Here is a good article explaining the rise of the quants. The main thing that is going on with Nate Silver and others is that they are combining information from several polls and doing meta-analysis with Baysian statistics. These models predict an easy Obama victory but combining information from several state level polls.
The reason I still disagree with them is that they are vulnerable to polls that are all making one systematic error: over-estimating the turnout among Democrats. I still think that is what is going on and combining information from a lot of polls that are making the same error simply re-enforces the error rather than corrects for it.
I still think that looking at the internals of these polls shows that Romney has strong leads among independents and that with anything like normal (i.e., non-2008 turnout levels) among Democrats and the increased enthusiasm among Republicans will lead to a Romney victory.
The reason I still disagree with them is that they are vulnerable to polls that are all making one systematic error: over-estimating the turnout among Democrats. I still think that is what is going on and combining information from a lot of polls that are making the same error simply re-enforces the error rather than corrects for it.
I still think that looking at the internals of these polls shows that Romney has strong leads among independents and that with anything like normal (i.e., non-2008 turnout levels) among Democrats and the increased enthusiasm among Republicans will lead to a Romney victory.
No comments:
Post a Comment