Sometimes demography is the boring gray lining under the shimmering silver cloud of social change. The people who fill out surveys want demographic data to reflect themselves as individuals -- celebrating their hard-fought identities -- but demography wants to understand them as groups. Social scientists would rather have an accurate estimate of the population average than a correct measurement of any one individual.
Often these visions coincide; sometimes, they collide. A good example is the collection of multiple-race identities by the federal bureaucracy, which has added a new layer of data reporting on the "Two or More Races Population." But almost all African Americans, for example, share with Michelle Obama a genealogy that includes White slave masters, although they don't report themselves as both Black and White on the Census. And those who practice racism against them don't much recognize the nuance either. Still, there was enough ambiguity in the science, and enough political demand for the change, to make it happen.
There are revolutions in real life, and then there are demographic revolutions. Off the top of my head, two of the latter jump to mind. One is the integration of migration into demographic thinking. Before that, we could make assumptions about population growth based on counting people, births and deaths, like this:
Population in the future = Population now + births - deaths.
Consider the population of male Chinese immigrants in the 19th century, which was restricted by the Chinese Exclusion Act. Rather than wither away as a cut-off branch of the Chinese family tree, in fact Chinese men found creative means of avoiding the immigration rules, and replenished their population with a revolving door of migrant workers. Understanding migration is crucial to any demographic analysis. The formula now looks like this:
That requires asking more questions and some more complicated math.
Another demographic revolution that followed broad social change was the delinking of marriage and childbearing. Once upon a time, demographers interested in population growth simply studied the birth rates among married women. Needless to say, this doesn't work very well anymore. That recognition also required a change in scientific methods and a different set of assumptions.
Now we discover that not only do people have children outside of marriage, they increasingly form long-term commitments and even have children in same-sex pairs. Coming to grips with this will create quite a bit of turmoil in that gray cloud lining -- most of it invisible to the consumers of demographic products.
After Congress adopted the Defense of Marriage Act in 1996, federal agencies refused to report on married same-sex couples -- even those married in places that recognized their vows -- because they were instructed that such marriages were not to be viewed as marriages. Census data were edited accordingly. But legal opinions have a way of evolving when administrations change, and Census will now count same-sex couples as married if they say they are married.
Reversing course and collecting better data will take years, however -- maybe 10 years, in the case of the American Community Survey, the biggest ongoing survey. The good news is that Bureau and its director Robert Groves, an experienced survey methodologist, are actually trying to do it right. Groves has taken a personal interest in the issue, and authorized the resources necessary to at least get it started.
As promised, the Census Bureau will tally the count of same-sex couples in the 2010 Census, but probably not that well. That's because the new, stripped-down Census form doesn't include enough information to gather an accurate count -- the form includes no detailed question on marital status.
I previously passed along estimates of the number of same-sex couples that were reporting themselves married in the American Community Survey. The remarkable figures from 2005-2007 suggested that about half of the roughly 750,000 same-sex couples had listed themselves as married. That implied that hundreds of thousands of couples who were not legally married (which is still less than 100,000) were finding a way to say "They Did" on the form. But it turns out that most of those married couples were the result of coding errors -- the way data enterers counted multiple check marks and missing data, for example. Of the 150,000 or so who remain "married" after the corrections, an unknown number are still the result of one error or another.
So getting good numbers will take a while. We have to sort out how to sort out the errors from the true responses. We need to understand the language couples use in and around their varyingly-recognized relationships. Ideally, we would correctly identify unmarried couples as well as those legally married. And if people want to say they're married, we need to count them that way regardless of their legal status, just as we always have with opposite-sex couples. The Census Bureau is working on a series of focus group studies with different kinds of couples and single people around the country. From there they will do in-depth studies of how people interpret questions asked in different ways. That will lead to redesigned questions on their survey forms.
When Carl Sagan called science a "candle in the dark" he didn't mean it as an abstract, distant point of a light, but as a way of actually illuminating things we need to know. Allowing the Census Bureau to apply its scientific powers to the question of changing families is a breath of fresh air from which demographers, and the public that relies on their answers, will ultimately benefit.
Disclosure: I am a former Census analyst and an unpaid adviser to the Census Bureau on the "Relationship Study Project," which is working on this issue.