With Census, Accuracy Counts

The U.S. Census Bureau faces two major obstacles when it comes to counting the homeless: where to look and how to win the trust of a population that doesn’t always want to be found.

Ten years ago, it didn’t do a very good job at either. The 1990 census counted 11,790 homeless people in Los Angeles County. A report the following year by a homeless support organization put the number at 36,800 to 59,100.

How to explain the discrepancy? Census workers failed to look much beyond homeless shelters and skid rows. And they conducted the homeless count between 2 and 4 a.m., not exactly a time people are willing to answer a survey about income, race and education.

In a practice census undertaken by Los Angeles officials and social service agencies in March, teams of outreach workers fanned out between 5 p.m. and midnight to parks, alleys and freeway embankments in six census tracts. In five of the six tracts, the teams counted more homeless people than were counted in 1990.

In a report issued last week, the demonstration project listed recommendations the Census Bureau would do well to heed: Conduct a count during more reasonable hours and in the winter when city shelters are still open. Hire homeless outreach workers and homeless people themselves to help find hidden camps. Use incentives such as sack lunches to get street people to talk with census counters.

The costs of a miscount are high. The census count dictates funding for social service agencies and shelters. Ten years ago these agencies had to make do with less while continuing to serve a population that, counted or not, still needed help.