In the run-up to the Wisconsin primary, RealClearPolitics, the top polling aggregator in the United States, predicted that Bernie Sanders would win Wisconsin by 2.6 points. Harry Enten, one of the top analysts for the nation's top standalone polling website, FiveThirtyEight.com, was a little more generous: he predicted a 5-point Sanders win.
Based on the data analyses I've been conducting over my last nineteen articles for The Huffington Post (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19), I predicted on social media yesterday afternoon -- in preparation for this, my twentieth article on the Democratic primary race -- that Bernie Sanders would defeat Hillary Clinton in the Wisconsin primary by 16 points (58 percent to 42 percent).
It may have seemed, on the face of it, a blindly partisan prediction, given that not a single poll in America had ever shown Sanders winning Wisconsin by more than 8 points -- and indeed two polls released in just the last week actually showed Hillary Clinton in the lead in Wisconsin. While it's true that I've been very transparent about supporting Senator Sanders in the primary race, my projection for Wisconsin was based on data -- granted, data the mainstream media has ignored and various media personalities (Philip Bump of The Washington Post, Gideon Resnick of The Daily Beast, Nate Cohn of The New York Times, and the aforementioned Harry Enten, to name a few) have snarked repeatedly -- and the data said to me that Sanders would win Wisconsin by far more than any poll anywhere had ever predicted.
Sixteen points, I said: 58 percent to 42 percent.
In the end, Bernie Sanders won Wisconsin 56.5 percent to 43.1 percent.
In fact, he won 69 of the state's 72 counties.
That's 96 percent of Wisconsin counties.
And in two of the three counties Clinton won, she won by just a few hundred votes out of many thousands cast.
Counting state result-pledged "PLEOs" (Party Leaders and Elected Officials), Sanders also took home a net delegate haul (+20) that was 250 percent larger than the supposedly optimistic one half-seriously given to him by CNN's John King in his many "Magic Wall" demonstrations.
Sanders doubled the already rosy net delegate target set for him by FiveThirtyEight.com.
He even approached the 15-point threshold that the contrarian Enten -- a strong believer in Clinton's inevitability -- had set for a "game-changing" performance. If Sanders were to win by 15 points or more, said Enten on Tuesday morning, "It would mean that maybe something has changed. Maybe Sanders has a real shot at winning this thing." Sanders fell short of Enten's threshold by just 1.6 percent.
Artificial threshold met or missed, the fact is that Sanders does have a real shot at winning this thing.
And the same data that gave a more accurate prediction of the Wisconsin results than the mainstream media -- and, indeed, a spot-on prediction of the caucus results in Alaska, Hawaii, and Washington, even as the media was between 20 and 30 points off in predicting Sanders' margins of victory in those states -- is the data that says Bernie Sanders and the Democrats are headed to a contested convention in Philadelphia this summer.
The first reason is hard data -- specifically, contrasts between early voting and Election Day voting in individual states, nonwhite voting patterns over time, and large-scale, macroanalytical data about how Senator Sanders performed in each primary and caucus over the course of March.
The second reason it's being predicted here, and not elsewhere, that the Sanders campaign is correct in stating that we're headed to a contested Democratic convention is that the 2016 Democratic primary race is absolutely nothing like the 2008 race, no matter what anyone in the mainstream media has been saying.
The hard data simply doesn't support that analogy -- and neither will the results.
Indeed, the constant drumbeat from the mainstream media to analogize 2016 to 2008 grievously misunderstands a simple fact: the situation Bernie Sanders is in right now is nothing like the situation Hillary Clinton was in eight years ago.
As I've analyzed early and nonwhite voting trends in my previous articles -- see the links above -- here I focus only on the second issue too little discussed by the media: broader conceptual models that help us predict how a given primary battle will play out.
In lay terms, there are many different models for how a primary race can unfold. And the model we saw when Clinton was running against Obama is entirely different from the model now being exhibited -- and with perfect precision -- in the Clinton-Sanders race.
So let's compare the two primary races, using easy-to-distinguish "headlines" for each model:
Primary Model: "Seesaw of Equals"
Year: 2008
Candidates: Barack Obama and Hillary Clinton
Primary Results (in % of Delegates Won):
Super-delegates Pre-Iowa (link): Clinton 53 percent (169), Obama 20 percent (63)
Month 1: Clinton 50.4 percent, Obama 49.6 percent
Month 2: Obama 63.3 percent, Clinton 36.7 percent
Month 3: Obama 50.6 percent, Clinton 49.4 percent
Months 4-6: Clinton 53.9 percent, Obama 46.1 percent
Key Features of This Model:
- Both candidates appear viable at all points in the race.
- Super-delegates are therefore looking seriously at both candidates at all points.
- Super-delegates flip one at a time, if at all, after tough personal deliberations.
Result:
In 2008, the Democratic candidates needed 2,117 votes to clinch the nomination. In the end, President Obama was able to do this only through super-delegates; he didn't clinch the nomination using exclusively pledged delegates earned from primaries and caucuses. In fact, he actually lost the popular vote to Clinton -- by well over a quarter of a million votes -- and only was announced as the presumptive nominee on the final day of the campaign season, when he rolled out 60 super-delegate endorsements all at once. These endorsements put him over the 2,117 mark.
At the Democratic National Convention, Clinton received more than 1,000 delegates' votes -- perhaps because, despite endorsing Obama in June, in August she permitted her name to be entered into the official convention roll call as a still-eligible presidential candidate.
Primary Model: "Collapse of the Prohibitive Favorite"
Year: 2016
Candidates: Bernie Sanders and Hillary Clinton
Primary Results (in % of Delegates Won):
Super-delegates Pre-Iowa (link): Clinton 97.8 percent (359), Sanders 2.2 percent (8)
Month 1: Clinton 59.4 percent (607), Sanders 40.6 percent (414)
Month 2: Clinton 51.2 percent (657), Sanders 48.8 percent (626)
Month 3 (ongoing): Sanders 63 percent (63), Clinton 37 percent (37)*
Months 4-5: N/A
* = Figure includes actual results from Wisconsin and projected results from Wyoming based on previous caucus results (all media outlets nationally predict a substantial Sanders win in Wyoming). States left to vote in this portion of the election cycle include New York, Connecticut, Delaware, Maryland, Pennsylvania, Rhode Island, and Indiana.
Key Features of This Model:
- Only one candidate appears viable until nearly the very end of the process.
- Delegates do not seriously consider the underdog candidate until the very end.
- Super-delegates flip in packs, in response to a growing media narrative and events in the final two months of voting.
Result:
As much as the media reports that Bernie Sanders is behind Hillary Clinton by "263" delegates, in fact prior to Wisconsin that number was, at a minimum, 16.5 percent less: 220, per FiveThirtyEight.com's latest count. I say "at a minimum" here because Sanders is still expected to gain two to six net delegates on Clinton after all provisional ballots are counted (and delegates allotted in response) in Arizona. So Clinton's actual lead is somewhere between 214 and 218. To put this into perspective, at this point in the 2008 primary race Barack Obama's pledged lead over Hillary Clinton fluctuated between 120 and 140 delegates, according to MSNBC. So Sanders is 66 percent further behind than Clinton was at this point eight years ago. (Chuck Todd wrongly wrote on Tuesday that Sanders is "twice" as far behind, likely because he -- like the rest of the mainstream media -- is for some reason using the now-outdated "263" figure for Clinton's lead).
After Wisconsin, Clinton is ahead of Sanders by somewhere between 194 and 198 delegates.
After Wyoming, Clinton's lead will almost certainly be between 184 and 188 delegates.
There are two full months of primaries and caucuses after Wyoming, including votes in some of the most delegate-rich states in America: California, Pennsylvania, New York, New Jersey, Connecticut, Maryland, and Indiana among them.
When Secretary Clinton was defeating Sanders throughout the Deep South, she was doing so on the strength of an average margin of victory among nonwhite voters of 70 percent. Today, her average lead over Senator Sanders among nonwhite voters, in both polling and actual vote tallies, is about 14 percent -- one-fifth of her previous advantage. It won't be enough to repeat the results Clinton achieved in the South, or indeed anything like them. Sanders will win a number of the large states listed above.
Going back for a moment to the two-model theory discussed above, the difference between Sanders in 2016 and Clinton in 2008 is that Hillary Clinton started out the present election cycle with a lead over Sanders that would have been unimaginable eight years ago. Before anyone in America had cast even a single vote, Clinton had a 351-delegate lead on her next-closest contender -- who was, not surprisingly, a political Independent, as Clinton had hoped to use her super-delegate lead to scare all other credible Democratic contenders from the nominating process. At least as to the Democrats, it worked; the Clinton campaign didn't count on a challenge from a progressive Independent, however.
Right now, the Clinton campaign is sending out its surrogates to tell the nation that a 200-delegate lead is basically "insurmountable." This is the same logic they used -- to be clear -- to try to ensure, via a 351-superdelegate lead, that Clinton wouldn't have a primary challenger in the first instance, as it was never in the Clintons' plans to let Democratic voters have options. Moreover, the Clintons were confident that the national media would provide assistance with their plan by always reporting super-delegates as though they were pledged delegates (which they aren't). That hope and expectation has been, of course, richly rewarded.
And in the first month of the primary season, Clinton performed so well that she built upon her existing 351-delegate lead with an additional 193-delegate lead among pledged delegates. While the virtual ties in Iowa and Nevada, and the shellacking in New Hampshire, were not according to Clinton's plan, her victories in the South -- South Carolina, then all the states in the so-called "SEC Primary" -- gave Clinton, by the end of the first month of voting, a larger pledged delegate lead than Obama had ever had on her (193) and a larger overall delegate lead (543) than basically anyone has ever had after just one month of an election cycle.
Over time -- one might even say glacially -- the mighty have fallen.
Clinton's 97.8 percent "win percentage" among super-delegates pre-Iowa was severely undercut -- if only in perception -- when she won just 59.4 percent of the pledged delegates in the first month of the primary season. Some may well have wondered (though almost none in the media did) why Democratic elites saw Clinton as best-qualified to lead the Democratic Party at a 97.8 percent clip, while actual Democrats in early voting states -- and not just any states, but the very best states for Clinton, demographically, of any in the U.S. -- only came to the same conclusion at a 59.4 percent rate. Still, 59.4 percent ain't bad. Disappointing, but not bad at all.
51.2 percent was a different story.
After leaving the South, it was clear that the rest of the American Democratic Party supported Clinton at only a roughly 50-50 rate. It turned out that Iowa and Nevada were much more representative of Democrats' opinion outside the South than was (frankly) either New Hampshire -- a big Sanders win -- or South Carolina, a big Clinton win. Still, 51.2 percent is a majority, and with the entire Democratic Party apparatus behind her, a Hillary Clinton who wins the support of a bare majority of American Democrats outside the solid-red states of the Deep South would certainly be a strong enough candidate to maintain super-delegate support.
But then we started to see the cracks in Clinton's "inevitability" narrative, and it became clear that this wasn't a third possible model for a primary race ("Essentially Uncontested," as was the case in 2000 with Al Gore and Bill Bradley) but in fact the worst possible one for the Democrats: "Collapse of the Favorite."
So while the media continued saying, "It's all about the math, stupid," a small cadre of political observers -- along with the almost 50 percent of Democrats who now support Senator Sanders -- began to see the obvious: the math wasn't going to get Clinton to a point where she could clinch the Democratic nomination with just pledged delegates.
That's worth repeating: Hillary Clinton will not clinch the nomination through primaries and caucuses.
Observers, this writer included, began to understand what the Sanders campaign had long understood, which is (a) the model for this primary race is (as we're calling it here for the sake of simplicity) "Collapse of the Favorite," and (b) with no clinched nomination prior to the Democratic National Convention, a contested convention in which both of two candidates make their case directly to super-delegates is inevitable.
Having said this, let's be clear: if and when Hillary Clinton gets to 2,383 delegates via pledged and super-delegates, the mainstream media will call this race for her.
Immediately.
You will see a blue-backgrounded graphic on CNN that says "Presumptive Nominee" and has Hillary Clinton's smiling face next to it.
Meanwhile, back in America, people will be recalling how many super-delegates switched allegiances in 2008. People like this writer will be pointing out -- as I have already -- that the "Collapse of the Favorite" narrative wasn't just a two-month narrative, but indeed one that played out in many of the states Clinton won. Accepting for a moment (though it's a grave over-simplification) that early votes largely went for Clinton this year solely because older voters vote early, why did Clinton win Floridians who voted two to four weeks early by 36 points, and those who voted one week early by 13 points? Why did the RealClearPolitics average of polling prior to Election Day in North Carolina predict a 24-point Clinton victory, when what the Secretary got instead was a 13-point victory and a near-tie (a 52 percent to 48 percent margin) on Election Day? Why was the difference between early and Election Day voting in Illinois and Ohio so vast -- in the former case about 30 points different, and in the latter about 20? It's one thing for early voters for favor Clinton, it's another for early and "live" (Election Day) voting to be essentially two different elections. Is it possible that Clinton was actually losing support between early voting periods and Election Days across the country?
The questions then start to come faster: why did 700 more Sanders delegates show up to the Clark County (NV) Democratic Convention than Clinton delegates, when Clinton "won" Nevada on Election Day by 5.5 points? Why did Clinton win nonwhite voters by an average of 70 points in the Deep South, and then start winning them by smaller and smaller margins everywhere else -- with a current margin of (depending upon the national or state poll) somewhere between zero and 19 points? Why did Secretary Clinton lose Wisconsin, a state any solid front-runner this late in the election season would expect to carry handily -- as now-President Obama did in 2008 -- by 13.4 points?
The answer: this isn't 2008. This is the slow-motion collapse of a front-runner's campaign.
A 48-point lead in New York is now down to just 10 points two and a half weeks later. Senator Sanders has won seven of the last eight contests, and -- lest we forget -- actually won the voting in Arizona on Election Day 52 percent to 48 percent after all the votes had been counted. In other words, he's won the last eight Election Days in a row, and it'll be nine on Saturday, when he wins in Wyoming by forty points or more.
So, let me ask this: has any presumptive major-party nominee ever lost nine straight Election Days in the latter half of the nomination season?
In 2008, the most losses in a row ever suffered by Barack Obama -- the eventual winner -- was two.
And that was considered a very, very tight contest.
Today, surrogates for the candidate who's about to lose nine Election Days in a row are saying that it's their candidate who's ahead -- and by a mile. Because (they say) "math."
But super-delegates don't cast votes based on math, as Hillary Clinton herself discovered in 2008, when super-delegates voted against the popular vote winner.
Let's say you're not convinced by this, however. You say -- as some reading this undoubtedly would -- that elections are at base pretty simple things: get the most votes, you win. Or, here, delegates. For those whose doubts about the above run in this direction I'll paint a different picture for you, one that explains not only why Sanders is still running but why pundits who say outlandish things, like that Sanders needs to win all the big states remaining 60 percent to 40 percent, are dead wrong.
So here's the picture I'd like to paint for you: it's June 7th, and the California primary has just finished. The results are in. Bernie wins. This is just another win in a long string of them: Sanders, after winning nine straight Election Days heading into New York, beat Hillary Clinton in her home state -- where the second-largest haul of Democratic delegates is -- 51 percent to 49 percent.
The Clinton camp responded, predictably, with just one word: "math."
But then Sanders went on to win Pennsylvania 52 to 48, New Jersey 51 to 49, and California 54 to 46. And in between those wins he nailed down victories in Connecticut, Rhode Island, Indiana, Kentucky, West Virginia, New Mexico, Montana, North Dakota, and South Dakota. Clinton walked away with victories in Maryland, the Virgin Islands, and Guam.
So, to get back to the picture being painted here -- one fully supported by the same data that predicted the results in Alaska, Hawaii, Washington, and Wisconsin, as well as the Election Day surprises in Arizona and North Carolina -- the California polls have just closed and delivered Clinton an eight-point loss, and, though Clinton hasn't (and won't be able to) get to the 2,383-delegate mark via pledged delegates -- and though at this point she's lost 22 of the final 25 votes (88 percent) -- she still leads Sanders in delegates by slightly less than 100.
She's lost the delegate race in the entire second half of the election season by about seven points -- 53 percent to 46 percent -- and, in addition to losing her home state, also lost Pennsylvania and California (among many other states). Her trajectory among Democrats nationally has been consistently downward, such that she's now behind her opponent in the national polling 52 percent to 47 percent and writers are coming out of the woodwork to point out the hollowness of her "wins" in Nevada, North Carolina, Arizona, Iowa, Massachusetts, and elsewhere. Even Florida and Ohio are looking like states that were closer than polls and early voting suggested they should have been. The narrative of a slow-motion campaign collapse seems fairly clear -- mathematical, even.
Under these circumstances, with Clinton having just lost California but clinched a delegate lead over Sanders (though not a majority of all delegates in the nomination fight), does she:
(a) Give a speech in Los Angeles announcing that, though she's just lost California, she has -- in fact -- actually won the nomination!
(b) Contact the press backchannel and insist that they put up a graphic announcing her as the presumptive Democratic nominee, despite her super-delegates having many weeks left to make their final decision about who to support?
(c) Not announce victory but call on Sanders -- who's won 22 of the last 25 votes and has weeks to make his case to super-delegates -- to concede?
You see now the problem with the whole "it's the math, stupid" wing of the national punditry.
So here's what the Sanders campaign and certain commentators have been trying to tell us about "the math": the math shows that the Clinton campaign is collapsing. The math shows that super-delegates backed Clinton before any votes had been cast and before her campaign had imploded -- an implosion that was not so much caused by unforced errors but by the historic strength of the Sanders campaign, as evidenced by its mobilization of young voters, record-breaking fundraising, and, most importantly, upward trajectory in the actual voting across the course of a five-month presidential primary season. And the math will show that Bernie Sanders polls much better than Clinton against every possible Republican candidate both nationally and in every battleground state.
But most importantly, the math would show, in this scenario, that not only had all the comparisons to the 2008 primary race been inapt, but that all the references to primary "math" as algebraic had occluded the fact that, in actuality, what we're dealing with here is a very complicated -- and deeply human -- calculus.
And that calculus is leading the Democratic Party straight to a contested convention in Philadelphia this summer.
Seth Abramson is the Series Editor for Best American Experimental Writing (Wesleyan University) and the author, most recently, of DATA (BlazeVOX, 2016).