With so many close races, and polls less accurate than they have ever been, it's hard to guess how Tuesday's results will come out. Hard, but irresistible.

Sometimes a strong systematic error can be surmised in advance. In 1980, the polls said that Reagan-Carter was "too close to call", but I was quite sure Reagan was going to win. I just wish I'd thought to place some friendly bets that last week: at least one University of Chicago professor of my acquaintance took several hundred dollars off his colleagues that way.

Here is how I knew. At the time, I was working for a small company in San Francisco that measured air pollution, sometimes for private companies, sometimes for the EPA or similar state agencies. (We were the scientists, not the lawyers, and provided data for the lawyers on whichever side hired us.) In the weeks leading up to the election, all but one of my fellow employees -- 5 or 6 in all -- came up to me at one time or another and said "Don't tell Lee, but I'm voting for Reagan". Lee was the owner of the company, and a big Carter fan. The only other one who voted for Carter was our meteorologist, and his aversion to Republicans in general and Reagan in particular probably had something to do with his being gay.

I'm too lazy to look it up, but as I recall, Mrs. Thatcher's election the previous year was also supposed to be "too close to call", but she won handily. The pollsters also said that Reagan's reelection and Thatcher's first reelection were not in doubt, but would not be landslides. In fact, both were landslides. Why the consistent errors, and all in the same direction? My hypothesis at the time was that polls are particularly inaccurate when one candidate (Reagan or Thatcher) appears more competent, while the other one (Carter or Callahan) appears 'nicer'. When it comes right down to the actual decision, voters get more serious. That is also why third-party candidates tend to shrivel in the end. I've never been sure about how and when this turn to seriousness occurs:

1. As in my San Francisco experience, some people simply lie to many of their friends and coworkers about how they plan to vote, and probably lie to pollsters as well.
2. I suspect that some also lie to themselves, and go into the voting booth intending to vote for the trendier or nicer candidate, but end up choosing dull competence.
3. Of course, prejudice on the part of pollsters and reporters is also a likely factor. This is not in every case prejudice towards one side of the political spectrum. An awful lot of races are said to be 'tightening up' this weekend, and reporters do have an incentive to overestimate the tightness and with it the suspense: that's what sells newspapers and keeps eyes glued to TV screens.

So how does this apply to Tuesday's election? It's seldom so clearcut as in 1980, but here are some totally unqualified off-the-cuff predictions:

I'm guessing that Jean Carnahan will lose by a fair margin. She seems nice enough, but surely Missouri voters will want someone with a little more talent (pun intended) to represent them for the next six years.

Will dull but worthy candidates like Allard in Colorado get a small bounce on Tuesday, as voters decide that worthy is more important than dull? I certainly hope so, and even a small bounce should suffice, if the polls are at all accurate.

Will Bob Smith voters in New Hampshire turn to Sununu in the end? Smith is not on the ballot, but he is in effect a third-party write-in candidate, and I expect his actual votes to be quite a bit fewer than whatever the polls are predicting.

Will Minnesota voters actually elect an old guy who can't be bothered to campaign or debate in prime time or tell us what he thinks of the issues, just to show their sympathy for the surviving Wellstones? I certainly hope not.

Will competence beat a Kennedy in Maryland? Again, I would expect a shift of a point or two away from the glamorous but flaky Kennedy.

My prediction: the Republicans take the Senate, picking up 1-5 seats. We shall soon see. I would think that the fact that there's a war on would help the Republicans.

Update: (11/4, 11:59 PM)

Last paragraph corrected to read "take the Senate" instead of "keep" it. Thanks, P.Y. and C.A. Somehow I sometimes think that the Republicans already control the Senate: probably something to do with the fact that they controlled it after the last election, and the only Senator who has died since then was a Democrat.