Sunday, October 31, 2004 ... Deutsch/Español/Related posts from blogosphere

Bush and Kerry: estimates

I am not really allowed to influence politics in the USA, and in some sense it is a great opportunity to keep this blog fair and balanced - one of the very few fair and balanced blogs around. No official endorsement will be revealed in this article.

According to my private index which is a weighted combination of available indices, the support for Bush and Kerry seems to be, three days before the elections:

  • 46% for Bush vs. 46% for Kerry overall; the race is dead even; Ralph Nader at 2% (less than in 2000)

The following categories are ordered from the most pro-Bush ones to the most pro-Kerry ones:

  • 92% faithful Republicans for Bush
  • 80% of born-again Christians for Bush
  • 75% of those who changed their mind after the new bin Laden tape for Bush
  • 72% conservatives for Bush
  • 71% of the "Iraq war is the top issue" voters for Bush vs. 25% for Kerry
  • 70% of investors for Bush
  • 54% voters who regularly attend religious services for Bush
  • 52% of married men for Bush
  • 50% married women for Bush
  • 49% men for Bush
  • 44% self-identified independents for Bush vs. 41% for Kerry
  • 49% newly registered voters for Kerry vs. 42% for Bush
  • 53% single women for Kerry
  • 57% of Northeast for Kerry
  • 70% of Hispanics for Kerry
  • 80% liberals for Kerry
  • 84% of black voters for Kerry (was: 90% for Gore in 2000)
  • 86% faithful Democrats for Kerry
  • 90% of Ivy League faculty for Kerry
  • 103% of anti-Bush bloggers for Kerry

The errors in these numbers are roughly 3%, which is the percentage of undecided voters. The error of the least reliable categories are biased and they can indeed reach 3%.

Add to del.icio.us Digg this Add to reddit

snail feedback (2) :


reader Anonymous said...

Polling is done via landline. Almost 2/3 of the calls are refused. Unlike as with neutrinos, most of which refuse to react with our detectors, it is not possible to say whether the political opinions of the sample that refused to be polled are statistically the same as those who did take the poll. Many people no longer have a landline, relying entirely on wireless, but no poll phone calls come on that.

So, instead of saying Kerry 46, Bush 46, the polls should read:
Kerry 16, Bush 16, undecided 2, refused to answer 66
and that will correctly reflect the uncertainty in this election.
Doing anything else would be string-theory-esque :) :)
-Arun


reader CarlBrannen said...

In getting ready to return to grad school, I came upon this link with regard to women and the physics GRE: The Consequences of Physics GRE Requirements for Graduate School... I did a search for austin+women+physics on the blog and didn't get a hit so I'm assuming you haven't read this link.