Learn from Our Mistakes: Challenges and Opportunities in Randomized Housing-First Communities

Photograph of an apartment building
Figure 1 Photograph by Alex Borland. CCO.

Submitted by: Molly Brown and Camilla Cummings


Sometimes community-based research studies fail.
We can learn from our mistakes through a process of honest reflection and publicizing our lessons learned.

There is a robust evidence base for the effectiveness of the Housing First model to address the needs of individuals experiencing chronic homelessness (Aubry, Nelson, & Tsemberis, 2015).

Housing First is a model of permanent supportive housing without preconditions for treatment participation or abstinence from drugs or alcohol. Wraparound support services are offered to tenants aiming to meet their person-centered goals. However, important gaps in the literature remain, such as examinations of differential outcomes in various Housing First models. For example, differences in quality of life outcomes in apartment buildings with higher versus lower concentrations of individuals with histories of homelessness has not yet been rigorously studied.

“…we hope our analysis of the shortcomings within our project illuminates for future researchers the current real-world issues that may impact Housing First studies.”

We want to better understand these differing outcomes between people and communities using rigorous research designs. A multidisciplinary research team collaborated with Downtown Emergency Service Center (DESC), a large homeless service provider in Seattle, to conduct a mixed method randomized trial of two models of Housing First

Individuals experiencing chronic homelessness were randomly assigned to one of two type of units:

  • single-site Housing First (i.e., housing in which all apartments are located in the same building and support services are provided on-site)
  • scattered-site Housing First (i.e., apartments located throughout a community in buildings consisting of partially non-Housing First apartments with service teams delivering support in the community)

We drew on person-environment fit theory, which suggests that alignment between an individual’s characteristics and the characteristics of their environment helps buffer stress and enhance well-being (Caplan & Harrison, 1993). The study aimed to examine housing and quality of life outcomes among individuals in the two housing conditions, with the alignment of tenant housing preference and their housing environment being a key predictor.

However, the study faced numerous barriers to its success. For instance, the study occurred in the midst of the community’s transition to a coordinated entry system, leading to our partnering agency’s loss of control over their housing procedures; and there were challenges with staff buy-in for the random assignment of tenants to the housing interventions. These challenges resulted in low enrollment and led to the study’s premature termination.

Despite our study’s failure, our experience sheds light on challenges to anticipate and overcome when developing Housing First trials with limited resources. Secondarily, our first person account unabashedly presents an example of failed research. Such reports are rarely depicted in academic literature. This paper acknowledges the difficulty of conducting rigorous research in community settings and provides some practical considerations for others engaged in university-community partnership research endeavors.

First-Person Accounts

In contrast to a traditional research article, this paper is written as a “First Person Account” in which written, personal accounts from all key stakeholders on the research team are presented. The accounts provide stakeholders’ perspectives on the challenges and facilitators that impacted the study and its success.


Community psychologists have played a key role in developing the evidence base for the Housing First model, and their work served as the basis for the study we aimed to conduct. Moreover, the field of Community Psychology has produced substantial literature on community-based research and strategies for effective collaboration with community stakeholders. One of our primary pitfalls was the lack of time we had available to develop a strong foundation for our study in the community which created a barrier to following best practices of the field. As such, we draw heavily on the Community Psychology literature for framing our recommendations to future researchers.

Lessons Learned
Experimental research on Housing First relying on existing community housing resources may face complex barriers. Lessons learned include:

  • Buy-in and trust with organization- and policy-level stakeholders is crucial.
  • Tensions between organizational culture and the research protocol should be addressed, such as reluctance to support a randomized design.
  • Research success may be promoted by balancing flexibility and structure in the research activities, such as flexibly adapting to challenges that arise with study activities, while maintaining important structures, such as training and supervision, to ensure the safety of research assistants and participants.
  • In the context of the current homelessness service landscape, researchers will likely need to navigate their research protocols within coordinated entry systems, which manage housing placements at the city- or county-level. In other words, researchers will need to collaborate with larger homeless service coordinating bodies, as opposed to solely collaborating with service organizations, to implement studies.

What Does This Mean For?

Research and Evaluation: Community-based experimental research is challenging. Community-based studies often do fail; however, these failures are rarely, if ever, articulated in the academic literature. Our paper aims to generate dialogue among Community Psychologists who have experienced struggles in their research. Although barriers to research are fluid and depend on the ecological context in which the study is situated, we aimed to develop lessons learned and recommendations that will translate to future studies.

Original Citation: Brown, M., Tran, A., Cummings, C., Fay, L., Malone, D., Fyall, R., & Tsemberis, S. (2020). Attempting randomized Housing First research in a community context: Reflections on failure. American Journal of Community Psychology. Advance online publication. doi:10.1002/ajcp.12421


Aubry, T., Nelson, G., & Tsemberis, S. (2015). Housing first for people with severe mental illness who are homeless: A review of the research and findings from the At Home—Chez Soi demonstration project. The Canadian Journal of Psychiatry60(11), 467-474. https://doi.org/10.1177/070674371506001102

Caplan, R. D., & Harrison, R. V. (1993). Person‐environment fit theory: Some history, recent developments, and future directions. Journal of Social Issues, 49, 253–275. https://doi.org/10.1111/j.1540-4560.1993.tb01192.x

Download a .pdf of this page here.


Contact Us