What race and age would you guess this person to be?
Photo: JOHN ALLE SANTA MONICA COALITION
When the Los Angeles Homeless Services Authority (LASHA) released the most recent point in count for the Homeless, the results also stated how many Hispanic, Black, Asian and White people were homeless. The specific ages of the homeless were also included.
Having done the count for almost 10 years in Pacific Palisades, the rules are clear. One does not “knock” on a tent, nor an RV, to see how many people might be living inside. Given that you’re not even supposed to engage the homeless, how did the agency come up with that information?
LAHSA’s Deputy Chief External Relations Officer Paul Rubinstein sent this editor to USC’s Dr. Ben Henwood for the explanation. USC was paid $1.6 million to extrapolate data for LAHSA. On the Westside it was determined there were 5,235 unsheltered homeless.
The races of the homeless was specific and included: Hispanic/Latino 1,153, Not Hispanic/Latino 4,082, American Indian/Alaska Native 67, Asian 163, Black/African American 1,414, Hawaiian 45, White 2,186 and multiple races 207.
The ages of the Westside Homeless were given as: Under 18 (138), 18-24 (108), 25-34 (1,220), 35-44 (1,445), 45-54 (1,030), 55-59 (517), 60-61 (151), 62-64 (289), 65-69 (160), 70-79 (146) and 80 and over (31).
The genders of homeless for the Westside were male 3,776, female 1,353, gender non-binary 85, questioning 21 and transgender 78.
Did USC teams actually speak to the 5,235 people to reach that conclusion? No.
More broadly, did the USC survey teams count each of the 50,000 unsheltered adults (+ or -1,500 people) in L.A. County? No. But over three months the USC teams randomly went to 584 of the census tracts in L.A. County.
Henwood said his teams canvased by foot and/or vehicle and invited anyone they encountered to participate in the survey. The team knocked on doors of vehicles, tents and make-shift shelters to ask people to participate. Of the 50,000 homeless, about 3,473 participated in a 59-question survey.
With only about seven percent surveyed how was the USC study able to say with certainty that 360 of the people in SPA 5 (seven percent) were in the 65-69 age range or that 207 of the people were multiple races?
Henwood said, “Our surveyors are also asked to record perceived characteristics of things like age, race/ethnicity, gender, etc., before they approach a person, so we have some info on people who refuse or are unable to engage with us for various reasons.
As Tim Campbell, who managed a municipal performance audit program pointed out, “You can’t get more subjective than guessing someone’s ethnicity based on observation.”
As far as age, people who live on the street generally look far older because of the lack of nutrition, alcohol/drug use, and the weather.
Henwood said, “We use this information [guessing] to attempt to account for or adjust for any ‘non-response.’ Non-response individuals are those who were not approached, refused or declined to give consent, or did not provide age and housing status information to determine eligibility. This is a statistical approach that would require more technical expertise than I can provide in this email, but hopefully it gives you some sense of how we approach the estimates of race/ethnicity that you asked about for the unsheltered population.”
Having taken statistics in college, this editor asked Henwood for the methodology for reaching the final totals that determined that there were 78 transgender individuals or 4,082 Not Hispanic/Latinos on the Westside.
He wrote, “For each point of non-response, we distribute the weight of NR individuals to responding ones using a nearest neighbor matching approach.”
Okay, if they used my nearest neighbors to determine what race and age I am, based on neighbors in Pacific Palisades, it might be a safe bet that it is mostly white and middle aged. (Surprisingly, I’m in a relatively heterogenous neighborhood that includes Indian, Persian, Asian and Black).
But to bet on the race and age of people living in a homeless encampment say in Venice, based on their homeless neighbor, might be more difficult.
Henwood continued to explain the scientific method USC used. “We distribute the weight for each NR individual to those with similar predicted values. We then post-stratify the weights of the eligible respondents to the number of households observed in the PIT count. More information regarding our statistical approach can be found in our methodology report here: click here..”
It seems that the $1.6 million (taxpayer dollars) that LAHSA paid to USC to classify the homeless, produced a fine-looking 25-page guestimate document with 11 appendices.
What’s your guess for the age and race of this homeless person?
Photo: JOHN ALLE SANTA MONICA COALITION
The survey by USC is insulting our intelligence! And the fact that they got paid $1.6 million of our hard earned tax money. Thank you for writing about this wasteful Statistical inaccuracy. It sounds like racial and age profiling at the highest level!! We could better serve a lot of homeless people with $1.6 million, because It does not matter what race they are or how old they are… Just help them!! Actually, anyone under 18 needs immediate and priority attention!! Any homeless under 18 break all our hearts!
Very cogent critique. Once again we are asked to rely on “science “ that is dubious.
Thanks,
Sue-Your report on the USC homeless count and analysis is spot on. What kind of feedback are you getting?
And let’s not forget, the teams only talked to people who wanted or could talk to them. How many people do we encounter who are clearly delusional and unable to give someone a coherent response. RAND says at least 50% of the unhoused have mental illness or substance abuse problems. How many of them were or could be interviewed? USC’s methods are inherently inaccurate.