I have been rather harsh on my fellow political scientists regarding Donald Trump. None of our prominent models either predicted his success, nor even permitted it, and few of us put his chances of winning at anything significantly greater than zero. But let's be honest here. Bernie Sanders has also outperformed our models.
Bernie Sanders will lose. But, if prominent models were to be believed, Hillary should have crushed him like a drunken frat boy crushing a cheap beer can. Instead, she will limp to the finishing line.
What's going on here? Let's start with my favorite whipping boy, The Party Decides. I... don't like this book, and never did. The basic argument is that party elites control the nomination process by endorsing and directing money to their favorite candidates, thereby signaling to voters whom they should support. There is a logic to it. The primary determinant of vote choice is party identification, but in a primary, all of the candidates have the same party label, so voters don't necessarily know what to do. Following elite cues makes sense.
Except that the model never really worked, empirically. As I explained here, party elites frequently fail to get the kinds of presidential nominees they want anyway. And then there's that mechanism I mentioned. Voters in a primary don't know what to do because they don't have their normal cue: party labels. Except that for the Democrats in 2016, they did. Hillary is a Democrat. Sanders "caucuses" with the Democrats in the Senate, but has run for election and reelection in Vermont identifying as an "independent" candidate. Is he functionally a Democrat? Yes, but the party label cue should advantage Hillary.
But of course, that's not what The Party Decides is all about. The authors of the book argue that endorsements control the game. Well, take a look at this. In endorsements, Hillary is crushing Bernie. And yet, she is only going to limp to the finishing line. Sorry, Party Decides-apparatchiks, this shouldn't be happening.
Most other models, though, have little to say about Hillary v. Bernie. Hillary is in the middle of her party ideologically, but Bernie, while far to the left, isn't outside the range of what Democratic primary voters can accept. They split the early states like Iowa and New Hampshire, so momentum could carry either to victory. There just aren't all that many other coherent models relevant after that.
And yet, even a Party Decides refusenik like me thought Sanders would lose by a larger margin. Why? He just... stands out as a weird candidate. His connection to the party has always been at least slightly tenuous, he is older and less polished than most successful candidates... There's the "income inequality" thing, but as I argued here, the division between the parties has been about redistribution since FDR, regardless of whether or not people use the slogan, "income inequality." Why has Bernie out-performed? Eventually, we need to figure that out.
Bernie will lose. But he will lose narrowly. He should have been crushed. This isn't a failure on the scale of underestimating Trump, but we need to acknowledge it. And people should stop reading The Party Decides unless they are doing an investigation into the history of failed theories.