Saturday, January 26, 2008

SC results exceed polls by a huge margin

Several weeks ago in the New Hampshire Democratic primary, polls projected Barack Obama winning over Hillary Clinton by as much as 11 points, only to lose to Clinton by 2 points.

Compare that to today's South Carolina primary, where Obama had a 15 point lead in the polls but won by a massive 28-point margin.

This begs the question: why is the latest and most scientific polling wrong so often all of a sudden? Has the demographic changed more rapidly than the polling methods? Are pollsters relying on less accurate systems of polling? Has the importance of polls been over exaggerated by the media? I dunno, but it seems worthwhile to look into it.

Over on CNN's website they have some demographic breakdowns of the voting. Perhaps the most intriguing fact is that exit polls showed that 61% of the Democratic voters were women. Typically, that would be a big help to Clinton. However, for Obama to have beaten Clinton by more than 20 point, there must have been a significant number who voted for Obama. I wonder if these percentages will carry over to the huge Super Tuesday primaries?

John Edwards finished with less than 20% of the vote, and if he couldn't get a significant percentage of the vote in South Carolina then it's obvious that his campaign is effectively over. I was hoping for a miracle, but sadly, it's not going to happen now.

Seeing as we're approaching the Connecticut Primary in 10 days, I'm going to endorse Barack Obama for President. I'll post something more substantive in the next several days, but for now I'm going to put up a link to the Obama Campaign website. Please visit it and support Senator Obama.

1 comment:

Dave Mooney said...

The polling for Clinton and Edwards was pretty much right on target. It seems possible that the 15% or so that was undecided broke almost universally for Obama. Another factor is the unprecedented turnout, which seems to have heavily favored Obama. The art of polling is the likely voter model and no model being used publicly would have predicted this turnout.