David Blankehorn has an interesting article coming out in the Weekly Standard that appears to be a condensed version of some of the arguments made in his new book arguing against gay marriage, The Future of Marriage.
The article is interesting in part because he eschews the argument made by Stanley Kurtz that data from Europe demonstrates a correlation between gay marriage and a decline in marriage and other social ills. From this (flawed) correlation data, Kurtz argues that gay marriage must have caused the problems. Says Blankenhorn: “Neither Kurtz nor anyone else can scientifically prove that allowing gay marriage causes the institution of marriage to get weaker.” He suggests “giving up the search for causation.” Maggie Gallagher, too, has avoided relying on Kurtz. Robert George of Princeton has seemed agnostic about Kurtz’s claims. Now Blankenhorn rejects the Kurtz thesis. It is becoming difficult to find even opponents of same-sex marriage who think Kurtz is right.
Blankenhorn has a new twist on some survey data, however, that he believes does undermine the “conservative case” for gay marriage. Blankenhorn writes that a 2002 survey of attitudes about families and marriage from 35 countries around the world shows that the presence of gay marriage or civil unions in a country correlates strongly with a series of beliefs that he describes as, roughly speaking, anti-marriage. For example, people in countries with gay marriage or civil unions are more likely to agree with statements like, “One parent can bring up a child as well as two parents together,” or “It is allright for a couple to live together without intending to get married.” Conversely, people in countries with no recognition of gay relationships are more likely to agree with statements like, “Married people are generally happier than unmarried people,” or “The main purpose of marriage these days is to have children.” (To keep this post relatively short, I won’t quibble here with the numbers in the surveys, the methodology, or the age of the data. What would happen to the correlation, for example, if we added a country like South Africa, which recently recognized SSM, to the mix?)
All this seems to show is that the more socially conservative a country is the more likely it is to oppose gay marriage and vice-versa. That’s no surprise. It’s no accident, for example, that the first states to recognize gay relationships in some form have been blue states, and that the states where the strongest majorities have banned gay marriage have been deep red states.
So what does this correlation mean for the debate? To Blankenhorn, it suggests that support for gay marriage is one of a “cluster” of “mutually reinforcing” beliefs about family life that are anti-marriage. To make this argument he comes up with a revealing analogy:
Find some teenagers who smoke, and you can confidently predict that they are more likely to drink than their nonsmoking peers. Why? Because teen smoking and drinking tend to hang together. . . .
Because these behaviors correlate and tend to reinforce one another, it is virtually impossible for the researcher to pull out any one from the cluster and determine that it alone is causing or is likely to cause some personal or (even harder to measure) social result. All that can be said for sure is that these things go together. To the degree possible, parents hope that their children can avoid all of them, the entire syndrome–drinking, smoking, skipping school, missing sleep, and making friends with other children who get into trouble–in part because each of them increases exposure to the others.
It’s the same with marriage. Certain trends in values and attitudes tend to cluster with each other and with certain trends in behavior. . . . They are mutually reinforcing. [emphasis added]
Accept one of the beliefs (e.g., that there’s no need to get married if you want to raise kids) and you are likely to accept them all (including support for SSM). Reognize SSM and you are likely to have a populace with anti-marriage views. But this doesn’t work well as an argument against SSM.
First, while he convincingly rejects Kurtz’s correlation-as-causation argument, Blankenhorn himself slips between correlation and causation. Unless supporting SSM or recognizing SSM in a country somehow causes people to accept anti-marriage views, or at least causes them to be more likely to hold such views, then there is no “mutually reinforcing” “cluster” of beliefs to be worried about. There is only a coincidence of beliefs, which may or may not be causally related. Support for SSM among individuals or recognizing SSM in a country is no more tied to general anti-marriage views than is opposition to the death penalty (which I’d bet is another correlation we would find) or any number of other beliefs that might be correlated in traditionalist and non-traditionalist countries. Blankenhorn picks certain survey results out of all the questions asked and asserts, effectively, “these must go together, they must be part of the same world-view.”
To say that these beliefs are “mutually reinforing,” as Blankenhorn does, is just another way of saying that one bears a causal relationship to the other. But as Blankenhorn correctly notes, “Correlation does not imply causation. The relation between two correlated phenomena may be causal, or it may be random, or it may reflect some deeper cause producing both.”
Second, as Blankenhorn’s analogy to teen smoking and drinking reveals, his conclusion that SSM is a bad thing is embedded in his argument. Yes, teens who smoke are more likely to drink and both produce individual and social ills. But we know they produce social ills not because these activities are correlated, but because there is a demonstrated and distinctive harm that each produces. Smoking causes cancer. Drinking causes drunk driving. Similarly, having children out-of-wedlock demonstrably increases risks to them; SSM may or may not produce social ills, but this conclusion is not reached by noting a correlation with practices that do cause harm.
Think of it this way: Suppose I could show that people who attend church regularly are more apt to hold very traditionalist views about the role of women (e.g., that a woman should be a homemaker, not a professional, and should defer to her husband’s authority) or are more likely to be racist (e.g., they oppose interracial dating or marriage), and in fact, suppose the survey data further showed that the more often people attend church the more likely they are to harbor racist and sexist beliefs. Would I have shown that attending church is a bad thing?
Bringing this back to marriage, I’d bet that there was a correlation in 1900 between support for ending the marital rape exemption, support for equalizing women’s role within marriage, granting women the right to vote, and support for ending marriage altogether. This correlation, if it existed, would tell us nothing about whether ending the marital rape exemption or promoting women’s equality or enfranchising women were good ideas.
I suppose Blankenhorn could respond: “Kids shouldn’t smoke because this is more likely to cause them to drink, even if smoking itself weren’t harmful. Similarly, we shouldn’t have SSM because support for SSM is more likely to cause people to believe things that are demonstrably harmful to marriage, even if SSM by itself is not.” But this response would itself depend on a conclusion that there is some causal relationship between support for SSM and support for anti-marriage positions, and yet we cannot know this from a correlation.
A person who’s generally anti-marriage could believe, quite mistakenly, that SSM too is anti-marriage. Instead of deinstitutionalizing marriage, SSM could be a small part of reinstitutionalizing it, despite the marriage opponent’s most fervent hopes. Nothing in a series of correlations in survey data answers that question either way.
In another part of his Weekly Standard article, Blankenhorn quotes from a number of academic supporters of SSM who do indeed see it as a way of deinstitutionalizing marriage. He argues that this also shows that SSM and deinstitutionalizing marriage “are linked.” I’ll address that argument in a coming post.
UPDATE: Stanley Kurtz agrees with me that, even though Blankenhorn eschews causation from the correlations he presents, in fact he slips into causal arguments. This isn’t a problem for Kurtz, of course, but on Blankenhorn’s own logic it’s a big problem for his own argument.