There is a strong consensus among the election forecasting models: Republicans have a good chance of taking back the Senate.  The Election Lab forecast — the most optimistic for the GOP — puts that probability at 96%.  We’ll have a final recap of our forecast tomorrow morning, but in the meantime here is how we got to this point.  In graphs.


The president’s party tends to lose seats in midterm elections. Period.  The graph above is for the House but the same applies to the Senate.  The logic of midterm loss is laid out nicely in this paper by Robert Erikson.  Part of the story is that the president isn’t on the ballot, and he often supplies coattails for congressional candidates to ride on.  Part of the story is that the electorate may react against the ideological direction of the president’s party–as it has since Obama took office–and seeks to elect candidates from the other party.  In any case, the “midterm penalty” is important.  In elections from 1980-2012, we estimate that this penalty is 3 points of vote share.


Then there is the president’s popularity.  As political scientist Lynn Vavreck noted Sunday, “The economy elects presidents.  Presidents elect Congress.”  The graph above shows that Obama’s first quarter approval rating was the third lowest in recent midterm elections, as Ben Highton wrote about here.  (It hasn’t improved since then.) The main reason is that, although third quarter economic growth was relatively strong and the unemployment rate has decreased, other economic indicators are not as rosy — see the graph of real disposable income in Ben’s post — and economic confidence has been stagnant for most of 2014 except for a very recent uptick.


The Republicans also have a stroke of luck.  The Senate seats up for reelection this year are in disproportionately Republican states, as Ben Highton noted here.  The graph above shows how the 2012 presidential vote in each “class” of Senate seats compares to the national average.  Class 2 — this year’s class — is the most Republican.  2016 will present Democrats with a more favorable landscape, but for now they must compete on difficult turf in states like Alaska, Arkansas, Louisiana, Montana, South Dakota, and West Virginia.  Barring a Democratic pick-up, those red states along are enough to give the Republicans the Senate.

We noted nine months ago that taking account of these fundamentals made the Republicans favored to take the Senate.  The quality of the candidates that the GOP has recruited only increased their odds of winning.  In fact, when we debuted Election Lab on May 5, we estimated at that point that the GOP had a 77 percent chance of winning and was predicted to win 53 seats.  We predict 53 seats again today.  The only change is that Michigan and Colorado are flipped relative to that earlier forecast.

And so this leaves us with the last graph, courtesy of Vox.  Although the forecasting models differ in how certain they are in a GOP Senate majority, they see a very similar picture in terms of the most likely outcome: