Clinton, Trump, and the Median Voter Theorem

Since Donald Trump defeated Hillary Clinton in last Tuesday's presidential election many have wondered how the polls failed to predict a Republican win. On Tuesday FiveThirtyEight.com gave Clinton a 71% chance of victory, the New York Times gave her an 85% chance of victory, and the Huffington Post gave her a 98% chance of victory. I think that the Median Voter Theorem provides an answer to questions of how most major polls overestimated Clinton's chances of winning.

According to MVT, the median voter is decisive, and parties therefore seek to win the median vote. In this election major media outlets may have misunderstood distribution of American voters and therefore misjudged the location of the median voter. Polls may have assumed that American voters had a relatively normal distribution (as in the graph below), or even assumed that the voter distribution skewed to the left. The graph below shows how, in an assumed normal distribution, Clinton's policy's put her slightly left of center, while Trump's policies put him far to the right, approaching the right tail of the distribution. By this model she would have won the vast majority of votes to the left of the median and encroached quite far into the right side of the distribution. Trump would then be left with only the votes to his right (in the relatively insignificant tail of the distribution), and a portion of the votes to his left that does not reach the median. In this scenario it makes sense to think that Clinton was on track for a comfortable victory. But the results did not play out as expected, and the median voter is likely not where anyone thought was.

We can never know the exact location of the median voter, but Trump's victory suggests that it is likely much farther to the right than most pundits realized. This would mean that the distribution skews to the right (it may also be bi- or multimodal, but in either case it skews right). Under MVT even in a skewed distribution the median voter will be decisive. If we assume that candidates and parties are flexible in their policies and want above all to win votes, then I think it is fair to predict that we will see a shift in the policies of both parties over the next four to eight years. The Republican Party will adjust to appeal to the voters that are right of its traditional establishment, but I expect the Democratic Party to take efforts to win these votes as well. Whether through electing a more populist candidate in the future (someone like Bernie Sanders), or moving policies to the right, I think we can expect these election results to reshape the core stances of both parties.