Every election cycle is different. Each has a unique political environment and set of circumstances, all of which keep elections interesting. Aside from the extremely competitive fight over the Senate, the biggest change this time around is the multitude of Senate forecasts using quantitative election models of various types; these new players are joining the game alongside the more traditional qualitative — or maybe a better term is "multidisciplinary" — approach that has been around forever.
While political scientists, occasionally joined by economists, have modeled elections for years, their findings were largely ignored outside academia. In the last decade, however, stats whiz Nate Silver — who shifted his FiveThirtyEight.com site from The New York Times to an affiliation with ESPN — turned his sights from analyzing baseball to politics with considerable success, starting a cottage industry of political modeling for public consumption. This was followed by The Washington Post teaming with a group of political scientists on the Monkey Cage and The New York Times creating its own in-house team with a feature both in the paper and on the website called The Upshot under the direction of David Leonhardt. The models generally incorporate past election results in each state, the value of incumbency, and in some cases campaign fundraising and public polls, all based on past history, to project what may happen.
Already on the scene was my good friend and competitor, Stu Rothenberg, and his colleague Nathan Gonzales at The Rothenberg Political Report. Joining them are our friends and competitors down the road in Charlottesville at the University of Virginia, Larry Sabato and his colleague, Kyle Kondik, and their team at Sabato's " Crystal Ball," as well as my team at The Cook Political Report. These latter three obviously look at polls, money, and past election results, but — relying upon experience and judgment rather than number-crunching alone — they also more subjectively look at candidate and campaign quality and other unique circumstances that might be difficult to quantify.
Rothenberg, as well as Sabato, Kondik, and their Crystal Ball team, are currently projecting a GOP gain in the Senate of between four and eight seats. The Times's Upshot uses the same four-to-eight range, then narrows it to a six-seat gain as the most likely outcome. The Cook Political Report says that a GOP gain of between four and six is most probably but expects the final number to more likely go higher than lower. Jennifer Duffy, The Cook Political Report's senior editor, is putting Republican chances of taking a majority at 50-50. Personally, I am a bit more bullish on the GOP's chances, putting it closer to a 60 percent chance. Silver and Harry Enten at FiveThirtyEight will switch to the algorithm-driven model later this summer, but their July 15 informal assessment, based on basically the same factors that go into their full-blown model, point to a GOP gain of 5.7 seats, which rounds up to a six-seat gain, the minimum gain necessary for the GOP to achieve a majority this year.
As I look through the underlying forecasts in individual races, my hunch is that the quantitative approaches work very well in the aggregate, in terms of coming up with overall forecasts — although I wonder about some of the forecasting at the level of individual races. One forecaster might rate a Democrat's chances a bit higher than I would in one race, and a Republican's better than I perceive in another. Average it all together, though, and they pretty much make sense and don't differ greatly from the assessments of veteran election-watchers or from what most party strategists for each side privately say.
The outlier, however, is The Washington Post's Monkey Cage feature, the Election Lab. As of Monday, it rates the Republicans' chances of taking the Senate at 87 percent, and it is giving a number of races that the other forecasters expect to be very close or at least reasonably close some awfully large odds of going Republican, although in a couple of cases Democrats are rated far better off than they are by others. Monkey Cage gives Republicans a 79 percent chance of winning the Iowa open seat, an 81 percent chance of beating Mark Pryor in Arkansas, an 82 percent chance of defeating Mary Landrieu in Louisiana, and a 98 percent chance of winning the open Georgia seat.
Going the other way, the Monkey Cage estimates Kay Hagan has a 97 percent chance of winning reelection in North Carolina, Jeff Merkley a 99 percent chance in Oregon, Al Franken an 89 percent chance in Minnesota, Jeanne Shaheen a 99 percent chance in New Hampshire, and Mark Warner a 99 percent chance in Virginia.
We'll see what happens. But as professor Sabato is fond of saying, "Those who live by the crystal ball end up eating ground glass."