Revised 11/11/12
Remember how weird it was that the Republicans seemed to have drunk their own Kool-aid on the subject of polling and skewing, so that as Election Night wore on proving them utterly wrong they got seriously amazed and distressed? I'll get back to that presently.
Mark Liberman of Language Log has uncovered the following, from Mark Levin's reaction to the Obama election, on his syndicated radio program last Wednesday:
We conservatives, we do not accept bipartisanship in the pursuit of tyranny. Period. We will not negotiate the terms of our economic and political servitude. Period. We will not abandon our children to a dark and bleak future. We will not accept a fate that is alien to the legacy we inherited from every single future generation in this country.
Liberman is a linguist, of course, so he assumes inevitably that Levin is making a grammatical mistake here—that the "from" phrase is in the wrong place and using the wrong preposition. But those of us who have been hanging out here for a while know better, don't we? What we have here is yet another, rather refined instance of what we have called the retroactionary standpoint—the curious proclivity of Republicans to perceive time in reverse, to imagine themselves traveling from a remembered future into a dim but hopeful past (reactionaries merely try to make the future more like the past; retroactionaries see themselves as creating the past out of what the future has taught them).
Liberman found the case in a blog, previously unknown to me, called Headsup, where they're not afraid to use the term "time travel". I felt as if I'd been put in touch with an unsuspected sibling. Moreover, they had an example of the retroactionary in a context where I'd never imagined it: that of Skewgate.
It was in a post on how the wingers determined that there was something wrong with polls favoring Obama: Headsup had understood something about what was bugging them that had escaped everybody from me to Nate Silver himself: where they were finding the bias in the polling wasn't in the procedure but in the results: whatever went into a PPP or Times-CBS or whatever poll, except for Gallup and Rasmussen, too many Democrats were coming out. In other words a sample taken at time t was being skewed by its outcome at time t + 1, in a retroactive biasing effect!
"There’s nothing dumb, unscientific, or risible about questioning polls’ partisan samples [," wrote Katrina Trinko at The Corner.] OK, got me there. It makes perfect sense to believe that the result of a sample can bias the sampling procedure itself, as long as you believe in time travel. Time travel is a great way to get out of trouble faster than you got into it, but as a research method, no. It's dumb, unscientific and risible.How cool is that? Headsup has discovered that retroactionary thinking is so deeply embedded in the conservative brain as to operate unconsciously, in the understanding of simple mathematics, and at the same time found evidence that it really doesn't work: in the sense that when they "unskewed" the polls, all they got was bad predictions. (Unless you think backwards all the way through: that it was the voting on November 6 that caused the poll results of the previous week, that it was the results that caused the poll to be taken, that it was the taking of those final polls that caused the campaign to end, which caused it to start, which caused Obama to be elected for his first term, which caused the polling of 2008, and so on back to Magna Carta...)
