Dave Dyer: It looks more like a rant/conspiracy theory to me. He starts with the premise that the polls were correct, and mixes in complaints about a-priori manipulation by changing voting laws, which have nothing to do with statistical analysis.
--ok... 0. "looks more like a rant/conspiracy theory" -- examples? 1. would you prefer the premise "polls=incorrect"? If so, why? Would there be any circumstance in which you would accept an election was fraudulent, based on evidence from polls? If so, would you not be assuming polls mean something? 2. further, anyhow said premise simply was NOT used, i.e. the "nonparametric" tests used there are still valid even if polls are complete random garbage. (Which is quite a nice trick, you might naively have not thought this possible.) It was, however, used in the (other) tests that assumed normal errors. 3. "complaints about a-priori manipulation by changing voting laws" are indeed mentioned as one motivation... "which have nothing to do with statistical analysis" is correct -- they were not used in the statistical analysis. -- Warren D. Smith http://RangeVoting.org <-- add your endorsement (by clicking "endorse" as 1st step)