Los Angeles homeless count raises doubts about accuracy. Is it time for a new way?

Jose Luis Camargo washes out a skillet outside his makeshift tent in Los Angeles
Jose Luis Camargo washes a skillet outside his camp under an elevated ramp linking the 110 to the 105 Freeway in Los Angeles.
(Francine Orr / Los Angeles Times)

When the Los Angeles Homeless Services Authority this month published a spreadsheet breaking down its homeless count by every census tract in the county, those with knowledge of Venice were incredulous.

LAHSA said there were no unsheltered people — no tents, no inhabited cars or RVs and no people living outdoors — in the northwest quarter of Venice, which is notorious as ground zero for homelessness.

It wasn’t just the number — so wildly wrong it couldn’t be a statistical error — that shocked them.

How, they wondered, had LAHSA blown the most closely watched census tract in the county, one scoured regularly by residents conducting their own tallies and more recently by a team of professional surveyors working for the Rand Corp.?

LAHSA did not respond to questions about the Venice numbers but released a statement Friday evening defending the integrity of the count.

“During the Count, we received several reports of user and technological errors resulting from a lack of training and poor internet connectivity,” said Ahmad Chapman, LAHSA’s communications director, in the statement. “Despite these errors, we are confident in the accuracy of this year’s homeless Count because LAHSA and its partners took several steps to account for what was happening in the field.”

Whether the result of human error or a systemic flaw, the botched Venice count has heightened criticism of the annual point-in-time count as inaccurate.

“I just don’t think a point-in-time count makes sense of really helping people,” said Connie Brooks, a Venice resident who participates in the unofficial counts that consistently show more than 200 homeless people in the area.

In City Hall this week, Council President Nury Martinez introduced two motions asking for an audit of this year’s count, which was conducted over three days in February, and those of prior years and an evaluation of whether to hand future counts to a third party.

One of the motions cites wide disparities in the increase or decrease of homelessness in various parts of the city since the last count in 2020. Though the average across the city was a modest 1.7% increase, some council districts had large swings; the most extreme were an 80% increase in a west San Fernando Valley district and a 40% decrease in the Westside district that includes Venice.

Enforcement of Los Angeles’ revised anti-camping law rolls out in slow and uneven steps.

May 2, 2022


Those disparities raised concern that the count is failing to accurately represent the progress somecouncilmembers have made in providing shelter and housing and may reflect decreases in some districts resulting from shelters being opened in others. The distinctions matter because, under a settlement in a federal court case, the city is responsible for providing shelter or housing for 60% of the unsheltered homeless people found in each council district.

Councilman Kevin de León, who seconded the motion, questioned the new count’s finding that there were 231 fewer people in shelters in his Northeast district than in 2020, yet there were 1,818 more people on the street.

Between the two counts, De León opened hundreds of shelter beds, including a tiny home village in Highland Park, two Project Homekey motels in El Sereno and a 483-room shelter at the L.A. Grand Hotel. (A second tiny home village in Eagle Rock opened after the 2022 count.)

De León’s chief of staff, Jennifer Barraza, acknowledged that the “decompression” of shelters — the reduction of beds to allow for distancing during the pandemic — could explain some of the discrepancy, but she found other inconsistencies. For example, she said, she could not understand the 6% decrease in homeless people in skid row, where there are hundreds of shelter beds.

“Those beds were also decompressed,” Barraza said. The people in them “either went somewhere else or they weren’t counted. It does feel like there’s a lot of stuff that isn’t told in the data we got.”

Similarly, the count found an increase of 225 RVs in Boyle Heights, Barraza said.

“I’m not sure if the jump in RVs is accurate,” she said.

Councilman Bob Blumenfield said he was mystified by the 60% increase in unsheltered homeless people reported in his west San Fernando Valley district, even though more than 200 people got off the street into shelter.

“I go out in our community for homeless outreach at least once a month, and my staff is on the streets everyday,” Blumenfield said in a statement. “The increase of people who are unsheltered per the LAHSA Count does not reflect the reality that we see. More transparency over this process would be incredibly welcomed because we are simply not getting answers that add up.”


Blumenfield said his staff who volunteered for the count kept paper records because they found the phone app to be glitchy. They also found census tracts outside his district that were included in his count.

“Math is math and these discrepancies, paired with app glitches as well as map problems, raise some serious questions,” he said.

Jason Ward, associate director of the Rand Center for Housing and Homelessness in Los Angeles, said he launched his survey in three high-profile neighborhoods — Venice, Hollywood and skid row — to fill in the gap from when the 2021 count was canceled and to gather data that the once-a-year count miss.

“How does it change over time?” he wanted to know. “Does it change if you do it day or night?”

Rand’s team of experienced street workers had an advantage over the massive force of lightly trained volunteers who scour thousands of tracts for the annual point-in-time count.

“I feel like we’re basically doing the same thing as the [point in time] count but doing it in a systematic way, with well-trained teams that are doing it consistently over time.”

Ward said he was surprised that LAHSA had not flagged the Venice count as a problem when it showed that the number of unsheltered people had dropped from 509 to 0.

LAHSA had ample warning that something was amiss. Shortly after the street count in February, a blogger named Christopher LeGras published an account of software glitches he experienced as a volunteer counting the Venice tract. He said a new phone application allowing counters to tabulate their findings while in the field crashed. He instead texted his numbers to a LAHSA worker but was not certain they would be included.

LeGras posted a screen shot from his phone showing that he recorded 85 homeless people, 43 apparently inhabited cars, 29 vans, 27 campers/RVs and 67 tents or makeshift shelters, together representing close to 300 unsheltered people.

Those numbers closely matched the records of the Rand researchers who inspected that census tract over three nights in January. On average, they found 86 people, 38 cars/vans, 18 RVs and 55 tents or makeshift shelters.

In a post after the LAHSA results were published, LeGras unleashed a rebuke.

“The lived experiences of both housed and unhoused people in Venice leaves zero doubt that the crisis has continued at an extreme level of intensity despite high profile — not to mention highly expensive — efforts by the City to address it,” he wrote.


In its statement, LAHSA’s Chapman said in areas where the app or volunteers did not record the data, staff conducted their own count.

But that doesn’t explain how zero homeless people were counted in that critical census tract.

The botched Venice numbers may take some shine off the 38.5% decrease that Councilman Mike Bonin, in a celebratory tweet, credited to “the largest and most successful place-based homelessness intervention programs in Los Angeles, moving nearly 300 people indoors from tents on Venice Beach and in Westchester Park.”

Adding the 250 to 300 people missed in the Venice count would leave Bonin’s district with a still-impressive 29% improvement.

But the resident activists whose distrust of the point-in-time count led them to conduct their own see a broadly distorted picture. Though a ribbon of tents and makeshift shelters was largely removed from the Venice boardwalk in 2021, dazed people still wander the nearby streets, one of which is lined with RVs and another with tents.

To restore confidence in the count, they say, a different model is needed.

“We all feel that the way to help people is to get accurate numbers in real time,” said Brooks. “How does a onetime point-in-time count help get people into housing? I understand that’s how funds get allocated. But how does it help individuals on the street?”

They point to a model being developed by a national initiative called Built for Zero. The group is working with about 105 communities, including Jacksonville, Fla., and Denver, to help them use what they call “by name” data — information, gathered by shelter and outreach workers, about individual people — to maintain a continuous picture of homelessness.


“That’s the end state we want to get to — that communities have at their fingertips at any point in time how many people are experiencing homelessness,” Built for Zero Director Beth Sandor said. “It’s not a once-in-a-year phenomenon. It’s never going to be completely perfect. It’s a dynamic problem. People are always moving in and out of homelessness. But that there is some high degree of confidence that we are covering the geography and the majority of the population.”

For Los Angeles, that could be a big lift, said Ward, the Rand researcher.

“It seems really great when your population is manageable and small enough,” he said. “I don’t know whether that’s feasible in L.A. That’s a lot of people to know by name.”

Times staff writer Dave Zahniser contributed to this report