WSJ: Alternative methods of counting for the Census Bureau
Here’s a great article from the Wall Street Journal…Be careful, otherwise you might end up like William Rickenbacker:
By CARL BIALIK
Even in a mandatory census, there are conscientious objectors.
Completing a census form is required by law, but census takers haven’t been able to get any information from more than 500,000 U.S. households this year. While census evaders theoretically can be fined up to $5,000, in practice they are rarely penalized—none were in 2010 or in 2000—for fear of creating a public backlash.
Eddie Rickenbacker Papers, Auburn University Library Special CollectionsWilliam Rickenbacker was fined $100 for refusing to respond to a 1960 census questionnaire, making him one of the few Americans to ever face a penalty for noncompliance.
Instead, the Census Bureau combines threats of penalties with painstaking follow-up over the phone and in person, including interviews with neighbors of nonresponding households. That approach, backed by rapidly rising spending for advertising and census workers, yields near-complete coverage of the U.S. population.
That balancing act is costly, but it yields better statistics than a voluntary census, statisticians say. Calling a survey mandatory boosts participation significantly, they argue, even when enforcement is limited. That is why some data experts in Canada are assailing the government’s new plan to make many census questions there optional.
But as the 2010 U.S. decennial census winds down, a number of critics say even more reliable demographic data could be obtained at far less cost. They point to a system in some European countries that links personal identification numbers to government records on births, deaths, housing and other characteristics. That allows for an annual census rather than waiting every 10 years, and eliminates much of the error experts say arises when people self-report their information.
“The statistical systems in the Netherlands and Germany are much better than our statistical system, because they have a registration system,” said Kenneth Prewitt, the director of the Census Bureau during the 2000 census and now a professor of public affairs at Columbia University.
The decennial census, mandated by the Constitution, is used to apportion congressional seats. It also determines a lot of federal spending.
The census also needs to come as close as possible to complete coverage of the country so that other surveys, such as election polls, don’t have to.
If, for instance, 12% of voting-age Americans are under age 25, but just 6% of respondents to a political poll are in that age group, then the pollster might count their responses more heavily to make up for the shortage of young respondents.
“It is the basic amino acid of every pollster’s existence,” Jay Leve, founder of the polling company SurveyUSA, said of the census.
If the census were voluntary, researchers and pollsters say that their work would be compromised.
An experiment conducted in 2002 and 2003 in the U.S. demonstrated the potential for a lower response rate from a voluntary survey. When the American Community Survey—a detailed annual poll that replaced the long form on the decennial census—was launched nearly a decade ago, the Census Bureau tested two messages. One said replying was “required by law.” The other told recipients, “you may decline to answer any or all questions.”
The difference was stark. Of those who received the mandatory wording by mail, 59.5% responded. Just 38.8% of those who were told they didn’t have to respond did.
Even with the harsher wording, just three in five households mailed back the survey. The decennial census similarly has seen a decline in mail response rates, from 78% in 1970 to 65% in 1990. That decline prompted the Census Bureau to start spending on advertising and other promotional activities—$278.7 million in 2000, in current dollars, and $370.6 million for this year’s census. That has helped keep mail-back rates steady in 2000 and 2010, but it hasn’t led to an appreciable increase.