Each January, thousands of volunteers fan out across Los Angeles to count homeless people living in riverbeds and canyons, on street corners and in hidden encampments.

The count does not purport to tap every person who experiences bouts of homelessness over a year, but rather produces a snapshot of the street and shelter population that can be used to shape policy and distribute funding. A count released in May found 58,000 homeless people throughout the Los Angeles region this year.

Now a new study says L.A.’s count underestimates the number of homeless people and reports year-to-year shifts in gender, ethnicity and age breakdowns that are not “plausible.”

Demographic surveys used to collect race and gender data are not random, and the final numbers are so different from school and welfare homeless data that they must be wrong, the Economic Roundtable, a Los Angeles research group, said in its report.

“The homeless count is valuable for providing a fresh picture of homelessness,” the Economic Roundtable said in the 47-page report. “But the count data is not reliable enough to be used for comparing the number or population composition of homeless residents from different counts.”

The Los Angeles Homeless Services Authority, the city-county agency that oversees the count, disputed the group’s findings.

“While well-intentioned, the report does not reflect sufficient familiarity with how the count is conducted,” Peter Lynn, the services authority’s executive director, said in a statement.

Los Angeles, like other jurisdictions, adheres to a strict federal definition of homelessness, and school and welfare office data do not meet the standards, the services authority said.

The accuracy of the count is important, because it provides the only numerical benchmark to gauge if homelessness is growing and to assess whether regional strategies work.

A report this year from the real estate research firm Zillow also found L.A. was undercounting its homeless numbers, which the company estimated hit 61,000 this year.

The Economic Roundtable did not give an alternate figure but said L.A.’s methodology missed people.

The Economic Roundtable was an unsuccessful bidder to help the services authority design the count methodology last year. The job went to a USC team.

The roundtable makes a number of recommendations for sharpening the data, including better volunteer training and more foot counts. (Many neighborhoods are canvassed by car.)

It also suggested using smartphone GPS technology and sending people posing as homeless, or “decoys,” into neighborhoods, as officials do in New York and Toronto, to ensure that everyone is counted.

In its response, the services authority said it was not feasible to walk the 4,000 square miles in the Los Angeles region or send decoys into the vast area. More than 41,000 homeless people live outdoors in L.A., as opposed to 3,900 in New York, whose population is largely in shelters, the agency said.

GPS reception would be spotty in parts of the county, and its readings could not be verified.

The Economic Roundtable also recommended consulting public assistance, healthcare provider and school data to improve the count’s accuracy.

The local homeless agency said the welfare office’s definition of homelessness is too broad to meet federal standards but added that it was working with the Los Angeles County Office of Education to improve the youth homeless count.

gholland@latimes.com

Twitter: @geholland