Sunday, November 11, 2012

Who Voted for Obama?

The world, or at least the political world, would be a better place if journalists knew something about statistics. For example, we now know that the following groups voted heavily for Obama:

  • Single people
  • People with incomes below $50,000
  • Younger people
  • Gays and lesbians
  • Nonwhites.
But what does this actually tell us? One would guess that the young are more likely to be single and to have  incomes below $50,000. One would also guess that gays are less likely to be married, and that nonwhites tend to have lower incomes. So which of these things actually matters, controlling for everything else?

There is a relatively simple way to answer this question-- multiple regression. Essentially that gives us an estimate of the separate effect of each of these characteristics. If you gave me exit-poll data on an Excel spreadsheet, I could do it in a couple of hours. If I didn't have the software, I could still get pretty far with cross-tabs.

As it is, though, we'll never know, at least not until the political-science journal articles start appearing. Pity. In the meantime we'll have to listen to a lot of blather from pundits who don't even know what they don't know.

No comments:

Post a Comment