First some context, then some data.
Ruth Graham has a story in the Boston Globe about how liberals and conservatives — researchers as well as policy advocates — are starting to agree that marriage is good and policy should promote it. I’m quoted, but apparently as an example of what Andrew Cherlin refers to as someone standing at “a line some liberal sociologists won’t cross, that line of accepting marriage as the best arrangement.” This is part of a spate of stories in which journalists look for a new consensus on marriage. Previous entries include David Leonhardt in the New York Times saying liberals are wrong in attributing the decline of marriage to economics alone, and Brigid Schulte in the Washington Post reporting that Isabel Sawhill has given up on “trying to revive marriage.” The narrow consensus in policy terms involves a few things, like increasing the Earned Income Tax Credit and reducing marriage penalties in some parts of the safety net, along with trying to improve conditions at the low end of the labor market (see this Center for American Progress report for the liberal side of these policies).
From teen births to marriage promotion
The idea of a cultural revival of marriage has been the futile bleat of the family right for decades, most recently retooled by David Blankenhorn. And in recent years these ideologues have taken to using as an example the supposed success of the cultural intervention to reduce teen pregnancy, to show how we might increase marriage and reduce nonmarital birth rates. This has been a common refrain from Brad Wilcox, quoted here by Graham:
As evidence of his optimism, Wilcox points to teen pregnancy, which has dropped by more than 50 percent since the early 1990s. “Most people assumed you couldn’t do much around something related to sex and pregnancy and parenthood,” he said. “Then a consensus emerged across right and left, and that consensus was supported by public policy and social norms. … We were able to move the dial.”
I think that interpretation is not just wrong, it’s the opposite of right, as I’ll explain below.
I don’t know of any evidence that cultural intervention affected teen birth rates. Cultural intervention effects are different from cultural effects — of course cultural change is part of the trend in marriage and birth timing, but the commonly cited paper showing an apparent effect of 16 and Pregnant on teen births, for example, is not evidence that the campaign to reduce teen pregnancy was successful. There was a campaign to end teen pregnancy, and teen pregnancy declined. I think the trend might have happened for the same set of reasons the campaign happened — the same reasons for the decline in marriage and the shift toward later marriage. The campaign was one expression of shifting norms toward women’s independence, educational investment, and delayed family formation.
The myth of teen pregnancy
I’ve been trying to say this for a while, and it doesn’t seem to be taking. Maybe I’m wrong, but I’m not giving up yet. So here goes again.
If you had never heard of teen pregnancy, you would see the decline in births among teenagers as what it is: part of the general historic trend toward later births and later marriage. I tried to show this in a previous post. I’ll repeat that, and then give you the new data.
First, I showed that teen birth trends simply follow the overall trend toward later births. Few births at young ages, more at older ages:
It doesn’t look like anything special happening with teens. To show that a different way, I juxtaposed teen birth rates with the tendency of older women (25-34) to have births relative to younger women (20-24). This showed that teen births are less common where older births are more common:
In other words, teen births follow general trends toward older births.
Today’s data exercise
Here’s a more rigorous (deeper dive!) into the same question. I show here that teenage women are less likely to have a birth if they live in place with higher age at marriage, and if they live in a place with lower marriage rates. That is, lower teen births go along with the main historical trend: delayed and declining marriage.
So if you think declining teen births are an example of how a policy for “cultural” intervention can reverse the historical tide, you’re not just wrong, you’re the opposite of right. The campaign to reduce teen births succeeded in doing what was happening already. This is not a model for marriage promotion.
Here’s what I did. I used the 2009-2011 American Community Survey, distributed by IPUMS.org. For 283 metropolitan areas, accounting for 73% of all U.S. 15-19 year-old women, I calculated the odds of a teenage woman reporting a birth in the previous year, as a function of: (a) the median age of women who married in that area in the previous year, and (b) the proportion of women ages 18-54 that are currently married in that area. I adjusted these odds for age, race/ethnicity, and nativity (foreign born). I didn’t adjust for things that are co-determined with births among teens, such as marital status, education, and living arrangements (in other words there is plenty of room to dive deeper). All effects were statistically significant when entered simultaneously in a logistic regression model, with robust standard errors for metro area clustering.*
The figures show probabilities of having a birth in the last year, adjusted for those factors, with 95% confidence intervals:
- Teen births are a myth. There are just births to people ages 13 to 19.
- Teen births have fallen as people increasingly delay childbearing and marriage. Falling teen births are simply part of the historical trend on marriage: rising age at marriage, declining marriage rates.
- The campaign to prevent teen births coincided with the trends already underway. Any suggestion that this could be a model for promoting marriage — that is, a policy that goes against the historical tide on marriage — is hokum.
- There remains no evidence at all to support any policy intervention to promote marriage.
* Well, the age at marriage effect is on significant at p=.054 (two-tailed), but my hypothesis is directional — and that cluster adjustment is brutal! Anyway, happy to share code and output, just email me. Here’s the regression table: