One fun thing about posting on 3 blogs is that I can reach different audiences. Zillions of people read 538, but as one of the many second-bananas to Nate, I’m always careful about restricting my posts there to just the facts. I don’t do a lot of follow-up posting there, and the comments section at 538 is enough of a zoo that there’s no real room there for in-depth back-and-forth discussion.
My main blog gets lots of readers and informed comments, but right now I have a backlog of over a month, which means that it’s hard to squeeze in a lot of topical material. If I start running 5 posts a day, I’m afraid people will get overwhelmed.
So I often post political stuff right here on the Monkey Cage which appears to reach a news-media audience not completely covered by my other blogs. For example, John Sides pointed me to Matthew Yglesias’s link to my recent remarks that “elections are inherently more unstable when more than two candidates are involved.”
Yglesias writes:
That seems right on a theoretical level. . . . But for a political scientist’s post on a political science topic, Gelman’s was uncharacteristically lacking in specific references to research on the subject—I wonder what’s out there . . .
I have two quick responses:
1. Although I’m a professor of political science,I got my Ph.D. in statistics and am not very familiar with the political science literature. Just for example, I’ve never taught (or taken) an intro to American politics course. So, when it comes to the literature, I’ll have to defer to John Sides and my other co-bloggers here.
2. Just to complete my thoughts about stable and unstable elections: This is something that Gary and I thought a lot about when we were doing our research, 20 years ago. U.S. presidential election campaigns have a lot of things going for them to make the outcome more predictable:
(a) A long history: we can predict this year’s election, to some extent, from last year’s. The U.S. isn’t a country like Guatemala where they’ve only been having competitive elections for a few years.
(b) Clear separation between the parties. Talk about Tweedledee and Tweedledum aside, the Democrats and Republicans are, according to Huber and Stanig, further apart on economic issues than are left and right groupings in just about every other industrialized country.
(c) Only two major candidates. Anderson in 1980, Perot in 1992 and 1996: they came close but they didn’t quite make it a real three-candidate race. It basically worked to focus on the Democrat and the Republican.
(d) Equal resources. Not quite: Nixon reputedly massively outspent McGovern in 1972, Bush had the edge over Gore in 2000, and Obama had a few hundred million to spare in 2008. Still, compared to referenda and elections for congress and governor, presidential races are on a pretty level playing field.
(e) A clear schedule: Voters have many months to sort out the information and make up their minds.
Various other elections violate rules (a), (b), (c), (d) above, to the extent that it would not make sense to try to predict the results nakedly from the fundamentals without information about the campaign itself. Back in the early 1990s when we were working on that paper, I was living in California and was thinking a lot about referenda, which violate conditions (a) and (d) above: There’s no track record, and typically one side is much better funded than the other. This is not to say that campaign spending is all that matters, only that it does seem to matter, and also there’s basically no baseline, no expected outcome to compare to. (And, no, early polls don’t count.) The British election violated condition (c), and maybe some of the others also. Primary elections typically violate condition (b) and (d), often condition (c), and arguably violate condition (a) as well, as each primary is really a different story. Elections for Congress typically violate condition (d). Special elections violate condition (e).
I don’t know of any systematic research on the predictability of elections and how they relate to the above conditions. It would be an interesting topic.