« Newsweek National Poll Puts Kerry in the Lead | Main | New Jersey, Revisited »
Sunday, October 03, 2004
Undecideds Break for the Challenger
Posted by DavidNYCWe all know it's true, but Guy Molyneux, a Democratic pollster, delves into the details of what exactly this means over at the Prospect. I strongly advise everyone to read this article. Molyneux encourages political reporters - and by extension, I imagine, bloggers as well - to focus not on the spread in horserace matchups but rather the "50 percent line."
More specifically, in the four incumbent re-election campaigns in the last quarter century, the president came in, on average, half a point below his final poll results. Challengers, however, averaged a four-point gain. So, therefore, if George Bush is at 49%, and if John Kerry is within reasonable striking distance come Nov. 1, we're going to have a photo finish. If Bush is below 49%, he's almost certainly melba toast.
Posted at 02:46 AM in General | Technorati
Comments
Yes. I've noticed the similar thing.
Look at the last election, which is an open one though. Most of the polls on Nov.6,2000, which is one day before the election, showed that Bush lead by 2-3%. But as we all know that Gore got 0.5% more than Bush did. That is a 2.5% error.
This year I believe this number will be biger. Perhaps 3-3.5%,because Democrats are much more eager to come out and vote.
This is the year 2000 data(from pollingreport.com):
.........................BUSH GORE
CBS........................44 45
GALLUP...................47 45
VOTER.COM.............46 41
ABC.......................48 45
HOTLINE.................45 42
ICR........................46 44
NBC.......................47 44
PEW......................45 43
WASHINGTON POST 48 45
Posted by: brightwater at October 3, 2004 11:11 AM | Permalink | Edit Comment | Delete Comment
Yes, the polls in 2000 undercounted Dems by 2% to 3%. Kerry is going to pick up support on election day!
Posted by: Rock_nj at October 3, 2004 11:49 AM | Permalink | Edit Comment | Delete Comment