Tuesday, October 26, 2004
I am going to let it all hang out and give my official predictions for the Presidential election in a week. Hopefully any errors I make will be on the pessimistic side.
There are two basic strategies for calling this race. The first method is to look at national-level polling results and then apply those numbers to the state based on the 2000 results. So if Gore won a state by 1%, but Kerry does 2% better than Gore nation-wide, then we can estimate that Kerry will win by 3%. This is a historically-oriented model.
The second method would be to look at the state polls, and apply what we know about the nature of undecideds and the specific dynamics of this race to come up with an electoral college vote. I am not going to use this method, not only because it is very time consuming, but also because a) there is an underlying partisan structure to elections, and b) the state polls can be wildly divergent and unreliable. Also, I just don't know enough about the state polls to speak intelligently about them.
Okay, the major national polls are of two basic types: structural polls and horse race polls. Structural polls treat the election as a referendum on the incumbent, while horse race polls look at the current spread between the candidates. The structural polls are presidential approval, right-track/wrong-track, and generic re-elect numbers. The former has had Bush with average numbers around 49%- not so good. What is more revealing are the right track numbers, which average about 41% (Yikes!). Well over fifty percent view the country as going in the wrong direction, which always spells trouble for an incumbent. (The real question, how can the same voter think the President is doing an okay job AND that the country is going down the crapper?) A lot of pollsters think the key poll is the re-elect number ("do you think the President deserves to be re-elected?), which as best I can tell is about 46%, a number that crops up again later.
The horserace numbers can be analyzed one of two ways. If you are just comparing the two candidates, then Bush is right in this thing. The average of the major polls has Bush leading among likely voters by 2 points, 48-46. Among registered voters, it is a dead heat at 46%. On the other hand, since this is not an open seat race but a race with an incumbent President, I am going to employ a referendum model: Bush's numbers are the decisive ones. If we assume that the registered voter model is a better one (I expect Democratic turnout to be high) and that undecideds will break, say, 2-1 for Kerry, then it looks like Bush is going to lose 51-48, for a three-point Kerry victory.
The next step is to come up with an estimate of the electoral college vote. Here all I am going to do is add 2 points to Gore's margin in 2000. This means that I expect that Kerry will win all the states that Gore won, for a total of 260 electoral college votes. Adding three percentage points to Gore's numbers would flip Florida, New Hampshire into the Democratic column. I think Kerry could very well out-perform Gore in North Carolina, Virginia, Colorado, Arizona, and Nevada. I think he will do worse in Tennessee, for obvious reasons. I am also going to give Kerry Ohio, even though this violates by basic method. Gore withdrew from Ohio while Kerry is hotly contesting it. And my gut just tells me that Kerry is going to take that state.
When you add New Hampshire, Florida, and Ohio to Gore's states, we have a grand total of 311 electoral votes (to Bush's 227). This is an underperforming electoral vote, since the electoral college tends to magnify the margin of the winner, but there you have it. It is still more than enough to win.
So my official prediction: Kerry wins 51-48 in the nationwide popular vote, and carries the electoral college 311-227. I am fully prepared to eat my share of crow come November 3rd. I just don't want to. I don't like crow.