Ten Miles Square


November 26, 2012 11:30 AM Red State Blue State, or, States and Counties are not Persons

By Andrew Gelman

Tyler Cowen points to this news article by Lauren Sandler:

Stunningly, the postponement of marriage and parenting — the factors that shrink the birth rate — is the very best predictor of a person’s politics in the United States, over even income and education levels, a Belgian demographer named Ron Lesthaeghe [and coauthor Lisa Neidert] has discovered. Larger family size in America correlates to early marriage and childbirth, lower women’s employment, and opposition to gay rights — all social factors that lead voters to see red.

All the analysis in the linked paper is at the state and county level. That’s fine but this is not going to tell you what is a “predictor of a person’s politics.” Cowen labels his post “Sentences to ponder,” and what I want to ponder is that people are so quick to jump from aggregate to individual patterns.

And, yes, I know that aggregate patterns are related to individual patterns but they’re not the same. In particular, from the evidence we’ve seen (and which we presented in our book), social issues are important for voters at the high end of the income scale, not the low end.

David Brooks catches this—-in his op-ed from 2004 that Cowen links to, Brooks explicitly labels the conservative “natalists” as being high income (“when people get money, one of the first things they do is use it to try to protect their children from bad influences… . It costs a middle-class family upward of $200,000 to raise a child… .”). Brooks is getting it right that it is higher income voters who are central to the culture war.

In summary, I’m not trying to slam or “debunk” the Lesthaeghe and Neidert article. I just think it should be understood as an aggregate, not individual, pattern, and interpreted in light of what we already know.

[Cross-posted at The Monkey Cage]

Andrew Gelman is a professor of statistics and political science and director of the Applied Statistics Center at Columbia University.


  • Equal Opportunity Cynic on November 27, 2012 2:11 AM:

    I don't understand. Obviously as a stats professor you're well aware that aggregate data is often used to make inferences about the individuals involved. Why would this bother you here any more than any other quant analysis?

    Are you suggesting that the relationship between a county's family size and voting tendency is reflective of something besides a relationship in the underlying individuals or families? If so, could you please spell out more clearly what you're getting at?

  • Helen Bedd on November 27, 2012 6:35 AM:

    I'm dubious about most of this.

    Take Mormons and members of the "quiverfull" movement. They have larger families than the average American but are also more affluent than average and they are overwhelmingly conservative.

    Then you have Hispanics, who are [on average] less affluent, have the first child earlier, have larger families but vote Blue.

    In the "Red Sex/Blue Sex" arena among whites, the earlier the first birth the more likely for the mother to be poor and still vote red. But of lot of this is due to it being a southern phenomenon. [Lack of comprehensive sex education and aversion to abortion being the main reasons.] And those reason are tied to Fundamentalist Christians [which explains the anti-gay part.]

    For me, the very best predictor of a personís politics in the United States is still church attendance. Weekly church goers vote red 4 to 1, those who never attend vote blue 4 to 1.

  • Helen Bedd on November 27, 2012 6:42 AM:


    To me the entire red/blue war [and not just over social issues] is waged between southern white college grads and northern white college grads. Those are the groups most likely to be involved [and contribute to] political campaigns and vote.