One thing I find funny is the current habit to parse the freaking hell out of them like the pollsters were picking apart goat entrails. Gallup finds 1200 people who don't screen their landline calls. A third of them are "independents." A tenth of those are from "battleground states." Half of them are women. The resulting pool of voters is barely enough to field two competing baseball teams yet that's sufficient for political pundits to draw massive conclusions about what independent women in battleground states think.
Many pundits believe if you bundle all the polls together you somehow increase the accuracy. Scientifically, that is nonsense which is why you never see a margin of error on these "poll of polls." Then you have people like Unskewed Polls who parse polls into tiny pieces, give greater emphasis to some of the pieces and then reinterpret the results into what they want to see. Nate Silver uses the polls to handicap the election like a Hoboken bookie.
The ultimate in parsing a poll was this story that trumpeted a 0.07% shift in polling results. That's one person out of 1293 people in a poll with a margin of error of plus/minus 3%.
The Misquoted Yoda Effect.
Since I've already cut open the goat.
|An Etruscan chart for reading livers.|
If Barack wins it will have to be without Ohio or Florida. I am convinced that Romney will get those states by hook or by crook; if he doesn't win with the votes cast Romney will win with the votes counted (John Hustad and Rick Scott will see to that.). Expect lots of mischief if the final results in those states are delayed days or weeks.
You can trust my prediction over those of "scientific pollsters." I killed a sacred goat and used skills passed down countless generations. Pollsters merely annoy lots of people and plug the results into a spreadsheet.