Least Possible Election Night Surprise

[ian]

It seems as if Oklahoma is Republican central, but polls show that it is increasingly growing Democratic at a fast pace and are trending towards John Kerry.  Now, I will be the first to tell you that this probably won't happen, but there is a very good website that, based completely on polls, predicts that Oklahoma will go to Kerry in November.  The website is http://www.electoral-vote.com/
It is in the section labeled "Projected Final Map".  Please visit and enjoy.

[agcatter]

Oklahoma trending towards John Kerry?  Man alive.  What the hell are you smoking?  Bush probably won't win there by more than 25 points.

[A18]

Trends don't touch the center of the map.

[AuH2O]

That's based on a linear regression formula that is easily confused if you have 2 polls, the first of which is higher than the 2nd.

In other words, by that method, Bush could lose Texas. Uhhh...

ian gets troll points for this one.

[WalterMitty]

ive heard some stupid things on this board from the other party:

'kerry is going to win virginia'

'in 10 years, idaho will be voting just like oregon'

but to suggest that kerry will win oklahoma is....i dont even know what to compare it to.  

suggestion for the dems:  try winning iowa, wisconsin, pennsylvania, before you set your sights on oklahoma.  just some friendly advice from good ol' walter mitty.

[Defarge]

Kerry winning in TN but losing PA...

Zogby like insanity, just with more pro-Bush sentiment

[Gabu]

Well, it is the least possible.

[Gabu]

It seems as if Oklahoma is Republican central, but polls show that it is increasingly growing Democratic at a fast pace and are trending towards John Kerry.  Now, I will be the first to tell you that this probably won't happen, but there is a very good website that, based completely on polls, predicts that Oklahoma will go to Kerry in November.  The website is http://www.electoral-vote.com/
It is in the section labeled "Projected Final Map".  Please visit and enjoy.

I don't think you've read his explanation of the linear regression lines.  If a candidate has made substantial gains in a state, the line assumes that that candidate will continue to make gains at an equivalent rate, which we know in reality probably won't happen.  Oklahoma is closer than it was before, hence the positive slope of the linear regression line, but it probably won't get any closer than it already has.

That map has Tennessee as "Strong Kerry" but Pennsylvania as "Barely Bush".  The webmaster has openly stated that one should not pay much attention to it until mid-October.

[King]

Well, it is the least possible.

Actually this is the least possible:

[ATFFL]

Kerry winning in TN but losing PA...

Zogby like insanity, just with more pro-Bush sentiment

You hit the nail on the head, they use Zogby polls as part of their "analysis" screwing things up wickedly.

Bush is up slightly in PA in 2 recent polls from reputable firms (though Strategic Vision is earning its reputation this cycle.)  

I would still give the edge to Kerry in PA.

[Democratic Hawk]

I can't see Oklahoma going Democrat in November.

Dave

[Light Touch]

OK, UT, and DC are as definite as it gets.  I'm surprised Bush is even on the ballot in DC.  Why bother?

[© tweed]

OK, UT, and DC are as definite as it gets.  I'm surprised Bush is even on the ballot in DC.  Why bother?

If your name isn't on the ballot, it discourages people from giving money.  Bush gets more money from DC than Kerry does.

[??????????]

That website has Connecticut as weak Kerry!!! LOL LOL

[??????????]

Here are some good results they predict!

Oklahoma:

Kerry 50
Bush  49

Tennessee

Kerry 59
Bush 39

Indiana

Bush 53
Kerry 46

South Carolina

Bush 53
Kerry 45

Connecticut

Kerry 53
Bush 45


 This is their map!

[ian]

I didn't mean to tell you that I thought Kerry would take Oklahoma this election; I was just saying that according to this website, the possibility is there, and trends are happening at a rapid pace there.  I don't mean to actually give credit of substantiality to Survey USA or independent/college polls because I don't, but, in case they ARE correct, then OK is trending more and more towards John Kerry, and that is all I was getting at by this post.  Boy, some of you are so CRITICAL!

[Gabu]

Here are some good results they predict!

Oklahoma:

Kerry 50
Bush  49

Tennessee

Kerry 59
Bush 39

Indiana

Bush 53
Kerry 46

South Carolina

Bush 53
Kerry 45

Connecticut

Kerry 53
Bush 45


 This is their map!

Had you read what I wrote, you would have seen the explanation for those exceedingly strange predictions.  They're not what the webmaster honestly predicts; they're automated results based on linear regression lines fitted to polling graphs.

The webmaster said that you shouldn't pay much attention to it until mid-October.

[Gabu]

Well, it is the least possible.

Actually this is the least possible:

Good point.

[ATFFL]

Not anymore, they have a new poll qhich changes teh Bush trend.

Their analytical method is so flawed it is not even funny.  Off the top of my head I can think of the following problems:

They use any poll, including some wildly disreputable ones.

They assume all trends will continue.

They compare between two differnet polls that may, and often do, use different methodologies.

They have real problems if both candidates gain support between two polls.  WIll we suddenly get more than 100% vote totals?

That is off the top o fmy head.  If i really thought about it I could probably find some more.

[??????????]

I think this guy is using a President Forever results map!

[Volrath50]

Not anymore, they have a new poll qhich changes teh Bush trend.

Their analytical method is so flawed it is not even funny.  Off the top of my head I can think of the following problems:

They use any poll, including some wildly disreputable ones.

They assume all trends will continue.

They compare between two differnet polls that may, and often do, use different methodologies.

They have real problems if both candidates gain support between two polls.  WIll we suddenly get more than 100% vote totals?

That is off the top o fmy head.  If i really thought about it I could probably find some more.

I don't really see a problem with using any poll result. There are many such websites, this guy uses any poll, others have strange formulas, and such, if you don't like his site, ignore it.

And the projections are automatically generated. It isn't necissarily what the guy thinks. Okalahoma is Kerry because it went from extremly Bush, to very Bush.

[ATFFL]

I don't really see a problem with using any poll result. There are many such websites, this guy uses any poll, others have strange formulas, and such, if you don't like his site, ignore it.

And the projections are automatically generated. It isn't necissarily what the guy thinks. Okalahoma is Kerry because it went from extremly Bush, to very Bush.

Let me explain why you can't use any old poll for the type of analysis he is attempting.  Lets take 2 companies, one of which is Opinion Dynamics and the other is POS.  OD does not push leaners at all, POS does.  SO they both poll a state, POS first then OD.  The OD poll has many more  undecideds, so the support for each candidate is lower.

This guy is going to plug the numebrs into his formula and decide that each candidate is losing support, when that is not the case.

And can you compare Zogby's Internet poll with any poll that has a reasonable methodology?

[Volrath50]

I don't really see a problem with using any poll result. There are many such websites, this guy uses any poll, others have strange formulas, and such, if you don't like his site, ignore it.

And the projections are automatically generated. It isn't necissarily what the guy thinks. Okalahoma is Kerry because it went from extremly Bush, to very Bush.

Let me explain why you can't use any old poll for the type of analysis he is attempting.  Lets take 2 companies, one of which is Opinion Dynamics and the other is POS.  OD does not push leaners at all, POS does.  SO they both poll a state, POS first then OD.  The OD poll has many more  undecideds, so the support for each candidate is lower.

This guy is going to plug the numebrs into his formula and decide that each candidate is losing support, when that is not the case.

And can you compare Zogby's Internet poll with any poll that has a reasonable methodology?

What I was saying is that he is not attempting any analysis. All he does is get the numbers, put them up, and let whatever scripts he has do the rest. He has no real input in this, its all done by polls and his computer. And while Zogby's polls are crap, they are, at least nominally polls. If I were him, I wouldn't include them, but he has decided to go with any poll. This should defeat criticism, as he is essentially a common carrier of polls, like an ISP or something. You can't really criticise an ISP for carrying bad stuff, as they are just the messenger, like this guy.

Sorry for rambling so horribly, but it is 1:52 AM, and I can't sleep.