Maybe, but I just don’t buy the idea that all these pollsters are destroying their reputation by reporting something that doesn’t exist. In 2008 was the same story. All the polls were wrong... until they were right. I believe there are a lot of pollsters who have their reputations at stake and are reporting reality, whether we like it or not.
No, I don’t think most pollsters try to skew their polls, but the climate today (I don’t mean Sandy) makes it very difficult to conduct polls. With a response rate of something between 10 - 20% according to what some pollsters recounted it is very difficult to get a non-skewed sample - and how do you know that the minority who actually responds to polls is really representative of the whole sample? Well, you don’t.
Some pollsters correct (recalculate) their results according to what they know (or expect a correct sample would look like) others do not.
An example was the Battleground poll released yesterday. The basic result was that Obama had the lead with 49/48, however, the resulting election day result prediction was Romney by 52/47. How come?
When they looked at the different age categories there Romney was in the lead in everyone except the youngest one (18 - 25 I think). Obviously for Obama to be in the lead in the whole sample his lead in this group must have been huge. However, as we know, and as the pollsters know, this is also the group that has the least tendency to actually go to the polls. So, given the reply to some other questions, which would help them to quantify the voting behavior they applied correction factors that gave the above result.
Rasmussen’s Ohio report, also out yesterday, is another case of a similar kind. He reported that more than 30% of his sample had voted. That subset was very much in Obama’s favor. However, he could have compared with the actual number of early votes which indicates that less than 20% of the expected votes have been cast so far. Correcting his results for this skewness would have given Romney a lead of 52/45 instead of 50/48 but Rasmussen did not do that sort of correction.
So which polls to trust? Basically, you have to look at the sample data and compare with what you know. So what do we know?
Well, there have been two large investigations by two acknowledged polling institutes (Rasmussen and Gallup) showing Republicans in the lead by 1-3% (nation wide), so that is what you can expect on election day.
The racial, age, and gender distribution is also pretty well known, and pollsters ought to be able to handle that.
At the moment ca 18% of the expected turn-out has voted.
Polls that confirm to those figures in their samples are probably closest to the truth, but there is of course also the polling history of the different outfits to take into account - Gallup, Rasmussen, Battleground are usually in the top.
Given the recent results from those and some other polling institutes I am pretty confident that Romney and Ryan will win the popular vote with a significant margin, large enough so that will also ensure a win in the elctoral college.
However, polling is not an exact science (to say the least) and the only poll that really matters is the one on the election day.