Finding the important races to watch

This is a follow-up to yesterday's post on the relative meaninglessness of the Cruz/O'Rourke contest in Texas. The basic problem you have as an observer of the House/Senate landscape is that you usually observe it from the perspective of journalists, and journalists require narratives. Journalists aren't statisticians, but looking at legislative elections and trying to figure out the likelihood of either chamber flipping requires a statistician's approach. It requires understanding the underlying characteristics of the constituencies that are likely to flip, the characteristics of a contest that contribute to a seat flipping, and how to aggregate those factors across the country in any given year. There are a bunch of models for how to do each of these things, and none of these things are what journalists do. Journalists tell stories, and they like flashy stories, like Cruz versus O'Rourke. That's the problem with relying on journalistic accounts of the legislative election landscape. Texas stays GOP, but it draws journalistic attention beyond its probability of flipping, with the problem being that journalistic attention is finite, and by drawing attention towards that race, the Cruz/O'Rourke contest draws attention away from more important stuff, like Heller/Rosen. Journalists are drawn to the big name (Cruz), the big money (expensive state with lots of media markets), and the big, Texas-sized chip on the shoulder that Texans have about Texas being Texas. (Get over yourselves.) All of that makes it easier for story-tellers to tell stories, but those stories aren't what you need if you want to know if the Senate flips. It probably won't, but the fact that Nevada is more within reach for the Democrats means that's a better place for your attention, which you'd never know if you just let the press dictate your attention. OHHH! Beto was in a PUNK band! (I always hated punk. Learn to play your instruments, people.)

So, how should you go about figuring how where to focus your attention? First, obviously, the polls. RealClearPolitics is a first-stop for me every morning, and the polling averages make it blindingly obvious that Heller should worry more than Cruz. That leads to the second factor. Apply some historical knowledge. Nevada's electoral history makes it a swing state. Texas? Not so much. You do need to look at some historical data. It helps at the Senate level to know something about the states. What about the House? There is very little point trying to find a handful of races for your attention, except for the entertainment value of the truly bonkers candidates, whom you know I love. With 435 districts, there's going to be random error. At the level of any individual race, the incumbent wins. No incumbent? The more experienced candidate wins. Beyond that? You're kind of wasting your time trying to guess without more detailed knowledge than is worth your time, as a casual observer, to gather. We just don't have enough House-level polling data.

My point is that these narratives that draw your attention-- yeah, they can be entertaining, but they aren't really relevant. Once the cycle is over, I'm going to send an email to Gary Jacobson, who compiles congressional election data, and distributes the data set to anyone who asks. I'm going to run some statistical models predicting outcomes based on candidate quality, spending, and underlying district partisanship. The models will work very, very well. The equations will come out pretty much the same way they do every year because math works. Social science works, when done properly. I'm looking at you, psychology! What I'm talking about here? It's called "replication!" Look into it! This process, though, isn't flashy, and it doesn't have a narrative.

But it's right. So, have fun reading about Beto O'Rourke skateboarding, or whatever. It doesn't matter. Personally, I'm just frustrated by the fact that I can't get up-to-the-second spending figures from non-incumbent candidates. (Although reporting has gotten better even since I was in grad school.) Remember, incumbent spending doesn't matter all that much, but challenger spending sort of does, even though most lose anyway.