Tuesday, October 22, 2002

A few quick bits and pieces.

Amidst all the puzzlement as to why Saddam Hussein would release nearly all prisoners from Iraqi jails, something struck me that I haven't seen even mentioned anywhere. As invasion becomes more and more likely, I could imagine SH trying to deter an American takeover by making Iraw as unappealing, ungovernable a place as possible. And if it doesn't deter, it will at least give him the satisfaction of spiting the postwar American administration and making their life more difficult. I have no idea how many of the released prisoners are actual nogoodniks, as opposed to innocent folks imprisoned by the regime for dissenting. But releasing the nogoodniks seems to me like a nasty, clever little bit of self-interest.

Robert Novak reports on an astonishing polling night for Bill Simon last Thursday-- Simon up by a point, and down by less than three over the three-night average. Something to remember: the margin of error on a poll isn't the only part of the poll's overall level of uncertainty. The margin of error usually represents the 95% confidence interval. That is, a poll result of 37% plus-or-minus 4% means that there is a 95% chance that the level of support lies somewhere between 33% and 41%. What goes unmentioned is the 5% chance that it does not-- and the result that, on average, once out of twenty times the true level of support will lie outside that confidence interval. A nightly tracking poll will get a clunker of a result one out of twenty nights; and that will skew three-day rolling average for, well, three days. (Rolling averages are a better guide than the snapshot numbers, since the chance that all three nights will be clunkers is very low-- 1/8000. But they mitigate the seriousness of the clunker night by stretching its effect out.)

Gray Davis deserves to lose, and I wouldn't at all mind seeing California voters turn him out. But I wouldn't change my bets just yet.

In New Hampshire, on the other hand: I'm going to predict Sununu beats Shaheen. Not because of the upcoming Bush visit to the state; the truth is that New Hampshire has never loved Bushes (remember Reagan '80, Buchanan '92, McCain '00), and that W carried the state by a breath and a whisker in the general election. But a) New Hampshire almost always picks a Republican if he (usually he) is even remotely plausible and credible, and sometimes (*coughBobSmithcough*) when he's not; and b) I have a good barometer to check. If Shaheen were particularly likely to win, we'd by now be seeing daily front-page screeching-howler-money-hysterical editorials from The Manchester Union Leader. The Union-Leader can't get its own preferred candidate elected in multicandidate primaries; but it specializes in driving up the negatives, and driving down the approval, of some especially-hated target. "Especially-hated" usually means "especially likely to beat the preferred candidate." If this bit of uncharacteristically goo-goo campaign refereeing, buried back on the editorial page, is the extent of the Union Leader's current attack on Shaheen, then they're not very worried. (If non-New Hampshirites read this editorial and think it does sound pretty harsh, all I can do is assure you: this is nothing by the Union-Leader's standards, and the decision whether to place editorials on the editorial or the front page really is a good measurement.)

No comments: