The predictions of the polls for this election are all over the place. Although in general a Republican win is expected, the scope and breadth of that win is highly disputed.
Liberal rags and bloggers say that the Republicans have peaked and that Democrats will stage a surprising resurgence by election day. They are trying to rally the troops, of course, and it might represent whistling past the graveyard.
Then again, they might be correct. Who knows?
Who knows? Why, the pollsters, of course. Except when they don’t.
Polls are always somewhat unreliable. It’s only ex post facto that we learn which ones were right and which ones wrong. But this year they are probably more unreliable than ever, because of that all-important variable: voter turnout.
Turnout always matters, and especially in off-years where it is usually low and therefore small variations matter even more. But in 2010, turnout is an especially wild wild-card.
We know that voter enthusiasm is higher on the Republican side, and we know that it will almost certainly make a difference. But how much higher, and how much difference? The gap is highlighted by this recent Gallup poll, the first this year to take into account likely voters rather than registered voters, and the results are shocking for Democrats and heartening for Republicans:
Under its “high turnout model” Republicans lead 53%-40%. Under its “low turnout model” Republicans lead 56%-38%.
That’s not only astounding, it’s an outlier. The volatility of the polls is further reflected in the flux of the Rasmussen polling of likely voters: the Republicans have only a 3-point lead this week although they had a 10-point lead a mere two weeks ago.
What does this mean? It means that each side will fasten on the polls that are more encouraging to its own agenda. It means that polls are likely to keep changing right up till election day. It means that in general Republicans are ahead. And that’s all I can figure it means.
Polls are good at prediction, to a certain extent. But they are acknowledged to be flawed, and there is always a fairly significant margin of error that in close races can be determinative. This year I believe the margin of error is even greater than usual, because this year the mood of the electorate is profoundly different than in previous years—even previous “throw the bums out” years such as 1994.
There is a disgust towards both parties the depth of which I’ve never seen before. There are more entrenched incumbents at risk. The energy level of the electorate is stronger and more seriously angry. So pollsters, beware. The usual rules may not apply.

