For decades political scientists have tried to predicatively model presidential elections. Until recently, these models have relied on what many refer to as the “fundamentals,” factors like war and economic growth to predict the outcome of elections. As predictive analytics has advanced, those earlier fundamental-based models have largely been debunked for their weak correlation to actual election results (Nate Silver provides a good explanation of the why, here). In their place has grown a new means of election prediction rooted in harnessing publicly available polling data.
But while polling is indeed critical (as a pollster myself, I say that not for selfish reasons!), fundamentals still hold an important place in understanding why a candidate rises, falls, or, in the case of Bernie Sanders, plateaus. The accepted conventional wisdom of this Democratic primary has been that Sanders started from nothing and built a juggernaut campaign that threatened to take down the establishment’s preferred candidate, Hillary Clinton. In fact, it’s an oft touted point in Sanders’ stump speech: that he started off at 3 percent in the polls and was able to close the race to a draw.
So what can explain this supposed meteoric rise, and was there anything that could have predicted it ahead of time? In fact, there is: “the fundamentals.”
A few months ago I built out a model using a set of predictors from past Democratic primaries. It was pretty straight-forward, with state-based predictors including percentages of white men, white women, African Americans, self-identified Democrats, self-identified Liberals, and non-college white voters in the electorate.
My goal at the time was pretty simple. I set out building the model to provide a rough idea of the primary outcome in states like Rhode Island, where polling was sparse. And through the primaries it has effectively done just that: correctly predicting Sanders’ 5-point win in Indiana (where public polling had Clinton up by an average +6.8%), his “surprising” win in Michigan, and his narrow loss in Kentucky.
But as the Democratic primary comes to a close, the model is better viewed as validating a more important point: that Bernie Sanders (or any candidate running with his platform) was always going to pull roughly 40% of the national Democratic primary vote against Hillary Clinton. Or so the demographic fundamentals argued. On the flip side, Sanders was fundamentally capped in his nomination quest by those same demographic-based structural realities. This time, from the disadvantages he faced in the electorate: mainly, his weakness among non-white voters.
What’s most noticeable from the model is that it casts doubt on a core Sanders campaign narrative: that their support has been constantly growing as the electorate gets to know him and his platform. And if you were to look solely at national public polling, you might come away with the same conclusion.
The problem with the argument, though, is that Sanders performed similarly against the demographically-based predictive vote model in both early states (New Hampshire and South Carolina) and the most recent set of primaries. A candidate who was growing their support over time would begin to outpace such a model, and Sanders has failed to do that. What does this mean? Well, put simply, Sanders has maintained his core base of younger, white voters, but has fundamentally failed to remedy his disadvantages among voters of color.
Sanders deserves credit for being able to build a campaign that allowed him to reach his maximum feasible support in the Democratic primary. But fundamentals are that for a reason: you either harness a winning demographic coalition or you find a way to change the demographic makeup of the electorate.
Sanders was not able to do either and that, more than anything to do with the much-hawked “process,” is the reason he is not the presumptive Democratic nominee.