Streets to Shelter: Understanding Attribute-Based Decisions in Shelter

• In Event: Unsheltered Homelessness, Shelter Attributes, and Housing Stability: Evidence from Policy, Practice, and Lived Experience

Abstract

The 2024 Supreme Court decision in the Grants Pass v Johnson case overturned a prior ruling the temporarily prohibited cities from clearing homeless tent encampments without the offer of a viable alternative shelter. In the wake of that decision, California governor Gavin Newsom ordered the clearing of all California homeless encampments. Proponents of tent encampment clearing argue that some individuals are resistant to offers of shelter, but there is limited quantitative data to substantiate those claims. At the same time, very little is known about the demand-side of housing solutions for people experiencing unsheltered homelessness. 

Shelter options vary dramatically, but the typical shelter available for chronically homeless single adults is a congregate facility, meaning no private locking door and shared sleeping spaces. These facilities also often have several other attributes which may be undesirable, such as requirements to check in every day by a certain time, curfews, no ability to stay with a romantic or life partner, time limits on total residence, or locations far from the tent community where one might be most comfortable. 

This paper combines a discrete choice survey experiment with supplemental survey evidence to provide causal insights into the factors influencing individuals’ decisions to stay in shelter. We address two questions: (1) Which shelter attributes are relatively most important to individuals? And (2) Which attributes most strongly affect shelter uptake? We examine preferences for features including length of stay, room privacy, presence of security guards, presence of case managers, proximity to transit and current location, substance use policy, resident participation in rulemaking, curfew policy, pet policy, partner policy, and cleanliness. 

Our data come from 605 interviews conducted with unsheltered individuals in the summer of 2025 across three Bay Area counties. We find that, when comparing preferences between shelters, nearly all defined attributes play an important role. On average, individuals prefer shelters that offer private rooms, have on-site case managers, do not impose a curfew, are located near both their current location and public transit, allow couples and pets, have no substance use restrictions, and maintain clean facilities. The only attributes that were not rated as important were the presence of on-site security and opportunities for resident participation in rulemaking. 

However, when respondents consider whether to enter a shelter relative to where they stayed the previous night, only policies related to privacy and personal freedom significantly influence decisions. Having a private bedroom, the absence of curfew or substance use restrictions, and the allowance of couples are key factors. More broadly, we find a strong desire among respondents to exit homelessness, 90% reported wanting to be off the streets within the next year. This closely mirrors what we find in the conjoint analysis where individuals most commonly prefer shelter to their previous night accommodation, selecting shelter approximately 70% of the time. Our work contributes to literature on unsheltered homelessness, while offering insights for policymakers into the mismatch between available housing options and housing options that meet the needs of unsheltered populations.

Author

• Stephanie Casey Pierce, University of Tennessee (Knoxville)
  Presenting Author