Exclusive: There were significant problems with the 2022 Los Angeles Homeless Count

Experts and count volunteers express skepticism about the results; The math doesn’t add up; Millions of dollars in homeless funding hang in the balance

[Publisher’s note: the all aspect report has been on hiatus as we focus our efforts on the upcoming November elections. We are making an exception for this important story. As always, the facts are rock solid and the opinions and experiences are solely our own.]

Late Thursday afternoon, the Los Angeles Homeless Services Authority (“LAHSA”) released the results of the 2022 the annual point in time (“PIT”) homeless count. If you followed the ensuing social media and email chatter you would have discovered that few Angelenos have much faith in the results. Concerns about the count’s accuracy are long-standing and well-documented. The difference this year was the introduction of several novel variables (more accurately, they are what statisticians call confounders, factors that taint data) into the already confusing concoction. The COVID pandemic, technological issues, and an unusual cold snap all contributed to a particularly confusing count.

Officials at the Los Angeles Homeless Services Authority (LAHSA), the City-County joint powers authority responsible for administering homeless services and funding, as well as the count itself, should have seen it coming. They had two years to prepare. The pandemic took a significant toll on people’s public spiritedness, not to mention their willingness to give up an evening of their time. Yet not only was the process unchanged from previous years — except for an unfortunate attempt at technology by LAHSA, more on that in a moment — it was if anything less well-organized. After two years in suspension due to COVID, at the last minute in January LAHSA delayed the count an extra month over concerns related to the omicron outbreak. Between the shifting schedule and the fact that the count took place during one of the coldest snaps in recent Southern California history the volunteer turnout was positively desultory.

My friends and I volunteered for the count in Venice. When we arrived at the LAHSA staging area we were the only volunteers there, and we noticed the sign-in sheets were almost completely blank. We were the only signatures on our page of volunteers, a situation that remained unchanged when we returned two and a half hours later with our results. This suggested that many parts of our section of Venice did not have any volunteer counters at all.

Federal law requires LAHSA to report PIT count results within three months. Yet LAHSA repeatedly delayed releasing this year’s results, offering vague explanations that like the count itself raised more questions than answers. The count is conducted by two to three thousand volunteers (though again, the turnout this year was likely considerably lower), who fan throughout the City in teams of two or three counting individuals, vehicles, and makeshift shelters. The results are tallied by census tract, then what are called Service Areas, and finally City Council Districts and the City itself. LAHSA and outside experts take the raw numbers as reported, and extrapolate an overall estimate of the total number of homeless. Worth noting is that LAHSA and the outside experts often tweak the methodology from year to year based on experience and volunteer feedback. While well-intentioned the changes also make meaningful year to year comparisons difficult if not impossible.

Why did an agency with an $808 million budget and nearly 500 employees, working with a federal agency with a budget of $68.7 billion and some 9,700 employees, need seven months to do arithmetic?

A broken fix for what wasn’t broken in the first place

One clue might be the new technology LAHSA introduced this year in the form of a smartphone app called “Akido.” In the past, volunteers used clipboards and pencils and paper to record the count. Teams are assigned by census tract, which depending on population density and other factors can be as small as a few blocks or as large as a square mile. Smaller and safer tracts are assigned teams of two, a counter and a recorder who count on foot. Larger and/or less safe tracts are assigned teams of three in cars, adding a driver. Some tracts use a combination of the two approaches with three-person teams.

My team was assigned to one such tract: Two of us walked Speedway on the Venice Boardwalk between Navy Street/Santa Monica and Venice Boulevard, then we got in the car with our third friend driving and covered the same north-south section on the streets east to Pacific Avenue.

Ironically, in the past the actual physical count itself was one of the few aspects of the process that was relatively straightforward: Drive or walk every block of your assigned tract, observe homeless people with your own eyes, and mark chits for the results. Federal standards require volunteers to count according to categories: Individual adults, individual youths (defined as ages 18-24), as well as people in cars, vans, campers/RVs, makeshift shelters, and tents. For safety reasons volunteers cannot approach vehicles or tents/shelters, and the regulations assume sole occupants, which is often the exception to the norm and further depresses accurate counting.

In contrast to straightforward pencil and paper the app required a significantly higher degree of focus to ensure correct counting. The text is small, the boxes counters tap with their fingers are small and often poorly responsive. These limitations were distractions as my friend and I walked part our assigned census tract along the Boardwalk. I found that I often had to tap the box marked “+” several times before it recognized the input. More frustratingly, sometimes I would tap the box several times and there would be a delayed response, requiring me to take another step by deleting the extra entries via the equally fussy “-“ box. The process was frustrating and inefficient while walking, trying to use it while driving down the street at 20 or 30 miles an hour proved nearly impossible. I previously wrote about the frustrations in detail.

The fact that the app solved a non-existent problem with an inferior product was not the end of the problems. When we returned to the LAHSA staging area for our part of Venice, Aikido crashed altogether. Each time I attempted to submit our data the app simply froze, until it locked up my iPhone completely. Fortunately, I was able to take screenshots of our count and text them to the personal cell phone of one of the LASHA workers.

For seven months I’ve wondered whether our count ever made it into the system. Yesterday’s results prove they did not. The screenshots below are from the Aikido app on my phone at the end of the count. We observed 85 individuals (61 adults and 24 youths age 18-24) as well as 30 cars, 29 vans, 16 campers/RVs, 54 tents, and 67 makeshift shelters. LAHSA counts each car, van, tent, and makeshift shelter as one person, while campers and RVs are counted as two. Thus, our total count for our census tract was 297 people. Even assuming the (highly unlikely) average of one individual in every camper and RV we still counted 281 unsheltered homeless people living in a small section of Venice on the coldest night in decades. Images from the app at the end of our count are below.

When LAHSA released the final tally yesterday, it showed our tract as having 77 homeless people, all of them counted as sheltered. For the statistically inclined that’s a delta of 380% from what my team and I observed. But wait, as the old commercials used to say, there’s more: According to the official LAHSA numbers there were zero homeless people living on the street the night we performed the count. All 77 people are listed as “sheltered.”

According to the fine print on the lower left of the LAHSA web page above there’s a disclaimer saying that individuals receiving motel vouchers are not included in the count. So who, exactly, are these 77 people? Where are they sheltered, and who counted them? That last question is key: There is only one team per census tract, so LAHSA did not receive conflicting data from different teams. By all appearances they simply made up a number. It is possibly the number of people who were staying in the infamous, troubled Rose Avenue Bridge Home Shelter.

Significant implications for policy and funding next year

These discrepancies aren’t merely the results of conflicting tallies. The different numbers reflect different realities. The lived experiences of both housed and unhoused people in Venice leave 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. As between the two numbers it’s a safe bet you’ll be hard-pressed to find many folks in Venice who believe the official number over ours.

Meanwhile, in the bizarro parallel reality of Los Angeles politics, Councilman Mike Bonin issued a self-congratulatory press release about how his policies – which amount to $1 million studio apartments that take years and decades to build, for people who often cannot so much as dress themselves anymore – have resulted in the biggest drop in homeless in the entire City. According to the count CD11 experienced a 29% drop in homelessness between 2020 and 2022. Venice saw a full 50% decrease. These numbers do not just defy people’s everyday experience. They also defy statistical probability.

Dan Flaming, President of the Economic Roundtable, points out that, “The tract your team counted had 509 homeless people in 2020, compared to zero this year.” He was the lead author on a November 2017 study called “Who Counts: Assessing the accuracy of L.A.’s homeless count” that identified methodological weaknesses in the PIT process. In July 2018 his team also partnered with a big data company to crunch that year’s count, concluding that the actual number of homeless people in L.A. was two to three times larger than what the count revealed. He and his team are delving into this year’s count. He already has suggested that an independent audit may be necessary to see just what went wrong, and how significant the problems are.

In addition to the basic indignity of ignoring hundreds of homeless people in our tract and likely others, LAHSA’s number and Mr. Bonin’s claims to success have major policy implications. Funding for homeless services, shelter, and programs by definition depends on the PIT count results. The numbers LAHSA announced on Thursday will be used to develop next year’s budget, including the dollars allocated to Venice Beach for services, shelter, and caare.

Suffice it to say we are not going to help 297 people living on the streets with a budget for 77 people who are sheltered. Yet 77 is the official number, for now.

Counting fractions of human beings

The problems with the Venice Beach count may only be the tip of the iceberg. In Playa del Rey, Census Tract 278102a includes West Jefferson Blvd. through Ballona Wetlands. Since at least Spring 2019 there has been a major encampment of RVs/campers along the creek. At this point there are at least 100. Yet according to the 2022 count there were only 14.9 unsheltered homeless people on that stretch. In 2020 the count only reported 20.5 homeless people in that tract (the fractional counts are the result of multipliers assigned to different unsheltered living situations to produce the final total estimates).

As noted these numbers will largely determine L.A.’s 2023-34 homeless budget. California’s Homeless Emergency Aid Program (HEAP) provides annual funding for homeless services in all 58 counties. HEAP grants rely on PIT count data. Worse, this divergence of budget and reality will be felt as the last COVID protections end, and at a time when inflation is still taking money out of working people’s pockets. If anything Los Angeles, like cities nationwide, needs to prepare for another wave of displacement, potentially a significant one. Instead, the City is acting as though the situation is well in hand. It is a form of madness.

For now, the question is whether the public, the media, policymakers, and candidates will demand those answers. So far the only candidate to have publicly raised the issue is CD11 City Council candidate Traci Park, who on Friday tweeted about other discrepancies in the count, specifically the RV encampment at Ballona Creek (full disclosure: I provide policy advice for the Park campaign). As a close ally and chosen successor of Mike Bonin, it will be interesting to see what position the other candidate, Erin Darling, will take on the issue.

Regardless of the CD11 race, hopefully others will join Ms. Park’s leadership and force LAHSA to explain the myriad discrepancies in the 2022 Homeless Count. The public health of our City, not to mention countless individual lives, depend on it.

Calls and emails to LASHA were not returned. This story will be updated as new information becomes available.