The Magazine

Start a Family...

And before you know it, you'll be voting for the GOP.

Apr 22, 2013, Vol. 18, No. 30 • By JONATHAN V. LAST
Widget tooltip
Audio version Single Page Print Larger Text Smaller Text Alerts

In 2005, Steve Sailer wrote a cover story for the American Conservative theorizing that the divide between red and blue states was driven in large part by the cost of family formation. Sailer dubbed this the “Dirt Gap” (referring to the price of homes with yards), and his general thesis was that affordable family formation—and the attendant bourgeois life which it enabled—was the source of our political divisions.


In February, George Hawley, a political science professor at the University of Houston, decided to test Sailer’s theory. In a paper published in the peer-reviewed journal Party Politics, Hawley built a model which ought to be studied by every Republican political operative in the country. Because it shows not only that Sailer was correct—lower median home values are closely linked to Republican voting—but that one of the key factors linking home values and Republican voting is marriage.

Hawley’s model has an elegant simplicity. He focused on the 2000 presidential election because (1) it was very close and, more important, (2) it coincided with a census. The two data sets, voting and demography, were captured almost simultaneously, lending more certainty to the findings.

Hawley didn’t just plop the election results on top of the census data and look for patterns. He instituted a number of controls in order to isolate the effects of different variables. For instance, home prices are driven to a large degree by a locality’s median income—so Hawley created a control for that. He also controlled for a county’s median age and for the percentage of African Americans and Hispanics. He controlled for rural living and college education and poverty. His goal in all of this was to isolate the effect of two factors: median home price and marriage rate.

That second variable, however, presented a problem. The census reports the median age at first marriage for states, but doesn’t break it down by county. So Hawley created a nifty little proxy. He took the “ever-married rate”—that is, the percentage who have ever been married, even if they are now separated, widowed, or divorced—for just one cohort: women aged 25 to 30. The higher the ever-married numbers for women in this group, the lower the average age of first marriage will be. And vice versa. It was an inspired bit of analysis.

And so, with his model complete, Hawley set his computer to crunching the numbers.

It turns out that, as Sailer proposed, the cost of housing does influence vote-choice. Hawley found that every $10,000 increase in median home value in a county resulted in a 0.3 percentage point decline of the vote for George W. Bush. Peering into the data, he noticed something else interesting: Even though median income and median home value are highly correlated, those two variables have opposite effects on voting. Higher house prices made people less likely to vote for Bush, but higher incomes made them more likely to vote for him.

The really eye-popping results, however, came when Hawley ran the model looking at ever-married 25- to 30-year-old women. It turned out that the marriage rate for these women was a greater influence on vote choice than any other variable Hawley measured. For every 1 percentage point rise in the ever-married rate, Bush gained 0.2 percentage points of the vote. In counties where the ever-married rate was far above the average (two standard deviations, for any stat-nerds keeping score at home), Bush’s share of the vote shot up a whopping 5.5 percentage points.

The average age at first marriage (which is what Hawley was indirectly measuring) was such a powerful driver of voting that it overwhelmed all sorts of other characteristics. For instance, it has long been assumed that the further people go up the education scale, the more likely they are to vote for Democrats. But in Hawley’s model the marriage rate almost completely eliminated this “education effect.” Which suggests that it is less the “education” that drives vote preference than the effect of higher education in pushing up the age of marriage for young adults. Once Hawley took marriage into account, education ceased to have any statistical significance in predicting votes.

Marriage even dramatically lessened the very robust effect of median home prices. But it didn’t eliminate it altogether, which suggests that housing price has an effect on voting independent of marriage rates.

Recent Blog Posts

The Weekly Standard Archives

Browse 19 Years of the Weekly Standard

Old covers