Skip to main content
Sanders — And The Media — Learned The Wrong Lessons From Trump In 2016

“The […] pitch is a mix of idealism and a shouting anger about the system, but at its heart is a hard-nosed math: He’s the only candidate with a sizable chunk of the electorate that won’t waver, no matter what, so a field that keeps growing and splitting support keeps making things easier.”

As I’m sure you’ve probably guessed, this passage seems like it could be about Donald Trump’s campaign for the Republican nomination in 20161 — but it’s actually about Bernie Sanders’s quest for the Democratic nomination this year. It comes from an April 2019 feature by the Atlantic’s Edward-Isaac Dovere that I consider to be a lodestar for Sanders’s strategy since it contains a lot of on-the-record reporting from Sanders campaign.2

The article also reveals how the conventional wisdom about how primary campaigns have changed, in light of Trump winning the Republican nomination in 2016 and Sanders’s better-than-expected performance against Hillary Clinton that year. In particular, it reflects several common assumptions in media coverage of the race this year:

  • In a divided field, the goal is to have the largest, most enthusiastic base.
  • Reaching out to different factions of your party is not that important — and may even dilute your differentiation from the other candidates. As long as you’re winning the plurality of votes, you’ll win enough states and the rest of the party will eventually come around.
  • There is no need to make peace with the party establishment. In our populist era, nominations are won from the grassroots up. Voters don’t care about things like endorsements, and “party elites” are largely feckless.

Indeed, if you’d beamed down to Planet Earth from some alien civilization in January 2016 and only seen that year’s primaries, those might be the lessons that you’d take away about how primaries work. They’re a reasonably apt description of the GOP primary, where the Republican establishment belly flopped in trying to stop Trump. The Democratic primary was a bit more complicated. Clinton beat Sanders by a rather clear margin in the end.3 Still, Sanders overperformed expectations by enough as to at least not contradict the notion that 2016 was the new normal.

But focusing on only what happened in the most recent election is a dumb way to do analysis. The media and campaign professionals spend endless amounts of time trying to draw “lessons” from elections because they are so consequential, but you can’t escape the fact that presidential elections happen only once every four years. No matter how many lessons you might try to draw from any one example, you’re still dealing with a sample size of one.

Rather, the history of the primaries — both in terms of how the media covers them and how campaigns wage them — is often one of fighting the last war and overcompensating for perceived mistakes from four years earlier. I’m not claiming to be immune from this either. One of the reasons that I initially bought so heavily into “The Party Decides” hypothesis in 2016 — which implied that Trump’s lack of support from the Republican party establishment would eventually doom him — was because of the Republican primary in 2012, when a series of insurgent candidates (Newt Gingrich, Herman Cain, Rick Santorum) briefly surged in the polls only to eventually lose to the slow-and-steady (and establishment-backed) Mitt Romney.

But Romney’s win in 2012 was in line with the longer history of the primaries, which underscores that building coalitions across the different wings of your party (Romney tried to unite conservatives and moderates, for instance) is usually a good strategy for winning instead of relying on just one faction. And as you can see in the table below, the establishment usually does win out in the end:

The party usually does decide, after all

Endorsement leaders before the Iowa caucuses where no incumbent president was running for that party’s nomination, 1972 to 2020

Year Party Endorsement leader before Iowa Did they have a clear endorsement lead? Did they win the nomination?
1972 D Ed Muskie
1976 D Fred Harris
1980 R Ronald Reagan
1984 D Walter Mondale
1988 D Dick Gephardt
1988 R George H.W. Bush
1992 D Bill Clinton
1996 R Bob Dole
2000 D Al Gore
2000 R George W. Bush
2004 D Howard Dean
2008 D Hillary Clinton
2008 R John McCain
2012 R Mitt Romney
2016 D Hillary Clinton
2016 R Jeb Bush
2020 D Joe Biden*

*Biden has not been formally nominated, but is the presumptive nominee.

Endorsement leaders from 1980 to 2016 are based on FiveThirtyEight’s 2016 system for calculating endorsement points. Endorsement leaders in 1972 and 1976 are based on “The Party Decides.”

Sources: “The Party Decides,” News accounts

Of course, a result like Romney’s win in 2012 that confirms your theories and priors can create its own issues — it can make you overconfident. You may mistakenly assume, for instance, that something is true 100 percent of the time when it’s actually only true, say, 80 percent of the time.

But while overconfidence is a serious problem, it’s probably better than assuming that an abnormal result (say, something that is true 20 percent of the time) is the new normal. You can see lots of this upside-down thinking in media coverage of elections since 2016. Because Trump won the general election as a (modest) underdog in 2016, a lot have inferred that underdogs not only can win, but that they usually win. Before the Virginia gubernatorial election in 2017, for instance, a lot of pundits predicted Republican Ed Gillespie would win despite being narrowly behind Democrat Ralph Northam in the polls. Northam won by 9 points instead.

It’s not just the media who makes these mistakes, however. It’s also the campaigns themselves. So while I don’t intend to do a long post-mortem of the Sanders campaign — instead, see my colleague Perry Bacon Jr.’s comprehensive analysis on what went wrong and why — I do want to take a quick inventory of the ways that the 2020 Democratic primary was different from the Republican primary in 2016.

Most of these problems were fairly predictable and known in advance. To ensure you that I’m not engaging in too much hindsight bias — because it’s always easier to say what went wrong with campaigns once you know who won and who lost — I’d point you toward this cautiously pessimistic take on Sanders from April 2019 and this cautiously optimistic one from just after the New Hampshire primary, which reflects how my thinking about the Sanders’s campaign evolved as it was happening in real-time. I started out thinking that Sanders’s factional, us-versus-them, screw-the-establishment strategy was a poor one. But it seemed to be going about as well as it possibly could after New Hampshire with the rest of the field divided.

So here are some potential flaws in Sanders’s strategy — and reasons why the precedent set by Trump’s win in 2016 might not have applied to his chances this year:

First, Democratic primaries and caucuses do not use winner-take-all rules, while many Republican states do. That makes it much harder for someone to be in a commanding position by narrowly winning primaries. Instead, in a multi-candidate race, you have to eventually broaden your coalition to avoid a contested convention.

In consideration of this, some elements of Sanders’s strategy simply never made a lot of sense. Here’s another key passage from that Atlantic article:

He’s counting on winning Iowa and New Hampshire, where he was already surprisingly strong in 2016, and hoping that Cory Booker and Kamala Harris will split the black electorate in South Carolina and give him a path to slip through there, too. And then, Sanders aides believe, he’ll easily win enough delegates to put him into contention at the convention. They say they don’t need him to get more than 30 percent to make that happen.

Sanders’s campaign was right that it could win around 30 percent of the vote with their base and that they could win early states with that amount … but it never really had a plan for what came after that. Dovere’s article makes it sound as though Sanders was actually counting on a contested convention, an awfully risky approach given that contested conventions are: (i) highly unpredictable and (ii) not likely to be favorable terrain for Sanders, given his lack of support from the establishment and that superdelegates would be allowed to vote on the second and subsequent ballots.

Second, Trump’s ideological ambiguity may have helped him in 2016, while Sanders got stuck in a “lane” this year. Trump may not have done that much traditional outreach to different parts of the GOP, but Republican voters from several wings of the party, perhaps not being quite sure what to make of a candidate like Trump who was reactionary on some issues (say, immigration) and quite moderate on others (say, gay marriage), were nonetheless Trump-curious. And that range of appeal was reflected in the primary results. In the Michigan Republican primary for instance, Trump won the support of 35 percent of “very conservative” voters, 37 percent of “somewhat conservative” voters and 37 percent of moderate voters, according to exit polls.

By contrast Sanders ran further to the left in 2020 than he did in 2016, with a message that shifted from economic populism to a broader and more “intersectional” leftism. This shift was reflected at the ballot box, too. For instance, Sanders won just 24 percent of moderates in the Michigan Democratic primary this year (but 63 percent of “very liberal” voters) — as compared 44 percent of moderates and 59 percent of “very liberal” voters in 2016.

Third, the importance that Democratic voters attached to “electability” and to beating Trump may have made them more willing to take cues from the establishment. Electability is a whole subject unto itself that I’m not going to do justice to here. I do think electability may be a teensy, tiny bit overrated as a reason for why Sanders lost. While perceptions of electability probably hurt Sanders relative to Joe Biden, it also probably helped him relative to candidates who weren’t white men: most notably, Elizabeth Warren.

Still, electability was important. In that Michigan exit poll, 58 percent of Democratic voters said they preferred a candidate that could beat Trump to someone who agreed with them on the issues — and 62 percent of the “beat Trump” voters went for Biden.

Fourth, Democratic party elites had an opportunity to learn from Republicans in 2016, and this may have made them more aggressive in backing Biden after South Carolina. One reason to be careful of declaring a “new normal” in any trend involving human behavior is that humans have an opportunity to learn from what happened last time and change course. The establishment’s failures in 2016 may have made partly elites reluctant to weigh in early on in the race this time around. But once they perceived Sanders as the true front-runner after Nevada, and Biden began to gain a little bit of momentum as he surged to a lead in South Carolina polls, party elites got behind Biden at extremely rapid rates. It’s hard to know if this would have played out the same way if there wasn’t 2016 to learn from.

Fifth, Trump’s success in 2016 may have reflected an unusually weak field of alternatives rather than a new paradigm. I’m not going to spend much time on this one, because it’s rather subjective. Were Warren, Biden, Pete Buttigieg, Kamala Harris, etc. better candidates than Ted Cruz, Marco Rubio, Jeb Bush, John Kasich, etc.? I don’t know. But Democratic voters were certainly much happier with their field of candidates in 2020 than Republicans were with theirs in 2016, which is some indication that the field might have been stronger.

Sixth, Democrats are a more racially-diverse party, which makes coalition building more essential than it is among Republicans. Sanders didn’t do badly with nonwhite voters — winning Nevada largely on the basis of Hispanic support while also winning the support of many younger black voters. That definitely made him a more plausible nominee than Buttigieg, whose support was extremely white, for example. Still, nonwhite support wasn’t a strength for Sanders in a head-to-head race against Biden, who generally beat Sanders by wide margins among black voters and held his own among Hispanics in many states. And Sanders’s outreach to African American leaders didn’t always go smoothly, with many members of the Congressional Black Caucus backing Biden instead.

Seventh, relying on a fixed base of voters can leave you vulnerable to a turnout surge from other candidates. One rather explicit assumption made by the Sanders campaign was that it could turn out more voters than its opponents. Hence, the campaign believed there was not much need to widen their circle to include others. Even their outreach efforts to Sanders-skeptical party leaders focused on trying to browbeat them into explaining why they didn’t support Sanders rather than explaining why they should support him. Again, from that Atlantic article:

That’s the case Sanders and his aides have been making as they’ve undertaken an outreach effort unlike anything from the last campaign, spending hours on phone calls trying to talk political leaders down from being completely opposed to him. They have a basic script: Start out asking where the support for Sanders is among their constituents, then ask why they think those people support him, and then ask the leaders to explain their own skepticism.

One problem with this attitude is that turnout in an election isn’t fixed — especially in primaries and caucuses where participation is low. You may turn out a certain number of people, but your base will not be constant as a percentage of the electorate if turnout goes up. In Michigan, for example, Sanders got nearly as many votes in 2020 (576,754) as in 2016 (599,000). But Biden got 838,555 votes in 2020, as compared to just 581,775 for Clinton in 2016 — part of a general pattern where turnout surged more in Biden’s best areas. So Sanders went from narrowly winning Michigan to losing it by 17 percentage points.

Perhaps Sanders was a victim of his own success in 2016. That is, he did such a good job of turning out voters in 2016 that it was hard to improve on his performance in 2020. Still, there was never really any recognition that his base alone might not be enough to win — and there were never many efforts to expand beyond his base. Indeed, it’s interesting that Sanders doubled down even more on his screw-the-establishment messaging amidst his success in the first three states — rather than trying to pivot to a more inclusive campaign and unify the party behind him.

Maybe, if a few things had gone differently, it would have worked. Sanders did outlast more than two dozen other candidates, after all. Maybe if Michael Bloomberg hadn’t qualified for the Nevada debate and stumbled so badly there, moderate support would have remained divided between Bloomberg and Biden, and Sanders would have had a lot more wins on Super Tuesday. Maybe Sanders’s “60 Minutes” interview hurt him. Maybe he could have gotten Warren to endorse him. Who knows.

But the Sanders campaign was always pursuing a strategy that claimed to defy the odds, that assumed the old rules no longer applied, and that stuck its thumb in the eye of the establishment. It would have been remarkable if it had succeeded. But it’s no surprise that it didn’t.


  1. Well, maybe apart from the part about “idealism.”

  2. Without, frankly, a lot of pushback, either from the reporter himself or from rival campaigns.

  3. Even though Sanders won a number of states in 2016 that didn’t translate into an especially close race for votes or delegates. Furthermore, at least some of Sanders’s success came from #NeverHillary voters who weren’t necessarily on board with Sanders’s platform but were just looking for some alternative to Clinton.

Nate Silver is the founder and editor in chief of FiveThirtyEight.