|
Edited on Sun Nov-05-06 01:44 PM by longship
I disagree with you on the absolute accuracy in these late polls. The pollsters are using historical models and party distributions to scale their data. But this year especially those models are likely to be incorrect. We see this in the flipping back and forth of the polls for a particular race. Allen's ahead; Webb is ahead; Allen is ahead again; They're tied now; etc. But the undecideds remain at about the same percentage. It's inconceivable to me that a high number of people are that fickle, changing their minds back and forth on a daily basis. Yet that's exactly what all these polls are indicating. It cannot be correct. The only conclusion is that the measurement of public opinion is an inaccurate science. We all know that already, but when it comes to the news media reporting it, such things are never spoken.
So, what do these polls tell us? Do they have any value at all? You bet!
Take the accumulated polls as a set and draw conclusions from them, but only where there are consistent trends and measurements. I don't see a lot of this in these contested Senate races. That's why I am tending to draw my conclusions on these races from other data. This is a far more inexact methodology but in absense of concrete, repeatable polls that's what one is left with.
I think the following conclusions can be drawn from what we're seeing:
1. Every single one of the large scale indicators are so biased towards the Democrats that one must color ones predictions in that direction. The accumulated data comes from the large issue polls, which *do* seem to have some stability in both numbers and trends, so they can be used. Even though individual race polls may not be accurate, the number of competitive races indicated by those polls is important--almost all of which are Republican-held seats. The number of these competitive races has been increasing throughout the election season, another important trend--again, always in favor of Democrats. Finally, although not always scientifically measurable, there is a huge accumulation of anecdotal evidence. All these large scale indicators are very important this year. They point towards something very important happening.
2. The candidate polls are counting "likely voters". But this is an measurement whose outcome is extremely sensitive--a small inaccuracy is likely to have a disproportional effect on the outcome, especially if Republicans are overstated and Democrats and independents are understated. In a year like this year where a disproportional number may be voting for Democratic candidates, the swing of miscounting this could be significant.
3. There seem to be many undecideds this year. We're a couple of days from the election and too many polls are still in the low-to-mid forties. I haven't checked, but I suspect that there are more than usual. I'm no expert on this but I'll bet that they will significantly break for the Democrats, just as the Independents will this year. Just maybe there are many people who don't want to tell pollsters how they will be voting.
When I look at all this stuff I *must* discount the individual candidate polls and even the accumulated polls in a specific race where there is no consensus. No races are worse in this than the three of four critical Senate races. That's why I'm not trusting them.
In short, I think Webb is ahead in VA, McCaskill is ahead in MO, and Tester is ahead in MT. I do not know at all what is going on in TN, but Dems could be ahead there too and the polls might not see it. Everything is skewed in the Dem's direction this year. That's not happened in a long, long time. The pollster's models are going to be inaccurate. "Likely voter" is not likely to be accurate, either. IMHO.
|