Our new Senate forecast gives the Republicans a 77 percent chance of a Senate majority. That's down slightly from our previous estimate, which pegged the GOP's chances of a takeover at 82 percent. So, what's changed?
The new projection reflects the conclusion of several primary elections -- North Carolina, Nebraska, and West Virginia -- and, more importantly, the inclusion of candidate fundraising through the first quarter of 2014 and, for some races, through the end of April. (The relevant Federal Election Commission data are here.)
Our earliest forecast showed that Republicans were already heavily favored due to the national landscape and the partisan complexion of the states holding Senate elections this year. We then showed that incorporating a measure of the "quality" of the candidates -- prior experience in elective office -- made things even more favorable to Republicans. Chris Cillizza and I discussed that forecast here. As we would expect, Republicans are recruiting and nominating relatively experienced and therefore more electable candidates.
Now, with fundraising in the model, the results are marginally more favorable to Democrats, but not by much. This means that Democrats are mustering some advantages in fundraising, but not particularly large ones.
Because so many states have not had primaries, we measure fundraising by summing up all the fundraising by Senate candidates in each party. We expect that many donors to primary candidates who lose will likely end up supporting the candidate who wins, as will donors who didn't contribute in the primary. For example, between January 2013 and the middle of April, North Carolina Sen. Kay Hagan (D) raised about $11 million. Her opponent, Thom Tillis, raised $3.2 million. Is Hagan likely to out-raise Tillis nearly 4-1 until November? Most likely not. In fact, our analysis of past elections suggests that at this stage party fundraising is a better predictor than fundraising by the primary candidates that go on to compete in the general election. (My collaborator and fellow political scientist Eric McGhee will say more about this in a post on the Monkey Cage Blog next week.)
By including fundraising, a host of states that were already looking good for Democrats now look even better. For example, in Oregon, Delaware, New Jersey, Minnesota, and New Mexico, the chances of a Democratic victory are 10 points greater than our last forecast. We estimate that in all these states the Democrats’ chances of winning are about 90 percent or better.
At the same time, our forecast in the key competitive races is less optimistic for Democrats. The model still sees Democrats as slight underdogs in Arkansas, Alaska, and Michigan. But "slight" is the operative word.
In Iowa, including fundraising in the model does improve the chances of Democrat Bruce Braley, but the model still sees Iowa as a 50-50 toss-up. The resolution of the primary there will likely increase Braley's chances, since the Republican nominee will have less prior political experience than the model currently assumes. (We use historical averages from open-seat races to generate an estimated level of candidate experience in races where the primary hasn't been held. All of the GOP Senate candidates in Iowa have less prior political experience than the historical average.) However, the model is still likely to see the race as fairly competitive.
Only in North Carolina is the model relatively favorable to the Democrats, giving Hagan a very solid shot at defeating Tillis. This largely reflects historical precedent: it is very rare for a state legislator to defeat an incumbent Senator. Of course, that doesn't mean it won't happen. But Tillis may have his own challenges too.
Also in contrast to current polls, the model sees two races as very long shots for Democrats: Republican-held seats in Kentucky and Georgia. Again, "long shot" does not mean "impossibility." But for Democratic candidates to win in red states given current national conditions -- and given the presence of an incumbent in Kentucky -- would be unusual.
A central reason our model gives the Republicans such a good chance of taking the Senate is the midterm penalty. Based on elections from 1980-2012, we estimate the midterm penalty to be 3 percentage points of the vote, on average. This penalty captures the well-known historical tendency for the president's party to lose seats in midterm years. The main question is how big the penalty will be in 2014. If the Democrats were to get lucky and only suffer a 1 percentage point midterm penalty, then we estimate that they would have about even odds (53 percent) of retaining a majority.
The midterm penalty – whatever it ends up being on Election Day – is typically not fully evident in polling this early. See, for example, this research by political scientists Joseph Bafumi, Robert Erikson, and Chris Wlezien. They conclude that "the electorate’s vote trajectory over the course of the campaign that follows is toward the out party" -- in 2014, toward the Republicans. So, we would take with a grain of salt polls in states like Georgia that show Democratic candidates making headway.
As always, the model's predictions could change -- as indeed they should, if conditions on the ground change. What we're giving you is a sense of where things stand today based on the factors in the model. We will continue to update the model as primary elections occur and as new fundraising numbers come in. We'll also build in polling numbers as more general election candidates become known. Stay tuned!