Nate Silver over at fivethirtyeight.com (soon to be a part of The New York Times) is doing some lengthy analysis of the accuracy of the various pollsters. If you head over there, you may find yourself pretty deep in the weeds and I'd be fibbing if I claimed that I know exactly what he's talking about every moment. My math skills are about on a par with my talent for Kabuki Dancing.
But I do understand some things and one is that when pollsters brag about their accuracy level, what they're talking about is how close their final polling came to the final outcome.They can be wildly off until a day or so before the election and still claim they called it within the margin of error. Silver cites this example…
In the 2008 Democratic primary in Wisconsin, for instance, which Barack Obama won by 17 points, American Research Group had released a poll on the Saturday prior to the election showing Obama losing to Hillary Clinton by 6 points; it then released a new poll 48 hours later showing Obama beating Clinton by 10 points. (It is very unlikely that there was in fact such dramatic late movement toward Obama, as most other pollsters had shown him well ahead the whole time).
In other words, I can now confidently predict for the next two years and five months that in the next presidential election, there will be a massive write-in vote for the robot from the old Lost in Space show and he'll win with 99% of the vote.Then a day or two before the election, I'll switch to whatever Gallup says and I'll probably be able to claim a pretty good batting average for that contest and boast of my accuracy.
A lot of folks claim that certain polls are slanted to please certain clients that subscribe to them. Sometimes, that's just a childish way of denying that the election does not seem to be going your way. Now and then, there may be some level of validity to the charge, which I especially see made against the Rasmussen Poll. It often (not always) looks like an outlier in giving far rosier forecasts for Republicans than all the others.
When Rasmussen is called out for deliberate bias, the answer is usually to point out how accurate their polls have been lately, meaning that they were on target just before the actual voting. I'm not interested in debating that. I'm just pointing out that you can be wildly off — deliberately or not — until very late in the game and still claim vindication 'n' victory when the final numbers are tallied.