Posted on 10/14/2008 5:09:38 AM PDT by tomymind
plenty of white-guilty in NoVA suburbs of DC, where i live, and yet still i see signs of hope for McCain.
For ex: in 1996, all the polls---all---were off the final in Clinton's direction. Some were close (1%), but the point is, if it's just random error (as you see in some of the 2004 polls), the other guy will come out ahead once in a while (as Kerry does in some of these 2004 polls). One statistician calculated that the likelihood of that being an accident was 240,000:1. None of the pollsters caught the 1994 House revolution.
But most of them were close in 2004, and almost all the polls were dead on in 2006. However, they were NOT catching the trends, which was toward the GOP. This was a very close election, in which I think the number was that 35,000 votes nationally gave the Dems the House and Senate. Michael Barone, who I think is the absolute best at analysis, said if the campaign had gone on one more week, the Republicans would have held both houses, and the polls would have been hugely wrong.
But the polls weren't always accurate: Ken Blackwell on RCP was between -6 and -14, but lost by 25!. "Bradley effect," anyone?
But the point is, when voter breakdown in 2004 was 39%/39% Dem Republican, is it really sensible to think that Dems suddenly outnumber Republicans by 10%? Or 14% as one VA poll recently does? Perhaps 3% since 2004. Not 5-10. And that's a huge error in polling. So, yes, if the polls are poorly constructed, they may well be flawed.
Now, do you have any evidence in 2004 that the polls were using anything OTHER than reasonable voter ID breakdowns? I don't. So obviously something has changed since 2006.
It's entirely possible all these polls are right (although an internal poll I happen to know about here in OH says they are complete hogwash) and that McCain will be crushed. Guess we'll see. My point is that there no reason to accept these voter ID breakdowns as "givens," which in turn shapes ALL the polls.
“Now, do you have any evidence in 2004 that the polls were using anything OTHER than reasonable voter ID breakdowns?”
As far as party ID goes, Ras uses dynamic weighting, Gallup doesn’t weight at all and Zogby weights according to exit polls from the previous election (the Freeper preferred method.) Right now, using three widely different party weighting methodologies, those three polls have Obama at +5, +6 and +6. No statistical difference between the outcomes of the three methods.
I came across an article yesterday from October 2006 where Republicans were complaining about the same thing. The polls showing the Pubbies losing had to be wrong, since they were oversampling Democrats. Those who will not learn from history...
Party ID was +15 for the Dems in 1976, when Carter won, and +15 for the Dems again in 1980, when Reagan won by 9 points. Party ID as a factor in who wins is a highly overrated metric.
The fact is, party ID numbers are based on exit polling and post-election interviewing, two of the most flawed polling methods. Criticising polls today because they will not weight based on flawed polls of yesterday is nonsensical.
“It does not need a graduate level training in statistics to prove this very simple fact.”
Nope, all one really needs is a willingness to read pollsters’ methodology statements, something that most non- “full of themselves” Freepers seem loathe to do.
Gallup doesn’t weight for party ID. And their LV poll results are right in line with the polls that do.
Party ID numbers are based on exit polling, which polling experts will tell you is 50-80% more flawed than scientific, pre-election polling. The average state poll has a margin of error of 4 points, so the average exit poll of a state has an MoE of 6 to 7.2 points, meaning the swing between exit polls and reality can be as high as 14.4 points. How many states are won by more than 14 points? Darn few. And therein lies the problem with relying on exit polls to determine party weighting, which is why most pollsters won’t do it, and NONE of the good ones do.
I understand you were stung by 2006. I was too. You seem to miss the point that trends mean a lot, and that everyone from Karl Rove to Michael Barone agree that the 2006 election came at exactly the wrong time, and that another month would have produced a massively different outcome, given that so many races were extremely close.
But you did not answer my question: what evidence do you have that in 2004 these relatively accurate final polls used such massively distorted voter ID and still ended up correct?
I think the debates were more significant back then, too. I remember (vaguely) Nixon skillfully redirecting every question he was asked into something he wanted to talk about.
The GOP has stopped teaching that level of debating skill, counting more on “Get out the Vote” than anything else.
Which brings the conversation around to who in the RNC and GOP is deciding on all this new strategy? Obviously, “Get Out the Vote” worked to get Bush into the White House, but the lack of attention to other details killed it for Congressional and Senatorial candidates.
Have you once heard McCain or Palin say send me to the White House and send my support to Congress? Nope.
You cannot compare an incumbent race to a race with both parties on equal footing.
I thought 2000 was pretty close as was 2004. Just 500 and some votes kept Gore from being president in 2000.
The electoral college in 1968 was not close. 301 for Nixon. 191 for Humphreys and 46 for George Wallace.
In 2000 it was 271 Bush and 266 Gore.. a much closer election and about as close as one can get with a bit over 500 voters in Florida deciding the election.
In 2004 it was 286 for Bush and 251 for Kerry and much closer than 1968.
“But you did not answer my question: what evidence do you have that in 2004 these relatively accurate final polls used such massively distorted voter ID and still ended up correct?”
I answered your question. You ASSUME that party ID is massively distorted because you BELIEVE the exit polling. That is your opinion; it is not fact. I believe your assumption is flawed. We have no way of knowing with any reasonable certainty what the party ID was for any election. All we can possibly know for sure is how people voted.
“You seem to miss the point that trends mean a lot...”
I’m not missing anything. We have three weeks left, and anything could happen. If I’ve posted that once, I’ve posted it 50 times in the last month. Just because I largely believe the polls now doesn’t mean that I think they can’t change. They can and most likely will.
But I’m also not making the massive assumptions that you apparently are—that ALL the polls are flawed, that even the polls that don’t weight by party ID are still somehow oversampling Dems, that McCain will make a huge move in the final few days that won’t be picked up in the polling.
Those are all assumptions. Give me facts.
LOL, you have it all wrong. My post was in response to another and I was questioning HIS information on these County polls being validated by others.
Really, you ought lose the “ready, fire, aim” attitude, I’m on your side.
Assuming the poll is accurate, O up 1, with 6 undecided means Fauxbama loses. Undecides this late in the game will break for the known element over the unknown 2 to 1. If not 3 to 1.
Why not?
Besides, you miss my point. Time and time again the tinfoil hat Freepers assert that ALL the polls are ALWAYS flawed and ALWAYS show the Dem winning, until the Pubbie miraculously wins. That is not true, and has never been true.
Some polls are better than others, but in recent Presidential elections we have had enough data points that by the time the election has rolled around, the majority have reached a consensus, and it has been the correct one.
And, with very few exceptions, that has been true since the advent of modern polling methods, incumbent or no incumbent.
http://www.gallup.com/poll/110548/Gallup-Presidential-Election-TrialHeat-Trends-19362004.aspx#1
“... the polls had Kerry WINNING up to and including ELECTION DAY...”
No, they didn’t. Of the 85 polls in the RCP average in the last two months before the election, only 7 had Kerry up, and only one pollster, Marist, called the election for Kerry. The Dems were in deep denial about the polls in 2004, just like Freepers in 2006. Both time the ideologues were wrong and the polls were right.
http://www.realclearpolitics.com/bush_vs_kerry.html
Obama does not have inexhaustible funds, in fact his war chest is being consumed at a good clip. McCain's own camp has revealed that the pre-season is over and 400 million is getting ready to turned on Obama. McCain didn't adopt the Dole strategy of blowing his whole wad right after the convention. Now whether you think that's prudent or not is not important to your assumption here. McCain might just be correct that voters might be weary of all these Obama-mania and McCains contrary message at the 11th hour will shake enough folks out of their stupor.
As far as the MSM giving exclusive discounts to the Obama camp well that's illegal under current campaign finance laws.
Why don’t you believe exit polling? That seems sillier than believing “entrance” polling, ie. a pollster calls and says are you a “Dem, Republican,” whatever. Why wouldn’t a person be MORE truthful AFTER voting than before?
Thanx
you don’t need to show up to vote.
Nursing homes in FL are NOTORIOUS for voting absentee. The hospital staff will even fill out the form for them.
I’m sorry but personally I really don’t like the accusation that I am trying to demoralize the conservative base, posting some polls. Until I’m told otherwise, I’m a proud member of this base and I don’t need a tutor.
We’ll see.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.