He’s gotten Slashdotted.
(E.g., he got mentioned on the Slashdot site, which sends six trillion geeks to his servers, sending it to its knees 
)
Yeah, I mean why would - wait, what’s this?
That asshole! I’ve got an OS programming project due on Friday because of his minix! I’ll be sure to disregard his numbers.
That’s pretty neat, actually. I think it’s pretty cool that a very experienced and well-known person is behind the site.
And he uses the same host as I! The coincidences are eerie.
10% of 1/29 of 180,000 is a whopping 600-odd voters per state on average. That hardly balances the cell-phone-only contingent.
Which, of course, contray to Mr. Zogby’s statement –
I never cared for electoral-vote.com. I never doubted the Votemaster’s intentions; I just found his methodology to be way off kilter. Here it is, Election Eve, and his site is showing Kerry winning—which is what I hope will happen—but I still put no faith in his numbers, which are an awkward patchwork of different polls. I don’t think it’s worth attacking the site, but I do think there are much better places to go for your information. If you want unscientific numbers that are based on gut instinct, past elections and some facts, might I suggest http://www.geocities.com/nuclearfurniture/election2004.html ? Yeah, yeah… gratuitous plug for my own site, but hell, we were on the subject…
Anyway, I’m serious about electoral-vote.com: it’s not good information, and it fluctuates too wildly to be believed. It’s got good graphics, but that’s about it. Reality, I’m afraid, is usually much more boring than that.
The daily map is not an awkward patchwork- it illustrates the results of only the most recent state polls.
He’s never represented the site as anything other than a momentary window on the numbers.
The only problem with ev.com is that he shows polls within the margin of error as being “barely ______.” They should be shown as toss-ups. Any electoral vote predictor that shows a state as going to _____ however thin the margin is just giving the people what they want, perhaps, but it’s misleading to do so. EV.com DOES indicate that the tossup states are should be taken with a huge canister of Mortons.
The NYT electoral projection is: Kerry will get a minimum of 242 electoral votes. Bush will get a minimum of 227. The rest are tossups. I like the NYT because they have never shown a tossup or something within the margin of error as anything but a tossup.
The bad news for Bush is that the tossups have been turning into Kerry states lately.
Tannenbaum was one of my professors at Brooklyn College. Way to go, Professor.
What I think electoral-vote.com should do is have a page with an alternate map where toss-up states are shown as being for neither candidate. Then the site user can examine all the maps, and decide the significance of them.
My concern would be that he’s the only predictor out there coming up with such a massive Kerry win, and he’s a staunch Kerry supporter. I mean, I hope Kerry wins too, but it suggests bias in his method.
I have been logging on to his website daily for a long time, and for the most part, it has not been a happy experience for a rabid anti-Bush person like myself. To the contrary, he has been very honest about being pro-Kerry, but still putting up the bad news - even as he himself often questions the validity of the polls no matter who is up.
I of course want to believe his prediction simply because I want it to happen. However, if anybody seems to have poured over every scrap of polling information, Professor T. seems to be the one I would trust most to interpret the results.
I still think the polls are way off - and just think of all those people with caller ID who saw some unknown person phoning in and didn’t even bother to pick up the phone - another several million non-polled people.
That depends. You can do a ‘meta-analysis’ of polls that use similar methodology to reduce the margin of error. For example, if you have ten polls, all of which show Kerry leading, but all of which are within the margin of error, then it’s likely that Kerry is leading. Especially if they are just barely within the margin of error.
This meta-analysis only really holds for polls that use the same methodology. Since most polls have differences in how they sample and the weightings they use, it’s not an exact science. But it’s still valuable.
I have to wonder about these electoral maps. How can anyone right now objectively say that Ohio is going for Kerry? Both the above mentioned sites say that Kerry will take Ohio, and yet here are today’s polling numbers:
Zogby 11/01: Bush +6
FOX News 10/31: Bush +3
SurveyUSA 10/31: Bush +2
CNN/USAToday/Gallup 10/31: Kerry +4
U of C Ohio Poll 10/31: Bush +.9
Rasmussen 10/31 - Bush +4
Of the six polls taken in the last two days, Kerry is only showing as leading in one. The latest poll of the bunch has Bush leading by 6. The Bush +6 is over 34 points higher than the margin of error. And yet, the two electoral vote count sites linked on this page show a Kerry victory? How come?
I meant ‘over 3 points higher than the margin of error.’
The results of a meta-analysis depend on factors other than the difference.
“Meta-analysis”?
Hmm…But Zogby had Kerry up by 2% on Oct. 31. Obviously, these polls are less than ideal. I’d say Ohio is too close to call.
“analysis of the analysis,” one presumes.
That’s the process you use to combine polls/studies/whatever.
Electoral votes-wise: Zogby is showing Florida tied and Kerry winning 264 to 247. Popular vote-wise: Zogby is showing Bush 48% to 47%.
So Electoral-Votes is focusing on Zogby? Is there any evidence that he’s cherry-picking his polls to show the results he wants?
What makes the Zogby poll better than anyone else’s? As far as I’m concerned, Zogby is one of the most questionable polls. Yes, I know he was accurate in 2000, but that may be a fluke. His methodology is quite different than others. For one thing, he’s assigning most of the undecideds to Kerry - something that I’m not sure is valid. We have a small sample set of previous elections which show that trend.
And of course, his methodology seems to drive a lot of variance, since two polls of his taken one day apart differed by 15 points.