Friday, June 19, 2009

Counting Same Sex Couples in the Census

Potentially great news - the Obama administration, hoping to score some points with gay and lesbian voters after a series of profound let-downs, has instructed the Census to try to count married same sex couples.

I fully support making the effort to count married similar gender couples. Not counting them smacks of denying their existence, which is untenable. But, it's going to take more than jiggering the software to simply acknowledge married similar gender couples in the Census. It's going to require re-interviews, which is going to require funding and a plan.

And that's because of a simple statistical fact.

Most married same-sex couples on the Census are not married same-sex couples.

Huh? Understanding that statement takes a bit of background, don't worry, it won't be painful...

The census form is a piece of paper that gets mailed to most people's homes, or gets filled out in a number of other ways by census enumerators (people who go door to door and from underpass to underpass looking to try to count as many people as possible). But most of the time, the census form is filled out by whoever is living at home.

That piece of paper then gets scanned into a computer and tabulated, counted with a few other hundreds of millions of pieces of paper.

And every once in a while, there is an error.
Some of those errors come from the process of scanning it in, but most come during the process of filling out the piece of paper.
But let's take another step back in the process. Getting counted as a married same-sex couple relies on three pieces of information: 1) the reported sex of the householder, 2) the relationship of the householder to "person 2", and 3) the reported sex of "person 2".

For example, if the householder checked herself off as Female, her relationship to "person 2" as Husband or wife and the sex of "person 2" as Female, then they would be considered a same-sex married couple.
So far, all well and good.
But the problem comes in that sometimes (usually pretty rarely), someone makes an error somewhere along the line - and it's not the same-sex married couples we have to worry about here - it's the married mixed-sex couples.
Let's say a small minority (say one quarter of one percent) of married mixed-sex couples are confused by the form, or are inattentive. Perhaps the person filling out the form starts with "person 2" out as her husband, but then her daughter calls in from the other room, and when she comes back to the form, she's thinking about her daughter and fills in Female. An innocent enough mistake. There are a thousand other scenarios where someone could get confused, inattentive, or just mark the box incorrectly.

There are a lot of married mixed-sex couples in this country, in the 2000 Census, there were a bit over 70 million of them. So, if only one quarter of one percent of the 70 million made an error filling out the form there would be 175,000 couples miscounted as same-sex married couples.
Assuming that all the legally married same-sex couples identified themselves correctly, it still wouldn't come anywhere near that large a number. So, the number of apparent married same-sex couples would be much larger than it really is, and most of these would, in fact, be erroneously classified mixed-sex couples.

That's why when you look at Census numbers, there seem to be a lot more same-sex married couples than we know actually got married, and married same-sex couples look a lot like married mixed-sex couples. That's because most of them are married mixed-sex couples, incorrectly classified. In fact, the married same-sex couples tend to be older, which is completely the opposite of what every demographic study of lesbian and gay people has found. That makes sense, considering that elderly people are probably somewhat more likely to make errors when filling out the form.

So, what's the fix?
Well, the only way to resolve who really is a same-sex couple and who is not would be to go back and ask them.
It would be a lot more accurate, and it's not hard to do. But it does take a small commitment of funds and personnel to get back in touch with the couple of hundred thousand people who will call themselves same-sex married couples on the 2010 Census.

For an example of how well it works, consider the work done by Susan Cochran and Vickie Mays at UCLA. They had a bunch of in-depth questions that they wanted to ask about the health of same-gender-loving people identified in the California Health Interview Survey, so they planned a follow-up conversation with anyone who identified themselves as gay, lesbian, or bisexual, or who reported having sex with someone of the same sex.
Lo and behold, when they called these people back, a surprisingly large number of them were completely baffled - they were not gay, lesbian, or bisexual, nor had they had sex with someone of the same sex - somewhere during the interview, a small error had been made, and a very small proportion of the heterosexual majority had become a sizeable percentage of the sexual minority population inadvertently.
The largest group where these errors had occurred was among the elderly, so many that in future runs of the CHIS, they decided not to even ask anyone over 70 about their sexual orientation.

In conclusion, I'm not trying to throw a wet blanket on counting same-sex couples - just the opposite - I want to count them accurately. And an accurate count means going back to weed out the sizeable number of erroneously classified heterosexually married couples.

Make your voice on the issue heard! Support the NGLTF's petition


  1. I'm glad to see this, and hope it's true. But I'm still insulted and angry. Assume you've seen

  2. Thanks Woolverine for pointing me to
    Neat website.
    It's easy to get discouraged, but I try think of it that a President's usual course is to try to present a middle road, and if we're not vocal in making demands, then that road isn't going to take us into account, no matter what his campaign promises were.
    So, let's pour on the pressure!