Nobody knows what would’ve happened had Bernie Sanders been the Democratic nominee in 2016. My guess based on my reading of the political science literature following Steven Rosenstone’s classic 1983 book, Forecasting Presidential Elections, is that Sanders would’ve done a bit worse than Hillary Clinton, because Clinton is a centrist within the Democratic party and Sanders is more on the ideological extreme. This is similar to the reasoning that Ted Cruz, as the most conservative of the Republican candidates, would’ve been disadvantaged in the general election.
But I disagree with Kevin Drum, who writes, “Bernie Sanders Would Have Lost the Election in a Landslide.” Drum compares Sanders to failed Democratic candidates George McGovern, Walter Mondale, and Michael Dukakis—but they were all running against incumbent Republicans under economic conditions which were inauspicious for the Democratic opposition.
My guess would be that Sanders’s ideological extremism could’ve cost the Democrats a percentage or two of the vote. So, yes, a priori, before the campaign, I’d say that Hillary Clinton was the stronger general election candidate. And I agree with Drum that, just as lots of mud was thrown at Clinton, the Russians would’ve been able to find some dirt on Sanders too.
But here’s the thing. Hillary Clinton won the election by 3 million votes. Her votes were just not in the right places. Sanders could’ve won a million or two votes less than Clinton, and still won the election. Remember, John Kerry lost to George W. Bush by 3 million votes but still almost won in the Electoral College—he was short just 120,000 votes in Ohio.
So, even if Sanders was a weaker general election candidate than Clinton, he still could’ve won in this particular year.
Or, to put it another way, Donald Trump lost the presidential vote by 3 million votes but managed to win the election because of the vote distribution. A more mainstream Republican candidate could well have received more votes—even a plurality!—without winning the electoral college.
The 2016 election was just weird, and it’s reasonable to say that (a) Sanders would’ve been a weaker candidate than Clinton, but (b) in the event, he could’ve won.
P.S. Drum responds to my points above with a post entitled, “Bernie Woulda Lost.” Actually that title is misleading because then in his post he writes, “I won’t deny that Sanders could have won. Gelman is right that 2016 was a weird year, and you never know what might have happened.”
But here’s Drum’s summary:
Instead of Clinton’s 51-49 percent victory in the popular vote, my [Drum’s] guess is that Sanders would lost 47-53 or so.
Sanders would have found it almost impossible to win those working-class votes [in the midwest]. There’s no way he could have out-populisted Trump, and he had a ton of negatives to overcome.
We know that state votes generally follow the national vote, so if Sanders had lost 1-2 percentage points compared to Clinton, he most likely would have lost 1-2 percentage points in Wisconsin, Michigan, and Pennsylvania too. What’s the alternative? That he somehow loses a million votes in liberal California but gains half a million votes in a bunch of swing states in the Midwest? What’s the theory behind that?
OK, there are a few things going on here.
1. Where does that 47-53 estimate come from? Drum’s saying that Sanders would’ve done a full 4 percentage points worse than Clinton in the popular vote. 4 percentage points is huge. It’s huge historically—Rosenstone in his aforementioned book estimates the electoral penalty for ideological extremism to be much less than that—and even more huge today in our politically polarized environment. So I don’t really see where that 4 percentage points is coming from. 1 or 2 percentage points, sure, which is why in my post above I did not say that I thought Sanders necessarily would’ve won, I just say it could’ve happened, and my best guess is that the election would’ve been really close.
As I said, I see Sanders’s non-centered political positions as costing him votes, just not nearly as much as Drum is guessing. And, again, I have no idea where Drum’s estimated 4 percentage point shift is coming from. However, there is one other thing, which is that Sanders is a member of a religious minority. It’s said that Romney being a Mormon cost him a bunch of votes in 2012, and similarly it’s not unreasonable to assume that Sanders being Jewish would cost him too. It’s hard to say: one might guess that anyone who would vote against someone just for being a Jew would already be voting for Trump, but who knows?
2. Drum correctly points out that swings are national and of course I agree with that (see, for example, item 9 here), but of course there were some departures from uniform swing. Drum attributes this to Mitt Romney being “a pro-trade stiff who was easy to caricature as a private equity plutocrat”—but some of this characterization applied to Hillary Clinton too. So I don’t think we should take the Clinton-Trump results as a perfect template for what would’ve happened, had the candidates been different.
Here are the swings:
To put it another way: suppose Clinton had run against Scott Walker instead of Donald Trump. I’m guessing the popular vote totals might have been very similar to what actually happened, but with a different distribution of votes.
Drum writes, “if Sanders had lost 1-2 percentage points compared to Clinton, he most likely would have lost 1-2 percentage points in Wisconsin, Michigan, and Pennsylvania too. What’s the alternative? That he somehow loses a million votes in liberal California but gains half a million votes in a bunch of swing states in the Midwest? What’s the theory behind that?”
My response: The theory is not that Sanders “loses” a million votes in liberal California but that he doesn’t do as well there as Clinton did—not an unreasonable assumption given that Clinton won the Democratic primary there. Similarly with New York. Just cos California and New York are liberal states, that doesn’t mean that Sanders would outperform Clinton in those places in the general election: after all, the liberals in those states would be voting for either of them over the Republican. And, yes, I think the opposite could’ve happened in the Midwest. Clinton and Sanders won among different groups and in different states in the primaries, and the gender gap in the general election increased a lot in 2016, especially among older and better-educated voters, so there’s various evidence suggesting that the two candidates were appealing to different groups of voters. My point is not that Sanders was a stronger candidate than Clinton on an absolute scale—as I wrote above, I don’t know, but my guess is that he would’ve done a bit worse in the popular vote—but rather that the particular outcome we saw was a bit of a fluke, and I see no reason to think a Sanders candidacy would’ve seen the same state-by-state swings as happened to occur with Clinton. Drum considers the scenario suggested above to be “bizarre” but I think he’s making the mistake of taking the particular Clinton-Trump outcome as a baseline. If you take Obama-Romney as a starting point and go from there, everything looks different.
Finally, writes that my post “sounds like special pleading.” I looked up that term and it’s defined as “argument in which the speaker deliberately ignores aspects that are unfavorable to their point of view.” I don’t think I was doing that. I was just expressing uncertainty. Drum wrote the declarative sentence, “Bernie Sanders Would Have Lost the Election in a Landslide,” and I responded with doubt. My doubt regarding landslide claims is not new. For example, here I am on 31 Aug 2016:
Trump-Clinton Probably Won’t Be a Landslide. The Economy Says So.
I wasn’t specially pleading then, and I’m not specially pleading now. I’m just doing my best to assess the evidence.