Wednesday, November 5, 2008

Campaign effects?

The returns aren't quite all in, but from CNN's latest numbers, John McCain won 46.9% of the two-party vote. Back in the middle of the summer, I predicted McCain would get 47.7% of the two-party vote. That's pretty close.

Now, my point here is not to brag -- many political scientists forecast roughly the same thing. The question is, why were we able to do this? How could we come so close to predicting what would happen just using variables like the economy, incumbency, and the current president's popularity? And if we can do this, what does this say about the role campaigns play? The implication is that campaigns don't amount to much.

If ever we should have seen campaigns mattering, it was this year. I can't recall a presidential election in which there was such an asymmetry between the two major campaigns in terms of money, skill, message discipline, enthusiasm, volunteerism, and basic competence. And yet that all apparently amounted to less than one point.

Of course, maybe the campaign asymmetry really did matter, but the effect was countered by Obama's race, McCain's history of moderation, etc. But still.

The fact that the poll trends were essentially flat throughout October, amid the debates, Joe the Plumber, William Ayers, cries of socialism, and everything else, suggests that the fundamentals really held despite all the sound and fury.


Anonymous said...

Off by .8%? And you call yourself a scientist.

Seth Masket said...

A social scientist. Cut me some slack.

Anonymous said...

A scientist who studies socials? Like, ice cream?

Anonymous said...

I don't think this election is a great success for forecasters. While many of them ended up on the mark, they only did so after a huge, ginormous, monstrous exogenous event intervened: the record-breaking collapse of interbank lending, the crushing losses in the stock markets, and the consequent intervention.

If the economy had not had its sudden, steep tanking until December, then at the very least McCain would have done much better and perhaps even won. And, more to the point, nobody's model included a variable for "Did the economy go completely in the crapper?"

Put differently, it took a huge exogenous shock to bring this election into the zone predicted by models that had no such shocks.

Seth Masket said...

Yeah, that's one way of looking at it. Another is that the models were right despite the ginormous exogenous shock. We can't really know what would have happened in the election had their been no financial meltdown, but I'm guessing the outcome wouldn't have been enormously different. Maybe a few points in McCain's direction, but still an Obama win. All of which suggests that things like party affiliation really still matter.

Anonymous said...

Yeahbut, let's say I predict that the Cowboys/Lions game goes to the Cowboys by 18.

And then it goes to the Cowboys by 19... after a plane carrying the Lions is attacked by T-rexes flying F-14s, so they play entirely with their second string and some Elvis impersonators.

I would argue that my prediction, though it turned out correct, was really not spot-on.

But yeah, we can't know how things would have turned out if the markets hadn't tanked yet. Thank God.

Seth Masket said...

I like your analogy, except that football game outcomes are determined almost entirely by the skills and strategic and tactical choices of those in the game. Conversely, there's a lot that goes on outside a political campaign that affects an election's outcome. Hell, Dukakis came within 9 points of winning, and he basically didn't even campaign. If the Lions decided that negative campaigning was wrong and that they'd never tackle a Cowboy, they'd lose by a lot more than 9.