Who else thinks that the Eletoral Vote Predictor website is hooey?

[PlanetBev]

It has always seemed to me to be no better than consulting Ms. Cleo.

[freetobegay]

1. I have never thought it to be reliable

Liable...

[wtmusic]

2. It's no more accurate than polls

but polls are worth looking at, sometimes

[Hog lover]

3. Which one?

aren't there several?

[grytpype]

4. It's bad.

Taking the most recent poll in each state is a pretty poor methodology. You're mixing a bunch of different polls with different polling methodologies on the same map.

[Firespirit]

5. If you mean electoral-vote.com

then YES.

I much prefer race2004.net and Chris Bowers at MyDD.

E-V.com is statistically weak. Although I'm no statistician, I have taken stat classes and have some knowledge of what is sound and what isn't, and what this site does is not statistically valid.

The guy who runs it is a Dem, but he seems to be bending over backwards to avoid appearing "biased".... to the point of including Strategic Vision polls. I question ANY poll explicitly marked as R or D. The fact is, all polls are NOT created equal, but this guy treats them as though they were. So... if SV does a sweep of battleground state polls, and most of them (as is likely) show * ahead, E-V includes them anyway and his map shows a very misleading result.

Also, he throws out older polls if there is overlap. Again, this could result in a S-V or Gallup red sweep, and has, which is misleading. There's averaging only if there's a tie for the dates of the polls.

There's no re-weighting for wacky internals such as party ID and no means of dealing with outliers. IMO a good polling site should disregard outliers in its projected/tracking maps, but keep them in the data set in case they turn out not to be outliers in the future; i.e., if later data backs up the seemingly weird result. No attempt is made to do this.

All that said, I don't have a problem with his final prediction map. But his daily maps are not sound.

[PlanetBev]

7. Yes, firespirit, it's electoral-vote.com

If not Miss Cleo, then my old Ouija Board that I used as a kid might be just as useful.

[Firespirit]

10. Miss Cleo... heh

How about a Wheel of Fortune... with vote totals from 0 to 538 randomly placed around the circle.

More succinctly than the above jargon: I don't have much use for that site's daily maps... in terms of accuracy, about the equivalent of cherry-picking one's polls... in terms of stability, about the equivalent of * in a random sample of three faceoffs with future President Kerry.

[Starlight]

6. Ms. Cleo would be much more accurate IMO n/t

[PlanetBev]

8. After this election, my hope is that all polls are proven dead wrong

Maybe four years from now we won't be on the roller coaster we've been put on this year by these stupid polls.

[elperromagico]

9. If Kerry wins and does a good job, we may not be.

He may coast to reelection. We can only hope; I can't take this every four years...

[high density]

11. It's, well, interesting

I like the idea but with so many wacko polls it seems a bit impossible to get a true feel for the election from that site. Too often it's crammed with GOP-slanted polls that give EVs to Bush with leads in the margin of error.

These polls are so fluid that I wonder what sort of idiots are out there are sitting around at home waiting to answer these polls. I wonder what effect caller ID has on these polls.

[HelpIsOnTheWayDamnit]

12. It's unreliable

It's jumps back and forth so much and the polls are heavily weighted to one side or the other.