THOUGHTS AND PERSPECTIVES ON PSH FROM THE GROUND
Robert Mitchell | @robertmitchellv
2018-09-04
bash-3.2$ whoami
- I work at Skid Row Housing Trust as a data analyst
- (I'll just refer to it as SRHT from now on)
- I have been working in this role for around three years
- I am self taught as a programmer/analyst
- Background in philosophy, comp lit, french, information science
A BIT ABOUT SRHT
- We do three things:
Develop affordable housing; specifically permanent supportive housingProvide supportive services to the residents of our affordable housing projectsHave an affiliate manage the properties, repairs, et cetera
A BIT ABOUT SRHT
- We believe in the Housing First model
- We believe in Harm Reduction
- This is pretty much industry standard for the most part
WHAT I WANT TO COVER
- Research ➡️ policy threshold ➡️ providerspace feedback loop
- Building in measurement when there was none
- And the data infrastructure associated with this effort
- Current work and the direction SRHT is going in connection to data
WHY TALK ABOUT THIS FEEDBACK LOOP?
- It reflects a frustration that many providers feel
- It may not be something that is necessarily shared often and connects to the kind of policy advocacy SRHT would like to be a part of in the future
DEMONSTRATION STUDIES
- Pathways to Housing: Supported
Housing for Street-Dwelling Homeless
Individuals With Psychiatric Disabilities (Tsemberis, 2000)
- Most HF people are familiar with this study, it's widely used to justify HF, which the providerspace is behind
- These survival figures help us advocate for HF
DEMONSTRATION STUDIES
- But then we see this retention figure
- On the one hand, this shows HF efficacy
- On the other hand, will this be the bar providers are held to?
SHAPING PUBLIC OPINION
SHAPING PUBLIC OPINION
APPLES TO APPLES?
- Is New York like Los Angeles?
- Is Utah like Los Angeles?
- How does this shape public opinion?
IS THIS A GOOD MEASURE?
- 🙅♀️
- I don't think this is a good measure since it says nothing of provider effort or quality
- It's effectively housing churn until there is a uniform retention calculation that controls for success and can also control for effort on the side of the provider
WHAT DATA DID I INHERIT?
- Lots of tables of things; nothing that is connected by keys or can be aggregated meaningfully
- Data that is lost on paper somewhere or in excel in a lonely, forgotten directory
- 💡 are there mandatory reports required through some compliance function?
Is this even useful? What does it communicate?Yeah you HUD mandated Annual Progress Report (APR) 👀
- Is there research literature? What is cited in those studies?
Are there comparisons to draw? Can you chase the citations?
BUILDING MEASUREMENT
- There were no measurements–nothing that I could use to demonstrate change in housing over time in a way that could be abstracted to an aggregate level meaningfully
- I didn't have the right kind of data to demonstrate good outcomes
BUT WHAT ARE OUTCOMES?
- Please no eye-rolling for this slide 🙄
- This word takes on a different meaning in the provider space
- For many the meaning is:
We need to show funders that we are effective–we need to figure out how to demonstrate this with data
RETURNING TO THE FEEDBACK LOOP
- Providers worry about data they share because it can affect funding
- If policy thresholds are set via demonstration studies, has anyone asked:
- Is this realistic for CoCs and providers?
- Did a replica of this model scale in the real world?
- Does this model fit for our CoC?
- Is the study an apples to apples comparison? E.g., funding, population, weather, et cetera.
- Does this conscript providers to succeed a priori or lose funding?
HOW DOES A PROVIDER BREAK FREE?
- Collect our own data
- Use validated research instruments (don't reinvent the wheel)
- Think descriptive not prescriptive
- Have no agenda
- Observe what actually happens in PSH
WHY DO I THINK THIS WILL WORK?
- Unless providers are able to take part in helping to shape the narrative around expectation setting for PSH, we will continue to be a step behind unable to risk losing funding
- Since we don't have something we are trying to prove, we can just observe our residents and gain insight into how they chance while in PSH
- When we use data rather than anecdote, we hope our perspective will carry more weight
WHAT WE ARE USING CURRENTLY
- World Health Organization Disability Assessment Schedule 2.0
- Modified Colorado Symptom Index
- Patient Health Questionnaire
- Generalized Anxiety Disorder
- Hospital Social Functioning Questionnaire
- PTSD Checklist (Civilian)
- Slapped, Threatened, and Throw
- Hurt, Insulted, Threatened with Harm, and Screamed
- Suicide Behaviors Questionnaire-Revised
- Alcohol Use Disorders Identification Test
- Drug Abuse Screening Test
HOW OFTEN DO WE COLLECT DATA?
- We conduct a longer survey at move in and follow-ups every six months
- We are working toward full coverage of all residents but this is difficult since there are so many different funding sources for our affordable housing projects
- Maintaining uniformity across our portfolio is no small task, but it's a value of ours
Data infrastructure
- This was a long process since it involved getting rid of vendors that were taking advantage of us
- Reorienting folks to the fact that we can do a lot with code and open source projects
- That building these tools out is not as scary as it sounds
WHAT AM I WORKING ON?
- A logistic regression model to predict negative exits from housing
- The 💡 is that homelessness recidivism will be an important area of study soon
- A generalized linear mixed effects model for individual change in housing
- So that we can learn what an average trajectory even looks like
- Automated case plan triage
- Based on our assessment, put together an individualized case plan based on response data