PunditFact: Dropping questions on marriage from the census

Asked to discuss marriage on a swing through London in 2003, then-New York Sen. Hillary Clinton offered the following appraisal: "Every marriage is a mystery to me, even the one I'm in. So I'm no expert on it."

Soon, all of us might know even a little less.

The Census Bureau, the government’s principal pollster and demographer, is considering changes to one of its regular surveys that could impact what we know about the state of marriage in the United States. The bureau announced recently that it may drop a handful of questions from its annual American Community Survey, which polls over 3.5 million homes on everything from how many rooms you have in your house, to whether you have Internet, to how long it takes you to get to work.

Out of concern that the survey is too long, no longer would Washington probe the state of your marital bliss or personal loss. They wouldn’t ask if you got married, widowed or divorced in the past 12 months. The year you got married and the number of times you’ve been hitched would remain your own business.

Jim Treat, chief of the American Community Survey, denied his office has anything against marriage.

"No," Treat said. "Not at all."

The problem with the questions, Treat explained, is that when Census officials asked other government agencies what survey questions were most important, the five ones on marriage ranked low compared to the rest. (Also on the chopping block is a question about medical offices and another about your undergraduate major.)

Demographers, however, are leading the charge to save the marriage questions.

"The American Community Survey was a Godsend," said Steven Ruggles, president-elect of the Population Association of America, the country’s professional group for people who study population trends.

"Marriage patterns are changing more rapidly than in any time in our history," Ruggles said. "Without this data, we would have no idea that a third of the people who are 20 to 24 years old now will never get married. We wouldn’t know that divorce has surged among Baby Boomers."

Ruggles, who once channeled his inner Neil Sedaka to write the article Breaking Up is Hard to Count, has called on his colleagues to speak up and be heard. He recently sent an email blast marked "URGENT."

"If you believe as I do that this change would significantly harm the nation's statistical infrastructure, you should make your feelings known by email (to) the Department of Commerce," the federal department that oversees the Census Bureau, Ruggles wrote.

There’s more to the potential change than just the feelings of academia.

The Family Research Council, one of the largest advocacy groups opposed to gay marriage, also wants the questions to stay. Dr. Patrick Fagan, director of Family Research Council’s Marriage and Religion Research Institute, said getting rid of them would be like the Labor Department cutting its measures of productivity, income, and jobs.  

"It makes no sense to take away questions about the most foundational relationship out of which springs every community and society," Fagan said. " Weakening the American Community Survey is simply bad policy and takes major real estate away from marriage and family data."

Perhaps most critically, there is money at stake. Only four government agencies told the Census Bureau that they really need answers to all of the marriage questions, but one of them was that financial behemoth, the Social Security Administration. Depending on how things break, over the next 75 years, Social Security could be solvent or it might be headed for a shortfall of more than $5 trillion. Last year alone, it cut checks for over $800 billion.

Looking forward, the size of many of those benefit checks depends on whether people stay married, are widowed or remarry. There are rules for allocating the money if there’s any change in your marital status.

Samir Soneji at the Dartmouth Institute for Health Policy and Clinical Practice, along with Gary King, a professor of government at Harvard, has in the past pinpointed gaps in the models the agency uses to project future spending. Soneji worries about the agency’s assumptions on rates of divorce and such.

"Without the American Community Survey spousal questions, no one outside the Social Security Administration has any basis for evaluation," Soneji said. "Without those questions, even actuaries and economists inside the Social Security Administration can only rely on speculation."

It’s not as though there is no other source of data whatsoever. Justin Wolfers is a University of Michigan economist who focuses on the dynamics where love, family and money come together. Wolfers recently stirred the demographic pot with his analysis that showed divorce rates falling (there’s a running debate over this). Wolfers relied on information from something called the Survey of Income and Program Participation, which also comes from the Census Bureau.

"But it is run only every four or five years," Wolfers said. "The last one was in March 2009."

Plus, Ruggles noted that because the marriage questions come near the end of that survey, they often go unanswered. This produces a significant undercount, he said.

Even officials at the Census Bureau agree that nothing quite matches this data from the American Community Survey. On the other hand, not everyone who deals with family data relies on it. According to Nancy Bennett, a senior fellow at the American Academy of Actuaries, life insurance companies typically get the statistics they need from experience or studies specifically related to the lives that they cover.

Anyone who does care has about two months to give the bureau feedback on its proposed changes.

"I hope we get under water with the number of responses we get," Treat said. "If these questions truly are needed, we would recognize that. It’s hard for us to know who outside of government uses our data. This is our chance to find out."

The bureau may soon learn what every swain knows. Not every proposal is accepted.

Update (May 14, 2015) 

The Census Bureau has saved marriage, or at least the questions it asks about it. We received a celebratory email from the Minnesota Population Center, the group that had led the charge to retain the questions.

"We are pleased to inform you that the Census Bureau elected to retain the field of degree and marital history questions in the ACS," the email said. "Thank you for the overwhelming response to our request to write in support of keeping these questions."

But it doesn’t look like this is the end of the matter. Apparently the Census Bureau has suggested it might not ask the questions every year going forward. Them’s fighting words to the folks in Minnesota. They’re calling for their members to keep those emails flowing to the tally takers in Washington.

* * *

The five marriage questions the Census Bureau may eliminate

Person Question No. 21a - In the past 12 months did this person get married?

Person Question No. 21b - In the past 12 months did this person get widowed?

Person Question No. 21c - In the past 12 months did this person get divorced?

Person Question No. 22 - How many times has this person been married?

Person Question No. 23 - In what year did this person last get married?