“a conclusion that MIGHT be true but fir which there is no evidence to indicate WILL be true”
Okay, but why does erring in the side of caution always mean erring on the side of Democrats? Because of party affiliation? But they can’t judge when voters will jump parties, nor when turnout will be depressed, nor how indies will go. So they call it the best way they can, given their limitations. Sometimes they’re right, as when it’s painfully obvious in 08. And sometimes they miss it, as in 04, 10, or the Wisconsin recall.
My thinking is they don’t really try to predict the outcome, just be able to say they took all the precautions and followed the industry standards. So that we get this fantasy world of polls alongside the real world. And no one will care, because producing them and reporting them are done for their own sake, rather than for telling the future. So long as they’re not so far off should people remember they don’t have an excuse.
“Okay, but why does erring in the side of caution always mean erring on the side of Democrats? “
I think there are two reasons. First is that what history says. Since and including 1984, the average partisan breakdown for all elections was D+2.5. Ras is just using the historical average, which is a reasonable assumption.
I think the second is credibility. He’s a businessman. Being unusually wrong in this business is far more damaging than being unusually correct is beneficial. And, let’s face it, the Dems are better at pressuring media outlets, so being unusually wrong in favor of Republicans is even MORE damaging.
Ras is using a D+2.5 model while everyone else is using a D+7 or more model. There is zero evidence suggesting a historic Democratic wave, so Scott Rasmussen can be very confident that the turnout will be no more D than D+4 (2008 is the only time that EVER happened since Reagan was elected). Which means he’s pretty much guaranteed to have the closest turnout model. If he’s right, he’s the King of the Pollsters for nailing it. If he’s wrong and this election is even or has more Rs, he’s still far and away the closest and he’s STILL the King of the Pollsters. If he’s wrong and turnout equals the most Democratic it ever was other than 2008, he’s STILL the closest and is STILL the King of the Pollsters.
Basically, with a D+2.5 split, he’s guaranteed to be the closest under any normal circumstances AND has rock solid non-partisan justification for his model. If he models on an R+4, he opens himself up to risk of being a partisan hack and lowers his chances of being the closest. It’s kinda like the Price is Right strategy of betting one dollar more than the previous person.