…which also concerns Brad Wilcox, but that’s just a coincidence. I’m sorry I never got around to catching this one, from the June 17 New York Times.
In an interesting article on cultural trends of the well-to-do, the New York Times Fashion & Style section profiled divorced middle-class people who feel ostracized by their class, which has now rebounded against the anything-goes divorce culture of the 1970s.
To lend weight to the generalizations illustrated by the colorful anecdotes, Pamela Paul wrote: “From the 1970s to the 2000s, the percentage of highly educated Americans who believe that divorce should be made more difficult rose from 36 to 48 percent.”
So precise, and right to the point — anti-divorce sentiment increasing among the cultural elite.
Where did this come from? And what do answers to the question mean — “divorce should be made more difficult” — asked before and after no-fault divorce spread across the country? My concerns only deepened when I read, after a quote a few paragraphs later: “… said W. Bradford Wilcox, an associate professor of sociology at the University of Virginia and director of the National Marriage Project.”
The original statistic was not attributed, but now I went back up in the story, copied “from 36 to 48 percent” and Googled the phrase. Sure enough, there on the first page of hits was Wilcox’s 2010 “State of Our Unions” PDF (co-authored with Elizabeth Marquardt), saying:
…the percentage of American adults expressing the view that divorce should become more difficult fell from 53 to 40 percent among the least educated, stayed constant at 50 percent among the moderately educated, and rose from 36 to 48 percent among the highly educated.
Following the paragraph in the PDF is this infographic:
Fortunately, anyone with a microscope can see that the graph was made from data from the General Social Survey, which is available in a free, not-too-hard-to-use web-based utility here. So there is little reason to just quote a Templeton foundation-funded blog post, when you can go straight to the source – a product of peer-reviewed funding from the lowly National Science Foundation. In fact, an NYT intern could have included data from the 2010 GSS as well, since it was out by the time of this 2011 article — so there was no need to even consult Wilcox’s 2008 data.
GSS has asked the DIVLAW question since 1974: “Should divorce in this country be easier or more difficult to obtain than it is now?” California’s first-in-the-nation no-fault divorce law was in 1970, and it was spreading across the country at the time, so this was a big issue in 1974. What does it mean now? I’m not sure.
Anyway, just looking at the “highly educated,” by which Wilcox means BA degree or higher, this is the percentage answering the question, “more difficult”:
The GSS utility helpfully calculates confidence intervals, and I’m glad to report that if you pool the years 1974-1979 on one end, and 2000-2008 on the other, you do indeed get a statistically significant increase at 95% confidence in the percentage of people with BA degree or higher thinking divorce should be “more difficult.”
This fits with Wilcox’s narrative, in which liberal attitudes from the cultural elite have wrecked the families of the poor, after which the rich turned back to more sensible family values (and church) for their own purposes, further exacerbating the gaps between rich and poor. This is a simple argument which relies on tortured facts. Wilcox and Andrew Cherlin advanced it together in a Wall Street Journal Op-Ed last fall, with this helpful illustration:
But the figure reveals a wrinkle that would — sorry to say — complicate the story. Let’s call it a dramatic reversal of the trend in attitudes toward divorce!
Look back at that trend. Instead of focusing on 1970s versus 2000s, look instead at the “early 2000s” (2000, 2002, and 2004) versus the “late 2000s” (2006, 2008, 2010). By that breakdown, the percentage of BA+ folks saying divorce should be “more difficult” fell from 54.1% to 44.2% — a 10-point drop that is also statistically significant at the 95% confidence level. That change is almost as big as the whole 1970s-2000s swing of 12 points that Wilcox and the NYT reported.
This doesn’t fit the NYT story, which is attempting to capture a recent trend.
One problem may be that archaic question about how “difficult to obtain” divorce should be. Compare that with the Gallup poll, which instead asks “whether your personally believe that in general [divorce] is morally acceptable or morally wrong.” They didn’t break it down by education level, but the trend over the 2000s was toward “morally acceptable”:
The “then and now” story in which the 1970s serves as a post-1950s bogey man is easy. I think the reality is not.
Real research addendum
The Wilcox-Marquardt report actually cites a paper in Journal of Marriage and Family by Steven Martin and Sangeeta Parashar, “Women’s Changing Attitudes Toward Divorce, 1974–2002: Evidence for an Educational Crossover,” which broke the story of the diverging social-class trends in attitudes toward divorce (using the same DIVLAW question). Their analysis was published in 2006 based on GSS data through 2002 (before the crash in elite attitudes toward divorce). It’s a serious analysis that tries to explain — or at least put in context — why college-educated women might be moving more in the direction of opposing divorce.
They conclude that, as divorce becomes less common for people with college degrees — which it has — then the opposition to divorce among these cultural elites is increasingly abstract. Martin and Parashar presciently wrote:
If a low probability of divorce reduces the personal salience of divorce for college graduates … they might increasingly view divorce and its attendant hardships as a social problem caused by other people’s behavior. If this occurs, trends in divorce attitudes could exacerbate family inequalities and sharpen class delineations in the “culture wars” over the future of families.
That seems plausible to me. I don’t know if the post-2002 change in attitudes seriously alters this story — it’s ultimately too much weight for one quaint attitude question to bear — but it’s worth considering.
*In a normal news story on a poll, the “margin of error” is reported, which I think is the 95% confidence interval for sampling error. When quoting a “report,” though, apparently that is not required.