Has a data-driven approach to housing the homeless gone too far?

HOME BASE: A special census conducted in 2011 sent volunteers out to question the city's homeless about their medical conditions, substance abuse and how long they had been living on the streets.
Neal Santos

Over the past few years, Philadelphia has quietly transformed a system used to document the city’s homeless and mete out life-changing, but perpetually scarce, housing.

Since 2011, directors of major homeless-support agencies have worked together to improve a once scattered and largely ad hoc system for doling out subsidized housing. In its place is a unified list that uses gathered data to establish a “vulnerability index,” essentially a score that prioritizes housing for the “most” needy. 

At least, that’s how it’s supposed to work. But caseworkers and homeless advocates have taken issue with the city’s attempts at developing a more data-driven approach to serving the homeless community. Some say administrators have grown content with an improved but still-imperfect system that now places enormous weight on sometimes outdated and incomplete information. Others say the new list can fail to accurately represent vulnerability — and that failure can have potentially fatal consequences.

The Office of Supportive Housing (OSH), a municipal agency that collaborates with several nonprofits to implement citywide homeless-support efforts, says it is simply employing best practices used nationwide to address the difficult problem of long-term homelessness.

“We’re using the data we have about individuals to identify who is most vulnerable,” says Roberta Cancellier, deputy director at OSH. “Of course, data is never perfect. There’s no perfect way to figure out if Joe is more vulnerable than John.”

The need for any sort of prioritization emerged because of the city's shortage of housing units for the homeless. Advocates say there are roughly 500 individuals living on city streets at any given time, but wait times for homeless individuals seeking permanent housing, as opposed to shelters, can stretch from six months to two and half years. Although the city recently proposed a plan to create 1,500 units of affordable housing, most probably will not be specifically earmarked for homeless people.

Cancellier says the city developed its current housing-priority list, known informally by case-workers as “the list of lists,” by combining city data with a special homeless census conducted in 2011. That survey, conducted by the “100,000 Homes Campaign,” a national movement that seeks to end homelessness through affordable-housing development, doc-umented 528 homeless individuals in Philadelphia. Volunteers worked throughout the night, using a one-page quest-ionnaire to quiz the people they encountered on their medical conditions, substance-abuse issues and how long they had been living on the street. Those answers were used to assign each respondent a “vulnerability score,” on a scale from one to eight, with eight being the “most vulnerable.”

Out of the 528 surveyees, 268 homeless people scored high enough to meet a standard, developed by a Harvard professor, that placed them “at higher risk of death than members of the general population facing identical conditions.” Most in that group were given priority for placement in “permanent supportive housing units;” others were spun off into more specialized lists for programs that targeted specific groups, like veterans.

But some social workers, who all requested anonymity out of fear of professional repercussions, said that this method was inherently flawed because it relied on self-reporting by the homeless.

“You come to someone in the middle of the night and give them a survey, they’re not always going to tell you they’re a drug addict or an alcoholic because they figure that’s going to somehow get them into trouble,” said one caseworker at a major homeless-support agency. “A lot of those people may have gotten much lower scores because they didn’t report things like that.”

The same is true for medical conditions, as many chronically homeless may have undiagnosed or self-diagnosed ailments that could also artificially affect their score. 

Others noted the folly of relying on three-year-old data.

“This survey needs to go on every year. If you’re really trying to seek out the people who are struggling the most, a list of people who were struggling three years ago isn’t going to tell you everything,” said the same social worker.

OSH countered that its list does con-tinually incorporate new information — albeit from a different source.

“The most recent list uses city data based on what we know about people from street-outreach-worker contacts,” said Cancellier, referring to teams, managed by homeless non-profits, that regularly check in on men and women living on the street.

The city uses the frequency of outreach contacts as one measure of how long someone has been homeless — and, therefore, how at-risk they are. But this can result in the sometimes-accidental contact with street-outreach teams taking precedence over other measures of vulnerability. 

Intake forms from one homeless housing agency, obtained by City Paper, showed that a candidate who needed a permanent address in order to receive a life-saving organ transplant was passed over in favor of a healthier candidate who had simply had more contacts with outreach workers. The city later lost track of the terminally ill individual, whose ultimate fate remains unclear.

Other social workers interviewed for this story also expressed frustration that homeless individuals at many city shelters are unknowingly cut out of the list for housing. Because men and women in many city shelters are rarely resurveyed for vulnerability and do not have contact with street teams, they are sometimes excluded from the list.

Cancellier did not dispute that the system could be improved, but said the city only had so many resources it could devote to data collection, and that with limited housing options some individuals would always wind up being excluded.

“It’s difficult, because if you make a commitment to serve people on the list, it means you have to say no to other people,” she said. “That causes friction, and it’s hard if the person in front of you seems extremely vulnerable. Some of the tension around this is just what happens when you prioritize people. That’s hard for all of us in this line of work.”