I think that this is the key problem with the polls that you trusted. Statistical correction for known demographic parameters (age, race etc.) is relatively simple and defensible. Statistical correction for voter turnout is based on a poorly known parameter set because it is hard to measure intentions and hard to distinguish failure to vote from failure to support. Those pollsters who simply used historical turnout data did quite well for the most part. The other pollsters just assumed a turnout model based on their preconceptions of public opinion and voting intention, which effectively means that they assumed the outcome that they were trying to measure. It is simply not defensible as a statistical technique - it's statistical manipulation to achieve a result, and its flaws were fully exposed by the outcome of the election.
I don't think you understand the issue as well as you think you do. All polls are already statistically corrected. It's not like they went out and surveyed 40% Democrats and 30% Republicans and just left it that way. All credible polls survey a certain number of people...it might even be 50/50 in terms of partisan split. Then they up-sample or down-sample (or both) the various groups based on what they think turnout will be.
Yes, which all polls make. As I said, I think you're quite misinformed on how polling works.
Telling me that I don't understand how polling works is rather amusing given the title of this thread, but beyond that I would argue again that you are missing the distinction between dependent and independent variables, and that in fact it is you who are misunderstanding the basics of statistical analysis. I understand how polling is conducted and the various ways in which the resulting data are reduced and analyzed. I'm not trying to tell you how it is done - I'm trying to point out why some of the analysis techniques are flawed, and that those flaws are the explanation for the gulf between your expectations and the actual outcome. I know that you disagree and that the chances of me convincing you are less than Romney's chances in the election.
No, that's a fact. It's based on registered Republicans not voting and based on Republican support from 2008 and 2004.
Based on Republican support in 2004 and 2008? So it was fine arbitrarily to assume that Democratic support and turnout would plummet (i.e. ignoring the 2008 turnout data), while at the same time assuming that Republican support and turnout would be sustained or increased (based on 2008 data). Do you not see the level of aggregate assumption here?
Again, you're wrong and don't understand how polling works. Perhaps you will before you're done reading the above.
If we had been having this discussion before the election, I would have disagreed with you, but the outcome would still be unknown and you could reasonably have argued that you believed what you were saying. What completely baffles me is how you can still be making these arguments after they were shown to be erroneous.
The poll corrections that you are defending were demonstrably wrong. Now you appear to be saying that Republican votes were down not because Republican support was down on expected levels, but because for some reason they just didn't vote. Before the election you argued that the polls were skewed and really showed a big Republican lead if you unskewed them. Now you are arguing that the incorrectly unskewed polls were correct even though they weren't, and that the reason that they were wrong was that the turnout was wrong. But they weren't skewed and they reflected the election outcome rather well, especially when aggregated.
I'm certainly glad that I don't share your understanding of how polls work.
There was obviously less support for the party, hence the election result. On what do you base your claim that there is less support for conservatism in general?
As I said - I was referring to support for the Republican party. So you agree there is less support - I thought you were arguing that it was just turnout.
I made no claim that there is more or less support for conservatism in general - I didn't say that and I already corrected you when you asserted that I had. I did ask you why you thought there is not less support, but, not surprisingly, you chose to avoid the question by just turning it around.













Who do you listen to? And why are they a better representative than people I talk to? And what is "the mood" of the country, jimmac? I can't wait to read that...