The Washington PostDemocracy Dies in Darkness

New Election Lab forecast suggests 86 percent chance that GOP wins Senate

Election Lab, our midterm elections forecast, has been updated with a host of new features, including — most importantly — an updated forecast. Our model currently gives Republicans an 86 percent chance of winning a Senate majority.

Among Election Lab’s new features are a new balance-of-power bar that compares the forecast to the current Senate make-up, a view of both the House and Senate map that isolates seats expected to change hands, drop-down menus that take you to any race, and the ability to share individual races via social media.

The updated forecast for the House is nothing surprising. In line with what we’ve previously written, we estimate that the Democrats have less than a 1 percent chance of taking the House. Our model currently estimates that the Democrats will win 193 seats, down slightly from the 201 they controlled after the 2012 election and the 199 they currently control, given existing vacancies. We expect to update this forecast with additional data about the candidates once the primaries are over, and with polling data as well. But, given how strong the Republicans’ position is, we would be surprised if any new information significantly altered the strong odds of continued Republican control.

The updated forecast for the Senate is perhaps more striking, of course. Our Senate model includes the same factors noted previously, but now also includes a polling average from various races that currently have sufficient polling data. Last week I described what we have learned about early Senate polls and how we combine the model and forecast. (We are indebted to Mark Blumenthal and the folks at Pollster for making these polling data available to us for this purpose.)

By itself, our forecasting model has always indicated that the GOP had a good chance of retaking the Senate. Nothing in the political landscape has shifted the model’s forecast. National conditions continue to provide headwinds for Democratic candidates. Most importantly, President Obama’s approval rating continues to be middling at best, and may even have declined slightly. Senate primaries have not yet produced the sorts of Republican candidates that arguably cost their party Senate races in 2010 and 2012.

There was a time, though, when the polling data suggested more GOP vulnerability. Consider, for example, the Kentucky Senate race between Republican Mitch McConnell and Democrat Alison Lundergan Grimes. In late 2013 and the first few months of 2014, polls suggested the race was tied. But now the polling has begun to line up more cleanly with the forecast. McConnell has opened up a narrow lead that, in combination with the model, is sufficient for us to forecast a Republican victory there.

Thus, although current polls adjust our model’s forecast for individual races, the polls do not change our topline prediction, which is currently bullish for the GOP.

Why so bullish? Here’s an explanation in a nutshell. Most analysts give the GOP a very good shot at controlling at least 46 seats. See the list over at The Upshot. Control of the Senate depends on nine apparently competitive seats: Alaska, Arkansas, Colorado, Georgia, Iowa, Kentucky, Louisiana, North Carolina, and Michigan. The GOP needs to win at least five of those seats to control a majority (since the Democrats would presumably control the Senate with 50 seats, given Vice President Biden’s tie-breaking vote).

At the moment, our model suggests that the GOP has a very good chance of winning the Republican-leaning states: Alaska, Arkansas, Georgia, Kentucky, and Louisiana. That gives them five seats. They also have a better than 50-50 chance of winning Iowa, where Joni Ernst’s recent surge has made the race neck-and-neck—a trend that is consistent with what our model suggested about the Iowa race back in May. Meanwhile, Democrats have a good chance of winning Colorado, Michigan, and North Carolina.

Our forecast in states like Arkansas, Georgia, Kentucky, and Louisiana gives the GOP a much better chance than many observers do. These races are toss-ups according to the Cook Political Report, for example. The reason is that our model is very confident of a GOP win in all three campaigns, and the polls do not give us enough reason to question this for now.  At the same time, the Democrats have strong candidates in these races, so it is possible that the prediction could shift in their favor. However, absent a clear trend toward the Democrats in the polls, our forecast will continue to favor the GOP in these races.

Our model’s forecast may not reflect what current polls suggest in every race. But, as we noted last week, there is at least some chance that the polls will move toward what a very basic forecasting model would suggest: namely, that it is hard for Democrats to win seats in a midterm year when a Democratic president isn’t that popular and when they must contest seats in so many Republican-leaning states.  If the polls don’t move toward our models’ predictions, then as Election Day gets closer our predictions will shift and give more weight to what the polls indicate is likely to happen rather than the models.

Now, some caveats. First, as The Upshot’s Nate Cohn noted last week, there is certainly reason to take the Senate polls with a substantial grain of salt. This is one reason why we think it makes sense to combine a forecasting model with polling averages at this point in time.

Second, this is, as always, a forecast based on conditions today. If things shifted in the Democrats’ favor in one or two of these races that currently lean toward the Republicans, then Democrats could retain the Senate. Given the large number of competitive races, this is certainly a possibility. If this happens, our model will update accordingly.

But for now, both our model’s prediction and current polls suggest the GOP has a very solid chance of taking back the Senate.