The Denver Post is out with a new, fairly comprehensive poll of the state this morning, and claims that it spells trouble for President Bush. Maybe the Star-Tribune has been running seminars on polling and poll interpretation on the side, because the differences between 2000 and 2004 just aren't as stark as the Post says they are.

First, let's do the numbers. Most importantly, the vote-for number and the approval rating number. The Post has Bush up 48-43 in a state he won 49-40 in 2000. It also has him at 50% approval.

Now, Colorado's voter registration in Republican, 37-30, but the poll has a sample of 41-37 Republican. This isn't as bad as the LA Times poll, but it does skew the numbers in two ways. It narrows the party margin, and it significatly understates independents. Independents are split 44-43 for Bush. Neither Mason-Dixon nor the Post explains why Democrats would be more likely to vote that Republicans. In fact, in a poll of 800, all of whom "stated they regularly vote in state elections," 4% didn't vote last time. There's no analysis of these DNVs.

When adjusting for actual party registration, Bush is up 49-42, and his approval rating creeps up over 50%. Hardly cause for panic, either locally or nationally. The adjustments are small, but they're almost all Karen Crummy has when breathlessly announcing that Kerry is "within striking distance."

The Post's analysis is the real crime here.

"This election is about the incumbent. Voters are asking, 'How is his performance? Does he deserve to be rehired?"' said Larry Harris, a director of Mason-Dixon Polling & Research Inc., which conducted the poll. "Right now, it looks like Bush is vulnerable in Colorado, and that spells opportunity for Kerry."

News flash: elections are always about the incumbent. Even elections when the incumbent is term-out are about him. People voted for Bush 41 largely because they bought him as Reagan II. Gore was so afraid of the incumbent he spent more time running away from his than towards himself. In Ford's case, it was even a little about the incumbent's predecessor. Humphrey lost largely on the strength on Vietnam, Johnson's war. Eisenhower's lukewarm endorsement of Nixon may have cost him the election. And Eisenhower himself was elected largely because of Korea, which Adlai Stevenson didn't start. (Neither did Truman, of course, but it was his job to finish it.)

But even elections that are about the incumbent need a challenger worthy of the office. Or at least one with something for people to latch on to. Usually that something is optimism. Even Carter, who got more purse-lipped and dour as the world crashed around him, came into office as a breath of fresh air. One of the major travelling expenses for the Kerry campaign is air fresheners.

"If Kerry gets ahead in Colorado, it suggests a massive Democratic landslide," said Larry Sabato, director of the University of Virginia's Center for Politics. "If he is even leading by a point or two in Colorado, then (Bush) will be hard-pressed to win the country."

Now Sabato's as good as they get. What he says is absolutely, 100% correct, but it's also 100% obvious, and 100% irrelevant. He's following Senate and Presidential races all over the country, and probably a few House races and a few governor's races, too. This is the third time in the last week he's been quoted by a paper out here, and he's got to be wondering if there are any political scientists living in Colorado. It's a general statement, and its power is entirely based on numbers shifts that the Post is trying to manufacture.

The poll indicates Bush, who beat Al Gore in Colorado by nine percentage points in the 2000 presidential race, now has two obvious problems: He is under 50 percent in the race, and fewer than half the voters have a favorable impression of him.

Since 1948, incumbent presidents running for re-election with approval ratings under 50 percent in June or July of an election year have lost.

Two statement, two problems. First, Bush won Colorado with under 50% of the vote. Now, there's no question that almost all those Nader voters who bother to show up are going to vote for Kerry. But when you're at 49%, you don't need very many to get you over 50. Yes, incumbents would like to have a re-elect number over 50%; it's not fatal, and when you're up by seven points, it's not even serious.

As for those approval ratings, proper adjustment for party gets them up just over 50%. Unadjusted, you're looking at a margin of 1 voter. But more importantly, there's no reason to think that 50% is an important one, or even a stable one. The presidents who were re-elected all had approval ratings in the high 50s or higher, those that were defeated, in the low 40s or lower.

I suspect that these numbers tend to roll away from 50% as the election gets closer. An election is a choice, and there's a tendency to think better of the lesser of two evils, or to think worse of the guy you're voting against. Nobody likes to vote against someone they have to admit is doing a good job, and I suspect there's a certain feedback loop at work here. You persuade yourself that the incumbent is better or worse, in order to help justify your vote.

This is fine for an election year. What's poisonous is when it infects the other three years, too.