On January 30th, I woke up before dawn to drive around East Oakland and count all the homeless people I could find.
I was one of the 600 volunteers who participated in the 2019 Point-in-Time (PIT) homeless count for Alameda County, California. The PIT count is a nationwide effort to tally the number of unsheltered Americans living on the street on a single night in January. It began in 2003 and happens every two years. The Department of Housing and Urban Development (HUD) requires communities to complete the count in order to receive federal funding for homeless programs.
“This data will help us understand who’s out there and what they’re going through, and design better services for them,” says Elaine de Coligny, executive director of EveryOneHome, the non-profit organization that is responsible for facilitating Alameda County’s PIT count.
Individual people are counted one by one, but EveryOneHome uses a multiplier to determine how many people might be living inside tents and vehicles. This multiplier comes from field research conducted by outreach workers. A separate count is held to tally Alameda County’s homeless youth, and an additional survey is also used to gather more demographic information for the county’s homeless population, such as health conditions, veteran status, and housing history. This data, along with the number of people counted on the street by volunteers, comprises the totals reported in the PIT count.
By this method, 2,761 homeless people were tallied on the streets of Oakland in 2017—over 25 percent more than were counted in 2015. And in all of Alameda County, 5,629 homeless people were counted in 2017—up nearly 40 percent from the previous count.
But PIT counts are widely understood to undercount the number of people experiencing homelessness by a significant margin—some experts say by half or more. And after spending a few hours scouring the pre-dawn streets of Oakland, I have a better understanding why critics of the count say its numbers are both far too low and too unreliable to be used as a sole basis for understanding a region’s homelessness service needs.
I was assigned to a group with one other volunteer and a guide—a man who is currently homeless, who was tasked with pointing out the kinds of places where people might be sleeping. To boost the accuracy of the count, this year EveryOneHome hired 150 such guides, who were paid $15 per hour in cash for their efforts.
We set out at around 5:30 a.m. with a plan to cover an entire census tract—a swath of mostly residential neighborhoods in East Oakland curled up behind Lake Merritt. We were given a map that highlighted the tract and a tally sheet that prompted us to provide certain demographic information about each person we might see—gender, age, and dwelling type. The methods used in PIT counts vary somewhat from city to city—in some places, volunteers also interview the people they find. In Alameda County, however, we were instructed not to speak to anyone we encountered.
It was still dark outside and we drove slowly, illuminating the spaces in between buildings with our headlights. As we wound through the streets, my group and I made sure to canvas every corner. We got out of the car and traversed a small park by foot, checking behind the swing set and under wooden benches. We poked around some foliage above a freeway overpass. We spotted early morning commuters waiting for the bus and pajama-clad dog walkers. At one point, an older woman carrying a large garbage bag full of cans caught our attention. “Not homeless,” our guide told us, with certainty. We did not count her.
After three hours of driving around and searching, it was time to head back to Oakland City Hall to turn in our tally sheets. They were blank. “If you want, I can jump in that dumpster and start digging around so you can count at least one houseless person,” my guide joked.
How is this possible—especially in a city whose high-visibility homelessness problem was recently designated a humanitarian crisis by the United Nations? The fact that we could’t count a single homeless person in our patch of Oakland is, in part, a reflection of the stark spatial disparities in this city. We were in a relatively affluent residential area—we counted more than one BMW parked on the hilly streets. But we also probably missed many people who were hidden from view—they were in alcoves or cars, or in the homes and apartments of friends and relatives, sleeping on couches and floors.
The PIT count is low by design, to a degree. HUD requires that the count be held during the last 10 days of January so that it can account for people who cycle in and out of homelessness, and may be unable to pay for temporary shelter at the end of the month. HUD also says that holding the count “on one of the coldest nights of the year can be very effective in raising public awareness of the challenges faced by homeless people without shelter.” That makes it easier to recruit volunteers to conduct the counts.
But advocates and service providers argue that scheduling the event in the winter creates an undercount. “The count is during the winter early in the morning, when it’s harder to actually find folks because they’re seeking some sort of refuge. They want to stay out of sight in general for their own safety,” says Kelley Cutler, the human rights organizer at the Coalition on Homelessness.
One figure critics often use to highlight the undercount is the discrepancy between PIT count figures and the number of homeless children in public schools reported by the school districts. In 2015, the PIT count reported that the total number of Americans experiencing homelessness was 564,708. According to a report by the National Center for Educational Statistics, there were 1.3 million homeless children attending public schools across the country that same year.
Part of the reason for this discrepancy is that the PIT count and the Department of Education use different definitions of homelessness: In addition to the children who are included in the PIT because they are literally homeless, the Department of Education definition includes “children and youths who are sharing the housing of other persons due to loss of housing, economic hardship, or a similar reason; are living in motels, hotels, trailer parks, or camping grounds due to the lack of alternative adequate accommodations.”
“We know there’s an epidemic, right? You would have to be blind to not understand the nature of the epidemic,” says Margaretta Lin, executive director of the Dellums Institute for Social Justice. “But HUD defines homelessness as people who are literally homeless. People who are in a motel for that night or couch surfing for that night, under the HUD definitions, they are not considered homeless.”
Yet the budget for homelessness in any given community is heavily determined by PIT count data collected on this one night. This is especially true of some state-level programs. Take California’s Homeless Emergency Aid Program, which provides one-time funding for homeless services in individual counties. HEAP grants are calculated using PIT count data, and there is a direct correlation between the number of people experiencing homelessness in a place and the percentage of the funding it can receive.
HUD requires that any community that wishes to receive federal funding for homelessness programs must conduct a PIT count every two years. And, while the sheer number of people experiencing homelessness in a place has bearing on this funding, this relationship is a little more complicated at the federal level. “They also award bonus money based on progress,” de Coligny says. That is, if communities are determined to have done a good job with the funding they have received in the past, there is the potential to win additional HUD dollars.
Measuring that local progress is done using the more qualitative data yielded by PIT counts. After conducting the count, organizations like EveryOneHome answer a number of questions. Has your shelter population gone up or down? Has your unhoused population grown or shrunk? Are people staying in shelters for longer or shorter periods than they were during the last count? Each question is weighted differently, and, after submitting all of the answers to HUD, each geographical area gets a ranking. Depending on that ranking, a geographical area becomes eligible for more or less funding.
In January, it was announced that Alameda County was awarded $33.5 million from HUD to fund homeless services, an amount designated by the data from the 2017 PIT count.
Apart from the way funding is allocated, critics say that undercounts are harmful because of the way they skew the narrative around homelessness. “We’re constantly having to deal with people saying, ‘This is the budget, and this is how many homeless people we have.’ But they’re using really bad math,” Cutler said. “It’s such a political issue that different places may want to use the count to show that they’re making progress, even if it’s not the case.”
As an example of this, Cutler points to Utah, which declared that it had eliminated chronic homelessness in 2015. At the time, the state said that it had achieved “a functional zero” when it came to individuals experiencing chronic homelessness—a figure based on PIT count data. However, in the years since, that success story has been amended. The state has struggled to provide shelter beds for all the people who sleep on the street every night, suggesting that the state still has a homeless crisis.
De Coligny agrees that it is impossible for PIT counts to catch everyone who is experiencing homelessness. “It only counts a point in time,” she said. “If you’re really going to understand what your system needs to do, you have to understand how many people it serves over a year. A PIT count can’t answer that for you.”
However, she still believes the count provides a vital metric, and is especially useful when combined with other types of data.
“We continue to view it as most likely an undercount. But what it does help us look at is how it’s changing over time. So if you’re using the same method, you can see trend lines,” she said. Furthermore, de Coligny says, the numbers can paint a more complete picture when combined with other types of data, such as gaps analyses—the reports conducted to identify the unmet needs in housing and services systems.
For example, EveryOneHome conducted its own gaps analysis for Alameda County in 2018, which identifies the specific services the county is lacking. According to its report, those gaps are in homelessness prevention, street outreach, subsidized permanent housing, and permanent supportive housing.
“PIT counts inform gaps analyses, because a gaps analysis has to do with both data about the [homeless] community … and the capacity you have,” de Coligny says. So while the PIT count would ideally inform a city of how many people they need to provide services for, a gaps analysis would show what kinds of services are needed, and where those services are lacking.
On the morning of the count, a friend of mine wound through the streets of Berkeley, just a few miles away from where I was. His team had a drastically different experience: In three hours, they tallied 123 homeless people. In fact, his area was so densely populated with people sleeping outside that they were only able to cover half of their designated census tract. They ran out of time and had to head back to the EveryOneHome headquarters without even counting the other half, leaving another, very different gap in the data.
“It’s weird that we didn’t see anyone,” my guide said once we had finished surveying our Oakland census tract. “I guess for every one person you count, you’re just as likely to miss another.”
The results from the 2019 count will be released this summer.
This story originally appeared on CityLab, an editorial partner site. Subscribe to CityLab’s newsletters and follow CityLab on Facebook and Twitter.
