Friday, October 10, 2008

Bradley Effect Debunked


Here's a good analysis of the Bradley Effect: FiveThirtyEight.com: Electoral Projections Done Right: The ...


The Bradley effect is named after LA Mayor Tom Bradley who led the polls in the guv race in California in 1982, but lost. Bradley was black and his opponent was white. The lesson was that people lied to pollsters about whether they would vote for the black guy. The article linked to above makes the case that the Bradley effect used to exist, but no longer does. Obama actually outperformed the pre-primary polling.

Two observations: 2008 is not 1982 and Barack Obama is not Tom Bradley.

4 comments:

  1. I'd read Andrew Hacker's article on the Bradley Effect in the September 15th New York Review of Books before dismissing it as 'a way to fill space by journalists'. Hacker is scholarly and not given to any PC nonsense.

    ReplyDelete
  2. I have read it and I think he grossly overstates any possible effect when he says to deduct 7%.

    ReplyDelete
  3. http://www.cbsnews.com/blogs/2008/10/11/politics/fromtheroad/entry4515246.shtml

    ReplyDelete
  4. First, as small request: when you post a link, give a comment about why you think it is worth looking at.

    The link goes to a story about a racist incident at a Palin rally. The original post is about the Bradley effect, which posits that people lie to pollsters and that results in polls overstating support for black candidates.

    Questioning the Bradley effect or its size is not to question that there are people who will not vote for Obama because he is black. It's not PC nonsense. Whether the Bradely effect exists is a factual question and to me the more perusausive evidence is that it is at least much diminished.

    ReplyDelete