A Little More Eeyorism (Which I Mention Just To Make Sure You Vote)

I began wondering tonight if polls were overstating Republican support.

Polls generally understate Republican support, of course, to the tune of 3-4 points. But not always.

In Nate Silver's cover-your-ass column explaining that Republicans could gain a lot more than the 50 or so seats he's predicting (worth reading for the Democratic Doomsday Scenario laid out in the beginning, and that's not even a real CAC-level doomsday scenario), he mentions this:

If Republicans tend to overperform expectations in some races, they will probably also overperform in many, most, or maybe even almost all races. The same holds true for Democrats. (The most recent time something like this occurred was 1998, when polls underestimated the standing of Democrats by 4-5 points nationwide and in almost all individual races.)

I've been thinking like this: Why do polls usually include too many Democrats? Because people don't want to answer political questions if they're Republicans, as they know their answers will tend to be disfavored -- that is, they'd rather keep those answer to themselves.

This would tend to explain the 1998 aberration in which Democrats, rather than Republicans, polled more poorly than their ultimate performance. Consider that in 1998, people decided, yes, Bill Clinton had perjured himself, not once but several times, and obstructed justice. But they had also decided to give him a pass on serial lawbreaking because the economy was too good and vote for Democrats (in an effort to stop impeachment).

Some people might feel ashamed to take that position, putting, as it did, filthy lucre ahead of principle and legality -- and in that case, they might have decided their answers were the disfavored ones, and thus avoided pollsters, resulting in Republican support being overstated.

In most years, one can imagine liberals are more eager to talk with pollsters, as they expect a Scooby Snack, as it were, for mouthing the socially-favored answer. This might account for the persistent understating of conservatism's popularity.

But this year, conservatives might be so eager to proclaim their political views, and liberals so sheepish about it (their god is exposed as a weakling charlatan) that the typical situation is reversed, as was the case in 1998, and polls overstate rather than understate conservative strength, and understate rather than overstate liberal strength.

Even if this eeyore speculation is true, it wouldn't mean the GOP is in for a bad night -- just that they'd be in for a very good night rather than an epochal one. Republicans would still lead by 7-10 or so (which is itself unprecedented or almost so). But it would mean that the lead of 10-15 is overstated.

In addition, if any of this is true, of course it's also true that you can't tack on the typical "and give the Republican 3-4 points to offset polling's traditional understatement of Republican performance" corrective, as you'd really need to correct in the exact opposite way.

So what does this mean?

It means you have to vote. No matter what the polls say, you need to vote, and vote party-line, straight ticket.

posted by Ace at 11:05 PM

| Access Comments