Printer friendly version

May 21, 2004

Cherry Picking Marriage & Family Numbers

PROEM:
Click "Turn Light On" at the top of the left-hand column if you'd prefer a simpler layout that you may find more readable.


I made it through most of the day maintaining my resolve not to spend any time on Andrew Sullivan's reference to Lee Badgett's piece arguing against Stanley Kurtz's Scandinavian analyses. Although Sullivan — among others of his fellow SSM advocates — seems peculiarly anxious to declare that this "mini-debate, at least, is now over," my time is just too thinly stretched to do today what Kurtz has promised to do next week.

Still, one factoid that Sullivan quotes from Badgett struck me as high, and I couldn't resist checking it out:

In the Netherlands, a bit south of Scandinavia, 90 percent of heterosexual couples with kids are married.

Well, I can't argue with that. According to the Netherlands StatLine, in 2003, there were 1,878,713 married couples with children and only 213,941 unmarried couples with children, which is, indeed, a 90% marriage rate (actually 89.7%). But that doesn't tell quite the whole story. There were almost twice as many single-parent households in the Netherlands in 2003 than unmarried couples: 424,742. In other words, married couples only make up 74.6% of all households with children.

That's still relatively high, of course. However, since we're looking at these numbers within the context of historically recent events, the current number is less important than the trends since those events. And in 1995, the year the same-sex marriage ball started rolling in the government, the numbers were: married, 2,012,375; unmarried, 99,610; and single, 360,754. In other words, married couples were 95.3% of all couples with children and 81.4% of all households with children. The numbers have since dropped 5.6 and 6.8 percentage points, respectively.

Although married couples represent the only decreasing group, a dark silver lining exists in that unmarried couples are increasing at a faster rate than single parents. At least there are two parents in such homes. It should be noted, however, that this rate difference decreases for households with multiple children.

Once again, it strikes me as premature, to say the least, to declare the matter settled.

ADDENDUM:
While poking around the statistical Web site, I noticed a new dynamic that will surely come up in the future, so it's worth mentioning here. The divorce rate in the Netherlands is now lower than it was throughout most of the '90s. It had been hovering in the low-to-mid 30,000s and increased steadily after 1998 (the year same-sex partnerships actually entered the law), rising from 32,459 that year to 37,104 in 2001.

On April 1, 2001, marriage itself was opened up to same-sex couples, and since then, the number of divorces has fallen — to about 32,000 in 2003. However, at the same time, the law was changed to allow couples to switch between marriage and registered partnerships with merely a visit to a registrar. Unlike marriages, registered partnerships can be dissolved out of court, by a solicitor or notary. This has led to a phenomenon called "flash annulments," of which there were about 5,000 in 2003. In other words, married couples are getting divorced without its being counted among the divorce statistics. "About 60 percent of the conversions [from marriage to partnership] were dissolved within a month, 90 percent within six months."

Including flash annulments, divorces in the past three years have been well above any previous year, before or after registered same-sex couples of one form or another.

Posted by Justin Katz at May 21, 2004 8:31 PM
Marriage & Family
Comments

Including flash annulments, divorces in the past three years have been well above any previous year, before or after registered same-sex couples of one form or another.

But as the graph shows, divorces+flash annulments increased sharply between 1998 and 2001, when gay marriage was legalized, but have decreased slightly since that time. This implies that if there is a causal relationship between gay marriage and divorce, it is that gay marriage reduces divorce.

Posted by: Jon at May 21, 2004 9:02 PM

And in 1995, the year the same-sex marriage ball started rolling in the government, the numbers were: married, 2,012,375; unmarried, 99,610; and single, 360,754. In other words, married couples were 95.3% of all couples with children and 81.4% of all households with children. The numbers have since dropped 5.6 and 6.8 percentage points, respectively.

But again, you're cherry picking your own statistics to try and suggest a trend that simply isn't supported by the totality of the data. For example, marriages per thousand inhabitants increased between 2001, the year gay marriage was legalized, and 2002 from 5.1 to 5.3. Moreover, marriages + domestic partnerships per thousand in 2002 was more than 5.8, the highest number since 1992. And the numbers of men and women who are either divorced or never married declined substantially, in both absolute and relative terms, between 1991 and 2002.

But the larger problem with this whole ridiculous exercise is that whatever effect gay marriage may have on heterosexual marriage rates, divorce rates, birth rates, abortion rates, or anything else, you cannot possibly draw any even minimally reliable conclusions about those effects from the limited data in your citations. The fact that are so many other factors that affect the rates of these events, that they are subject to long-term trends that pre-date legal gay marriage or gay partnerships by decades, that year-to-year random fluctuations may swamp any other effects, and that gay marriage has been legal for such a short time, means that there is no credible basis for asserting that gay mariage either harms or helps marriage, children, families or societies. You are just desperately trying to come up with anything, no matter how strained or silly, that might possibly undermine gay marriage, and throwing it out there to see what sticks.

Posted by: Jon at May 21, 2004 9:37 PM

You know, Jon, just as I'm considering how to respond to your valid points, you slip into personal attacks that convince me anew of the uselessness of devoting time to arguing with you.

Posted by: Justin Katz at May 21, 2004 9:42 PM

But what I found in looking into your claims is too good not to pass on, lest anybody else believes you to have the upper hand (particularly given your tone).

But again, you're cherry picking your own statistics to try and suggest a trend that simply isn't supported by the totality of the data. For example, marriages per thousand inhabitants increased between 2001, the year gay marriage was legalized, and 2002 from 5.1 to 5.3.

— Talk about the "totality of the data"! The marriage rate per 1,000 inhabitants for 2001 (5.1) was an historical low and a drop from 5.5 the previous year. That it regained some ground the following year hardly represents a broad view. (Interestingly, the population is so small that it only takes about 800 couples to shift the number by 0.1 per 1,000, so same-sex marriages contributed about 0.3 to 2001 and 0.2 to 2002, making the heterosexual rate even more historically low.) Moreover, turning to some newer data, even as the population struggled upward from 15,987,075 in 2001 to 16,254,933 in 2004, married citizens dropped from 7,079,502 to 7,026,801, and divorced citizens rose from 886,439 to 965,093. It looks like things will prove to have worsened, not improved, since 2002. For reference for a subsequent point, let me point out that the number of "never married" citizens rose, 2001 to 2004, from 7,138,972 to 7,386,612.

Moreover, marriages + domestic partnerships per thousand in 2002 was more than 5.8, the highest number since 1992

— Yes, but you're forgetting that more than half of all registered partnerships in 2002 are actually downgraded marriages on their way to court-free divorce, so your combined number should be 5.55. Moreover, you didn't add domestic partnerships to any of the marriage rates after 1998, which would have given you:

1998: 5.79
1999: 5.91
2000: 5.68
2001: 5.31

And the numbers of men and women who are either divorced or never married declined substantially, in both absolute and relative terms, between 1991 and 2002.

— If you're talking about the "Marriage partners" data, which I believe you are, you need to read your definitions. The "Never married" on the table that you're citing is: "marriage partners who have never been married before." A drop in that number means an increase in divorces and subsequent remarriages. Similarly, a drop in this divorce number just means that fewer newlyweds were previously divorced. With the total number of currently divorced people continuing to climb and the number of marriages continuing to drop, that could just mean that divorcees aren't remarrying.

Posted by: Justin Katz at May 21, 2004 10:55 PM

The marriage rate drop between 2001 and 2002 (0.4) is less than half that between, for example, 1991 and 1995 (1.0), the period before the “same-sex marriage ball” even “started rolling,” let alone before gay marriage was actually legalized in the Netherlands. As for rates of the divorced and never-married, those rates are far lower in the years after gay marriage was legalized than they were in the early 90s. These facts demonstrate that other and more powerful forces than gay marriage must be at work shaping these numbers, and illustrates the basic point of my last post, a point you don’t seem to have understood, which is that none of these figures provide a basis for any meaningful inference about the impact of gay marriage. Marriage rates, cohabitation rates, birth rates, divorce rates, abortion rates, and all the rest of it, are subject to so many influences, and the effects of those influences are so hard to isolate and measure even over extended periods of time, let alone over periods of just a few years, during which random fluctuations may swamp any causal effect, that you cannot possibly draw reliable inferences about the effect of gay marriage on these outcomes by citing aggregate statistics over periods of two or three years. But even if those numbers were meaningful, they wouldn’t support your claims anyway. You are simply sifting through the totality of the data and cherry-picking the “good” statistics that are inversely correlated with gay marriage and the “bad” ones that are positively correlated with it. It’s a joke.

The "Never married" on the table that you're citing is: "marriage partners who have never been married before." A drop in that number means an increase in divorces and subsequent remarriages.

You’re wrong even about this. Since the definition takes no account of the number of previous marriages, a drop in “marriage partners who have never been married before” is consistent with a decrease in rates of divorce and remarriage. And the divorce statistics themselves show a decrease in divorce, both absolute and relative. This decrease pre-dates the advent of “flash annulments,” and, as I pointed out to you earlier, the combined rate of divorces and flash annulments began to decrease the year gay marriage was legalized, which is also inconsistent with the inference that gay marriage causes marital dissolution.

Posted by: Jon at May 22, 2004 12:32 AM

Make that "between 2000 and 2001" in the first line of my previous post.

As I noted earlier, the marriage rate increased following the legalization of gay marriage.

Posted by: Jon at May 22, 2004 12:38 AM

Before I respond, for your "never married" and absolute and relative divorce rates, are you looking at the table titled "Key figures marriages and partnership registrations"?

Posted by: Justin Katz at May 22, 2004 12:57 AM

Moreover, turning to some newer data, even as the population struggled upward from 15,987,075 in 2001 to 16,254,933 in 2004, married citizens dropped from 7,079,502 to 7,026,801, and divorced citizens rose from 886,439 to 965,093. It looks like things will prove to have worsened, not improved, since 2002. For reference for a subsequent point, let me point out that the number of "never married" citizens rose, 2001 to 2004, from 7,138,972 to 7,386,612.

These statistics are just irrelevant, too. Your argument is about the alleged effect of gay marriage and partnership on marriage and divorce. The numbers you give above would tell us precisely nothing about that effect even if it weren’t for all the other problems with your argument that I described previously. Since we know that the marriage rate increased in the year following the legalization of gay marriage, and that the divorce+anullment rate decreased in the year following the legalization of gay marriage, the reduction in the total number of married citizens over that period cannot be attributed to a decrease in marriages or an increase in divorces caused by the legalization of gay marriage, since there were no such changes in those rates. It must instead be caused by other factors, such as the naturalization or births of unmarried citizens, and the deaths or emigration of married ones or their spouses. Similarly, the number of divorced citizens is cumulative, and will increase even after the divorce rate falls, as long as absolute gains outnumber absolute losses. Since we know that the divorce rate did in fact fall after gay marriage was legalized, that must be the explanation in this case, at least for those years for which you have provided marriage and divorce rate data.

I have tried to explain to you at some length, in two separate posts, why none of the statistics you have cited provide any meaningful basis for conclusions about the effect of gay marriage on general marriage or divorce rates, but you seem so lost in your number-crunching that you are utterly oblivious to the question of meaning.

Posted by: Jon at May 22, 2004 5:38 AM

Jon,

You're all over the place. I'm probably being gullible in continuing to try to discuss this with you, but so be it. This original post doesn't represent an argument of causation, or even correlation (and I haven't moved in that direction with you because 1. your tone makes it amply evident that it would be a waste of time, a conclusion that is bolstered by 2. the fact that I can't even get you to agree with what a given number represents.)

This post investigates a specific claim from a piece by Lee Badgett that Andrew Sullivan declared to end the debate. The erroneousness of that declaration appears to be something with which you agree. (Although, perhaps not, since you're simultaneously declaring statistics useless and then using them for your own purposes.) Badgett's piece, in turn, was in response to a piece by Kurtz, which was in part a response to claims by others, such as Darren Spedale, that marriage in Scandinavia became stronger in the '90s. Before we can discuss what trends mean, we have to decide what trends are.

So, before I continue with you, you'll have to answer the simple question that I asked, above, to ensure that we're talking about the same numbers, at least: for your "never married" and absolute and relative divorce rates, are you looking at the table titled "Key figures marriages and partnership registrations"?

One point I will make before you answer that is that it isn't exactly true that "the marriage rate increased in the year following the legalization of gay marriage, and that the divorce+anullment rate decreased in the year following the legalization of gay marriage." Same-sex marriage arrived on April 1, 2001 — with nine months left in the year. In that year, marriage rates dropped, and divorce rates rose. In the following year, some of those losses and gains were made up, but not beyond preexisting levels. Moreover, the limited data out to 2004 suggests that things have worsened, not improved. (And you're ignoring the larger trend since same-sex marriage became de facto, anyway.)

But again: answer my question.

Posted by: Justin Katz at May 22, 2004 7:46 AM