On the one hand: Everything. When a candidate loses by such a close margin, absolutely everything could have made the difference. Had Clinton pulled resources from states that didn't wind up close and applied them to Pennsylvania, Wisconsin, and Michigan; had Clinton handled the email controversy a little bit better; had she run better ads; had Taylor Swift endorsed her. Even most of the things that political science hasn't been able to show make any difference at all might have been enough. Take, for example, early ads. Political scientists have found that the effects of ads fades rapidly, so that it's impossible to demonstrate that ads run weeks (or months or years) before an election have any eventual effect on vote choice at all. But the methods used are unable to distinguish very small effects from zero, and (quite sensibly) the norm is to treat anything which doesn't have a sufficiently demonstrated effect as if there's no effect at all. And yet when we're talking about extremely narrow margins such as the 2000 and 2016 presidential elections, it's quite possible -- I'd go so far as to say likely -- that some of those not-proven effects really did matter.
That's the context in which to assess claims about James Comey's investigation, especially the choices he made at the very end of the election, and of the press coverage of that investigation. In an election in which minor candidate gaffes or, for all we know, such unlikely suspects as billboards and lawn signs might have been enough to make a the difference, then of course Comey's letter to Congress about the investigation of a Clinton aide's laptop computer and the subsequent over-the-top press frenzy about it likely cost Clinton the election. And of course the media's year-long obsession with the email story at the expense of anything else Clinton said or did -- along with a press strategy by the candidate and her campaign which failed to convince the media to cover her popular policy ideas -- cost her the election.
But what of the argument by Clinton's critics that the election never should have been close?
That's where we get to the second answer to "What Happened?": Nothing happened at all. The idea that Clinton "should" have won easily was an illusion, and in fact the Republican candidate should have been considered the likely winner:
Political scientists and others have put in a great deal of effort to understand how general elections work. And the key variables such as the economy, the current president's popularity, and the length the incumbent party has been in the White House combined to predict a small win for the Republicans. Since in fact Clinton won the total vote by about 3 million votes and 2 percentage points, what needs to be explained is why she did so well, not why she did so badly.
I should back up a bit and some important caveats. There are multiple models, and they aren't perfect. Usually, looking at some combination of all the models is the best guide to what's really happening, but it's always possible that a true fundamentals model would have revealed that Clinton was a small or even a solid favorite who ran behind where she should have been.
And then there's the Trump factor. It seems intuitive to most people (myself included) that Trump should have been a weaker candidate than any generic Republican. But there's simply no way to objectively estimate how much weaker. Yes, we've seen buffoons at the state level finish well behind what a competent candidate would have done, but it's far from clear that Trump was equivalent to a Todd Akin or a Christine O'Donnell. And presidential elections may just be different. Republican party actors were willing to give up on Akin in order to win other contests. But not only is there no other contest that's nearly as important as the White House, but -- more importantly -- abandoning the presidential candidate is a formula for downballot devastation anyway. There's no triage when it comes to presidential elections.
It may not be very satisfying to answer "What Happened?" by saying that nothing happened; all the hoopla of a normal presidential campaign plus all the mishegoss unique to the 2016 contest all added up to Hillary Clinton doing a bit better in the national vote than she "should" have done while losing anyway thanks to a flukish distribution of the votes across the states. But it might well be correct.
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
Jonathan Bernstein is a Bloomberg View columnist. He taught political science at the University of Texas at San Antonio and DePauw University and wrote A Plain Blog About Politics.
The math holds up there is an observed third-term effect, holding everything else even. But with so few observations, it's quite possible that what's really happening here is no effect at all, with an illusion created by some offsetting, not-measured, effect. That's true for everything in these models, but the theory is in my view stronger for (say) economic effects, and so I'm less skeptical that the observed relationship is bogus.
For more columns from Bloomberg View, visit http://www.bloomberg.com/view.