It kinda blows my mind that the press obsesses over elections as horse races, yet can't even grasp some of the fundamentals of the horse race.
They continue to report on the pres polls fanatically, devoting a huge chunk of the coverage to that, yet they keep blowing it. They keep reporting how many points Kerry is ahead or behind Bush, when it's well known among pollsters, politicians and everyone in the business that that is a much less reliable indicator than a simpler number: how far Bush is over or under 50%.
It's well known and not controversial that elections involving incumbents have historically worked very differently than those without. With incumbents, most voters go through a two-stage process:
1. Rehire/fire the incumbent. For a president, this rehire/fire idea has been going on the back of our heads for four years, long before we think about who the challenger might be. Even once we know the challenger, it's not really about him. We already have a guy, and it's mainly about that guy.
2. If a voter begins to lean heavily toward firing, then and only then does he seriously consider the challenger. At that point, it's largely a question of whether the challenger is acceptable or not.
If a voter comes down on the rehire side, step 2 never happens. Brilliant challengers go down in flames, because much of the public never really bothers to get to know them, or gives them serious thought if they do. Even if a voter knows the challenger and likes him better, picking him is a risk. They already know who they have got, and unless he's been bad enough to fire, many people aren't going to take that risk. So the appeal of the challenger is just moot.
Now here's the key thing. It turns out--and we have data going back to at least Eisenhower on this--that most people have really made up their minds on the rehire/fire question by now, whether or not they're aware of it.
Or put another way, people who have had four years of the incumbent and still have not committed to voting for him almost never do.
Put still another way, the undecideds break overwhelmingly for the challenger. In fact, the data we have suggests that either every one of them breaks for the challenger, or more likely most do, and some of the people who said they were for the incumbent either defect or stay home.
Whatever the internal machinations of voters moving around, it turns out that if you look at the final poll numbers in 1996 showed Clinton leading Dole 51% to 37%--as they did (that's an average of the final three network polls)--it would not be wise to do the subtraction and predict that Clinton would win by 14%. The so-called incumbent rule states that you'd be much closer simply taking the final Clinton number, assuming he would get 51% and add all the undecideds to the challenger(s), in that case, Dole and Perot.
It actually turned out even worse for Clinton than that. He ended up with 49%--gaining none of the undecideds, and losing 2% to boot--while Dole scooped up 4% of the undecideds to end with 41%, and Perot got a few of the undecideds as well.
This is not a 1996 phenom. There's a great piece in The American Prospect laying out all the arguments, with a lot more data, and even more data--going back to Eisenhower--at this Mystery Pollster site.
With only slight discrepancies--e.g., Reagan picked up one whole percentage point of the undecideds; that's the outlier in the past quarter century--the rule always seems to hold.
Meaning if George Bush is below 50% this year, he's in deep shit. (Or below 49% if you think Nader will actually get his 2% this time, and/or that Bush will be another outlier to pick up 1% of the undecideds.)
Bush running 7-8 points ahead a few weeks ago, indicated Bush would win if the election were held then. But Kerry running neck and neck in the polls now actually indicates he would win easily if the election were held today.
So why is the press not reporting it this way? Too complicated for the public to swallow? It's really not that hard a concept to grasp.
Either they are being really dense, or assuming you are.
They make me so mad.
