## Monday, May 19, 2008

### Electoral College Math 0.1

The polls at this point in time give a good indication of how the Electoral College math might stack up if the general election comes down to a tactical result.

Basically, you can think of any Presidential election result (with two major parties splitting the Electoral votes) as coming down to three types:
Strategic result--"a landslide"--over 10% lead in popular vote; an Electoral College win over 100.
Decisive result--a lead of 4-10% in the popular vote and a safe Electoral College win (think Carter over Ford, or Clinton over Dole).
Tactical result--a lead of less than 4% in the popular vote and a "close race" in the Electoral college.

Right now, the national polls show the projected popular vote to be a virtual tie. This may not be at all the case at some points in the general election campaign just now beginning. I expect Obama to get a major bump from the Democratic convention and have an 8-10% lead at that point (around Labor Day). What happens afterwards depends on how well McCain and the Republicans respond to the situation, and whether they can dislodge Obama and the Democrats from the roll they would be on if that occurred. If they muck it up too much (like get Bush and Cheney involved in the campaign), such a lead could even expand to the strategic (wouldn't that be nice!)

I see only one reason to think McCain & Co. could succeed, then--this is the American electorate we are talking about. They will be selling fear, and there may be buyers. In that case, the decisive lead could shrink to an insignificant one, in which the vagaries of the Electoral College will be decisive (just like '00 and '04).

At any rate, recent polls give a good indication which will be the key states (not necessarily the same as always, surprisingly!) and what the final score might look like. Rasmussen is coy about it in their forecast, showing large numbers of leaners and a reluctance to move states from their '04 results (understandably), and CQ will be the same.

We have no obligation to be coy, Roy; so we will go out there now with our prediction for a tactical outcome: 293-245 Obama.

We have the following states changing from the 252-286 loss Kerry had to Dubya in '04: Ohio, Colorado, New Mexico, and Iowa. That's it.

We're going against the direction of the most recent polling on New Hampshire (currently showing a significant McCain lead), Ohio (on the strength of the Republican party collapse shown in '06), Michigan (like New Hampshire, still processing some Obama resentment), and surprisingly, Wisconsin. We're only going against the Rasmussen Market opinion in NH, where it's pretty much even, as it is for Nevada (looks like a tough slog to win that one for Obama). There are a lot of states looking close right now, though: 14 of the 34 or so states with polls recently published have raw margins less than 5% (so none of them would be considered "statistically significant" by the pollsters).

New Hampshire turns out to be important, because with it, Obama doesn't lose if he falters in Ohio, Michigan, or Wisconsin. New Mexico we have to assume will go to Obama, and Colorado, too.

Techy: My bets on states--Old ones: against Republicans in NJ at 17%; for Democrats in NM at 63%; for Democrats in Ohio at 60.4; against Republicans in ME at 19%; for Dems in Kansas at 12.5% (looking for Sebelius here!).

New bets/bids: for an upset in Alaska at 6%; sell vs. Republicans looking for an upset in CA at 8.9% (easy money); Buy MO at 37.4% (I'm going to regret that one; hopefully I'll sell it for a small profit and regret that, rather than the usual practice of holding onto the end there and losing); and sell NH vs. Republicans in NH at 48.2 (again, going against the current polling).

Ed Stoner said...

Love your analysis but I have to confess that I do not understand your betting percentages and what they mean. I guess I need to spend more time in Las Vegas! I like the idea of Obama picking a woman as a running mate as long as she is not married to Bill Clinton.

Chin Shih Tang said...

"Bets on states": On the Rasmussen Markets site, one can take a position on each of the states, and to which party it will go. They've been out there for months or years.

Those bets are now looking more meaningful since--practically speaking--we are down to the two candidates.

It's really more like a commodity exchange than a casino.

Let's take a state--Florida, for no particular reason. Right now, the chart looks like this:
Dem 23.1 25.0 23.1
Rep 73.0 75.0 75.0
Field 0.1 5.0 0.3

There are six different possible bets you could make: for the Dems to win, for them to lose (you're "selling" on them), for the Republicans to win or lose, or for anyone else (Field).

Right now, the Dems are about 25-75 to win the state in this market, though about 2% of the theoretical probability of the outcome is unaccounted for in the contracts made, and 4% in the bidding for winners.

Chin Shih Tang said...

Taking a look at this posting from the present--June 14--I have to say I didn't properly identify my predicted outcome as a "tactical" result. That result--if it's a secure one, as opposed to a terrible shaky outcome in the key states--is really at the decisive level, reflective of a 4-7% popular vote lead.

That's where we are now in terms of popular opinion and its messed-up translation into Presidential scorekeeping. Obama would win if he can put Michigan and Ohio away, leave plenty of opportunitites for upside surprises, and keep his cool.

If it tightens up (I think it's going to go the other way, at least until the climactic final weeks), Obama could stil get the same Electoral College outcome, just having a lot more closely contested states.