February 14, 2011 | Houston Chronicle | Original Article

Did national head count overlook too many people?

When the U.S. Census Bureau releases its first detailed report on Texas later this week, the numbers will describe the state's surge of young Latinos, its aging Anglo population and the continued exodus of people from West Texas.

But a series of reports by the Office of Inspector General raises questions about the accuracy of the data and even suggests that one of the major bragging points for Census Director Robert Groves is less clear-cut than it seems.

Groves said last summer that the 2010 census would return $1.6 billion to the U.S. Treasury — that later grew to $1.9 billion, savings attributed to more people than expected returning their forms by mail, requiring fewer face-to-face interviews, and the fact that $800 million set aside for a technological or weather-related disaster wasn't needed.

But the latest report by the Inspector General, sent to Congress last month, contends that wasn't necessarily good news, but rather the result of an "inability to adequately estimate costs."

It also found that technical problems increased costs and delayed the submission of data, which in turn slowed efforts to ensure accuracy. That, along with shortcuts taken by some census workers, increased the risk of errors, according to the Inspector General.

The associate director for the decennial census disputed the report.

"That is the OIG's opinion," Arnold Jackson said. "We don't agree. We were on time, under budget and, so far, have had very, very high-quality metrics."

Checking its numbers

The bureau conducts its own fact-check, known as the Census Coverage Measurement. About 170,000 housing units were re-surveyed in August and September; the results will be released in mid-2012.

Some errors are unavoidable.

"It's impossible to have a perfect census," said Arturo Vargas, executive director of the National Association of Latino Elected Officials Educational Fund. "It doesn't exist."

The bigger concern is whether members of certain groups — such as African-Americans or Latinos - were missed, he said.

That's important for redistricting, which relies upon census data to determine where to draw political boundaries. The Census Coverage Measurement will determine whether there was an undercount, among other things.

Steve Murdock, a sociologist at Rice University and former director of the Census Bureau, said any errors are likely to be small, with an undercount of perhaps 1 percent or 2 percent.

The effort was complicated by anti-immigration rhetoric, which may have made some people reluctant to participate, he said. (The census does not ask immigration status.)

The housing crisis, with historic numbers of homeowners in foreclosure, also complicated things.

Jackson said census workers relied upon community and religious organizations to ease fears in immigrant communities but conceded that foreclosures made it more difficult to determine which homes were vacant and where two or more families were living in one house or apartment.

Texas was the big winner when the first results of the 2010 census were released in December: Its population grew more than twice as fast as that of the nation as a whole, to 25.1 million.

Texas was the big winner when the first results of the 2010 census were released in December: Its population grew more than twice as fast as that of the nation as a whole, to 25.1 million.

As a result, the state will gain four Congressional seats.

The national population grew 9.7 percent, to 308.7 million.

A Latino boom

Estimates indicate 61 percent of population growth in Texas has been among Latinos, Murdock said. All minority groups together accounted for 85 percent of population growth in Texas between 2000 and 2009.

While most demographers expect Texas' African-American population to remain stable - it made up 11.5 percent of the population in 2000 - they predict an increase in residents of Asian ancestry, a category that includes people of Chinese, Vietnamese and Korean descent, as well as those from South Asian countries such as India and Pakistan.

Most Asian-Americans are concentrated in a handful of Texas counties, including Fort Bend, Ellis and Williamson, Murdock said.

The increase could affect redistricting, with some groups calling for an Asian-centric district in Fort Bend County, said Lloyd Potter, the state demographer and director of the Texas State Data Center.

Population loss in West Texas will be a factor in redistricting, too, Potter said, as political districts there will have to be geographically larger to encompass the same number of people.

Between 1990 and 2000, 68 of Texas' 254 counties lost population. Murdock said recent estimates suggest that increased to 119 counties between 2000 and 2010, despite the overall population growth.

Questions on accuracy

Efforts to check the count's accuracy are under way.

David Whitford, chief of the bureau's Decennial Statistical Studies Division, said about 170,000 housing units were resurveyed in August and September. That's 0.13 percent of the nation's households.

The results will be compared with those gathered last spring to determine how close the 2010 census came to reaching everyone in the United States.

But the official 2010 numbers will remain the basis for apportioning congressional representation, redrawing political boundaries and distributing billions of dollars in federal funding.

Jackson conceded there were problems with the paper-based data collection system, which the Inspector General said adversely affected efforts to follow up with the 50 million households that did not return forms by mail and hampered quality control.

Jackson said the problems were resolved within three or four weeks, allowing the follow-up to be completed on time.

The bureau turned to the paper-based system after dropping plans to use hand-held computers because of problems with the software.

There were glitches, but Jackson said the report "overstated" the problem.

The report also found that some census workers ignored the bureau-approved script when interviewing residents. In some cases, they assumed the race or ethnicity of the resident, rather than asking, or relied upon online directories and other Internet sources to complete the forms.

Jackson acknowledged that, with 500,000 temporary workers, some won't follow procedure.

"We think the number of people found doing that was not high," he said. "The IG may think it's a lot."

Potter said the initial population figures released late last year eased some of his concerns about accuracy.

"For me, that was mildly reassuring that our undercount problem wasn't as bad as I and perhaps others had concerns it might be," he said.

"That's not to say we don't have an undercount, but certainly it wasn't so significant that ... we only got three (additional congressional) representatives, as opposed to four."