If the 2016 presidential vote is evenly split between the parties, which one is more likely to win the Electoral College and therefore the presidency? I estimate that the Democrats’ chances of winning the Electoral College vote are between 83 and 89 percent, giving them a significant advantage. This argument contrasts with those who are cautious of a Democratic advantage, such as Jonathan Bernstein and Harry Enten. The reason I predict such a significant advantage is because of ongoing, long-term trends altering the electoral outlook in a number of key swing states.
To make predictions for 2016, I analyzed how the popular vote margin (the Democratic minus the Republican percentage of the vote) compared to the national vote in every state from 1992 through 2012. I examined the states individually to detect any long-term trends. For example, while Oklahoma was already significantly more Republican than the nation in 1992, it steadily became even more Republican over time.
The key to the predictions for 2016 is taking these long-term trends into account. There is also uncertainty to factor in. There is uncertainty about the trajectories of the states. And, even if states were not changing, there would be uncertainty in predicting state outcomes due to idiosyncratic factors that cannot be anticipated in advance. Both types of uncertainty can be estimated and taken into account when predicting state popular vote shares and the eventual distribution of Electoral College votes. (I follow the approach developed by Andrew Gelman and Gary King in this article.)
In short, the uncertainty is incorporated by simulating thousands of hypothetical elections where the national party division of the vote is split equally and then counting the Electoral votes across the 50 states and the District of Columbia each time. In a given simulation, if the Democrats receive more than half the Electoral votes, they are the simulated winner. If the Republicans receive more than half the Electoral votes, they are the winner.
The distribution of Electoral votes for 10,000 sets simulations is shown below. Across the simulations the Democrats receive a majority of the Electoral votes 89 percent of the time. (0.5 percent of the simulations produced an Electoral College tie, 269-269.)
Even if the trends across the states slow to – on average – half their current rates, things still look very good for the Democrats. Under this scenario, the probability of the Democrats winning the Electoral vote is 83 percent. Taken together, these two sets of simulations suggest that if the national vote is evenly split, then the Democrats’ chances of winning the Electoral College vote are between 83 and 89 percent
As suggested above, the key to the Democratic advantage are the trends underway in some key states. While Oklahoma has moved significantly in the Republican direction, it was already strongly Republican. But, consider the 10 states whose 2012 presidential margin was within five points of the national margin. In five of them (Florida, Iowa, Minnesota, Ohio and Pennsylvania) the trends are modest in size and hard to separate from random noise. In Colorado, a more significant trend in the Democrats’ favor appears underway, but it has been uneven, which makes predictions for Colorado in 2016 more uncertain. In the remaining four states (Nevada, New Hampshire, Virginia and Wisconsin), the trends are clearer, more substantial, and all favor the Democrats.
Virginia is especially notable. In every presidential election from 1952 through 2008, the presidential margin in Virginia was more Republican than the national presidential vote. But, the gap has been narrowing. In 2012, the gap closed completely with Obama beating Romney by 3.9 percentage points in Virginia, which exactly matched Obama’s national margin of victory. If the trend continues, then in 2016 Virginia would be expected to be more Democratic than the national vote. This implies that if the national vote were split evenly, then Virginia would be expected to give its Electoral College votes to the Democratic candidate.
This is what the simulations suggest. Across the 10,000 simulations the average Democratic margin in the Virginia is +2.2 percent more Democratic than the country, with the Democrats winning it 85 percent of the time. Even under the more conservative set of simulations based on the notion that the movement in Virginia and other states slows to half the current rates, Virginia’s Electoral College votes go to the Democrats 69 percent of the time.
In future posts I’ll examine the underlying details of the Electoral College math, but for now, it looks like the Democrats might have a sizable “structural” advantage heading into 2016.