First look: 2013 divorce rates show big drop

The 2013 American Community Survey (ACS) is out, and with it the numbers we need to update the divorce rate trend.*

These are the estimates based on the giant sample: In 2013 there were 63,951,934 married men, and  63,619,135 married women. In 2013, 1,071,278 men and 1,197,095 women reported getting divorced in the 12 months prior to their interview. That means the refined divorce rates — divorces per 1,000 married people — were 16.8 for men and 18.8 for women. Wow! Look at the trend now:

That’s a very big drop, almost as big as the 2009 drop. What does this mean? It’s too early to say without more investigation, but consider it in light of my analysis of the recession period. Starting with the 2008 data, it looked to me like there was a big drop at the start of the recession — which we figure is related to the costs of divorce (legal fees, real estate, other transition costs) — and then a rebound as the postponed divorces start to materialize.

I didn’t just guess that based on the trends. I used the variables associated with divorce in 2008 (age at marriage, marital duration, education, race/ethnicity, etc.) to predict divorces in the later years, and found that the prediction was higher than the observed number of divorces, suggesting a deficit of realized divorces.

That interpretation might still be true (which is good, because the paper was published less than a year ago). But the ACS marital events (e.g. ,divorce) data only go back to 2008, so it’s difficult to evaluate the 2009 drop in historical perspective. Now we have to wonder: What if 2009 was really closer to the “normal” rate of divorce decline, and really the recession just gave us the 2010-2012 increase, and no drop? I’m not ready to conclude that, of course. My analysis still makes sense. And there is previous research (cited in the paper) that shows declines in divorce during recessions. But 2013 is going to have to be explained somehow.

I have an idea. I’ll just hop over to the government’s official divorce statistics page to compare these ACS numbers with the actual number of divorces recorded. Wait, what?

Hm. Well, OK, there’s no “detailed data” from the vital statistics system anymore. But surely there is at least a simple count available? I’ll just click on “National marriage and divorce rate trends,” to get those. Um…

So these numbers only go through 2011, and they exclude 6 states, which together account for 20% of all US divorces. Here’s a good data exercise: find another rich country that doesn’t have a count of its own marriages and divorces.

As I have pointed out in increasingly alarmed tones (in this post and earlier), the ACS marital events and marital history questions have been slated for removal by the budget powers that be.** Because if there’s one thing we don’t want to spend money on, it’s information. Why bother? We can just do a Google search or use Big Data to count up #imfinallydivorced hashtags. Yes, I just made that up. But the way things are going we may soon be begging Facebook and Google to tell us what’s going on. As Vonnegut might say, “Good luck, America!”

Notes

* The public use files are up on the IPUMS.org site, and the online analysis tool is available for quick analysis, but for this I used the numbers from the full survey, available on the Census Bureau’s American FactFinder, tables 1YR_B12503 and 1YR_B12001. I realize it’s odd that the rates for men and women here differ (by more than would be possible even if lots of women are divorcing other women). This is a survey question, not a count of legal divorces.

**The information about the planned cuts to the American Community Survey is here: https://www.federalregister.gov/articles/2014/10/31/2014-25912/proposed-information-collection-comment-request-the-american-community-survey-content-review-results:

Direct all written comments to Jennifer Jessup, Departmental Paperwork Clearance Officer, Department of Commerce, Room 6616, 14th and Constitution Avenue NW., Washington, DC 20230 (or via the Internet at jjessup@doc.gov).

Comments will be accepted until December 30.

Marriage rates among people with disabilities (save the data edition)

Cross posted on the Families as They Really Are blog.

Disability is a very broad concept, representing a wide array of conditions that are not easily captured in a simple demographic survey. However, disabilities are very prevalent, especially in an aging society, and the people who experience disabilities differ in important ways from those who do not. Previously I reported — in a preliminary way — that people with disabilities are much more likely to divorce than those without. Here I present some numbers on marriage rates.

This isn’t the kind of thorough, probing analysis this subject requires. But I have two reasons to do it now. First is that I hope to motivate other people to pursue this issue in greater depth. And second, I want to highlight the importance of the data I’m using — the American Community Survey (ACS) — because it might be not available for much longer. These questions have been slated for demolition by the U.S. Census Bureau on cost-saving grounds. I put details about this issue — and how to register your opinion with the federal government — at the end of the post.

Disabilities

The ACS asks five disability questions (I put the shorthand label after each):

1. Is this person deaf or does he/she have serious difficulty hearing? (Hearing)
2. Is this person blind or does he/she have serious difficulty seeing even when wearing glasses? (Vision)
3. Because of a physical, mental, or emotional condition, does this person have serious difficulty concentrating, remembering, or making decisions? (Cognitive)
4. Does this person have serious difficulty walking or climbing stairs? (Ambulatory)
5. Does this person have difficulty dressing or bathing? (Independent living)

These aren’t perfect questions, but they cover a lot of ground, and the ACS — which involves about 3 million households — can’t get into too much detail.

One great thing about having these questions on the giant ACS is you can use the data to get all the way down to the local level, or into small race/ethnic groups. And with the marital events questions, you can combine disability information and marriage information.

First-marriage rates

Using marital events (did you get married in the last year), marital history (how many times have you been married), detailed race and ethnicity breakdowns, and the disability questions above, I produced the following figure. This uses the combined 2008-2012 ACS data because these are small groups, but even with five years of data these groups get quite small. There are about 90,000 non-Hispanic Whites with a cognitive disability in my sample, but only 356 people who are both White and American Indian with a hearing disability (the smallest group I included). This sample is people ages 18-49 who have never been married (or just got married).

The overall first-marriage rate for people ages 18-49 is 71.8 per 1,000. For people with disabilities it’s 41.1 (shown by the blue line). So that’s much lower than for the general population. But there is a very wide variation across these groups, from 15.5 per thousand for Blacks with disabilities in independent living all the way up to above the national average for Whites and White/American Indians with hearing disabilities. (For every condition, Blacks with disabilities have the lowest marriage rates.)

I don’t draw any conclusions here, except that this is an important subject and I hope more people will study it. Also, we need data like this.

In previous posts demonstrating the value of this data source, I wrote about:

• Divorce generally
• Divorce for “millennials” in 25 cities
• Divorce for Asian Americans from different national origins

Whether you are a researcher or some other member of the concerned public, I hope you will consider dropping the government a line about this before the end of the year.

The information about the planned cuts to the American Community Survey is here: https://www.federalregister.gov/articles/2014/10/31/2014-25912/proposed-information-collection-comment-request-the-american-community-survey-content-review-results:

Direct all written comments to Jennifer Jessup, Departmental Paperwork Clearance Officer, Department of Commerce, Room 6616, 14th and Constitution Avenue NW., Washington, DC 20230 (or via the Internet at jjessup@doc.gov).

Comments will be accepted until December 30.

Divorce, animated

Next up the in The Story Behind the Numbers series of animations for my book: Divorce (Chapter 10, “Divorce, Remarriage, and Blended Families). Using the demographic characteristics associated with divorce, from my paper here, the artists at Kiss Me I’m Polish set the story in charming abstract creatureville:

Here’s the relevant table from the paper. Positive coefficients mean the variable is associated with increased odds of divorce, negative is the reverse:

Meanwhile, I have written several posts about the planned cuts to the American Community Survey, which include the questions necessary to conduct this analysis: marital events (did you get divorced in the last year) and marital history (how many times have you been married, and when was the last time).

Here’s the information on how to register your opinion on these cuts:

The information about the planned cuts to the American Community Survey is here: https://www.federalregister.gov/articles/2014/10/31/2014-25912/proposed-information-collection-comment-request-the-american-community-survey-content-review-results:

Direct all written comments to Jennifer Jessup, Departmental Paperwork Clearance Officer, Department of Commerce, Room 6616, 14th and Constitution Avenue NW., Washington, DC 20230 (or via the Internet at jjessup@doc.gov).

Comments will be accepted until December 30, 2014.

So you want to know the Asian divorce rate (save the ACS marital events edition)

One of the most popular posts ever on this blog is about Asian incomes, and especially the variation in average incomes across Asian national-origin groups and cities. Turns out the diverse Asian groups have different divorce rates as well. Why not? It would be nuts to assume the immigrants and their descendants from everywhere from Bangladesh to Japan had common family practices and behavior.

We can figure this out with the American Community Survey (see below; data from which is provided by IPUMS.org). The ACS is big enough to measure divorce rates for Asian subgroups if you pool together a few years — for this I use the 2008-2012 file. For reliability, here I am just showing those groups that had a sample of at least 1,000 married people. And I’m including as separate groups those that selected more than one “race” — Japanese-White, Korean-White, and Filipino-White (you’ll see why I separated them out). Note these are multiple-race individuals, not couples in which the two spouses reported different races.

The national refined divorce rate — divorces per 1,000 people — fell from 20 to 18 at the start of the recession in 2008, before rebounding back up to 19 by 2012. So compare these numbers with about 19 as the national average divorce rate (click to enlarge).

Look at that spread! Now won’t you feel a little foolish for even asking what the “Asian” divorce rate is? I leave the interpretation to the relevant experts (media note: but I’ll be happy to speculate if it will help you get your story past the editor).

A further wrinkle: gender. Unfortunately, because the ACS is a household survey, if someone is divorced, the person they divorced is usually not living in the same household, which means we don’t know who they divorced (or even the other spouse’s gender!). Naturally, men and women in the same ethnic group can have different divorce rates to the extent that they marry outside their own group (or get gay divorced at different rates).

So here are the divorce rates for the same groups, but separately by gender. Groups above the line have higher divorce rates for men (Pakistanis, Cambodians), those below the line have higher divorce rates for women (Korean, Vietnamese, Korean-White). Click to enlarge:
By now you’ve realized what a wonderful treasure-trove of data this is for understanding the incredibly expanding family complexity that pulses all around us. Or, as they say, “Pretty nice data you got there. I’d hate to see something happen to it.” Read on.

Speak up

Last week I reported “millennial” generation divorce rates for 25 metropolitan areas. That’s something you can only get from the very large American Community Survey (because we have no national registry of marital events).

In addition to local areas, however, the vast size of the ACS lets us drill down into very small groups in the population — like small Asian subgroups. For another example, remember the big ruckus over same-sex marriage (you know, homogamy)? I for one would love to have good national data on same-sex marriage patterns when the equality-deniers finally lope back into their caves and the dust settles.

But now the feds are proposing to scrap the marital events (did you get married, divorced, or widowed last year?) and marital history (how many times have you been married, and when was the last time?) questions from the ACS just to save a few million dollars. I hope you’ll help demographic science convince them not to. (In the previous post I listed a bunch of divorce facts we only know because of the ACS questions.)

The information about the planned cuts to the American Community Survey is here: https://www.federalregister.gov/articles/2014/10/31/2014-25912/proposed-information-collection-comment-request-the-american-community-survey-content-review-results:

Direct all written comments to Jennifer Jessup, Departmental Paperwork Clearance Officer, Department of Commerce, Room 6616, 14th and Constitution Avenue NW., Washington, DC 20230 (or via the Internet at jjessup@doc.gov).

Comments will be accepted until December 30.

* Contrary to popular belief, there is no “Asian” category on the Census/ACS form. People are identified as Asian if they pick any of the Asian national origins listed on the “race” question. It’s all pretty American-exceptional. Here is the question, from this form:

Top 25 cities for Millennial divorce (save the American Community Survey marital history and events questions edition)

First some news, then the urgent story behind the news.

All marriages and divorces are local. Whether and when people marry is dependent on who they meet and the conditions under which their relationships develop — social, economic, and even political. And divorce depends on local factors as well, such as the likelihood or feasibility of meeting alternative partners, the costs and consequences of divorce, and the norms and laws regulating divorce. (The same is true for forming and ending non-marital relationships, but those are harder to measure and their dynamics are different).

The Millennial Divorce Capital

So, I know the question you’re dying to have answered is: Where do Millennials get divorced? This question is so compelling that I’ve suspended my normal objection to arbitrary generational definitions, and let Mellennials be defined as anyone born in 1980 or later — so this is the divorce rate among people roughly ages 15 to 31 in the 2009-2011 American Community Survey.

Measuring divorce is complicated, because you’ve got a lot of choices. Here are two simple ways: The number of divorces among Millennials occurring in the previous year per 1,000 Millennials in the population (the crude Millennial divorce rate), and the number of those divorces per 100 married Millennials (the refined Millennial divorce rate). Think of the crude rate as the chance of meeting a Millennial who just got divorced walking down the street, and the refined rate as the chance that one of your married friends just got divorced. I’ve ranked them by the refined rate for this table, which we can use to crown Portland, Oregon the Millennial Divorce Capital of the United States (it’s #1 on both measures).

This is just the 25 largest Millennial population centers, for which we have the most reliable estimates of divorce rates. Nationally, 6.2 out of every 1,000 living Millennials reported getting divorced in the previous 12 months.

It’s complicated

The differences in the two rates I show can be very important. For example, if I expanded the list to the top 50, you would see that the city in which you are most likely to bump into a divorced Millennial at random (spilling your non-caffeinated beverage) is Salt Lake City, where an astonishing 9.7 out of every 1,000 Millennials got divorced in the past year. That’s not because they love divorce, however, it’s because they love marriage. An amazing 34% of Salt Lake City Millennials in 2009-2011 had already been married, compared to just 23% in Divorce Capital Portland.

On the other hand, it’s not just that more married people means more people available for divorce. It’s also the case that early marriage increases the risk of divorce. And more than that, places where early marriage is common have higher rates, even for people who get married at older ages. In this figure I show an adjusted divorce rate (technically, the predicted chance of divorcing in year 5 given marrying at age 23) by the average age at first marriage in each metro area: Divorce is less common in the late-marriage cities*:

(Note that, to produce this figure, you need a survey that asks millions of people how many times they’ve been married, the year they most recently got married, and whether they got divorced in the past year. You don’t just type this into a Google search box.)

But wait, I’m afraid it’s more complicated than that. And here I’m moving toward the urgent story behind the news.

People move around. Divorce may occur in a split second, but what demographers call “relationship dissolution” unfolds over time. People move after they divorce, they divorce after they move, and they may even get divorced in places other than where they live. The ACS data I’m using here help sort this out. This divorce incidence measure is based on a survey question, not a legal record. As with all the other questions on the survey (age, race, income, education, etc.), we more or less have to trust the answers people give (some implausible answers are edited out by the Census Bureau). If we really want to understand how and where divorce (or marriage) happens, we need to be creative and careful, and use the best data. And this is the best data.

Here’s a simple illustration. This figure shows the percentage of two groups of Millennials in each state who arrived in the past year: Those who are married, and those who just got divorced. For example, in Oregon, home of the Millennial Divorce Capital, 17% of the divorced Millennials lived in a different state last year. So either moving to Oregon led to their divorce, or their divorce led to an interstate migration. In contrast, only 7% of Oregon’s married Millennial population just got there (click to enlarge):

The red line is the diagonal, so states above the line — most states — have more divorced arrivals than married arrivals (I excluded a few states with few cases in the data). There are, naturally, a lot of fascinating ways you might approach these questions. Which brings me to the urgent news.

Save the American Community Survey marital events and history data

I know from experience that some of you are thinking things like, “Break it down by race!”, “What about gay couples?”, and “What about hypergamy?” If you want those answers, get out your wallets. This information doesn’t just happen, it’s garnered through a massive federal data collection, without which our ability to know ourselves and our society would be severely compromised. And that’s what might be about to happen.

The data I just showed came from the American Community Survey (ACS), the large Census Bureau survey that replaced the “long form” of the decennial census in the 2000s. (The data are wonderfully prepared and distributed by the good people at IPUMS.org.) Unlike a simple national random survey — which is a major undertaking in itself — the ACS uses a sophisticated rotating geographic design that samples from all around the country to gather the information we need for all levels of geographic detail – down to the neighborhood.

Filling out this survey, in 3.5 million households, is estimated to take about 2.3 million hours of the American people’s time and cost a fortune. Now the federal government is reviewing the different parts of the survey looking for unnecessary parts, and they have identified 7 questions that could be cut, including the ones I’ve been using here: marital events (did you get married, divorce, or widowed in the last year), marital history (how many times have you been married, when did you get married most recently), and a couple others. So I’m trying to convince you to submit a public comment urging them not to make the cuts.

Why do we need this?

Believe it or not, there is no national count of marriages and divorces. That’s right, your government cannot tell you how many legal marriages and divorces there are. They used to collect this from every state, but now they don’t. States collect this information, but it’s not standardized, and it’s not collected together. And, even if it were, we wouldn’t be able to analyze it with all the detail I’ve used here — using marriage duration, age at marriage, and other important factors. So, even at the national level, this is all we have.

However, just for national marriage and divorce statistics, we wouldn’t need the ACS. We could use a smaller survey, like the Current Population Survey or others. If they wanted to work out one of those alternatives before canceling these questions, that would be OK for national statistics.

However, for smaller populations — state and local populations, minority groups, gay and lesbian couples — there is no alternative. If we lose these questions on the ACS, we lose the ability to do all that. Unfortunately, there is no legal or legislative mandate to collect this information down to the local level, which is why it’s on the chopping block. It’s just super interesting and important, not legally required. So we need to communicate that up the chain of command and hope they listen.

To help motivate and inform you toward that end, here’s a list of what we would not know about divorce without the ACS marital events and history questions, and then the information for contacting the federal government with your comment. These are just from my blog, I haven’t done an exhaustive search. The point is not that I’ve done so much, but that there is so much of vital importance that we can learn from this data.

• The refined divorce rate in 2012 was 19 per 1,000 married people.
• The overall projected divorce rate for couples marrying in 2012 is about 50%. This requires using marriage, divorce, and widowhood incidence to calculate competing risks. You need all the ACS questions for that.
• We lost about 150,000 divorces during and after the recession. Then divorce rebounded to catch up to its (declining) trend. That’s the result of my analysis published in Population Research and Policy Review, which relied on a model using all the ACS individual data in all 50 states.
• People with disabilities are much more likely to get divorced than people without. The magnitude of the difference depends on the type of disability.
• Divorce rates in first marriages are more than three-times as high for Black women as for Asian women in the U.S.
• The 2008-09 refined divorce rate by state was correlated with Google searches for “colt 45 automatic” at .86 (on a scale of -1 to 1).
• The 2011 crude divorce rate by state was correlated with Google searches for “vasectomy reversal” at .79.
• The changing pattern of same-sex marriage across states and local areas. We don’t know this yet, but we should, and we’ll want to, and the ACS is the only way we’ll be able to.

Speak up

The information about the planned cuts to the American Community Survey is here: https://www.federalregister.gov/articles/2014/10/31/2014-25912/proposed-information-collection-comment-request-the-american-community-survey-content-review-results:

Direct all written comments to Jennifer Jessup, Departmental Paperwork Clearance Officer, Department of Commerce, Room 6616, 14th and Constitution Avenue NW., Washington, DC 20230 (or via the Internet at jjessup@doc.gov).

Comments will be accepted until December 30.

* Here’s the mixed-effect multilevel regression testing the relationship between average age at marriage (meanagemarr) and odds of divorcing, controlling for age at marriage and duration of marriage, for 262,269 married people in 283 metro areas:

If you want to see serious research into the effects of age at marriage, local age at marriage, and religion, on divorce, this paper by Glass and Levchak is the right place to start.

Changing Hispanic racial identity, or not

Hector Cordero-Guzman called my attention to a controversy over Hispanics changing their racial identities. Here is a quick rehash and a few comments. (Spoiler: the New York Times ran a bad story.)

At the Population Association of America, Carolyn Liebler, a sociologist at the University of Minnesota, and James Noon, who works on administrative records at the Census Bureau, presented preliminary results from a comparison of individual race/ethnic responses to the 2000 and 2010 Decennial Censuses. After analyzing millions of individual Census responses, they reported in their abstract that 6% of people changed their race or Hispanic origin classification between 2000 and 2010.

Details of the analysis apparently are not publicly available, but D’Vera Cohn, a writer at the Pew Research Center, reported on their findings, under the headline, “Millions of Americans changed their racial or ethnic identity from one census to the next.” Is this a lot of change? It’s hard to say without a comparison (and without the analysis details). “Millions” does not really mean “a lot,” but it sounds like it does. If the Census race/ethnic identity questions don’t fit people’s self-concept very well then a certain amount of bouncing around is to be expected.

The focus was on Hispanics, whose place in the racial classification scheme is squishy (including immigrants who came at different ages from countries with different racial schemes and ancestral origins, living in different parts of the country with different racial attitudes, some concentrated in dense communities and some dispersed, some economically marginalized and some much more upwardly mobile, etc.). According to D’vera Cohn, 2.5 million Hispanics were “some other race” in 2000 and “white” in 2010, while 1.3 million were “white” in 2000 and “some other race” in 2010.

I might conclude from that that it’s messy and the categories don’t work very well. But it’s also possible that this reflects fluid identities, and people actually change how they see themselves in a systematic way over time. The PAA abstract says “responses and corresponding identities can change over time,” which leaves open the possibility that the change is in measurement in addition to identity, but the hypothesis they suggest are about identity (hypothesizing that women, young people, and people in the West have more complex or less stable identities).

Identity shift is how New York Times Upshot writer Nate Cohn interpreted the D’Vera Cohn report. Under the headline, “More Hispanics Declaring Themselves White,” he converted that bidirectional flow into “net 1.2 million” changing from “some other race” to “white,” and proceeded to run away with the implications. It’s a good example of using any number greater than zero to confirm something you already believe. For example, he wrote:

The data also call into question whether America is destined to become a so-called minority-majority nation, where whites represent a minority of the nation’s population. Those projections assume that Hispanics aren’t white, but if Hispanics ultimately identify as white Americans, then whites will remain the majority for the foreseeable future.

Hm. The “net” flow from “some other race” to “white” is 1.2 million. That is 3% of the 2000 Hispanic population, or 2% of the 2010 population. So even if it’s truly an identity change, does that save the White majority in the long run?

Anyway, as Cordero-Guzman points out in a detailed discussion, referring to a post by Manuel Pastor, the Census questions changed between 2000 and 2010, with Census adding, in bold, “For this census, Hispanic origins are not races” to the form in 2010. Since many Hispanics write “Hispanic” under “some other race,” this probably discouraged them from choosing “some other race” in 2010.

Cordero-Guzman also points out that the context in which the question is asked (and in which the respondents live) is important. For example, 82% of Puerto Ricans on the island use “white” on the American Community Survey, while in New York City only 45% do. Does their identity — in the sense of how they really think of themselves — change when they are in New York, or do they interpret the question differently because they are answering a question in a different social setting? You can’t quantify that difference, probably, but I wouldn’t assume it’s just an identity change.

In a follow-up post, Nate Cohn acknowledges the wording changes — “an important detail” — but returns to the assimilation-upward mobility story. He should have just acknowledged that he jumped to conclusions in the first post and overreached in the race to produce an important, “data-driven” post. (Nate Cohn may have consulted actual experts, but if he did he didn’t include them in the post. It’s just data, I guess.)

The information economy did it

There is a lesson here in the new information economy. Academic conferences used to be less in the public eye. A preliminary analysis, shared with other researchers, then a Pew writer posts on the results, and the Times splashes them all over, all before a paper is even available. I think the Times story is basically wrong — the data as reported are not independent evidence of “assimilation.” (So, the person with the biggest megaphone was the person who was most wrong — surprise!) But the analysis could well be an important piece of research in a larger literature, and I think it’s good to present preliminary research at conferences. You can’t stop reporters from racing to be wrong, but I do think it would be better to distribute the paper publicly when it’s presented.

Sexual minority counts

One of the big happenings at the Population Association of American (PAA) conference, just completed, was news of progress toward collecting better data on sexual diversity.

Photo of PAA 2014 by Philip Cohen from Flickr Creative Commons

Call it weakness if you like, but in this area I am prone to viewing modernity as a march of progress from a dark past toward a half-full glass of bright future, with popular politics driving widening notions of human rights, motivating legal reforms, compelling the adoption of state bureaucracies to progressive social reality, and gradually incorporating us into a new world order more or less of our own creation.

That last part – about the bureaucracies incorporating the public – might not be the most complicated, but it is still pretty thorny. (And from here till the next subhead it gets technical.)

Good news bad news

The good news is that we have great new data collections coming along. Virginia Cain from the National Center for Health Statistics reported on their new sexual orientation question for the National Health Interview Survey, the largest federal health survey (the paper doesn’t seem to be available yet). This is already yielding important data on health disparities for sexual minorities, which is vital for policy responses to inequality.

Tim Vizard from the UK Office of National Statistics also reported on his agency’s new sexual identity question, which has been tested for several years on a few hundred thousand people each year. The latest numbers show 1.5% of adults self-identifying as lesbian, gay, or bisexual. They get these low numbers because they ask a very simple, narrow question, only on sexual identity rather than sexual attraction or sexual behavior (see other studies for the range of estimates).* Importantly, less than 4% of the UK respondents are refusing to answer, and the question is not affecting overall response rates – two big fears in the statistical agencies that appear to be receding with these and other results. Here’s how they ask it (semi-confidentially, so that in theory a husband and wife taking the survey together could both tell the interviewer they’re gay without either knowing what the other said):

The other good news is that the U.S. Census Bureau is making great strides (which I first praised here), on several tracks. First, they are working on the same-sex married couple data from the American Community Survey (ACS). At present they only release aggregate estimates of same-sex couples, differentiating between those that are married versus cohabiting (explained here).

A big reason we don’t have more data is the bad news: In another paper (just an abstract is posted, but you can ask the authors for a copy), Census analysts Daphne Lofquist and Jamie Lewis reported on their investigation into possible errors in the same-sex couple data the ACS has collected.

The background is that in a 2011 paper (linked here) Census analysts showed that a lot of seemingly same-sex couples were actually different-sex couples in which someone’s sex was miscoded.** If even a tiny percentage of different-sex couples make a mistake on the form – say, 1-in-1000 – then you would roughly double the number of same-sex couples. And they do. The paper used name-gender associations to reveal that, for example, in Texas 29% of supposedly male-male couples had one partner with a name that was used by women 95% of the time in that state – probably women accidentally marked as male.

But that 95% cutoff is a conservative estimate of the error. In the new analysis Lofquist and Lewis went further and checked same-sex couples against their Social Security records to see what sex they had recorded there. The result was shocking: 72.5% of the same-sex couples had a member whose sex didn’t match the Social Security record. Yes, some people change their sex/gender, and some people’s Social Security Records are wrong, but not that many. The much more likely culprit is simply a tiny number of straight people mismarking the sex box (there are some other technical possibilities, too).

The great thing about just asking people their marital status and sex is that you can count gay and lesbian couples without changing anything about the form (such as asking about sexual identity or orientation). That’s what all the people want who think I’m backward for worrying about couple-sex gender terminology. “C’mon!” they say, “Why do you have to label marriage as homogamous or heterogamous – just call it marriage!” Maybe someday, but at the moment that approach is producing an accuracy-crushing level of noise in the same-sex couple data.

Fortunately, Census is also moving forward with other improvements to fix this. The most important change is probably to the basic relationship question, which will soon look something like this, with couples labeled “opposite-sex” or “same-sex,” and the gender-neutral “spouse” added beside “husband/wife.” This will allow Census to check those couples that are reported as married to see if their same/opposite relationship identification matches what they reported for their sexes:

If we end up with a question like that, which seems most likely (the Census testing and development is quite far along), then we should be able to much more reliably identify same-sex couples (both married and cohabiting).

We’ll get used to this

That proposed new relationship question has 17 categories. That’s a long way from these six, in 1960 (the whole series of Census forms is here):

That goes to show you that family diversity is a state of collective mind as well as a structural reality. Building bureaucratic bins into which we pour data describing the various aspects of our lives is one of the defining elements of modern life. Eventually, I am pretty sure people will become disciplined by the new bureaucratic reality, and identities will calcify around checkboxes. That’s life under the modern state. (Even most haters, once they realize the data is being collected, will want to answer the questions accurately so they don’t get counted as gay – although, just as a few people refuse to answer race questions, there will be holdouts.)

* Identifying transgender people is much more complicated and difficult. The number of required questions and categories increases as the size of the groups in question grows smaller. This is feasible for smaller, more targeted surveys, but not in the immediate cards for the big ones (see Gary Gates’s presentation at PAA for more on this).

** I’m pretty sure Gary Gates was the first person to identify this problem, but can’t remember which paper it was in.