A more rational prediction from electoral-vote.com....

[stry_cat]

http://www.electoral-vote.com/pred/

This seems like more reasonable prediction criteria that their pervious method.  Any thoughts?

[J. J.]

http://www.electoral-vote.com/pred/

This seems like more reasonable prediction criteria that their pervious method.  Any thoughts?

Not even close.  It is exceptionally unlikely that any cadidate will win the big trifecta of  FL, OH, PA.  It is also unlikely for any candidate to win the little trifecta of WI, IA, and MN.

I would suggest that these be referred to as "FOP" and "WIM," respectively.

[Silent Hunter]

He's using Zogby for some of this and his predictions are just plain odd. Bush only +2 in TN? +3 in NC? Kerry actually winning FL and practically all of the swingers?

Is the 2:1 for the challenger prediction going to hold? I don't think so. The 2:1 calculation only goes back to 1976 and 2004 is different to any election since then.

This guy is a partisan Democrat. It's getting to his analysis, IMO.

If he's right, this is a 2000 in reverse and the Kerry first term will be a nightmare.

[MODU]

That is the problem with thier website.  They don't look at the source of the poll figures, and assume they are all accurate. 

Terrible projection.

[Giant Saguaro]

Well I don't know which is worse, missing UT and MD or missing everything BUT UT and MD.

No, seriously, a lot wrong with this map from where I am. This would seem to reflect a Kerry +1 to +4 national instead of a Bush +1 to +4 or +5, which is what we've been getting. AZ, NM, IA, WI, AR, TN, FL, NC, VA, OH, and that's first glance. I suspect not screening for bad polls is the result. MN is likely not strong Kerry either.

[J-Mann]

He certainly picks and chooses his polls based on which ones are showing Kerry doing better.  There was a time in September where I'm sure he wasn't too happy.

[Blue Rectangle]

No, it is still a flawed concept that can only produce poor results.  The assumption that states have trends that are accurately measured by polls and which therefore can be extrapolated is baseless.  A good example is AZ, which is shown as "weak Bush".  I have yet to see evidence that AZ has changed much from 2000, or that Kerry could be gaining ground/Bush losing ground.  The polls, on average, have shown a strong Bush lead, but the random nature of the poll results and timing is enough for EV.com to claim a trend.  This site substitutes blind faith in extrapolation for reason.

[dougrhess]

Still doesn't make sense to me.

Nader got 6% in ME in 2000? Wow. What's he polling there now?

[Nym90]

He certainly picks and chooses his polls based on which ones are showing Kerry doing better.  There was a time in September where I'm sure he wasn't too happy.

No, he doesn't. His methodology is clearly explained on his site. Can you provide proof of a time when he broke the rules?

[Nym90]

No, it is still a flawed concept that can only produce poor results.  The assumption that states have trends that are accurately measured by polls and which therefore can be extrapolated is baseless.  A good example is AZ, which is shown as "weak Bush".  I have yet to see evidence that AZ has changed much from 2000, or that Kerry could be gaining ground/Bush losing ground.  The polls, on average, have shown a strong Bush lead, but the random nature of the poll results and timing is enough for EV.com to claim a trend.  This site substitutes blind faith in extrapolation for reason.

Agreed, the projected map based on linear regressions is junk. Too few data points for many states, and trends tend to be national moreso than local.