Model update: No one knows who will win the House

October 8, 2012

No, I'm not above using this album cover again.

If you've been following along with the Election in Numbers, you know that most models are predicting that President Obama will win reelection comfortably (more than 300 electoral votes, in many estimates) and Democrats will retain the Senate, with 52 to 54 seats.

The House, however, is a bit of a mystery. Democrats are leading in generic ballot polling, which prompted Sam Wang to predict that they're favored to retake the majority there. And political scientists Robert Erikson and Christopher Wlezien have found that in midterm elections, generic ballot polling (even from a year ahead of time) is a pretty good predictor, bolstering Wang's case.

But John Sides, Eric McGhee and Ben Highton at the Monkey Cage have built a model using economic data that is more bearish on Democrats' chances, forecasting a net gain of only one seat, with Republicans retaining their majority. And now, Wang is coming closer to their view.

Wang has altered his model to account for Republicans' twin advantages from (a) being the incumbent party in power and (b) controlling most state legislatures and thus spearheading this decade's redistricting wave. To retake the House, he concludes, Democrats need to win the popular vote by about 2.5 percentage points. And that's exactly where they are in polling. The median poll since the convention gave Democrats a lead of precisely 2.5 points.

So Wang concludes the race is tied. His prediction estimates that both parties will get 217.5 seats, with a standard deviation of 11 seats in either direction. So at a 95 percent confidence level, the results range from a 43-seat margin for Democrats to a 43-seat margin for Republicans. That's a more optimistic result for Democrats than the Monkey Cage model, which predicts a Republican margin of 47 seats, an outcome that Wang's model rules out. But it's a lot less promising for Democrats than Wang's previous forecast.

All of which is to say that it's really hard to predict House elections. And this race could be an interesting test. If Republicans largely preserve their majority, then the economic fundamentals-based approach of Sides, McGhee and Highton is vindicated, especially if the margin remains above the 43-seat limit that Wang predicts. If the results are razor-thin, or if Democrats win, then using generic ballot polls, as Wang does, might be a better bet.

Comments
Show Comments
Most Read Business
Next Story
Ezra Klein · October 8, 2012