Errors Bedevil Census Bureau in Drawing U.S. Population Portrait
Hundreds of thousands of U.S. couples, as well as Americans living beyond the age 100, simply disappeared last year. At least they did according to recent statistical errors by the Census Bureau.
The agency last week acknowledged it had over-counted same- sex households by almost 30 percent, and its initial prediction for the number of centenarians fell short last month by more than 50 percent. The bureau is also facing challenges from cities claiming under-counting in last year’s decennial tally, as well as criticism of its flagship annual survey.
The glitches were both an embarrassment for the bureau and a reminder of just how hard it is to get an army of temporary workers -- or even the agency’s professionals -- to capture an accurate portrait of a changing country of 308.7 million.
“If you don’t have the number right, you can’t make the right policy with the resources that you have,” said Phil Sparks, co-director of the Washington-based Census Project, a nonpartisan coalition of groups advocating for a fair and accurate census. “If it’s a big city, we could be talking about billions of federal dollars spread over 10 years.”
The agency on Sept. 27 said the number of same-sex households grew to 646,464 in 2010 from 358,390 in 2000. Officials said they had revised down the 2010 figure, citing flaws in the earlier count. The original count had found an increase to 901,997 from 594,391.
The questionnaire’s layout confused some survey participants into checking the wrong sex, said the officials at the agency, whose overall record is generally so good that challenges to previous census counts have yielded minuscule changes to the tally.
“It looks like the census form in 2010 could’ve had an effect on artificially inflating the number of same-sex spouses,” Martin O’Connell, chief of the fertility and family statistics branch of the bureau’s Social, Economic and Housing Statistics Division, said in a telephone interview.
The corrected figures won’t affect government funding because the federal Defense of Marriage Act already prohibits extending benefits such as medical care and housing allowances to same-sex couples. Still, the mistake provided ammunition to gay-marriage foes, who said the smaller same-sex population should make politicians less wary about opposing such unions.
Years to Fix
Census data is the most complete measure of same-sex couples. Researchers use it to extrapolate information such as the size of the gay community, said Gary Gates, a demographer at the UCLA School of Law’s Williams Institute, which studies sexual orientation and gender identity law and public policy.
“The need for accuracy is even more acute because of the limited data resources on the LGBT population,” Gates said.
Because of the complexity of changing the forms that were at the heart of the problem, it may be years before the Suitland, Maryland-based bureau is able to yield more accurate data, he said.
The Census Bureau has no plans to reclassify the opposite- sex couples who were initially categorized as same-sex, according to Michael Cook, a spokesman. That means about 255,000 people have disappeared from some demographic data.
The agency also overestimated the number of Americans thought to be living 100 years or longer. Six years ago the census predicted that the country would be home to 114,000 centenarians by 2010. Census officials in 2009 lowered their estimate to 64,024, though even that projection was high. The actual number was 53,364.
While this mistake also won’t have an impact on federal funding, other inaccuracies can. Alleged undercounts in the 2010 tally have led cities such as New York and Miami to challenge the results.
New York Mayor Michael Bloomberg claimed in March that the Census Bureau missed counting as many as 225,000 residents. A shift of only 107,060 people would have resulted in the loss of one congressional seat from New York, instead of two.
Census counts determine the distribution of federal aid, which can affect how a municipality provides services from job training to health care to education, said Mark Jaffe, president and chief executive officer of the Greater New York Chamber of Commerce.
“That matters to businesses because even though they’re not required to provide health care, it is important to have a healthy, well-trained workforce,” he said.
Bloomberg is founder and majority owner of Bloomberg News parent Bloomberg LP.
‘Far From Perfect’
The likely problem with the city counts were the temporary census workers tasked with tracking down residents who didn’t respond to the initial mail-in form, said Patricia Becker, head of the Southeast Michigan Census Council, a group that promotes access to and use of demographic statistics.
“It’s a lot easier to mark something vacant than it is to find someone who lives there,” Becker said. “Temporary enumerators are far from perfect.”
The census bureau’s Count Question Resolution appeals process, which began in June, may determine how much of $400 billion in federal funds cities and states get based on their population. There were 73 challenges from jurisdictions as of Sept. 30.
Following the 2000 census, challenges unearthed discrepancies in 1,183 of the 39,000 jurisdictions, or less than 3 percent of all U.S. jurisdictions, according to the census bureau. The corrections resulted in a net gain of just 2,697 residents. In the case of California, the challenge resulted in the addition of five residents.
‘Honest and Transparent’
Adjustments to the 1990 Census added about 8,400 people to the official tally. The final results measuring the accuracy of the 2010 Census won’t be available until next year.
“We at the census bureau, from our evaluations of past censuses, know that no census of the U.S. is perfect,” Robert Groves, director of the agency, wrote May 16 on the bureau’s web site. “Whatever happens during Count Question Resolution, we seek to be cooperative, honest, and transparent in what we do.”
Some jurisdictions cited wide variations between the 2009 population estimates in the American Community Survey and the 2010 results.
“The ACS has its problems because it doesn’t have enough of a sample size,” Becker said. “This is something that’s fixable.”
‘The Gold Standard’
For all its problems, the agency remains the premier population-data center in the world, said William Frey, a demographer at the Brookings Institution in Washington who has been studying the census for more than 30 years.
“The U.S. Census is considered to be the gold standard,” Frey said. “Scientifically they’ve done as well as they can. They have a very good research staff there.”
Still, the recent embarrassments didn’t end with bad counts. Last week, the bureau had to confess to a problem with the work habits of its employees.
“Recently, management and the union received concerns from staff who observed a few staff members who appeared to be napping in public areas, such as the lobby or library,” the agency said in a Sept. 28 statement. “That behavior is unacceptable.”