1. Presented by:
Janet Salm
Regional Homelessness & Housing Services Evaluator, King County
Supporting System-Level Thinking to
End Homelessness in King County
Supporting system-level thinking to end homelessness in King County
Introduction
I know that homelessness isn’t usually a topic you’d associate with hope. I want to share with you my journey with Tableau, how I’ve used it to understand our data better, and as a result, to present compelling evidence to stakeholders to help them make better decisions. Along the way, I will try to share a few bright spots from the world of homelessness – there are a number of bright spots and trends right now that really do bring hope.
This is Rex Hohlbein. He’s an architect. He had an office in Fremont, near the canal, near Tableau’s offices, in fact. He started opening his door to homeless folks, offering a hot beverage, socks, a kind word, a listening ear. With a designer’s eye, he took great photos and shared them on Facebook, and started to connect, REALLY CONNECT, our homeless neighbors with people who care.
Really, it was an opportunity to let technology serve a higher purpose: connecting humans.
Stories like: replacing a car battery. Sometimes it’s not a lot of money. The connection is what matters.
Once you have a connection, it’s human to human, and only natural to give back: play a song on the guitar, paint a picture, carve a pendant…
Rex has now started a nonprofit to do this work full time, and this is one of the most exciting and inspirational things I’ve seen lately – I encourage you to go check out the Facebook page and get involved if you’re so moved…
I also wanted to start with these images and stories because I recognize that what we’re talking about here is data, but all the data is composed of individual stories, and we never want to forget that.
What Rex’s work shows is how strongly we yearn to help our neighbors and to connect on an individual basis. While the experience of homelessness is all too personal, if you experience it, the factors that play into an experience of homelessness are more likely system failures, not individual failures:
lack of housing affordability,
jobs that pay a living wage,
cuts to services like mental health,
and so on.
Luckily, there are systems set up to help people who have fallen through the cracks. Federal, state, local and philanthropic funding to address homelessness is not insubstantial. While it may not be enough, knowing what the resources are can help us tune our systems, make sure they’re producing all the benefits they were designed to deliver.
We collected data on all the public funding in our community, cleaned it up, annualized it, etc.
First, we used Excel and pivot tables. We got pretty far, but it took us a long time. For those of us who “think in spreadsheets”, it was awesome. For others, we translated our findings into Powerpoint, and we all know what Edward Tufte would say to THAT.
When we put it in Tableau, suddenly we could see: RELATIONSHIP (suburban vs rural, singles versus families);
And GEOGRAPHY: Being able to see where units are located gives the viewer an immediate sense of coverage across the county AND concentration in the downtown core.
Once we overlaid Council district boundaries, we could tell councilmembers how many units and dollars are in their jurisdictions.
INTERDEPENDENCE (federal / state / local sources)
And we got a whole new sense of the COMPLEXITY of the system (braided funding, sheer number of programs).
[foreshadow: there are a LOT of programs here – are they all doing the same thing? Producing the same results?]
Pause point: quick story. Ira Glass and creativity (gist: when you’re a beginner and you LOVE music, it is incredibly painful to hear your crappy efforts when you know the transcendent beauty you WANT to create – that distance will only be filled in with practice, and you must have courage) – by way of explaining how I have this picture in my head of the data visualizations I _want_ to create, and what I actually have the time to create doesn’t come anywhere close to the picture in my head. But I am going to get there!!
3 examples in this section, that are really 3 bright spots: families, single adults and youth and young adult populations
Families: Family Characteristics dashboard: Getting a view into this data for the first time took me weeks of work – first a week to clean the data, then endless SPSS and Excel analyses. We got Tableau after that, and now, whenever we want an update, we import the new data and it’s updated and ready to go. Before Tableau, we could only afford to look at this data once every 6 months or year. Now, the effort to look at the data is trivial.
Bright spot: With this view of family characteristics (first time homeless, strengths, low housing barriers, and they all want to WORK), we could see that this population really would succeed with a new model at the national level called rapid rehousing – as the name implies, it moves families back to housing quickly, in the private rental market with a short term rental subsidy and some case management.
Seeing the data like this is what convinced funders and providers to adopt this approach.
Rapid Rehousing moves families back to stability in an average of 38 days, instead of the 2 year stay in our transitional housing, after which they still need to move back to the private rental market.
Better outcomes for families, cheaper for the system, which means more families can be helped.
And btw, we used Tableau to put together dashboards on the project which we can share with the 6 funders and 4 providers on a regular basis!
Single adults: map of where people come from. Bright spot: with ONE visual, put to rest the whole “they come here for the services” or “they’re from Des Moines” thing.
Before: Excel analysis, recoding the zip codes by hand.
Now: Visually plotted on the dashboard, updated, can SEE – and seeing is believing.
Youth and young adult populations: YHC Dashboard on timing intervals – bright spot: understanding process and flow / timelags so we can address them
In Excel, this was a project that I’d take a run at every few months – and once I’d figured out one aspect (are providers slow in responding?), we would be unable to use the data because we hadn’t analyzed whether the central coordinating system had been slow in providing referrals. It simply never took off, because it spurred the blame game without a full picture. Now we’ve got it automated, full picture, and can update it regularly.
A bonus was that we could quickly toggle the information to analyze fairness of our system to see if youth of color were having different outcomes or taking longer to be connected to housing. Equity and social justice are so important to us at King County that we would have invested the time to get this project done, but I’m not sure how long it would have taken, or what else the project would have had to bump off the workplan.
Feds – HEARTH – focus on systems performance
Local leadership: RARE, BRIEF AND ONE TIME
System view: RARE BRIEF ONE TIME
Key performance indicators
Danger when you look at them from this level: either the audacious goals set are demoralizing and ignored; or they become too incremental (10% YOY that will take care of itself with minimal effort)
How do you make the system view REAL?
LIVE DEMO HERE: Show the three measures
Demo quartiles and reference lines
Show attempts at program type, population
When you get down to the program level, we saw a lot of variation.
BENEFITS:
Know more stuff – data exploration “cost” is trivial. Examples: could see those who stayed longest in shelter, begin to visualize client movement between programs
And externally, we are using automation to FREE THE DATA!
Supported by local Foundation (Bill & Melinda Gates), using some consultant resources (VizTric) – shoutouts
For the first time: access, real-time access for providers, funders, governing board members, journalists, and the public. Transparency while still keeping our client-level data safe.
In conclusion – I hope this was useful for you!
For me: With the time saved, and the power of telling stories, my time will be freed up to do more of what will truly benefit our system: I’m planning to join our financials to our performance data to see per-successful outcomes cost and our next big project will be to design a system flow model that balances housing stocks and people flowing through the system (recursively) to help maximize the system resources.
For the audience: I hope you heard a few examples of how we at King County are using data and some bright spots that inspire you to believe we can to make homelessness RARE, BRIEF and ONE-TIME