Introduction

Thousands of persons experiencing homelessness (PEH) die each year in the United States from a myriad of preventable causes, including sanitation issues, overdoses, extreme weather, drowning, and violence (Boulder County, 2022; McCormick, 2022; Waugh, 2023). There is a large literature on the criminal victimization of unhoused persons and the mechanisms through which fear of victimization impact them (Ellsworth, 2019). Underlying issues with these concerns involve homeless outreach efforts and difficulties connecting persons to relevant services. Unsheltered PEH can be mobile in ways that jeopardize their inclusion in point-in-time counts, or their ability to maintain contact with service providers and receive assistance (Gibson, 2007).

Due to the many kinds of problems experienced by PEH, there are often multi-agency partnerships to address homelessness holistically. For example, these partnerships often involve local police departments in an enforcement capacity, service navigation capacity, or both (Brown, 2023; Diamond et al., 2022; Hipple, 2017; Simpson, 2015; Townley et al., 2022). Police departments increasingly have homeless outreach teams, provide support in citywide cleaning efforts, and contribute to non-traditional first response models. These response models include co-responder or alternative response teams (Dee & Payne, 2022; Marcus & Stergiopoulos, 2022) designed to better address PEH, persons with substance-use disorders, and individuals experiencing mental or behavioral health crises.

Foundational to the present study are concerns about crime near homeless encampments, and the underutilization of police data as a tool to contact unhoused persons with a history of camping on public property. We begin by examining the degree to which homeless encampments are present in areas with substantial crime concerns. There are multiple possibilities, two of them are that homeless encampments maybe relegated to or allowed to exist in high-crime areas, or encampments may contribute to area-crime problems.

The focus then shifts to whether persons identified at encampments can be more easily contacted in a city based on the totality of their known encampment geographies. For example, can police camping tickets help a Housing and Human Services case manager contact a particular John Doe? The utility of this stems from contact difficulties for outreach workers seeking to connect with persons potentially without fixed addresses, reliable phone numbers, or stable employment. An implicit assumption here is that if an individual has been issued ten camping tickets, all within the same city block, the individual’s location is more likely to be in a known and spatially fixed area compared to elsewhere in a city. While this paper draws on the previous literature about homeless mobility, encampment locations, and victimization concerns, existing police data do not seem to have been used as an outreach tool to contact chronic inhabitants of homeless encampments.

The rest of the present study combines multiple literatures before turning toward the study location, data, analysis, discussion, and conclusion. Location quotients are used to assess crime concentration and dispersion at varying distances, and the haversine equation is presented as the chosen method of coordinate pair distance measurement between encampment inhabitants multiple past encampment locations. The discussion focuses on how the current work supplements existing research and limitations with the present study.

Literature review

These [PEH] had often spent years on the streets and in an attempt to cope with the threatening (e.g., been robbed or assaulted) and often hostile (e.g., been verbally harassed by property owners) environment they had geographically withdrawn to the perceived safe confines of the “nooks and crannies” that exist in the built environment.

- Rollinson (1998, p.113)

There is a large literature on harms experienced by unhoused persons in different settings, and many studies focus on criminal victimization (Ellsworth, 2019, 2021a, 2021b, 2022; Gaetz, 2004; Garland et al., 2010; Lee & Schreck, 2005; Roy et al., 2014). Crime appears to concentrate near homeless encampments when encampments are identified through online reporting mechanisms (Russell, 2020). Persons turned away from shelters appear to be at greater risk of victimization (Garland et al., 2010), and spatial predictors of homeless victimization are similar to predictors of homeless offending (Yoo & Wheeler, 2019). Russell (2020) found that crime concentrated closer to where community members reported homeless encampments to the city. Within one city block of reported encampments, person crimes (e.g., assault) were 3.6 times more concentrated than in the city overall, and overall crime was 2.9 times more concentrated. Moreover, crime was observed as being less concentrated the further one travelled from encampments until reaching three city blocks, at which point crime was approximately as dispersed as in the rest of the city (Russell, 2020).

While few other studies have examined crime near encampments, other locations and approaches have been taken. When emergency winter homeless shelters open, some crimes appear to increase nearby, while others decrease (Faraji et al., 2018). While Faraji et al. (2018) did not use location quotients, they did find a concentration to dispersion dynamic, or distance decay, that mirrored other studies examining the spatial extent of criminogenic facilities and crime places (e.g., Groff, 2011; Ratcliffe, 2012). Multiple studies point to mobility considerations for risk of victimization (Ellsworth, 2019; Puente Guerrero, 2023; Tyler & Beal, 2010), the assertion being that not all PEH are at equal risk of victimization, and greater time spent in unguarded public places increases this risk (Hindelang et al., 1978). Mobility, geography, and outdoor exposure seem to matter.

Despite the importance of outdoor locations and whether homeless encampments are likely to increase risk of victimization, limited research exists on encampment geographies (Chamard, 2012, 2015, 2010; Harrell, 2019; Russell, 2020). Targeted efforts to disperse encampments in Los Angeles appear to have significantly reduced crime in some areas (Burk & McDonald, 2010), but encampments share the same public spaces as other activities, and the research is far from conclusive (e.g., Allen & Nolan, 2022). Encampments commonly appear close to transportation routes, parks, trails, homeless service providers, and grocery stores (Chamard, 2012, 2015, 2010; Harrell, 2019). Campsites are significantly more likely to be near transit nodes, service providers, and in mixed-use residential zoning areas in Portland, Oregon (Harrell, 2019). Chamard (2015) found that nearly half of encampments identified in police calls for service appeared in right-of-way land use areas, such as along roadways. Encampments obscured from view, within abandoned buildings, spatially isolated, or otherwise camouflaged may be less likely to be reported and less likely to be adjacent to roads (see e.g., Devuono-Powell, 2013). Chamard (2015, pp.158–159) also found that significant geographic differences exist between encampment locations, residential land use parcels, and multifamily land use parcels. For example, on average, encampments were significantly closer to streams, bars, gas stations, and tobacco stores than multifamily residential parcels.

Research on PEH mobility commonly differentiates involuntary and voluntary movement. Involuntary homeless movement typically focuses on law enforcement, as an extension of the city, forcing PEH to certain areas or out of certain areas. For example, Snow & Mulcahy (2001) distinguish containment, displacement, and exclusion strategies, all possessing within-category nuance. Wardhaugh (1996) similarly indicates a spatial confinement associated with a combination of necessary service locations coupled with continuous spatial displacement. Research on spatial marginalization has expanded greatly in recent decades (see e.g., Kaufman, 2022; Rose, 2019). Recent simulation models found that involuntary displacement can especially worsen health outcomes and overdose concerns among unhoused persons (Barocas et al., 2023). However, interaction of law enforcement with unhoused persons is often not a discretionary matter, but rather, the result of inadequate solutions for a situation that businesses and residents complain to the police about (Herring, 2019, 2021). It also appears that the police do not believe homelessness is a problem that they are adequately prepared to handle, and officer decision making maybe the same as that used by non-police homeless outreach workers (Brown, 2023).

The voluntary movement research regularly finds a pattern to unhoused movement, based on land features, everyday necessities, service use, and social networks (Cloke et al., 2003; Iwata & Karato, 2011; Jocoy & Del Casino, 2010; May, 2003; Rollinson, 1998; Suzuki, 2008; Whynes, 1991; Wolch et al., 1993). Voluntary unhoused movement tends to be constrained because of service use requirements or practical considerations about where income can be secured in relation to available sleeping places (Jocoy & Del Casino, 2010; Rollinson, 1998). While most PEH mobility research relies on surveying or interviews with the unhoused, studies have also used Global Positioning System (GPS) devices to generate real-time and potentially more objective mobility data (see Šimon et al., 2020). There seem to be differences in mobility when differentiating small town versus large city mobility, the use of public transportation, economic opportunity, and the housing conditions of people (Jocoy & Del Casino, 2010; Šimon et al., 2020). While a substantial number of determinants of unhoused mobility have been observed in the literature, to our knowledge, they have never considered mobility regularity in relation to data collected by law enforcement agencies. The present study asserts that geographic police data may be leveraged to identify places where specific PEH may be contacted for service provision and navigational support.

The present study

This study has two objectives and three corresponding hypotheses. The first is to analyze the concentration of crime, as a potential harm, immediately adjacent to two operationalizations of unhoused encampments. One operationalization is of resident reported encampments to the city, and the second is of encampments identified via police tickets issued for camping on public property. This has the potential to affirm previous findings that crime concentrates around encampments (Russell, 2020), while contributing to considerations about how resident encampment reporting may differ from police enforcement. This also provides added support to concerns that persons inhabiting encampments may be at greater risk of victimization than elsewhere in the city. This study employs location quotients to establish whether crime is more concentrated adjacent to encampments than comparably sized areas elsewhere in the city.

Hypothesis 1

Crime concentrates the closer one travels to identified unhoused encampments.

The second objective is to identify whether tickets issued by police for camping on public property could be useful for outreach efforts. In order to be useful, one would expect that camping tickets are concentrated among some persons and the camping tickets occur close enough to each other that recommendations are not to search large geographic areas.Footnote 1 The concentration of tickets among some people suggests that some are more exposed to the potential harms as identified through lifestyle exposure theory (Ellsworth, 2019), and resource allocation can occur in an evidence-based manner rather than randomly. The proximity of camping tickets near past tickets suggests patterned mobility and underlying reasons that some areas would be selected. Knowing this, searching for someone in a city is easier than searching a broader geographic area, or searching randomly. If the same PEH are not habitually ticketed with camping on public property, or if repeated tickets are issued in spatially distant locations throughout the city, camping ticket locations could be of limited value as an additional outreach tool.

Hypothesis 2

Camping tickets concentrate among a small number of individuals.

Hypothesis 3

A person’s camping tickets overtime are spatially proximate to their past tickets.

This study provides insight for practitioners attempting to contact persons experiencing homelessness, and it furthers academic discussion in multiple ways. Testing the hypotheses provides additional insight into harms near encampments, and a previously underexplored benefit of cross-agency partnerships for understanding homelessness in a community. The stability of these geographies could suggest nuanced decision-making about optimal encampment location. It also may shed light on constraints of the built environment regarding service utilization locations. This study represents a substantive addition to research on the geographic mobility of PEHs given that previous research did not approach unhoused sleeping locations from the perspective of formal police responses to camping on public property.

Methods

Data

The location for the present study is Boulder, Colorado, from November 2021 through October 2022. Boulder has a temperate climate, approximately 110,000 residents according to the US Census, and is located at the base of a mountain range in northern Colorado. For identifying homeless encampments, prior research tended to use either calls for service to police departments (Chamard, 2015) or resident reported encampments to the city (Russell, 2020). This study uses both police data and resident reported encampment data. First, Boulder Colorado provides an online reporting tool for community members to report various kinds of perceived problems, including unsanctioned camping in the city. Community member reported unsanctioned camping data from November 2021 through October 2022 were used. After geocoding all reports within the city limits, there were 4299 unsanctioned camping reports, or approximately 358 per month during this period. Those data are routed to the city Utility Department that prioritizes inspections, notifications, and cleanups in accordance with city policy. The Boulder Police Department (BPD) works with homeless outreach workers and local service providers to connect PEH at encampments to services whenever possible; however, BPD is not made aware of community member reports of unsanctioned camping, and thus community members’ reports of encampments are wholly separate from encampment tickets issued by officers.Footnote 2 Additionally, community member reports do not contain individual-level data pertaining to the persons residing in encampments, unlike police ticket data.

The second data source are all BPD tickets issued during this 12-month period for camping on public property. Officers issued 664 camping tickets in the city, or approximately 55 per month during this period. Officers ticket individuals for camping on public property during many kinds of officer-resident encounters, including to clear flashflood areas, building entrances, and pedestrian-trail underpasses. BPD camping ticket data include several variables of interest for the present study. Unique person identifiers, person demographics, unique ticket identifiers, and both spatial and temporal data are available. Spatial data include longitude and latitude values used to geocode circumstances in ArcGIS Pro, and the person identifiers are used to distinguish persons ticketed.

The third data source are BPD crime offenses for the 12-month period of November 2021 through October 2022. This study uses all non-traffic and non-camping offenses occurring within the city. Offense data were categorized to be compliant with the National Incident Based Reporting System (NIBRS), and include categories for offenses pertaining to persons, property, society, and all other offenses (FBI, 2018). Person crimes include offenses like assault, intimidation, and homicide, whereas property crimes include offenses like burglary, motor vehicle theft, and shoplifting. Society offenses include drug crimes, animal crimes, and disorderly conduct, whereas all other offenses can include serving warrants, protection order violations, or obstructing government operations. The city experienced 1205 person crimes, 5361 property crimes, 1783 society crimes, and 3048 other crimes during this time. These data are compared against both methods of identifying encampments, community member reports and BPD issued tickets for camping, to measure geographic crime concentration.

Analysis

Research has already demonstrated that homeless encampments, offending, and victimization do not occur randomly in urban settings (Chamard, 2015; Faraji et al., 2018; Harrell, 2019; Rusell, 2020; Yoo & Wheeler, 2019) but are instead concentrated, typically close to transit nodes, homeless service providers, greenspace, and some kinds of businesses. Given this, the analysis begins with location quotients to consider the concentration of offenses near encampment sites identified through community members’ reported unsanctioned camps, and through BPD tickets data.

Location quotients are commonly used as a density measure calculating one rate, counts per area, compared to another rate, such as counts per area elsewhere (Hodgen & Wuschke, 2023; McCord & Ratcliffe, 2007; Russell, 2020). They are often expressed as:

$$LQ = \frac{ci/ai}{{cr/ar}}$$

where ci is the crime that occurs adjacent to encampments, ai is the area of this measured adjacency, cr is the crime that occurs in the city, and ar is the area of the entire city. This equation is then used to estimate the crime per area in a select place in relation to the crime per area for the city. Buffers at 300, 600, and 900 feet were used. These distances were used because they are equal distances from the origin points, and because these distance thresholds are similar to other studies (Faraji et al., 2018; McCord & Ratcliffe, 2007; Russell, 2020). A LQ value of two or greater has been used to differentiate consequential concentrations of crime, though LQs are purely descriptive (Groff, 2011; McCord & Ratcliffe, 2007; Russell, 2020).

For the second step of the analysis, using BPD individual data to understand the geography of repeat camping ticket locations, an expanded table was created such that each of the 664 encampment charge coordinate pairs was compared to every other coordinate pair in Microsoft: Excel. This resulted in 440,896 rows of data where Column1 contained one a total of 664 times, then two a total of 664 times, and so forth. Column2 contained numbers one through 664 sequentially, 664 times. Because unique tickets and unique persons were identified per ticket, the data were then cleaned to remove every row where there was duplication by Column1: Ticket1-Person1 to Column2: Ticket1-Person1. This ensured that unique individuals’ camping tickets were compared 1–2, 1–3, 1–4, and not double counting 1–1, 2–2, and so forth.

In order to then calculate coordinate pair distances, the Haversine formula was used (Chopde & Nichat, 2013; Maria et al., 2020; Shylaja, 2015). The Haversine formula is one of several choices to estimate distances, and it tends to yield similar results to other calculations (Maria et al., 2020). The equation tends to be expressed as below (e.g., Chopde & Nichat, 2013) in which phi are latitude values and psi are longitude values.

$${\text{Distance}} = 2r\sin^{ - 1} \left( {\sqrt {\sin 2 \left( {\left( {\Phi 2 {-} \Phi 1} \right)/2} \right) + \cos \left( {\Phi 1} \right)\cos \left( {\Phi 2} \right)\sin 2\left( {\left( {\uppsi 2 {-} \uppsi 1} \right)/2} \right)} } \right)$$

This equation was recreated in Microsoft: Excel with the Earth radial value expressed in feet to be consistent with the location quotient areas similarly measured in feet. The Excel function constructed Haversine Formula is:

$$\begin{aligned} {\mathbf{Distance = }} & {\mathbf{20925722*2*ATAN2(SQRT(1}} - {\mathbf{(SIN(((Lat2*PI()/180)}} \\ & - \,{\mathbf{(Lat1*PI()/180))/2)}}^{{\mathbf{2}}} {\mathbf{ + COS(Lat1*PI()/180)}} \\ & {\mathbf{*}}\,{\mathbf{COS(Lat2*PI()/180)*SIN(((Long2*PI()/180)}} \\ & - \,{\mathbf{(Long1*PI()/180))/2)}}^{{\mathbf{2}}} {\mathbf{)),}}\,{\mathbf{SQRT(SIN(((Lat2*PI()/180)}} - {\mathbf{(Lat1*PI()/180))/2)}}^{{\mathbf{2}}} \\ & + \,{\mathbf{COS(Lat1*PI()/180)*COS(Lat2*PI()/180)}} \\ & {\mathbf{*}}\,{\mathbf{SIN(((Long2*PI()/180)}} - {\mathbf{(Long1*PI()/180))/2)}}^{{\mathbf{2}}} {\mathbf{)}}) \\ \end{aligned}$$

In this function Lat1, Long1 are associated with the first latitude and longitude coordinate pair (e.g., Person1-Ticket1), and Lat2, Long2 are associated with the second latitude and longitude coordinate pair (e.g., Person1-Ticket2). The final distance values were randomly sampled and verified for accuracy in ArcGIS Pro using the Measure Distance tool.

Results

Crime is substantially more concentrated near Boulder resident encampment reports and police camping tickets than they are concentrated elsewhere in the city. Additionally, using multiple buffered distances, crime is observed to be less concentrated per foot the further distance travelled from these reports or tickets. Table 1 below provides the LQ values for crime near these circumstances. The aggregation of all offense types, all crime, was approximately 2.96 times more concentrated in the 300’ areas around resident identified encampments than crime elsewhere in the city. All crime was 6.67 times more concentrated within 300’ of all officer encampment tickets in the city during this time period. Each LQ for offenses around officer encampment tickets was greater than LQ values for offenses around resident identified encampments. At the 300’ buffers, person crimes appear most concentrated, compared to the other offense categories near resident identified encampments, whereas society crimes were most concentrated near officer identified encampments.

Table 1 Crime location quotient values near homeless encampments
Full size table

Identifying that crime is more concentrated near two datasets pertaining to homeless encampment locations, the focus now turns to individuals noted in the police department encampment ticket data. In the 12-month period, there were 237 distinct persons issued the 664 total encampment tickets. Most of these individuals were identified by officers as being white, male, and an average of 42 years old. The number of tickets issued to each person followed a pareto principle, or j-curve (Martinez et al., 2017; Weisburd, 2015), with 2.8 mean tickets per person and a median value of 1 ticket per person. However, the most ticketed individual received 37 encampment tickets, or about three per month. The 90th percentile of ticketed individuals all received six or more tickets but were demographically comparable to the entire population of individuals. These 24 most-ticketed individuals represented 10% of all persons ticketed but accounted for approximately 45% of all tickets issued. Table 2 below provides descriptive information on the individuals who received camping tickets by police officers.

Table 2 Descriptive statistics of ticketed individuals
Full size table

Despite the large number of camps associated with the 24 individuals identified in the police camping tickets, there were fewer unique camp locations. These most ticketed individuals all had at least two camps in the same location. Table 3 below provides greater detail on the 24 most ticketed persons. The most ticketed person received 37 tickets at seven unique places, receiving about five tickets at different times at each of their frequented camping places. Among all 24 people, individuals camped an average of more than twice at the same location. Moreover, these unique locations were often spatially proximate to each other. The most ticketed individual, a 39-year-old white female with 37 tickets in 12 months, on average built camps within 663’ of her other camps. Half of the most ticketed persons camped an average of 835’ or less to their other encampment locations. In the city’s downtown area, the 835’ represents approximately 2.5 city blocks. Every one of these persons had more camping tickets than unique camping locations, in other words, every person used at least one location to camp multiple times.

Table 3 Descriptive Statistics on individuals with most camping tickets
Full size table

Discussion

The results of the present study suggest crime is concentrated near encampments, and police issued camping tickets are concentrated among certain people in certain places. This indicates general support for the idea that the police collect data that can help with homeless outreach. Hypothesis one is affirmed by the finding that crime becomes more concentrated as one travels toward encampments, from 900 to 600 feet to 300 feet. This is true when considering resident reported encampments and when considering police tickets for camping on public property. The finding also holds true regardless of the crime category. Crime is 2.96 times more concentrated within 300 feet of resident identified encampments and 6.67 times more concentrated within 300 feet of officer issued camping tickets compared to the concentration of crime elsewhere in the city. This supports the limited previous research in other study locations (Russell, 2020) while also finding that crime is more concentrated closer to officer tickets than resident reports of camping.

Several potential reasons exist for the heightened concentration of crime near officer issued camping tickets compared to resident reported encampments. One explanation is that police officers are deployed in areas already experiencing other crimes where camping is co-observed, and tickets are issued (Braga et al., 2012; Martinez et al., 2017; Weisburd, 2015). Due to the nonrandom nature of crime in a city, officers are regularly assigned to engage in proactive patrols in areas that have historically experienced crime—such as areas with liquor selling establishments or near off-campus university student housing (Cozens et al., 2019). Proximity of those areas to places likely to have encampments is less clear. A second explanation is that when officers contact persons at encampments, on-scene warrant checks conducted by officers reveal other crimes that would not be readily visible to passersby that report the encampment to the city. Broadly, it is possible that the same conditions affecting crime reporting also affect encampment reporting, but those factors do not impact officer issuance of tickets.

It is also important to note that while other studies use similar methods to assess whether places cause crime (Faraji et al., 2018), the present study makes no such claim. It is possible that encampments increase crime (Russell, 2020), or that encampments are relegated to parts of a city already experiencing elevated levels of crime, pushed to those areas due to municipal poverty governance practices (Kaufman, 2022; Rose, 2019; Snow & Mulcahy, 2001). To disentangle this relationship further longitudinal methods using time and space variables would be necessary, and such a study on crime and unhoused encampments has not yet been published. Simultaneity is also possible, with crime in public places increasing the likelihood of built encampments and encampments increasing crime, requiring statistical methods not frequently seen in this literature.

The second hypothesis is supported that police camping tickets are concentrated among a small number of individuals. The median number of charges among all persons issued camping charges is one, but the overall distribution of charges to persons is heavily skewed. The 10% of persons with the most camping charges received approximately 45% of all camping charges issued by this agency in a 12-month period. These 24 individuals of the 237 total unique persons, received 301 of the 664 total camping charges issued. On average each member of this group of frequently ticketed persons received a camping ticket every other month. The most charged person received a camping charge about every 10 days during this 12-month period.

Persons with police camping tickets during this time are somewhat unique compared to the 2020 point-in-time (PIT) count of PEH in the county that includes the study city. According to the PIT count, there were 668 individuals counted, among them about 62% were male, and about 70% of them were White (City of Boulder, 2023). Among the 237 individuals ticketed by the police with camping in public places from November 2021 to October 2022, 72% were male, and about 87% were white. The average age of persons ticketed, 42 years old, is like the greatest proportion of chronic PEH being in the 45–54 age category (City of Boulder, 2023). However, none of the ticketed persons were under the age of 30. Altogether this suggests that persons ticketed with camping by the police in this jurisdiction were more likely to be male, white, and older than would be expected by the PIT count in the county.

Similar to other spatial crime concentration research (Martinez et al., 2017; Weisburd, 2015), persons who received multiple camping tickets were often ticketed in nearby places overtime. The third hypothesis is supported by results that regularly ticketed persons each camped in at least one place multiple times. The individual with the most camping tickets received 37 tickets in seven places, with the seven places being an average of 663 feet from each other. Among the top 24 individuals, each of their camping tickets commonly occurred within three city blocks of their other camping charges. This suggests that police ticket geographies may be of some utility when attempting to contact PEH who sleep in public places. Moreover, camping tickets regularly occurred near the city’s Business Improvement District, near public transit routes and along a greenspace with includes parks and a creek—this is consistent with previous research on homeless encampment geographies (Chamard, 2010, 2015; Harrell, 2019).

Multiple underlying factors may be responsible for the finding that persons are ticketed with camping in nearby places overtime. First, officers issue tickets and officers are often assigned to high crime locations as apart of hot-spot policing crime reduction strategies (Braga et al., 2012), and the police may be present in areas with frequent community member calls for service (Herring, 2019, 2021). Underlying geographic reasons also exist for why persons would stay in the same places outdoors overtime. The nearness to services, places to avoid bad weather, proximity to transportation routes, and nearby available income sources are all likely important for the everyday geographies that result in persons sleeping in certain places where they may also be issued tickets (Chamard, 2015; Rollinson, 1998; Wardhaugh, 1996).

Future research and limitations

A substantial literature exists regarding the philosophical and legal considerations about whether sleeping in public places, or other PEH activities, should be criminalized (see e.g., Mitchell, 1997, 1998; Mitchell & Heynen, 2009; Smith, 2005). While the present study focuses on whether already collected data are of geographic utility for PEH outreach, not on whether the data should be collected, we raise several questions that future research should consider addressing. One question is whether tickets for camping in public places are “effective” for their intended purpose. We assume that the intended purpose is to decrease the prevalence of persons sleeping in outdoor spaces and direct the ticketed to relevant services and housing. To our knowledge this has not been well established, with some literature positing that negative consequences outweigh the benefits (e.g., Robinson, 2019). Examining effectiveness would also likely require court data to understand ticket dismissals, alternative sentencing options, and individual successes when receiving non-criminal justice services. Similarly, few studies have assessed whether public space enforcement efforts increase safety, particularly for the people sleeping outdoors that are prone to victimization and premature mortality (Boulder County, 2022; Waugh, 2023; McCormick, 2022; Ellsworth, 2019).

Future research could also shed greater clarity on outdoor space cohabitation and the role of social networks. The present study focused on individual persons ticketed at certain places, but prior PEH research point to the importance of social networks for where one travels (Iwata & Karato, 2011; Peters & Robbilard, 2009; Wolch et al., 1993). Where someone’s friends sleep outdoors could be substantially more important than nearness to a bus stop or homeless service provider. By analyzing the most network-central persons, service providers may identify the largest influencers of PEH mobility and sleeping locations.

There are several limitations of the present study that warrant discussion. First, the data used only included individual identifiers and geographies associated with the police issued camping tickets, but it is possible that persons ticketed with camping were victims or offenders associated with other police-documented situations. A more inclusive approach may further refine individuals’ geographies with more police data and marry those data to information available through service providers, or the city’s Housing and Human Services to understand where gaps exist.

A second limitation associated with identifying persons from police camping tickets is that it is unclear if discretionary or non-discretionary considerations are associated with certain people receiving these tickets or receiving them in certain places. For example, given concerns about PEH drowning (e.g., Boulder City of Boulder, 2023; County, 2022), it is possible that officers are more likely to ticket persons near waterways to prevent camping there compared to outdoor locations distant to creeks. While some research suggests officer discretion is secondary to PEH-related community member calls for service (Herring, 2019, 2021), additional research is needed in police discretion during PEH encounters (Brown, 2023; Noe, 2017; Reinhard, 2023).

Conclusion

Collectively, this study’s results suggest that camping tickets are concentrated among some persons, PEH have geographically patterned sleeping habits, and these outdoor sleeping areas may have more concentrated crime occurring nearby. Harm mitigation efforts and outdoor homeless outreach strategies could benefit from taking this into account, and law enforcement agencies appear to collect data that can be leveraged to help problem-solving partnerships. While the present study only used camping ticket geographies, and not the geographies of all tickets and reports an individual’s associated with, using all data could add further geographic specificity for where someone may be successfully contacted. Homelessness issues are complex, require interagency partnerships to address, and are aided by the utilization of many sources of information. This includes information inherently collected by local police departments.