Telling Stories About Polls
Roger Kimball rails against “The Narrative” that says Obama is very likely to win the election:
Ah, the polls. I pointed out, as I have often pointed out here, that polls are often fragile, unreliable constructs: more the product of hope than the evidence of fact.
This is a revealing line, since Kimball is the one replacing evidence with wishful thinking here. He believes and hopes that Romney will “not only win but win big,” but there is not much evidence he can cite that would support this hope. So he pretends that facts he dislikes are “more the product of hope,” when it is he who is relying almost entirely on hope (or delusion). There is some unfounded storytelling going on here, but it’s not happening the way Kimball describes.
Polls are measures of voter preferences. While imperfect, they do measure voters’ political leanings reasonably well. If they were as frequently misleading as Kimball hopes they are, no one would pay for them or pay any attention to them. Kimball repeats the standard “oversampling” argument, which has been thoroughly debunked. Here is Silver:
But pollsters, at least if they are following the industry’s standard guidelines, do not choose how many Democrats, Republicans or independent voters to put into their samples — any more than they choose the number of voters for Mr. Obama or Mitt Romney. Instead, this is determined by the responses of the voters that they reach after calling random numbers from telephone directories or registered voter lists.
Kimball also pushes the idea that “most polls (Rasmussen is an exception) canvass registered rather than likely voters.” Paul Rahe makes the same weird argument elsewhere:
Scott Rasmussen’s polls differ from those of his rivals in one particular. He samples likely, not registered voters. Come October, however, many of the other pollsters shift their sampling and focus, as Rasmussen does, onto likely voters.
What Kimball and Rahe describe may have been true in late July or early August, but it ceased to be true five or six weeks ago. It wasn’t true for the entire month of September. Just review the national surveys that have been released since the end of August, and you will find that virtually all surveys except for Gallup are polls of likely voters. Almost every polling outfit made that change long before now, and the only reason that someone wouldn’t realize this is if he has been carefully ignoring non-Rasmussen polls and clinging to the illusion that only Rasmussen can be trusted.