both total crime and property crime. For violent and disorder crime, the coefficients suggest 2.7% and 1.4% decreases in crime, respectively. However, the p-values are 0.12 and 0.32, respectively, suggesting that this effect is insignificant. Table 3 Quasi-Poisson Models with Fixed Effects Total Crime Violent Crime Property Crime Disorder Crime Emergency Hotels -0.0275 -0.0217 -0.0275 -0.0140 Standard Error 0.0141 0.0141 0.0141 0.0141 p-value 0.0506 0.1239 0.0510 0.3214 IRR 0.9728 0.9786 0.9729 0.9861 Percent Change -2.7160 -2.1440 -2.7120 -1.3870 Standard Shelters 0.0120 0.0047 0.0172 0.0131 Standard Error 0.0138 0.0138 0.0138 0.0138 p-value 0.3858 0.7347 0.2120 0.3402 IRR 1.0120 1.0047 1.0173 1.0132 Percent Change 1.2010 0.4680 1.7340 1.3220 Number of Observations 354 354 354 354 Number of Districts 59 59 59 59 Number of Months 6 6 6 6 McFadden's R 2 0.9152 0.9387 0.8567 0.9039 Discussion This study examined the relationships between homeless shelters, specifically emergency hotel shelters introduced during the COVID-19 pandemic, and crime rates across NYC community districts. In contrast to long-standing public fears that shelters may increase neighborhood crime, the findings from this analysis suggest a more nuanced and possible reassuring conclusion. While standard congregate shelters were not significantly associated with crime rates, the presence of emergency hotel shelters was associated with a modest decrease in both total and property crimes, with p-values approaching statistical significance at the 0.05 threshold. These findings provide early evidence that emergency hotel shelters designed to improve privacy, autonomy, and health outcomes during a public health crisis may not only avoid increasing crime but may also be linked to slight reductions in certain types of crime at the neighborhood level. These results align with a growing body of literature suggesting that fears of rising crime tied to homelessness and shelter expansion may be overstated or rooted more in stigma than empirical observation (Link et al., 1995 ; Tsai et al., 2017 ). Rather than supporting the predictions of broken windows theory, which would anticipate a rise in the presence of disorder-related cues such as shelters, this study instead suggests that certain forms of temporary housing, especially those that promote stability, privacy, and health, may mitigate crime risks or even contribute to safer community conditions. The modest reduction in property and total crime associated with emergency hotel shelters could stem from several mechanisms. These shelters reduce crowding, minimize exposure to high-risk street environments, and are generally accompanied by support services (Padgett et al., 2023 ). By providing safer and more stable accommodations, these shelters may have reduced residents’ involvement in survival-related offenses (e.g., theft) while also lowering their risk of victimization. Moreover, the physical structure and operation of hotel shelters—often discrete, monitored, and embedded in commercial zones—may have altered the opportunity structures for crime in ways not captured by congregate shelter models. However, the interpretation of these results requires caution because of several methodological limitations. First, the analysis was limited by a lack of variation in standard shelters over time, making it difficult to isolate their effects. Unlike emergency hotel shelters, which were introduced and phased out during the study period, standard shelters remained relatively constant within each community district. As a result, the observed null effects may reflect statistical constraints rather than the true absence of an association. Second, the study is limited by its sample size. With only 59 community districts and six time points, the total sample size of 354, while sufficient for the analysis conducted, may lack the statistical power needed to detect smaller or more nuanced effects. Relatedly, while the quasi-Poisson model appropriately addresses overdispersion in the crime count data, this model does not allow for traditional goodness-of-fit measures, such as likelihood-based R-squared values, limiting interpretability of model fit. Third, the data do not account for the size or capacity of each shelter, which is an important omission, as larger shelters may have different impacts on community dynamics than smaller shelters. Nor was it possible to account for differences in shelter management, support services, or resident composition, all of which may influence outcomes in meaningful ways. Additionally, unmeasured factors such as foot traffic, nearby businesses, and policing practices could not be directly controlled, although the inclusion of fixed effects for months and community districts mitigated some of these concerns. Despite these limitations, this study provides valuable insight into the ongoing policy debate over the placement and design of shelters. The evidence that hotel-based emergency shelters did not lead to crime increases and may have had a small protective effect challenges common assumptions in both public discourse and urban policymaking. As NYC and other cities continue to explore models of supportive and traditional housing in the post-pandemic era, the results underscore the importance of not only whether shelters are provided but also how they are designed and implemented. Future research should investigate the long-term effects of both traditional congregate shelters and emergency hotel-based shelters on crime, health, and housing stability. Future studies could also incorporate qualitative data from residents, service providers, and community members to better understand the lived experience of shelter use and its social consequences, such as that used by Padgette and colleagues (2023) to assess the impact of hotel-based shelters on residents. In addition, access to more granular data, such as shelter size, specific support services provided, and resident demographics, would help disentangle the mechanisms by which different shelter types can influence crime and public safety. Conclusion This study offers new evidence on the relationship between emergency homeless shelter placement and crime, with a focus on NYC during the COVID-19 pandemic. Contrary to prevailing public concerns and longstanding assumptions that associate homelessness and shelter expansion with increased crime, the findings suggest that emergency hotel shelters are not linked to higher crime rates and may, in fact, have been associated with slight reductions in total and property crime. By leveraging community district-level data and applying a quasi-Poisson model with fixed effects, this analysis provides a more robust understanding of the shelter-crime relationship, particularly distinguishing between traditional congregate shelters and temporary hotel-based shelters. While standard shelters were not significantly associated with crime, the results for emergency hotel shelters underscore the importance of shelter design, resident privacy, and supportive infrastructure in shaping neighborhood outcomes. Although limited by data availability and modeling constraints, this research challenges simplistic narratives that frame homeless shelters as inherently criminogenic. Instead, it highlights how thoughtful, health-oriented interventions, such as hotel-based shelters, can support vulnerable populations without undermining public safety. As cities continue to face intersecting crises of homelessness, housing affordability, and community trust, these findings highlight the potential benefits of rethinking how homeless shelters are conceptualized, designed, and integrated into urban landscapes. Continued research in this area is vital. A deeper understanding of the structural, social, and spatial factors that moderate the relationship between homelessness and crime can guide evidence-based policymaking and help cities implement interventions that are both compassionate and effective. In doing so, we may move closer to solutions that serve the needs of both housed and unhoused residents while promoting safer and more inclusive urban environments for all. References Berge L (2018). \"Efficient estimation of maximum likelihood models with multiple fixed-effects: the R package FENmlm.\" CREA Discussion Papers . Calsyn, R. J., & Winter, J. P. (2002). Social support, psychiatric symptoms, and housing: A causal analysis. Journal of Community Psychology, 30 (3), 247–259. https://doi.org/10.1002/jcop.10004 Coalition for the Homeless (2020). Testimony on The Disparate Impact of COVID-19 on Homeless People in New York City before the NYS Legislature. Retrieved from https://www.nysenate.gov/sites/default/files/covid-19_testimony_may_18_2020_final.pdf Cohen, L. E., & Felson, M. (1979). Social Change and Crime Rate Trends: A Routine Activity Approach. American Sociological Review, 44 (4), 588–608. https://doi.org/10.2307/2094589 Ee, M., & Zhang, Y. (2024). Homelessness and Crime in Neighborhoods. Crime and Delinquency, 70 (8), 2195–2218. https://doi.org/10.1177/00111287221140835 Ellsworth, J. T. (2019). Street Crime Victimization Among Homeless Adults: A Review of the Literature. Victims & Offenders, 14 (1), 96–118. https://doi.org/10.1080/15564886.2018.1547997 Faraji, S.-L., Ridgeway, G., & Wu, Y. (2018). Effect of emergency winter homeless shelters on property crime. Journal of Experimental Criminology, 14 (2), 129–140. https://doi.org/10.1007/s11292-017-9320-4 Ferguson, K. M., Bender, K., Thompson, S. J., Xie, B., & Pollio, D. (2012). Exploration of Arrest Activity among Homeless Young Adults in Four U.S. Cities. Social Work Research, 36 (3), 233–238. https://doi.org/10.1093/swr/svs023 Fischer, S. N., Shinn, M., Shrout, P., & Tsemberis, S. (2008). Homelessness, Mental Illness, and Criminal Activity: Examining Patterns Over Time. American Journal of Community Psychology, 42 (3–4), 251–265. https://doi.org/10.1007/s10464-008-9210-z Link, B. G., Monahan, J., Stueve, A., & Cullen, F. T. (1999). Real in Their Consequences: A Sociological Approach to Understanding the Association between Psychotic Symptoms and Violence. American Sociological Review, 64 (2), 316–332. https://doi.org/10.1177/000312249906400213 Link, B. G., Schwartz, S., Moore, R., Phelan, J., Struening, E., Stueve, A., & Colten, M. E. (1995). Public knowledge, attitudes, and beliefs about homeless people: Evidence for compassion fatigue? American Journal of Community Psychology, 23 (4), 533–555. https://doi.org/10.1007/BF02506967 Markowitz, F. E. (2024). Community-Level Relationships Between Homelessness and Crime in Finland. Crime and Delinquency . https://doi.org/10.1177/00111287241231746 McGuire, J. F., & Rosenheck, R. A. (2004). Criminal History as a Prognostic Indicator in the Treatment of Homeless People With Severe Mental Illness. Psychiatric Services, 55 (1), 42–48. https://doi.org/10.1176/appi.ps.55.1.42 Mechanic D., McAlpine D. D., Rochefort D. A. (2014). Mental health and social policy: Beyond managed care (6th ed.). Pearson. Meyer, B. D., Wyse, A., Corinth, K. (2023). The size and Census coverage of the U.S. homeless population. Journal of Urban Economics, 136 , 103559. https://doi.org/10.1016/j.jue.2023.103559 Padgett, D. K., Bond, L., & Wusinich, C. (2023). From the streets to a hotel: a qualitative study of the experiences of homeless persons in the pandemic era. Journal of Social Distress and Homeless, 32 (2), 284-254. https://doi.org/10.1080/10530789.2021.2021362 Pebesma, E., 2018. Simple Features for R: Standardized Support for Spatial Vector Data. The R Journal, 10 (1), 439-446, https://doi.org/10.32614/RJ-2018-009 Pebesma, E., & Bivand, R. (2023). Spatial Data Science: With Applications in R . Chapman and Hall/CRC. https://doi.org/10.1201/9780429459016 R Core Team (2022). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria . https://www.R-project.org/ Silver E. (2002). Mental disorder and violent victimization: The mediating role of involvement in conflicted social relationships. Criminology, 40 (1), 191–212. https://doi.org/10.1111/j.1745-9125.2002.tb00954.x Snow, D. A., Baker, S. G., & Anderson, L. (1989). Criminality and Homeless Men: An Empirical Assessment. Social Problems, 36 (5), 532–549. https://doi.org/10.1525/sp.1989.36.5.03x0010j Tsai, J., Lee, C. Y. S., Byrne, T., Pietrzak, R. H., & Southwick, S. M. (2017). Changes in Public Attitudes and Perceptions about Homelessness Between 1990 and 2016. American Journal of Community Psychology, 60 (3–4), 599–606. https://doi.org/10.1002/ajcp.12198 Wilson, J. Q., & Kelling, G. L. (1982). The Police and Neighborhood Safety: Broken Windows. The Atlantic Monthly , 29-38. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {\"props\":{\"pageProps\":{\"initialData\":{\"identity\":\"rs-7151989\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":true,\"archivedVersions\":[],\"articleType\":\"Research Article\",\"associatedPublications\":[],\"authors\":[{\"id\":487455782,\"identity\":\"fe4058bc-0411-4339-a2fb-cab72b3037b6\",\"order_by\":0,\"name\":\"Brianna Camero\",\"email\":\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwklEQVRIiWNgGAWjYBACAyjNzA+lidfCLtlAqhZ+gwPEajFnP/x0w8c9NtLG51enSTBUWCc2ENJi2ZNmdnPGszRjsxtvt0kwnEknrMXgBoPZbZ4Dh5PNbpzdJsHYdpgYLezfbv85cLh+8wyQln9EaeExu81w4DCzAX8vUEsDEVose3LKbvYcSGOWuMG72SLhWLoxQS3m7Me33fhxwIaZv//sxhsfaqxlCWpBAIkEBoYE4pWDAP8B0tSPglEwCkbByAEAKHxDSGouykoAAAAASUVORK5CYII=\",\"orcid\":\"https://orcid.org/0000-0002-5756-0338\",\"institution\":\"George Mason University\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Brianna\",\"middleName\":\"\",\"lastName\":\"Camero\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2025-07-17 20:01:02\",\"currentVersionCode\":1,\"declarations\":\"\",\"doi\":\"10.21203/rs.3.rs-7151989/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-7151989/v1\",\"draftVersion\":[],\"editorialEvents\":[],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":87865555,\"identity\":\"64f2e9b3-1c8d-4892-8fbd-838fd7229977\",\"added_by\":\"auto\",\"created_at\":\"2025-07-29 20:03:35\",\"extension\":\"pdf\",\"order_by\":0,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"manuscript-pdf\",\"size\":399059,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"manuscript.pdf\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-7151989/v1/52659d31-dbb4-4893-81dd-24ddd57773eb.pdf\"}],\"financialInterests\":\"\",\"formattedTitle\":\"The Effects of Emergency Homeless Shelters on Crime during the COVID-19 Pandemic\",\"fulltext\":[{\"header\":\"Introduction\",\"content\":\"\\u003cp\\u003eHomelessness has long been a pressing social issue in the United States (US), with approximately 400,000 people living in shelters and an estimated 200,000 individuals sleeping on the streets on any given night, although the latter number is particularly difficult to assess with certainty (Meyer et al., \\u003cspan citationid=\\\"CR15\\\" class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e). In New York City (NYC), homelessness had already reached record highs prior to the COVID-19 pandemic, but the onset of this public health crisis exacerbated the situation. Owing to a lack of access to private spaces, individuals experiencing homelessness were at significantly greater risk of COVID-19 transmission, hospitalization, and death (Coalition for the Homeless, \\u003cspan citationid=\\\"CR3\\\" class=\\\"CitationRef\\\"\\u003e2020\\u003c/span\\u003e). In April 2020 alone, COVID-19-related deaths among individuals experiencing homelessness were 157% higher than the city’s average monthly homeless death toll for all causes in 2019, and the mortality rate among individuals in homeless shelters was 62% higher than that of NYC’s overall population (Coalition for the Homeless, \\u003cspan citationid=\\\"CR3\\\" class=\\\"CitationRef\\\"\\u003e2020\\u003c/span\\u003e).\\u003c/p\\u003e\\u003cp\\u003eIn response, NYC implemented emergency measures, including the leasing of commercial hotels to serve as temporary homeless shelters. These hotel-based shelters allowed for greater physical distancing and offered private rooms without the requirements, providing a degree of autonomy and safety not typically available in traditional congregate shelter settings. Unlike conventional shelters, hotel-based shelters often have fewer behavioral requirements, less oversight, and greater privacy (Padgett et al., \\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e). While these emergency shelters aimed to reduce viral transmission and support the well-being of vulnerable populations, their presence in residential and commercial neighborhoods also raised public concerns, particularly around crime.\\u003c/p\\u003e\\u003cp\\u003ePublic perceptions frequently link homelessness to crime, even though this association is often rooted more in stigma than in empirical evidence. Past national surveys have shown that many people perceive people experiencing homelessness as dangerous or unpredictable (Link et al., \\u003cspan citationid=\\\"CR11\\\" class=\\\"CitationRef\\\"\\u003e1995\\u003c/span\\u003e; Tsai et al., \\u003cspan citationid=\\\"CR22\\\" class=\\\"CitationRef\\\"\\u003e2017\\u003c/span\\u003e), reinforcing social stigma and shaping public discourse and policy. These perceptions are further reinforced by criminological theories such as broken windows theory (Wilson \\u0026amp; Kelling, \\u003cspan citationid=\\\"CR23\\\" class=\\\"CitationRef\\\"\\u003e1982\\u003c/span\\u003e), which suggests that visible signs of disorder, such as loitering, public intoxication, or vandalism, signal weakened informal social control, which can lead to further deviance. Under this framework, the presence of homeless shelters might be viewed as a form of social disorder that contributes to neighborhood decline and increased crime.\\u003c/p\\u003e\\u003cp\\u003eRelatedly, routine activity theory (Cohen \\u0026amp; Felson, \\u003cspan citationid=\\\"CR4\\\" class=\\\"CitationRef\\\"\\u003e1979\\u003c/span\\u003e) provides another lens through which to understand possible links between homelessness and crime. According to this theory, crime is likely to occur when a motivated offender encounters a suitable target in the absence of a capable guardian. The rapid introduction of emergency shelters may disrupt local routines, increase foot traffic, or draw individuals into new public spaces, thereby altering the opportunity structure for certain types of crime. Moreover, because many individuals experiencing homelessness are at heightened risk of victimization, particularly from assault, robbery, and theft (Ellsworth, \\u003cspan citationid=\\\"CR6\\\" class=\\\"CitationRef\\\"\\u003e2019\\u003c/span\\u003e), the increase in visible homelessness in certain areas may also reflect vulnerability rather than criminality.\\u003c/p\\u003e\\u003cp\\u003eEmpirical findings on the relationship between homelessness and crime, however, are mixed and context dependent. In Vancouver, Faraji and colleagues (\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e2018\\u003c/span\\u003e) reported that the introduction of temporary winter shelters was associated with a 56% increase in property crime in the immediate area, primarily driven by theft from vehicles, vandalism, and other thefts. However, the rates of breaking and entering commercial properties actually decreased by 34% in the immediate proximity of these shelters, suggesting a complex and uneven effect. Similarly, a study in Los Angeles reported that higher levels of homelessness were significantly associated with increased property crime in affected neighborhoods (Ee \\u0026amp; Zhang, 2022). Moreover, a longitudinal analysis in Finland linked homelessness to increased rates of violent crime, property crime, and public intoxication (Markowtiz, 2024). These findings underscore that the presence of shelters or unhoused individuals can influence crime patterns, but the mechanisms driving those patterns remain poorly understood and are likely moderated by broader social, environmental, and policy factors.\\u003c/p\\u003e\\u003cp\\u003eAt the individual level, however, there is evidence to suggest that certain characteristics associated with homelessness, particularly mental illness, substance use disorders, and survival behaviors, can increase the likelihood of involvement in criminal activity. Approximately one-third of individuals experiencing homelessness suffer from major mental illnesses, excluding substance-use disorders (Mechanic et al., \\u003cspan citationid=\\\"CR14\\\" class=\\\"CitationRef\\\"\\u003e2014\\u003c/span\\u003e). These conditions can hinder access to stable housing and employment and, when left untreated, may contribute to behaviors that bring individuals into contact with the criminal justice system (Calsyn \\u0026amp; Winter, \\u003cspan citationid=\\\"CR2\\\" class=\\\"CitationRef\\\"\\u003e2002\\u003c/span\\u003e). Research shows that individuals suffering from homelessness in conjunction with serious mental illness are more likely to be involved in aggressive or violent incidents (Link et al., \\u003cspan citationid=\\\"CR10\\\" class=\\\"CitationRef\\\"\\u003e1999\\u003c/span\\u003e; Silver, \\u003cspan citationid=\\\"CR20\\\" class=\\\"CitationRef\\\"\\u003e2002\\u003c/span\\u003e). Additionally, survival-related offenses, such as shoplifting, trespassing, loitering, and public urination—sometimes referred to as “quality of life” crimes or disorder offenses—are more commonly committed by unhoused individuals (Fischer et al., \\u003cspan citationid=\\\"CR9\\\" class=\\\"CitationRef\\\"\\u003e2008\\u003c/span\\u003e; McGuire \\u0026amp; Rosenbeck, 2004). These actions, while often nonviolent, contribute to the broader perception of individuals experiencing homelessness as criminal actors, although these behaviors often arise from necessity rather than intent to harm.\\u003c/p\\u003e\\u003cp\\u003eIndividuals experiencing homelessness are also disproportionately subject to arrest and law enforcement scrutiny. Snow and colleagues (\\u003cspan citationid=\\\"CR21\\\" class=\\\"CitationRef\\\"\\u003e1989\\u003c/span\\u003e) documented elevated arrest rates among individuals who are homeless, especially those with nonviolent offenses. Similarly, Ferguson and colleagues (\\u003cspan citationid=\\\"CR8\\\" class=\\\"CitationRef\\\"\\u003e2012\\u003c/span\\u003e) reported that compared with their housed counterparts, young adults who were homeless were significantly more likely to engage in property crime and drug usage. However, the high rates of criminal justice involvement among unhoused individuals do not necessarily indicate a direct causal link between homelessness and increased criminal behavior. Rather, they highlight the criminalization of homelessness and broader systemic failure in providing adequate housing, mental health care, and social services.\\u003c/p\\u003e\\u003cp\\u003eImportantly, the literature offers limited insight into the specific effects of emergency homeless shelters, particularly hotel-based models, on neighborhood-level crime. Faraji and colleagues’ (\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e2018\\u003c/span\\u003e) study remains one of the only peer-reviewed analyses to directly examine the short-term impacts of temporary shelters on crime, but it focused on congregate-style winter shelters and not the more private, decentralized hotel-based model implemented during the COVID-19 pandemic in NYC. Moreover, hotel-based shelters offered distinct benefits compared with traditional shelters: residents experienced improvements in physical and mental health, hygiene, safety, nutrition, and overall well-being (Padgette et al., 2023). These enhanced living conditions may serve as protective factors against both victimization and criminal involvement.\\u003c/p\\u003e\\u003cp\\u003eThe current study contributes to the growing but still nascent literature by examining how emergency hotel shelters, implemented as public health interventions during the COVID-19 pandemic, may have affected crime at the community district level in NYC. In doing so, it provides a unique opportunity to differentiate the impacts of congregate shelters and hotel-based shelters and to evaluate whether the design and conditions of temporary housing influence crime outcomes. By shedding light on this relationship, this study aims to inform both homelessness policy and public safety strategies in post-pandemic urban environments.\\u003c/p\\u003e\"},{\"header\":\"Methods\",\"content\":\"\\u003cp\\u003e\\u003cem\\u003eMeasures\\u003c/em\\u003e\\u003c/p\\u003e\\u003cp\\u003eTo conduct this study, I use data obtained from the Department of Homeless Services in New York City through a freedom of information law request. These data are broken down at the community district level (n = 59) according to the number of standard homeless shelters and the number of commercial hotels used as temporary emergency homeless shelters (referred to as emergency hotels in the results tables) to combat the spread of COVID-19 at six different time points: September 2020, January 2021, March 2021, April 2021, October 2021, and February 2022. With 59 community districts and six time points, the total sample size was 354.\\u003c/p\\u003e\\u003cp\\u003eTo assess the impact of these shelters on crime, I obtained data from the New York City open data website, which was geocoded into community districts using the geographic coordinates included in the crime data and a community district shapefile retrieved from New York City’s Department of City Planning website. This geocoding process was completed via the “sf” package (Pebesma \\u0026amp; Bivand, 2023; Pebsema, 2018), whereas the analysis was completed via the “fixest” package (Berge, \\u003cspan citationid=\\\"CR1\\\" class=\\\"CitationRef\\\"\\u003e2018\\u003c/span\\u003e), both of which are in RStudio (R Core Team, \\u003cspan citationid=\\\"CR19\\\" class=\\\"CitationRef\\\"\\u003e2022\\u003c/span\\u003e). I then aggregated the crime data by month into four categories: total crime, violent crime, property crime, and disorder crime. Total crime is a broad measure of all crime incidents that occur within each community district, including violent, property, and disorder crime. Violent crimes are crime incidents where a victim is harmed by or threatened with violence, such as robbery, aggravated assault, homicide, and rape. Property crimes are crime incidents where a victim’s property is stolen, damaged, or destroyed without physical harm to the victim; these include burglary, larceny, and vandalism. Finally, disorder crimes are relatively minor physical or social crime incidents where they may have caused a disruption to the community, such as noise disturbances, prostitution, loitering, and solicitation.\\u003c/p\\u003e\\u003cp\\u003eTable\\u0026nbsp;\\u003cspan refid=\\\"Tab1\\\" class=\\\"InternalRef\\\"\\u003e1\\u003c/span\\u003e presents the correlation matrix for the variables. This was largely done to ensure that emergency hotels and standard shelters were not highly related to each other and that similar phenomena were measured. This was not the case, with the correlation being approximately − 0.01, indicating that they are independent measures. Furthermore, standard shelters are moderately correlated with each of the crime types, whereas emergency hotels are weakly correlated, suggesting a different relationship for each type of shelter with crime. This weak correlation for emergency hotels, however, could be an artifact of low variability within the emergency hotel variable. Table\\u0026nbsp;\\u003cspan refid=\\\"Tab2\\\" class=\\\"InternalRef\\\"\\u003e2\\u003c/span\\u003e thus shows the descriptive statistics of each variable, with emergency hotels having a mean of 0.75 and a standard deviation of 1.73, suggesting fairly high variability. The standard shelters variable was notably very different from the emergency hotels variable, with a mean of 4.918 and a standard deviation of 5.482.\\u003c/p\\u003e\\u003cdiv class=\\\"gridtable\\\"\\u003e\\u003cdiv align=\\\"left\\\" class=\\\"colspec\\\" colname=\\\"c1\\\" colnum=\\\"1\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"char\\\" char=\\\".\\\" class=\\\"colspec\\\" colname=\\\"c2\\\" colnum=\\\"2\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"char\\\" char=\\\".\\\" class=\\\"colspec\\\" colname=\\\"c3\\\" colnum=\\\"3\\\"\\u003e\\u003c/div\\u003e\\u003ctable float=\\\"Yes\\\" id=\\\"Tab1\\\" border=\\\"1\\\"\\u003e\\u003ccaption language=\\\"En\\\"\\u003e\\u003cdiv class=\\\"CaptionNumber\\\"\\u003eTable 1\\u003c/div\\u003e\\u003cdiv class=\\\"CaptionContent\\\"\\u003e\\u003cp\\u003eCorrelation Matrix\\u003c/p\\u003e\\u003c/div\\u003e\\u003c/caption\\u003e\\u003ccolgroup cols=\\\"3\\\"\\u003e\\u003c/colgroup\\u003e\\u003cthead\\u003e\\u003ctr\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u0026nbsp;\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003eEmergency Hotels\\u003c/p\\u003e\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003eStandard Shelters\\u003c/p\\u003e\\u003c/th\\u003e\\u003c/tr\\u003e\\u003c/thead\\u003e\\u003ctbody\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eEmergency Hotels\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e1.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e-0.0132\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eStandard Shelters\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e-0.0132\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e1.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eTotal Crime\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.1490\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.4946\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eViolent Crime\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.0165\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.6695\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eProperty Crime\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.2707\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.1064\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eDisorder Crime\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.067\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.5374\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003c/tbody\\u003e\\u003c/table\\u003e\\u003c/div\\u003e\\u003cdiv class=\\\"gridtable\\\"\\u003e\\u003cdiv align=\\\"left\\\" class=\\\"colspec\\\" colname=\\\"c1\\\" colnum=\\\"1\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"char\\\" char=\\\".\\\" class=\\\"colspec\\\" colname=\\\"c2\\\" colnum=\\\"2\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"char\\\" char=\\\".\\\" class=\\\"colspec\\\" colname=\\\"c3\\\" colnum=\\\"3\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"char\\\" char=\\\".\\\" class=\\\"colspec\\\" colname=\\\"c4\\\" colnum=\\\"4\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"char\\\" char=\\\".\\\" class=\\\"colspec\\\" colname=\\\"c5\\\" colnum=\\\"5\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"char\\\" char=\\\".\\\" class=\\\"colspec\\\" colname=\\\"c6\\\" colnum=\\\"6\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"char\\\" char=\\\".\\\" class=\\\"colspec\\\" colname=\\\"c7\\\" colnum=\\\"7\\\"\\u003e\\u003c/div\\u003e\\u003ctable float=\\\"Yes\\\" id=\\\"Tab2\\\" border=\\\"1\\\"\\u003e\\u003ccaption language=\\\"En\\\"\\u003e\\u003cdiv class=\\\"CaptionNumber\\\"\\u003eTable 2\\u003c/div\\u003e\\u003cdiv class=\\\"CaptionContent\\\"\\u003e\\u003cp\\u003eDescriptive Statistics Table\\u003c/p\\u003e\\u003c/div\\u003e\\u003c/caption\\u003e\\u003ccolgroup cols=\\\"7\\\"\\u003e\\u003c/colgroup\\u003e\\u003cthead\\u003e\\u003ctr\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u0026nbsp;\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003eEmergency Hotels\\u003c/p\\u003e\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003eStandard Shelters\\u003c/p\\u003e\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003eTotal Crime\\u003c/p\\u003e\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003eViolent Crime\\u003c/p\\u003e\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c6\\\"\\u003e\\u003cp\\u003eProperty Crime\\u003c/p\\u003e\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c7\\\"\\u003e\\u003cp\\u003eDisorder Crime\\u003c/p\\u003e\\u003c/th\\u003e\\u003c/tr\\u003e\\u003c/thead\\u003e\\u003ctbody\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eMean\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.7542\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e4.9180\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e599.5000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e121.0600\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c6\\\"\\u003e\\u003cp\\u003e234.7000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c7\\\"\\u003e\\u003cp\\u003e191.9000\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eMedian\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e3.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e552.5000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e107.5000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c6\\\"\\u003e\\u003cp\\u003e205.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c7\\\"\\u003e\\u003cp\\u003e176.5000\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eStandard Deviation\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e1.7309\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e5.4815\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e230.5848\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e63.6986\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c6\\\"\\u003e\\u003cp\\u003e107.8252\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c7\\\"\\u003e\\u003cp\\u003e73.3715\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eMinimum\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e35.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e5.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c6\\\"\\u003e\\u003cp\\u003e26.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c7\\\"\\u003e\\u003cp\\u003e3.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eMaximum\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e11.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e22.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e1378.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e323.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c6\\\"\\u003e\\u003cp\\u003e786.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c7\\\"\\u003e\\u003cp\\u003e418.0000\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003c/tbody\\u003e\\u003c/table\\u003e\\u003c/div\\u003e\\u003cp\\u003e\\u003cem\\u003eAnalytic Strategy\\u003c/em\\u003e\\u003c/p\\u003e\\u003cp\\u003eBecause the dependent variables of crime are count variables, I initially set out to utilize a Poisson or negative binomial regression model. However, for both of these model types, the data did not pass the Kolmogorov-Smirnov test. With p-values of less than 0.001 for each crime type, the crime data significantly differ from the Poisson and negative binomial distributions. Because of this, I opted for a quasi-Poisson model after ensuring that the assumption of the variance being a linear function of the mean was met. The standard errors of these models are also clustered at the community district level due to the aggregation of the data. Furthermore, I use time and place fixed effects in the quasi-Poisson model to control for the differences between community districts and months that cannot be controlled for due to a lack of data availability, such as the amount of physical disorder in a place, collective efficacy within community districts, or the amount of foot traffic in a community district in a given month. Thus, by adding month and community district fixed effects, I can ensure a more accurate estimate of the actual relationship between homeless shelter types and crime.\\u003c/p\\u003e\"},{\"header\":\"Results\",\"content\":\"\\u003cp\\u003eTable\\u0026nbsp;\\u003cspan refid=\\\"Tab3\\\" class=\\\"InternalRef\\\"\\u003e3\\u003c/span\\u003e presents the results of all four regressions of crime type on emergency hotels and standard shelters. However, while I present McFadden\\u0026rsquo;s R\\u003csup\\u003e2\\u003c/sup\\u003e, this number is essentially uninterpretable because the fixed effects are included within the model as well. Thus, the amount of variance in crime accounted for by only the homeless shelters is unknown. Within each of the crime models, while standard shelters have consistently positive coefficients, they are insignificant in each model, with p-values ranging from 0.21 to 0.73. Thus, standard shelters appear to have no significant effect on crime. However, this may be due to the inability to isolate the effects of standard shelters, since it is unlikely that there would be any major change in the number of standard shelters in a community district between time points. For emergency hotels, however, the coefficients for total crime and property crime approach significance at p\\u0026thinsp;=\\u0026thinsp;0.0506 and 0.051, respectively. The effect of emergency hotels is also consistently negative, suggesting that the presence of more emergency hotels may result in a crime reduction effect, with a percentage decrease of approximately 2.7% for both total crime and property crime. For violent and disorder crime, the coefficients suggest 2.7% and 1.4% decreases in crime, respectively. However, the p-values are 0.12 and 0.32, respectively, suggesting that this effect is insignificant.\\u003c/p\\u003e\\u003cp\\u003e\\u003cdiv class=\\\"gridtable\\\"\\u003e\\u003ctable float=\\\"Yes\\\" id=\\\"Tab3\\\" border=\\\"1\\\"\\u003e\\u003ccaption language=\\\"En\\\"\\u003e\\u003cdiv class=\\\"CaptionNumber\\\"\\u003eTable 3\\u003c/div\\u003e\\u003cdiv class=\\\"CaptionContent\\\"\\u003e\\u003cp\\u003eQuasi-Poisson Models with Fixed Effects\\u003c/p\\u003e\\u003c/div\\u003e\\u003c/caption\\u003e\\u003ccolgroup cols=\\\"5\\\"\\u003e\\u003cdiv align=\\\"left\\\" class=\\\"colspec\\\" colname=\\\"c1\\\" colnum=\\\"1\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"left\\\" class=\\\"colspec\\\" colname=\\\"c2\\\" colnum=\\\"2\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"left\\\" class=\\\"colspec\\\" colname=\\\"c3\\\" colnum=\\\"3\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"left\\\" class=\\\"colspec\\\" colname=\\\"c4\\\" colnum=\\\"4\\\"\\u003e\\u003c/div\\u003e\\u003cdiv align=\\\"left\\\" class=\\\"colspec\\\" colname=\\\"c5\\\" colnum=\\\"5\\\"\\u003e\\u003c/div\\u003e\\u003cthead\\u003e\\u003ctr\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u0026nbsp;\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003eTotal\\u003c/p\\u003e\\u003cp\\u003eCrime\\u003c/p\\u003e\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003eViolent Crime\\u003c/p\\u003e\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003eProperty Crime\\u003c/p\\u003e\\u003c/th\\u003e\\u003cth align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003eDisorder Crime\\u003c/p\\u003e\\u003c/th\\u003e\\u003c/tr\\u003e\\u003c/thead\\u003e\\u003ctbody\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eEmergency Hotels\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e-0.0275\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e-0.0217\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e-0.0275\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e-0.0140\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eStandard Error\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.0141\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.0141\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e0.0141\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e0.0141\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003ep-value\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.0506\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.1239\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e0.0510\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e0.3214\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eIRR\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.9728\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.9786\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e0.9729\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e0.9861\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003ePercent Change\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e-2.7160\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e-2.1440\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e-2.7120\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e-1.3870\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eStandard Shelters\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.0120\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.0047\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e0.0172\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e0.0131\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eStandard Error\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.0138\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.0138\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e0.0138\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e0.0138\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003ep-value\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.3858\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.7347\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e0.2120\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e0.3402\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eIRR\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e1.0120\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e1.0047\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e1.0173\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e1.0132\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003ePercent Change\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e1.2010\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.4680\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e1.7340\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e1.3220\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eNumber of Observations\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e354\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e354\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e354\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e354\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eNumber of Districts\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e59\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e59\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e59\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e59\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eNumber of Months\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e6\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e6\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e6\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e6\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003ctr\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e\\u003cp\\u003eMcFadden's R\\u003csup\\u003e2\\u003c/sup\\u003e\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e\\u003cp\\u003e0.9152\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e\\u003cp\\u003e0.9387\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c4\\\"\\u003e\\u003cp\\u003e0.8567\\u003c/p\\u003e\\u003c/td\\u003e\\u003ctd align=\\\"left\\\" colname=\\\"c5\\\"\\u003e\\u003cp\\u003e0.9039\\u003c/p\\u003e\\u003c/td\\u003e\\u003c/tr\\u003e\\u003c/tbody\\u003e\\u003c/colgroup\\u003e\\u003c/table\\u003e\\u003c/div\\u003e\\u003c/p\\u003e\"},{\"header\":\"Discussion\",\"content\":\"\\u003cp\\u003eThis study examined the relationships between homeless shelters, specifically emergency hotel shelters introduced during the COVID-19 pandemic, and crime rates across NYC community districts. In contrast to long-standing public fears that shelters may increase neighborhood crime, the findings from this analysis suggest a more nuanced and possible reassuring conclusion. While standard congregate shelters were not significantly associated with crime rates, the presence of emergency hotel shelters was associated with a modest decrease in both total and property crimes, with p-values approaching statistical significance at the 0.05 threshold. These findings provide early evidence that emergency hotel shelters designed to improve privacy, autonomy, and health outcomes during a public health crisis may not only avoid increasing crime but may also be linked to slight reductions in certain types of crime at the neighborhood level.\\u003c/p\\u003e\\u003cp\\u003eThese results align with a growing body of literature suggesting that fears of rising crime tied to homelessness and shelter expansion may be overstated or rooted more in stigma than empirical observation (Link et al., \\u003cspan citationid=\\\"CR11\\\" class=\\\"CitationRef\\\"\\u003e1995\\u003c/span\\u003e; Tsai et al., \\u003cspan citationid=\\\"CR22\\\" class=\\\"CitationRef\\\"\\u003e2017\\u003c/span\\u003e). Rather than supporting the predictions of broken windows theory, which would anticipate a rise in the presence of disorder-related cues such as shelters, this study instead suggests that certain forms of temporary housing, especially those that promote stability, privacy, and health, may mitigate crime risks or even contribute to safer community conditions.\\u003c/p\\u003e\\u003cp\\u003eThe modest reduction in property and total crime associated with emergency hotel shelters could stem from several mechanisms. These shelters reduce crowding, minimize exposure to high-risk street environments, and are generally accompanied by support services (Padgett et al., \\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e). By providing safer and more stable accommodations, these shelters may have reduced residents\\u0026rsquo; involvement in survival-related offenses (e.g., theft) while also lowering their risk of victimization. Moreover, the physical structure and operation of hotel shelters\\u0026mdash;often discrete, monitored, and embedded in commercial zones\\u0026mdash;may have altered the opportunity structures for crime in ways not captured by congregate shelter models.\\u003c/p\\u003e\\u003cp\\u003eHowever, the interpretation of these results requires caution because of several methodological limitations. First, the analysis was limited by a lack of variation in standard shelters over time, making it difficult to isolate their effects. Unlike emergency hotel shelters, which were introduced and phased out during the study period, standard shelters remained relatively constant within each community district. As a result, the observed null effects may reflect statistical constraints rather than the true absence of an association.\\u003c/p\\u003e\\u003cp\\u003eSecond, the study is limited by its sample size. With only 59 community districts and six time points, the total sample size of 354, while sufficient for the analysis conducted, may lack the statistical power needed to detect smaller or more nuanced effects. Relatedly, while the quasi-Poisson model appropriately addresses overdispersion in the crime count data, this model does not allow for traditional goodness-of-fit measures, such as likelihood-based R-squared values, limiting interpretability of model fit.\\u003c/p\\u003e\\u003cp\\u003eThird, the data do not account for the size or capacity of each shelter, which is an important omission, as larger shelters may have different impacts on community dynamics than smaller shelters. Nor was it possible to account for differences in shelter management, support services, or resident composition, all of which may influence outcomes in meaningful ways. Additionally, unmeasured factors such as foot traffic, nearby businesses, and policing practices could not be directly controlled, although the inclusion of fixed effects for months and community districts mitigated some of these concerns.\\u003c/p\\u003e\\u003cp\\u003eDespite these limitations, this study provides valuable insight into the ongoing policy debate over the placement and design of shelters. The evidence that hotel-based emergency shelters did not lead to crime increases and may have had a small protective effect challenges common assumptions in both public discourse and urban policymaking. As NYC and other cities continue to explore models of supportive and traditional housing in the post-pandemic era, the results underscore the importance of not only whether shelters are provided but also how they are designed and implemented.\\u003c/p\\u003e\\u003cp\\u003eFuture research should investigate the long-term effects of both traditional congregate shelters and emergency hotel-based shelters on crime, health, and housing stability. Future studies could also incorporate qualitative data from residents, service providers, and community members to better understand the lived experience of shelter use and its social consequences, such as that used by Padgette and colleagues (2023) to assess the impact of hotel-based shelters on residents. In addition, access to more granular data, such as shelter size, specific support services provided, and resident demographics, would help disentangle the mechanisms by which different shelter types can influence crime and public safety.\\u003c/p\\u003e\"},{\"header\":\"Conclusion\",\"content\":\"\\u003cp\\u003eThis study offers new evidence on the relationship between emergency homeless shelter placement and crime, with a focus on NYC during the COVID-19 pandemic. Contrary to prevailing public concerns and longstanding assumptions that associate homelessness and shelter expansion with increased crime, the findings suggest that emergency hotel shelters are not linked to higher crime rates and may, in fact, have been associated with slight reductions in total and property crime.\\u003c/p\\u003e\\u003cp\\u003eBy leveraging community district-level data and applying a quasi-Poisson model with fixed effects, this analysis provides a more robust understanding of the shelter-crime relationship, particularly distinguishing between traditional congregate shelters and temporary hotel-based shelters. While standard shelters were not significantly associated with crime, the results for emergency hotel shelters underscore the importance of shelter design, resident privacy, and supportive infrastructure in shaping neighborhood outcomes.\\u003c/p\\u003e\\u003cp\\u003eAlthough limited by data availability and modeling constraints, this research challenges simplistic narratives that frame homeless shelters as inherently criminogenic. Instead, it highlights how thoughtful, health-oriented interventions, such as hotel-based shelters, can support vulnerable populations without undermining public safety. As cities continue to face intersecting crises of homelessness, housing affordability, and community trust, these findings highlight the potential benefits of rethinking how homeless shelters are conceptualized, designed, and integrated into urban landscapes.\\u003c/p\\u003e\\u003cp\\u003eContinued research in this area is vital. A deeper understanding of the structural, social, and spatial factors that moderate the relationship between homelessness and crime can guide evidence-based policymaking and help cities implement interventions that are both compassionate and effective. In doing so, we may move closer to solutions that serve the needs of both housed and unhoused residents while promoting safer and more inclusive urban environments for all.\\u003c/p\\u003e\"},{\"header\":\"References\",\"content\":\"\\u003col\\u003e\\n\\u003cli\\u003eBerge L (2018). \\u0026quot;Efficient estimation of maximum likelihood models with multiple fixed-effects: the R package FENmlm.\\u0026quot; \\u003cem\\u003eCREA Discussion Papers\\u003c/em\\u003e.\\u003c/li\\u003e\\n\\u003cli\\u003eCalsyn, R. J., \\u0026amp; Winter, J. P. (2002). Social support, psychiatric symptoms, and housing: A causal analysis. \\u003cem\\u003eJournal of Community Psychology, 30\\u003c/em\\u003e(3), 247\\u0026ndash;259. https://doi.org/10.1002/jcop.10004\\u003c/li\\u003e\\n\\u003cli\\u003eCoalition for the Homeless (2020). Testimony on The Disparate Impact of COVID-19 on Homeless People in New York City before the NYS Legislature. Retrieved from https://www.nysenate.gov/sites/default/files/covid-19_testimony_may_18_2020_final.pdf\\u003c/li\\u003e\\n\\u003cli\\u003eCohen, L. E., \\u0026amp; Felson, M. (1979). Social Change and Crime Rate Trends: A Routine Activity Approach. \\u003cem\\u003eAmerican Sociological Review, 44\\u003c/em\\u003e(4), 588\\u0026ndash;608. https://doi.org/10.2307/2094589\\u003c/li\\u003e\\n\\u003cli\\u003eEe, M., \\u0026amp; Zhang, Y. (2024). Homelessness and Crime in Neighborhoods. \\u003cem\\u003eCrime and Delinquency, 70\\u003c/em\\u003e(8), 2195\\u0026ndash;2218. https://doi.org/10.1177/00111287221140835\\u003c/li\\u003e\\n\\u003cli\\u003eEllsworth, J. T. (2019). Street Crime Victimization Among Homeless Adults: A Review of the Literature. \\u003cem\\u003eVictims \\u0026amp; Offenders, 14\\u003c/em\\u003e(1), 96\\u0026ndash;118. https://doi.org/10.1080/15564886.2018.1547997\\u003c/li\\u003e\\n\\u003cli\\u003eFaraji, S.-L., Ridgeway, G., \\u0026amp; Wu, Y. (2018). Effect of emergency winter homeless shelters on property crime. \\u003cem\\u003eJournal of Experimental Criminology, 14\\u003c/em\\u003e(2), 129\\u0026ndash;140. https://doi.org/10.1007/s11292-017-9320-4\\u003c/li\\u003e\\n\\u003cli\\u003eFerguson, K. M., Bender, K., Thompson, S. J., Xie, B., \\u0026amp; Pollio, D. (2012). Exploration of Arrest Activity among Homeless Young Adults in Four U.S. Cities. \\u003cem\\u003eSocial Work Research, 36\\u003c/em\\u003e(3), 233\\u0026ndash;238. https://doi.org/10.1093/swr/svs023\\u003c/li\\u003e\\n\\u003cli\\u003eFischer, S. N., Shinn, M., Shrout, P., \\u0026amp; Tsemberis, S. (2008). Homelessness, Mental Illness, and Criminal Activity: Examining Patterns Over Time. \\u003cem\\u003eAmerican Journal of Community Psychology, 42\\u003c/em\\u003e(3\\u0026ndash;4), 251\\u0026ndash;265. https://doi.org/10.1007/s10464-008-9210-z\\u003c/li\\u003e\\n\\u003cli\\u003eLink, B. G., Monahan, J., Stueve, A., \\u0026amp; Cullen, F. T. (1999). Real in Their Consequences: A Sociological Approach to Understanding the Association between Psychotic Symptoms and Violence. \\u003cem\\u003eAmerican Sociological Review, 64\\u003c/em\\u003e(2), 316\\u0026ndash;332. https://doi.org/10.1177/000312249906400213\\u003c/li\\u003e\\n\\u003cli\\u003eLink, B. G., Schwartz, S., Moore, R., Phelan, J., Struening, E., Stueve, A., \\u0026amp; Colten, M. E. (1995). Public knowledge, attitudes, and beliefs about homeless people: Evidence for compassion fatigue? \\u003cem\\u003eAmerican Journal of Community Psychology, 23\\u003c/em\\u003e(4), 533\\u0026ndash;555. https://doi.org/10.1007/BF02506967\\u003c/li\\u003e\\n\\u003cli\\u003eMarkowitz, F. E. (2024). Community-Level Relationships Between Homelessness and Crime in Finland. \\u003cem\\u003eCrime and Delinquency\\u003c/em\\u003e. https://doi.org/10.1177/00111287241231746\\u003c/li\\u003e\\n\\u003cli\\u003eMcGuire, J. F., \\u0026amp; Rosenheck, R. A. (2004). Criminal History as a Prognostic Indicator in the Treatment of Homeless People With Severe Mental Illness. \\u003cem\\u003ePsychiatric Services, 55\\u003c/em\\u003e(1), 42\\u0026ndash;48. https://doi.org/10.1176/appi.ps.55.1.42\\u003c/li\\u003e\\n\\u003cli\\u003eMechanic D., McAlpine D. D., Rochefort D. A. (2014). \\u003cem\\u003eMental health and social policy: Beyond managed care\\u003c/em\\u003e (6th ed.). Pearson.\\u003c/li\\u003e\\n\\u003cli\\u003eMeyer, B. D., Wyse, A., Corinth, K. (2023). The size and Census coverage of the U.S. homeless population. \\u003cem\\u003eJournal of Urban Economics, 136\\u003c/em\\u003e, 103559. https://doi.org/10.1016/j.jue.2023.103559\\u003c/li\\u003e\\n\\u003cli\\u003ePadgett, D. K., Bond, L., \\u0026amp; Wusinich, C. (2023). From the streets to a hotel: a qualitative study of the experiences of homeless persons in the pandemic era. \\u003cem\\u003eJournal of Social Distress and Homeless, 32\\u003c/em\\u003e(2), 284-254. https://doi.org/10.1080/10530789.2021.2021362\\u003c/li\\u003e\\n\\u003cli\\u003ePebesma, E., 2018. Simple Features for R: Standardized Support for Spatial Vector Data. \\u003cem\\u003eThe R Journal, 10\\u003c/em\\u003e(1), 439-446, https://doi.org/10.32614/RJ-2018-009\\u003c/li\\u003e\\n\\u003cli\\u003ePebesma, E., \\u0026amp; Bivand, R. (2023). \\u003cem\\u003eSpatial Data Science: With Applications in R\\u003c/em\\u003e. Chapman and Hall/CRC. https://doi.org/10.1201/9780429459016\\u003c/li\\u003e\\n\\u003cli\\u003eR Core Team (2022). R: A language and environment for statistical computing. \\u003cem\\u003eR Foundation for Statistical Computing, Vienna, Austria\\u003c/em\\u003e. https://www.R-project.org/\\u003c/li\\u003e\\n\\u003cli\\u003eSilver E. (2002). Mental disorder and violent victimization: The mediating role of involvement in conflicted social relationships. \\u003cem\\u003eCriminology, 40\\u003c/em\\u003e(1), 191\\u0026ndash;212. https://doi.org/10.1111/j.1745-9125.2002.tb00954.x\\u003c/li\\u003e\\n\\u003cli\\u003eSnow, D. A., Baker, S. G., \\u0026amp; Anderson, L. (1989). Criminality and Homeless Men: An Empirical Assessment. \\u003cem\\u003eSocial Problems, 36\\u003c/em\\u003e(5), 532\\u0026ndash;549. https://doi.org/10.1525/sp.1989.36.5.03x0010j\\u003c/li\\u003e\\n\\u003cli\\u003eTsai, J., Lee, C. Y. S., Byrne, T., Pietrzak, R. H., \\u0026amp; Southwick, S. M. (2017). Changes in Public Attitudes and Perceptions about Homelessness Between 1990 and 2016. \\u003cem\\u003eAmerican Journal of Community Psychology, 60\\u003c/em\\u003e(3\\u0026ndash;4), 599\\u0026ndash;606. https://doi.org/10.1002/ajcp.12198\\u003c/li\\u003e\\n\\u003cli\\u003eWilson, J. Q., \\u0026amp; Kelling, G. L. (1982). The Police and Neighborhood Safety: Broken Windows. \\u003cem\\u003eThe Atlantic Monthly\\u003c/em\\u003e, 29-38.\\u003c/li\\u003e\\n\\u003c/ol\\u003e\"}],\"fulltextSource\":\"\",\"fullText\":\"\",\"funders\":[],\"hasAdminPriorityOnWorkflow\":false,\"hasManuscriptDocX\":true,\"hasOptedInToPreprint\":true,\"hasPassedJournalQc\":\"\",\"hasAnyPriority\":false,\"hideJournal\":true,\"highlight\":\"\",\"institution\":\"\",\"isAcceptedByJournal\":false,\"isAuthorSuppliedPdf\":false,\"isDeskRejected\":\"\",\"isHiddenFromSearch\":false,\"isInQc\":false,\"isInWorkflow\":false,\"isPdf\":false,\"isPdfUpToDate\":true,\"isWithdrawnOrRetracted\":false,\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"researchsquare\",\"isNatureJournal\":false,\"hasQc\":true,\"allowDirectSubmit\":true,\"externalIdentity\":\"\",\"sideBox\":\"\",\"snPcode\":\"\",\"submissionUrl\":\"/submission\",\"title\":\"Research Square\",\"twitterHandle\":\"researchsquare\",\"acdcEnabled\":true,\"dfaEnabled\":false,\"editorialSystem\":\"\",\"reportingPortfolio\":\"\",\"inReviewEnabled\":false,\"inReviewRevisionsEnabled\":true},\"keywords\":\"Homelessness, Shelters, Crime, New York City, COVID-19\",\"lastPublishedDoi\":\"10.21203/rs.3.rs-7151989/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-7151989/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"\\u003cp\\u003eThis study examines the associations between emergency homeless shelters and crime in New York City (NYC) during the COVID-19 pandemic, when public health imperatives required rapid housing solutions for individuals experiencing homelessness. In response, the city leased commercial hotels to serve as temporary emergency shelters, raising public concerns about potential increases in neighborhood crime. Using data from the NYC Department of Homeless Services and publicly available crime data, this study assesses the effects of both traditional congregate shelters and hotel-based emergency shelters on crime across 59 community districts at six time points between 2020 and 2022. Crimes were categorized as total, violent, property, or disorder. Quasi-Poisson regression models with community and time fixed effects were used to estimate associations. The findings indicate that traditional shelters have no significant relationship with crime. Emergency hotel shelters, however, were associated with modest reductions in total and property crime, although these findings approached but did not reach conventional levels of statistical significance with p-values around 0.051. These results challenge common public perceptions linking homelessness interventions to increased crime and underscore the importance of shelter design and implementation context. While limitations include constraints in traditional shelter data variation and shelter size information, this study contributes to the evidence base informing public health and housing policy. This study suggests that hotel-based shelters may not only support vulnerable populations during public health crises but may also have neutral or even protective effects on neighborhood safety.\\u003c/p\\u003e\",\"manuscriptTitle\":\"The Effects of Emergency Homeless Shelters on Crime during the COVID-19 Pandemic\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2025-07-22 18:12:47\",\"doi\":\"10.21203/rs.3.rs-7151989/v1\",\"editorialEvents\":[{\"type\":\"communityComments\",\"content\":0}],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"researchsquare\",\"isNatureJournal\":false,\"hasQc\":true,\"allowDirectSubmit\":true,\"externalIdentity\":\"\",\"sideBox\":\"\",\"snPcode\":\"\",\"submissionUrl\":\"/submission\",\"title\":\"Research Square\",\"twitterHandle\":\"researchsquare\",\"acdcEnabled\":true,\"dfaEnabled\":false,\"editorialSystem\":\"\",\"reportingPortfolio\":\"\",\"inReviewEnabled\":false,\"inReviewRevisionsEnabled\":true}}],\"origin\":\"\",\"ownerIdentity\":\"56255979-6a18-44ce-969c-aa539f9d144c\",\"owner\":[],\"postedDate\":\"July 22nd, 2025\",\"published\":true,\"recentEditorialEvents\":[],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"posted\",\"subjectAreas\":[],\"tags\":[],\"updatedAt\":\"2025-07-29T19:55:29+00:00\",\"versionOfRecord\":[],\"versionCreatedAt\":\"2025-07-22 18:12:47\",\"video\":\"\",\"vorDoi\":\"\",\"vorDoiUrl\":\"\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-7151989\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-7151989\",\"identity\":\"rs-7151989\",\"version\":[\"v1\"]},\"buildId\":\"8U1c8b4HqxoKbykW_rLl7\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}