The United States Census Bureau has determined that its original count of the number of same-sex married households was nearly three times too high.
The Census Bureau originally showed 349,377 same-sex married couple households and 552,620 same-sex unmarried-partner households in the United States in 2010. The actual counts are now estimated to be 131,729 same-sex married couple households and 515,735 same-sex unmarried-partner households.
The actual number of same-sex households (married and unmarried), if the new estimates are correct, are 72 percent of what was originally reported, and married same-sex households are 38 percent of what was originally reported.
Robert Groves, director of the Census Bureau, explained the error in a blog post and YouTube video.
“Let me tell you a wonderful story, a statistical detective story of sorts,” Grove begins. He then explains how small errors can lead to large discrepancies when dealing with small groups within a large population.
Suppose, for instance, a population has 100,000 squares and 1,000 circles. If only one percent of squares were miscounted as circles, then the count would show 2,000 circles, or twice as many circles as actually exist. Since same-sex couples represent a small part of the overall population, an error in data collection can lead to a large miscount.
The Census Bureau suspected that their counts were off because they were much higher than what other surveys have found. They checked their results by looking at first names that were predominantly one gender. Over 95 percent of the time someone named Mary or Alicia are female, for instance. The Census Bureau's new estimates more closely match what other surveys have found.
While the Census Bureau does not seem to know for certain what caused the error, they suspect it resulted from confusion in the survey questionnaire. Respondents or the door-to-door census takers inaccurately recorded gender as a result.
Six states and the District of Columbia currently allow same-sex couples to enter a state-recognized marriage. The census bureau does not distinguish between same-sex married households that are officially recognized by their state and same-sex couples who are recognized as married by themselves or their religious affiliations but not their state.
Same-sex marriage has been a controversial policy in many states. The census bureau must have recognized the sensitivity of the topic when it carefully announced its error with the well-produced web video, complete with graphic illustrations.
Groves displayed pride that his bureau is careful to look for and correct errors when they are found. “This is the technical expertise of the Census Bureau at its finest – examining statistics for anomalies, detecting the cause of a found anomaly, and fixing mistakes from data collection when possible to give the country the best statistics possible,” Groves wrote.