Skip to main content
ABC News
A Sanders Comeback Would Be Unprecedented

Dear democratic socialists, political revolutionaries, Bern-feelers at large: We need to have a talk.

Let me begin by saying that I bear no ill will towards Mr. Sanders. Nothing that follows should be misconstrued as an attack on his policies, his track record, his electability in November or his character. I’m not a corporate media crony, or a plant from a pro-Hillary Clinton super PAC. I’m just a guy who believes in the predictive power of cold, hard data.

And the unsexy truth is that, barring some catastrophic news event, Sanders will not win the Democratic nomination for president in 2016. In fact, most past candidates in Sanders’s position dropped out long before this point in the race, and those who stayed in made little pretense of winning. (The Sanders campaign, which announced Wednesday it was laying off a ton of staff, may be recognizing this.)

Historically speaking, Democratic primary races do not have many twists and turns. Rather, the eventual winner tends to take an early lead — on or before Super Tuesday — and stay there. Runner-ups can kick for a while, but they tend to concede the race by February or early March.

As it stands, Sanders is firmly in runner-up territory. He is losing 9 million to 12 million among those who have already voted, and polls show him lagging by an average of 8.8 percentage points in the states yet to vote1. Sanders has gained substantially in national polls but is still the less popular candidate (outside of the Bernietopia that is social media2).


Sure, you may say, it’s unlikely, but he still could come back, right? I looked back through every Democratic3 race since 19724 to see if there’s any precedent for a late-stage political revolution. For each candidate, I charted what percentage of remaining delegates he would have needed in order to clinch the nomination5 at each point in the race — let’s call this a candidate’s “comeback score.” So each candidate starts at 50 percent, and as he wins (or loses) contests, his comeback score falls (or rises). Higher comeback scores are bad.

To be kind to the Sanders camp, I ignored superdelegates6 and demographics.7


The result is pretty striking: After the early days of the campaign, no underdog has ever won the Democratic nomination. A true come-from-behind victory would show up on this chart as a green line (winners) wandering above the 50 percent line (falling behind) before crossing back over (catching up) and veering toward the bottom of the chart. Instead, after the mad scramble for the first 10 percent of delegates, no candidate ever crosses over the 50 percent line. That is, the king stay the king. (Of course, there haven’t been that many Democratic primaries in the modern era, so I wouldn’t interpret this data as some type of iron-clad rule.)

The reason for this is pretty simple: Proportional allocation of delegates makes comebacks really, really hard. You can’t just notch wins in a string of states, as Sanders did in late March and early April. You have to start consistently trouncing your opponent by large margins in every contest. You need, well, a political revolution.

But what about Obama? Sanders supporters have compared their candidate’s current deficit to Obama’s in 2008, but at this point in that election Obama was actually winning by 143 pledged delegates — enough that Clinton, despite still holding a lead in superdelegates, was receiving pressure to drop out of the race. In fact, Obama was at no point in 2008 actually behind Clinton in pledged delegates. It’s just that the media usually included superdelegates in their counts in 2008, and the DNC has instructed them not to this time around. That’s because we’ve learned our lesson: Superdelegates can change their mind. Unfortunately for Sanders, pledged delegates can’t.

If Obama isn’t a good comparison for Sanders, who is? There’s no good answer to this question because most candidates in Sanders’s position dropped out long before this point in the race. The median comeback score for a candidate on the day he ends his campaign is 53 percent;8 Sanders is, as of today, over 58 percent. He crossed over the 53 percent mark on March 12, back when Marco Rubio was still in the Republican race.

To reiterate: I’m not counting superdelegates. If I were, his comeback score would be 83 percent.

Why has Sanders opted to stay in the race this long? Without a window into his campaign headquarters, I can only speculate. I suspect the uncertainty on the Republican side has bled over to media coverage of the Democrats, giving everyone the impression that a Sanders nomination is more likely than it really is. Beyond that, Sanders could be trying to pull Clinton, the de facto nominee, further to the left (though she’s pretty dang liberal already). Maybe he feels less responsibility to hold the Democratic Party together than previous runner-ups, having only joined it recently. Or, he may really just believe that Clinton is an unacceptable nominee, and won’t concede the race to her.

Whatever the reason, denying the delegate math does neither Sanders nor the Democratic Party much good. There are surely benefits to keeping a nominee accountable with a competitive race, but Team Bernie must argue that these pros outweigh the cons of a prolonged, increasingly negative campaign. It’s not really possible to tease out cause and effect here, but Clinton’s unfavorability ratings have risen substantially in the last several months. The “Bernie or bust” meme is almost surely an empty threat, but it remains to be seen how much of this Clinton enmity carries over to the general election.

And now, having roused the full ire of the internet, my spreadsheets and I will go into hiding until the dust settles around Democratic nominee Hillary Clinton.


  1. This average is weighted by delegate count, and excludes states with fewer than two polls.

  2. Regular internet users (and Redditors, in particular) are disproportionately white, young, male, city-dwelling, and liberal—all categories which support Sanders more than the Democratic Party at large. Demographically speaking, Sanders is just about the perfect viral candidate. Unfortunately, shares and retweets are not votes.

  3. I limit my analysis to Democrats because the Democratic nomination process assigns delegates proportionally, which makes comebacks harder. The Republican system, with some winner-takes-all states, can turn slight advantages in a few key states into huge delegate wins. See also: Donald Trump.

  4. 1972 is the typical start date for the “modern era” of presidential primaries, after the McGovern-Fraser Commission changed delegate rules to more or less their current form.

  5. Fans of statistical pedantry will note that these numbers specifically reflect the comeback necessary to clinch a plurality of delegates, not the majority needed for the nomination. Using the majority threshold as a target unfairly disadvantages candidates like Michael Dukakis in 1988, who was winning a multi-candidate field with 30-40 percent of the vote for much of the primary season. He needed to up his game to around 60 percent in order to hit the majority threshold, but since he was already winning, it would be misleading to say that he needed a comeback. Effectively, I’m assuming that the plurality winner gets the nomination — a generous assumption for a party outsider like Sanders. Credit is due to Dan Mendelsohn and Anders Dohlman for helping me refine this methodology.

  6. To be more precise, I imagined that all superdelegates vote at the end of the primary season, like one big final caucus. Of course, if “superdelegates” were an actual state, it would be bigger than California and currently polling 92–8 for Clinton.

  7. The remaining states are diverse, caucus-free, and mostly closed primaries. These are away games for Sanders; I ignored everything except delegate counts, tilting this analysis in his favor.

  8. This does not include candidates who never officially ended their campaigns (like Jackson in 1988) or ended their campaigns after all pledged delegates had been assigned (like Clinton in 2008).

Milo Beckman is a freelance writer for FiveThirtyEight. His work can be found at He also constructs crossword puzzles for The New York Times.