I love the new breed of celebrity pundits – the wonks who make their case with data rather than just opinions. Some, such as Nate Silver with his electoral predictions, have been spectacularly successful. I like Ezra Klein, too. His Wonkblog, though, is not 100% Ezra. There are other contributors. Recently, one of them, Dylan Matthews, sullied the blog and its wonky reputation. He made a rookie mistake in describing the relationship between getting married and getting happy.
The article was another entry in the current discussion about whether it is better to marry at a relatively younger or older age. (My answer: For some of us, it is better never to marry! Don’t expect to find that perspective in all that many places.)
I was especially let down by the article because it started out in such a promising way. It was as if the author actually understood something about statements of causality and the kind of methodology that supports such statements:
“None of the data we have on marriage are definitively causal. That’s a good thing. To have rock-solid evidence that marriage causes anything, we’d need to randomly require some people to marry at one age and others to marry at another age and then compare the results (and even that study design would have plenty of problems).”
“…So the finding that women who wait to get married make more money doesn’t mean that a particular woman would earn more if she got married at 29 rather than 23. It could be that there’s no effect of her age-at-marriage on income, and that the earnings differential between the groups just reflects other differences between their members.”
Methodologically, this is really basic stuff. You learned about it if you ever took even the lowest-level research methods class. But it is still worth applauding because when it comes to statements about the implications of marrying, too many journalists and even social scientists lose their minds. All of their training and good sense evaporates.
And sure enough, many paragraphs into the article, Dylan Matthews has lost any good sense he may have had. He knows he shouldn’t go there about getting married and getting happy (e.g., “Again, it could just be that happy people tend to get married earlier”), but he just can’t help himself. Ultimately, he offers this:
“But the differences are still striking. It’s uncontroversial at this point that marriage, in general, makes you happier, due to the work of Dartmouth’s David Blanchflower.”
Go ahead and click on the link if you want to see the work that has supposedly made the “get married and you’ll get happier” claim uncontroversial. It is astonishing in its unpersuasiveness.
- First, the study does not follow the same people over time. The authors are comparing different people who are currently married or currently unmarried. That means that we don’t know whether any differences in happiness have anything to do with marriage, or whether, for example, something associated with marriage accounts for the difference. Maybe married people have more access to the kinds of resources that bring happiness, such as the constant glorification (matrimania), the freedom from getting stigmatized (the singlism that singles endure), and the benefits and protections accorded by more than 1,000 federal laws.
- Second, the study uses the ever-popular cheater technique of comparing only those who are currently married to the always-single, the divorced, the separated, and the widowed. From that comparison, Dylan Matthews is happy to claim that getting married makes you happier. But the divorced and separated and widowed people did get married! Now look at the results in the tables at the end of the paper. All of those previously married people are less happy than the currently married – true – but they are also less happy than the people who stayed single! They did get married, and they ended up less happy than the people who stayed single. (Again, we can make no causal statements about this.)
- Finally, the paper is more than a decade old. That in itself is not a bad thing, but it is dated. We now know that at least 18 long-term studies have followed people over time to see whether those who got married also got happier. The results eviscerate Dylan Matthews’ claim that it is “uncontroversial at this point that marriage, in general, makes you happier.” You can read the original review here or my discussion of it here, but my bottom line summary is this: “In 18 studies, people reported their happiness or satisfaction starting before they got married and continuing for years afterwards. Getting married did not change people’s happiness.”
Dylan Matthews’ belief in the embarrassingly mediocre marriage and happiness evidence offered by Blanchard (which even he and his co-author cautioned about – see p. 17) was not his only act of matrimaniacal credulity. Look again at Figures 13A and 13B reprinted from the Knot Yet report. (The report was produced by the National Marriage Project, which should immediately set off your matrimania-detectors.) In those figures, the authors report just 25 percent of the data. What would you learn from the other 75 percent? In another article, I will offer a close reading and critique of a marital status and happiness study that has been subjected to the same sort of cherry-picking as we see here, and show how such selective reporting of the data greatly misrepresents the actual link between marriage and happiness. [Here it is.]
In the meantime, Ezra Klein, I’m pleading with you – please don’t compromise your great reputation or and Wonkblog’s by hosting scientifically embarrassing pieces by Dylan Matthews or anyone else.
And to Ezra Klein and Dylan Matthews and everyone else – including journalists, social scientists, and readers everywhere – whenever you read that research has shown that getting married makes you better in some way, beware! You are probably getting scammed.
[Notes. (1) I have been getting requests lately to explain how I go about figuring out what’s wrong with the many misleading studies and statements about the implications of marrying. I’m working on an article tentatively titled, “No study has ever shown that marriage makes you happier, and no study ever will.” If the current draft survives, it will end with a section, “Critiquing the next marriage study in 5 easy steps.” In the meantime, you can read some of my previous critiques at “On getting married and (not) getting happier: What we know.” (2) Thanks to my older brother for the heads-up about the Wonkblog post. (3) My latest elsewhere: “About those 10 things never to say to singles” and "How 20 million readers were misled about happiness."]