This is a guest post by political scientist Eric McGhee, one of the people behind our House and Senate forecasts.

A few weeks ago, we used our model of House elections to show the importance of the “midterm penalty” to our forecast.  All else equal, the president’s party usually does worse in midterm elections.  We showed that the penalty is large enough to prevent the Democrats from retaking the House under any reasonable set of national political conditions.

We have recently reported a forecast for a Democratic Senate that’s more pessimistic than most of the predictions out there now.  Over at, Harry Enten suggested that their own forecast—which leaned slightly toward a Republican takeover—might be too optimistic for Democrats, given that the midterm penalty was not yet part of their calculus.

What can we say about this idea?  How much is the midterm penalty really hurting the Democrats?

According to our model, the answer is: quite a bit.  The graph below replicates for Senate elections the one we created for the House.  Each line shows the odds of a Democratic Senate majority for various levels of Obama approval, first for a hypothetical 2014 that has a presidential election (the solid line) and then for the midterm election year it actually is (the dotted line).

The yawning gap between the two lines is immediately obvious.  At the current level of Obama approval, in a presidential election we would forecast a roughly three in four chance of Democrats holding the chamber.  But with the penalty, the probability is below one in five.

Unlike the House forecast, however, a Democratic Senate doesn’t seem as far-fetched.  The odds of a Democratic majority climb steeply with rising Obama approval, and reach about 50/50 at an Obama approval rating of 60%.  To be sure, we are not saying that the Democrats have little chance of maintaining a majority unless Obama’s approval rating hits 60%.  Instead, what this graph shows is how much more the playing field tilts against the president’s party in a midterm year.

But our forecast for the Senate is still fundamentally dependent on the dynamics of each race, more so even than our forecast for the House.  There are so few races that only a small number of outperforming Democrats (or underperforming Republicans) could shift the forecast and the result on Election Day.