Will the Democrats' advantage outlast Barack Obama's particular coalition? Photographer: Stan Honda/AFP/Getty Images
Will the Democrats' advantage outlast Barack Obama's particular coalition? Photographer: Stan Honda/AFP/Getty Images

After the 2012 election, I wrote that there seemed to be an emerging bias in the Electoral College that favored Democrats. I presented several possible alternative explanations, and predicted that political scientists would study the phenomenon. My buddy Ben Highton has done just that, and the results appear to be very good for the Democrats.

The idea of Electoral College bias is whether one party would have an advantage in electoral votes in a dead-even election. Bias is possible in a system of winner-take-all by state if one party’s votes are distributed more efficiently. Imagine, for example, that the U.S.had two kinds of states: overwhelmingly Democratic ones and dead-even ones. Under those circumstances, Democrats would “waste” votes in the solidly Democratic states, and Republicans would use their votes efficiently. In an extreme case, Republicans could win each of those toss-up states by a single vote and lose everywhere else by millions, and yet still win.

Political scientists studying earlier elections had found no structural Electoral College bias. For every New York where Democrats wasted lots of votes, there was a Texas where Republicans were inefficient. But there was no particular reason it worked this way, and now the pattern appears to be changing -- in favor of the Democrats.

Ben looked at the long-term trends (1992-2012) in each state, and found that Democratic states are moving toward the Democrats; Republican states are moving even more sharply toward the Republicans and, most critically, swing states are moving toward the Democrats. As a result, in those 14 swing states, only three (North Carolina, Ohio and Florida) are more likely to go Republican than Democratic in 2016 if the election is a dead heat. Moreover, comparing 2012 and 2016, only Pennsylvania and Minnesota are trending toward Republicans; the others are all moving in the Democrats' favor (Oregon and Michigan both have a less than 1 percent chance of lining up Republican in a 50/50 election in 2016, which is unchanged from 2012).

By adding it all up, including both uncertainty about the trends and about any particular state in any particular year, Ben estimates the odds of Democrats winning a dead-heat election in the Electoral College at more than four in five. Put another way, Republicans would have to win the popular vote by “between one and two percentage points” to have an even chance of winning the Electoral College.

So how solid is the finding?

First, there’s no mechanism identified for the trend, so there’s no particular reason it couldn’t tail off or even reverse.

Second, the concept of Electoral College bias is built around the idea of a 50/50 electorate, which isn’t necessarily accurate. Imagine, for example, that every Democrat in Texas flipped parties, and everyone else stayed the same. Then, Republicans would have a natural underlying majority in the nation, and the Electoral College would be unaffected. But if you (mistakenly) assumed a dead-heat electorate, with all else being equal, then this type of analysis would find an Electoral College bias, even though the chances of either party winning were completely unchanged. That seems unlikely to be happening, given that Democrats have won the popular vote in four of the five last elections and that no obvious Republican voting majority has been apparent in other contests.

Third, it’s still possible that something particular to Barack Obama’s electoral coalition is at work here, but the trend analysis doesn't just look at the last two or three elections (as I did back in 2012). The same probably goes for the possibility that a temporary advantage in electioneering creates an apparent bias (by systematically pushing swing states toward the Democrats). That wouldn't explain the long-term trends, either.

I’m increasingly convinced this is something real, and it’s a pretty big deal. As Ben says, that large a bias would almost certainly have flipped the 2000 election to the Democrats; other elections close enough to have been affected by a bias this large would include 1976, 1968, and 1960 (if the losing party had been helped by an Electoral College bias of this size).
If these results hold up through 2016, expect the parties to begin flipping their positions on the Electoral College, perhaps very rapidly.

To contact the author on this story:
Jonathan Bernstein at jbernstein62@bloomberg.net

To contact the editor on this story:
Max Berley at mberley@bloomberg.net