Sunday, August 17, 2008

NC and the Bradley Effect

At about the eight minute mark of this week's Headline Saturday David Crabtree said he recently spoke to a pollster who said that like in previous cycles with major black candidates on the ballot Barack Obama would end up under performing the polls because people, concerned about seeming racist, would lie to pollsters about their true voting intent.

I think we've already seen evidence this year that will not be the case in North Carolina. The average of surveys from various groups that polled in the Democratic primary for the state showed Obama defeating Clinton by an average of 7.3 points. Of course he ended up winning by 14 points, actually out performing the polling in the state by 6.7%.

The reason Obama did so much better than the numbers anticipated is that pollsters underestimated both the composition of black and young voters within the electorate, as well as the extent to which they would support Obama.

I don't see any reason that would be different in the general election. Most polling, including ours, is taking a conservative approach so far to forecasting turnout from the groups that support Obama at the highest levels.

Beyond that, I also think that voters in a Democratic primary are more likely to feel bad and lie to a pollster about their intent to vote for the black candidate than voters in the general election.

If anything I believe the polls are more likely to once again undersell Obama's performance in the state than they are to produce a Bradley or Wilder effect.

No comments:

Web Statistics