This week, the U.S. Census Bureau changed its mind about how many same-sex couples were identified in the 2010 Census. Instead of the 900,000 couples reported in data released this summer, the Bureau revised the figure downward to just under 650,000 couples. They also estimated that there were about 130,000 same-sex couples who described themselves as spouses living in every state in the Union, even though they can legally marry in only six states and the District of Columbia.
How did a quarter million same-sex couples suddenly disappear? And why are there so many married same-sex couples living in states where they can't get married?
Regarding the first question, allow me to indulge my inner statistical geek. It turns out that those disappearing 250,000 same-sex couples were really different-sex couples. They miscoded the sex of one spouse or partner, so it just looked like they were same-sex couples. To be fair, those miscodes don't happen very often and could include smudges or even errant pen strokes. But they don't have to be common to wreck havoc on same-sex couple numbers.
Statisticians call this a "false positive" problem: a few errors among a very large group make them appear to be part of a much smaller group and "contaminate" that group with incorrectly identified responses. There are 63 million different-sex couples compared to just 650,000 same-sex couples in the U.S. The Census Bureau estimates that less than 5 in 1,000 different-sex couples make mistakes while coding their sex, but that amounts to about 250,000 miscoded couples. Check out this great video that they did that discusses the problem.
The answer to why so many married same-sex couples live in states that don't recognize their marriages has nothing to do with statistical anomalies. The culprit here is the complex legal climate that same-sex couples face with regard to relationship recognition. Legally recognized same-sex relationships in the U.S. come in many varieties -- marriages, civil unions, registered domestic partnerships -- and states have different rules about what statuses they recognize. The federal government doesn't recognize any of them.
Census forms provide only two options for couples to indicate that their relationship is more than that of roommates. They can say that they are spouses (actually "husband/wife") or "unmarried partners." If you were married in Iowa but live in Nebraska, a state that does not recognize your marriage, are you spouses or unmarried partners? What about if you're in a civil union in Illinois? You aren't technically married, but you file your taxes jointly and you need a divorce to break up. If it walks like a duck...
The Williams Institute conducted a survey in which we asked same-sex couples what terms they used to describe their relationship in the Census and what their actual legal relationship status was. Among couples who called themselves spouses, 70 percent were legally married, 15 percent were in civil unions or registered domestic partnerships, and 15 percent were not in a legal relationship. That last group said they considered "husband" or "wife" to be the most appropriate description of their relationship.
Complicating things further, among couples who called themselves unmarried partners, 4 percent were legally married. They thought this was more accurate because the federal government or their state didn't recognize their marriage. About 17 percent of unmarried partners were in a civil union or registered domestic partnership.
So most same-sex "married" couples counted in the Census are in a legal relationship, but not all are married. Conversely, some "unmarried" couples are, in fact, married, and many are in legally recognized relationships that come with the rights and responsibilities of marriage.
Though the problems in counting same-sex couples can by complicated, the solutions are fairly simple. Census Bureau survey forms need to be changed. They are not up to the task of collecting clear and accurate data on same-sex couples.
Canada and the U.K. both collect similar data and use questions that reduce inaccurate or unclear responses. The U.S. Census Bureau has begun testing a version of these questions, but they need to move more quickly. They've done a great job of explaining just how bad the problems are. Now they need to fix them.
While they're at it, they should take another page from their Canadian and British colleagues and collect data on all LGBT people, not just same-sex couples. Federal surveys should routinely collect information about sexual orientation and gender identity. As the primary statistical agency charged with collecting data documenting the rich diversity of the American population, the invisibility of LGBT Americans in Census Bureau data is not acceptable. The reality of American politics is that you don't count unless you are counted. The time has come for the Census Bureau to count everyone in the LGBT community.