Moneyball

I finally saw Moneyball last night, how statistical analysis changed the game of baseball. Before the ideas presented in the book, scouts judged players on their knowledge and intuition. They just knew what they saw. In politics, the experts just know what they know. They too don't think statistical analysis applies, but it does too. Here are a few.

Individual Predictions are Dicey
What happens overall is far more predictable. While statistical analysis can tell us what's likely to happen, it can't tell us what will always happen. Nothing happens every time. Thus, predicting several races is far easier than predicting individual races. The experts predict Republicans will lose four seats in Illinois because Democrats will be favored in each district. Favorites don't always win in anything. Republicans will win 1 or 2 of these districts. Which ones? No idea. Based on the odds I think Democrats will win 2.4 districts and Republicans 1.6.

Districts that lean Republican will eventually elect a Republican
A Democrat may win an R+3 in a Democratic year, but the next election is unlikely to have a higher number of Democrats, lower number of Republicans, or strong independent advantage. When things return to normal, most of these districts will go back to Republicans.

The 2010 Republican wave was very predictable because the Democrats had 257 districts but only a Democratic PVI in 192. Republicans have a positive PVI in 234 districts. A Republican majority is the normal state of things.

Party registration is like batting average
Batting average used to be sacrosanct, but it really was an okay stat that doesn’t predict results as better stats. It’s not how many people that are registered that counts. It’s how many people that vote. This may sound obvious, but too often people look at a 5 point Democratic registration advantage and think that means the Democrat will win. Democrats have a 14 point registration advantage in Pennsylvania, but exit polls have shown that they never have had more than a 7 point advantage with voters. It’s not universal, but Democrats tend to register a lot of people who don’t vote very often.

A wave year will only happen under certain circumstances
There are certain factors that lead to a wave year. Absent of them, a wave is rare. Presidents make big gains in their first election but don’t make big gains in re-elections, no matter how they do.

Issues need to be current and active to impact elections
In 2010 Democrats ran on “Republicans are going to kill Social Security.” It didn’t resonate because Republicans hadn’t passed a Social Security bill and the issue wasn’t a big deal. Democrats, on the other hand, passed a health insurance bill and that bill was still a big deal.

At some point political experts will realize that past voting is highly predictable of future voting. Until then, we’ll just go on their feel.