Well, according to USA Today, Dean is definitely going to lose to Bush next November. After all, numbers never lie:

President Bush is ending his third year in office with 63% job approval, the highest rating of any president since Lyndon Johnson, who finished 1963 with a 74% rating a month after John F. Kennedy’s assassination.

Johnson went on to win the 1964 election 10 months later in a landslide over Republican Barry Goldwater.

With the exception of Jimmy Carter, every president since Franklin Roosevelt who ended his third year in office with job approval above 50% won the re-election he sought. Presidential job-approval polling began with Roosevelt.

Richard Nixon, who was at 50% at the end of his third year, also won. Carter was at 54% when the year ended.

Polling analysts and presidential scholars agree that it is too early to consider Bush a sure winner next year, despite his showing now. Things can change:

? Bush’s father was at 50% approval at the end of 1991, and he lost to Bill Clinton. A sour economy and a perception that he was at a loss to fix it helped do him in.

? Jimmy Carter ended 1979 with 54% approval and was defeated by Ronald Reagan in 1980. Carter’s response to the Iranian hostage crisis, which was seen as weak, and a senate primary challenge by Sen. Edward Kennedy eroded confidence in his leadership.

I’d trust Miss Cleo’s predictions about the future as much as I would any political pollster. While it’s “fun” to use polling numbers to play fortune teller with political races, there are three big problems with these kinds of predictions.

The first is that the sample size for this is much too small to make accurate predictions. If the presidential approval ratings only date back to Roosevelt’s first term, then were only looking at eighteen total elections as our sample. But before you crunch the numbers, you have to remove the five elections that didn’t involve an incumbent (52, 60, 68, 88, 00). Then there were also three instances in which the incumbent wasn’t actually elected to the job he was trying to regain (48, 64, 76). When you remove all that, you’ve got a situation in which two of the ten elections in which an incumbent was seeking to remain president didn’t succeed. While you can spin this to say that Bush has an 80% chance of beating his challenger, there really isn’t enough data to make that determination.

Secondly, the article doesn’t give any information about the nature of the polls. Are the numbers they’re comparing coming from the same organization using the exact same methods over the last 70+ years? If not, any differences (no matter how subtle) could have an enormous impact on the poll results. Even if they were the same methods, the two out of ten that lost above were the only two incumbents that lost. While you can make the argument that Truman and Johnson’s failures to seek renomination were due to their polling numbers, there are no examples of an incumbent with low polling numbers seeking re-election and losing. The Carter and Bush Sr. losses mentioned in the article are aberrations to a trend that they never prove in the first place.

Finally, it’s not like these things happen in a vacuum. Presidential approval ratings are so hazy, that you can barely use them to make judgments about present performance, much less use them to predict the future. Why don’t they make the correlation that the only two exceptions to the rule were when the incumbent was seeking reelection with a crappy economy? That seems a lot more tangible than any anonymous surveys of registered voters too polite to hang up the phone on an annoying pollster.

In the end, it’s all just a numerology game that’s as silly as a sports fan’s obsession with statistics. So-and-so athlete has never missed a field-goal during a game with a full moon. Incumbents who poll over 50% a year prior to the election (usually) get reelected. Blah, blah, blah…