I see that Mickey Kaus is still pushing this line that the general inattention to the late generic polls showing a GOP surge was an example of liberal media bias. I've always thought that Mickey's is far too great a mind to waste -- even a part of it -- on the liberal media bias canard. But we can deal with that issue another time. The truth is that those late generic polls were on to something. But the reason people didn't pay more attention has nothing to do with liberal bias. It's rather more subtle than that -- and for that reason ignored.
To make sense of this you've got to go back to the 1998 midterm where an expected landslide for the Republicans turned into a small but significant Democratic victory. This was supposed to have been a great shocker. But if you were paying attention it really shouldn't have been.
At the time I was working at the now long-abandoned Cambridge offices of The American Prospect -- the then-bi-monthly, now bi-weekly, and soon to be monthly liberal policy mag. I was going around saying that I thought the Democrats would actually pick up seats and I wanted to write an article on the dynamics in play. That got vetoed by the higher-ups who thought we'd look stupid running an article talking about a good Democratic year after the Republicans had picked up forty seats.
Now as you can probably tell I'm rather proud of having gotten this one right. But the truth is that it was really only a matter of watching the polls. As I said before, the 1998 results were treated as a big upset. But if you looked at the polls it wasn't at all. The generic polls and those of individual races were really quite close to the mark. And at the end of the campaign they were switching over, if I remember correctly, into the Dems' column. The key was that everyone was so convinced that the Democrats were going to pay the price for Clinton's shenanigans that they found ways to argue themselves out of the what the polls were saying. Not just Republicans, but Democrats too. (See, it wasn't conservative media bias then either.)
The favored argument was that whatever the polls said, the massive turnout among aggrieved Christian-conservative whack-jobs would tip the scales in the Republicans' favor. Needless to say, that didn't happen.
And I think that's pretty much what happened this time too. Going into the weekend most people were pretty convinced that the Democrats were going to hold or pick up a few seats. That consensus in that direction was very strong. And since people didn't see an obvious reason for the late move in the Republicans' direction, they just ignored it.
The point, I think, is that group-think is often more powerful than actual data.