The nature and frequency of homeless contacts with the police

ABSTRACT

This study assesses differences between citizen and police initiated homeless-involved dispatch incidents, as well as the resulting police reports. This is accomplished by analyzing dispatch data and police reports identified as involving homeless individuals in a mid-sized city in the central United States from 2020 to 2021. Altogether 12,843 dispatch incidents and 1,762 (13.72%) reports are analyzed. This study found 93% of documented police contacts to be citizen initiated. Incidents that did not involve the homeless were substantially more likely to be officer initiated than homeless-involved incidents. Significant differences are found on the basis of initiation type and offense type for nine of the 18 offense categories. Police initiated incidents are more likely to produce reports about failure to appear or comply, homeless encampments, and drugs or alcohol. Citizen initiated incidents are more likely to produce reports about trespassing, assault or menacing, and theft or shoplifting. The results support prior research (Herring, 2019) and suggest that care should be taken regarding assumptions about why officers contact the homeless. Given the non-criminal nature of many citizen generated incidents, some support exists for an alternative response to address loitering, suspicious persons, or other non-serious and non-criminal homeless incidents.

KEYWORDS:

• Homelessness
• crime
• calls-for-service
• complaint-oriented policing‌
• police

Acknowledgements

I would like to thank Bethany Ciesielski for reviewing earlier drafts for grammar and structure. Sergeant Jeff Kessler deserves thanks for consultations about the intersectionality of dispatch and police data and how homelessness relates to both. Lastly, Dr. Troy Payne deserves thanks for pointing out additional literature that had been missed.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Numerous studies present discuss semantic complications and definitional issues with the term “homeless.” Cordray and Pion (1991) discuss some of these

2 Russell (2020) and Yoo and Wheeler (2019) both aggregated offense categories into three categories, and two categories, respectively. Snow et al. (1989) distinguished between all Part I and Part II Uniform Crime Report (UCR) types. Andresen and Linning (2012) discuss concerns about the aggregation of crime types at different units of analysis; it potentially conceals important distinctions, and in an understudied area like homeless offending, it is especially problematic.

3 The most recent homeless PIT count suggested 450 individuals according to the city website. This is in relation to the approximately 110,000 residents according to the US Census of the same year. Note that the PIT count is limited by several factors (see e.g., Cordray & Pion, 1991) and likely an undercount; the residential population estimate is also likely an undercount because of differences between residential and ambient populations.

4 While arrest data are not presented here, custodial arrests were uncommon in this agency during 2020–2021 among homeless involved incidents.

Additional information

Funding

No funding was secured to complete this research.

Notes on contributors

Daniel Reinhard

Daniel Reinhard, PhD, is the Chief Data Analyst for the Boulder Police Department and has spent years working with law enforcement in various capacities before being in his current role. He is an invited speaker on analyzing and responding to homelessness at the Problem Oriented Policing Conference, and some of his other homelessness research can be found in International Criminal Justice Review, and Journal of Qualitative Criminal Justice and Criminology. His other research is on varied subjects but typically involves environmental criminology and includes geographic information systems.