There are a rash of new state polls which make it look like a blow for O-bomba. The truth is a lot of polls were moving towards Obama last week, so there was movement in the electorate. But as I noticed two national polls showed a reverse trend, and some of these polls are clearly producing garbage.

I am truly surprised at how bad the Quinnipiac polls are. Does anyone really believe Obamba is up 8 points in Ohio or 15 points in Pennsylvania??? Without the Quinnipiac outlier in their mix of polls the RCP average for Ohio would be a perfect tie at 0%, not the +2% they have now for Obama. No way +8% for Obama.

The CNN/Time poll for Virginia is no better with a +9 for Obama. Obama could win VA, but if he does it will be by 2-3% points, not 9%.

These state numbers look really ridiculous when combined with the last ABC News/Washington Post and Gallup Daily Tracking Polls, each of which shows Obama leading by 4% nationally – and dropping. With Obama only ahead nationally by 4 there is no way he could be up 9% in conservative VA. Maybe in deep blue NY or CA, but not in VA. Or Ohio, or Florida.

How could all the state polls be seeing the same phenomena? Well, the scary answer for the GOP is there is a mass movement to Obama – which I doubt is happening in the magnitude of some of these polls given the national picture. One blogger thinks the pollsters are ‘cooking’ the voter models. Maybe on one poll, but not on all of them, and the trend is across AP, Quinnipiac, CNN/Time and others.

So, if it is not a huge wave towards Obama what could it be? Well, it could be (and I am speculating here) a small wave that has exposed the problem with the voter models used in polls AND something about voter engagement in the process. I have said this many times, a small error in party affiliation will cause a huge error in the result, amplifying one side over the other inaccurately.  It’s like trying to aim at something far away, where a small error in angle ends up being a large error in targeting once the projectile has propagated along its path.

The voter affiliation IS driving all these poll models, it has to. But what if one side has gone underground, gone quiet? Obama’s supporters are intense and energized and loud. And also too many times thuggish and rude and ‘in the face’. McCain-Palin supporters are quietly determined, but not less energized. They tend to have the optimistic steely eye whereas the liberals are the wildly angry type.

Polls are voluntary, they suffer from challenges all other scientific sampling and statistical processes do not have to deal with: truthfulness and the desire of the subjects to be sampled. I think they have enough mechanisms in to deal with truthfulness and it is probably a component of their margin of error. But participation is another deal all together.

It could be Obama supporters are just more willing to answer the phone, get in the pollsters’ face, and rage against the machine. They are high volume, but may not be a large crowd. Now, with the financial mess (which I still think Pelosi and the Dems orchestrated to lose and keep it the good news for Obama coming) there could be real anger building out there in the country. But I am not so sure that anger has picked a champion yet. It seems America is saying a pox on all your houses.

If we think of two equal groups where one is loud and one is quiet, if we tried to measure their group sizes based on volume we would get a biased result towards the loud group. I think that we are seeing something similar here. I was not sure until I saw those state polls jump out like they did, and this goes back to the voter model I mentioned.

If these models are trying to translate the intensity of the respondents – their volume – to gauge their group size, then that could be why the bottom line is off. There may be more of the left answering the and less of the right answering the phones. The left wants to yell out their anger, the right is quietly biding its time and watching things work out. Like I said, I was not sure this was more than speculation until I saw dropping national polls for Obama simultaneously with huge gains in state polls – where the voter models have a greater effect on the sample.

So, either Obama is heading for a blow out or the two sides of the electorate are, by the nature of their emotional approach to the election, are being measured differently in the polls producing a bias. Honestly I am not sure which is true, and it really could be that it is a bad year for the GOP.

Tags: 2008 Election, Polls