In the first post on our House forecasting model for the 2014 election, we estimated that, as of this moment, Republicans are likely to gain five seats in the House of Representatives. That estimate is in line with some others, such as this recent analysis from David Wasserman of the Cook Political Report, who projects that right now the Republicans are likely to gain between zero and 10 seats.
But as we noted in that post, any specific seat estimate comes with a lot of uncertainty. Any forecasting model is a simplification of reality and will not make perfect predictions, which makes uncertainty inevitable. As a result, it can be more useful to convert estimates to probabilities. This enables us to answer questions such as, “What is the chance that the Democrats retake the House?”
In short, what we do is lots of simulations based on our model’s results. (In statistical terms, we are sampling from the error distribution of the model.) The uncertainty underlying the model means that simulations will generate a range of forecasts. The question is, how many of those simulations forecast that the Democrats would win 218 or more seats?
The answer is very few. Just over 1 percent, in fact. This is a testament to the fact that current conditions in the country, and the presence of so many Republican incumbents, make it hard for the Democrats to pick up many seats. In order for that forecast to change measurably in the Democrats’ favor, the economy needs to grow more rapidly or President Obama needs to become more popular, or both. A few more Republican retirements and strong Democratic challengers wouldn’t hurt, either.
To be sure, this positive forecast for Republicans does not presage a “wave” in Republicans’ favor, as we argued in our first post. The National Journal’s Josh Kraushaar took issue with that, suggesting that certain statistics — like presidential approval and the House generic ballot — did suggest a Republican wave. Our forecast has already taken presidential approval into account.
We also estimated a different model that included the generic ballot in lieu of economic growth and presidential approval, relying on the last Gallup generic ballot reading for the elections when these polls are available (1972-2012). If we take the current Pollster average in the generic ballot, the model’s forecast is not much different — a Democratic loss of two seats, and only a 2 percent chance of gaining a House majority. Indeed, even when the Democratic share of the generic ballot peaked during the government shutdown, the model actually forecast no additional seats for the Democrats (and therefore none for Republicans either).
Kraushaar later said that he did not think any wave would “translate into as many House seats as 2006” because there are “fewer competitive seats.” That’s part of what our model is capturing, and it’s why the model suggests that, as of now, there will not be a wave. (Kraushaar also pointed to the Senate as the chamber to watch — and we agree. We plan to have a more formal forecast of the Senate in the new year. At the moment a GOP majority is absolutely a possibility.)
But as far as the House is concerned, 2014 is shaping up to be a status quo election. National conditions just aren’t that favorable to Democrats right now, at least when placed in historical context. If those conditions become more favorable, you’ll see seats become more competitive, funders on one side get excited, better challengers start to emerge, and so on.
A true partisan wave can overcome factors such as incumbency and district partisanship. In the months ahead, we’ll have a better sense of any changes in national conditions and of competition on the ground in districts — and thus our forecast may change. As of today, however, it’s not a good one for Democrats.