No one, with the exception of maybe the Vermont Senator himself, expected Bernie Sanders to win the Michigan primary last night. Talking heads on every major station assured a landslide victory for the Clinton camp. So did her operation's staff, who failed to set up enough campaign offices across the state. Even some democratic voters, convinced that Hillary had the win wrapped up, reportedly switched sides to vote against Trump in the GOP contest.
Now pundits are scrambling to account for what happened—claiming the upset owes to Clinton's support for NAFTA or her false charge that Sanders voted against the auto bailout; Bernie's slow gains with minority voters or continued dominance in college towns. In all this confusion, the one thing that remains abundantly clear is that no one was more wrong than the pollsters.
Early Tuesday, RealClearPolitics polling averages had Clinton winning the state by more than 20 points. Nate Silver and his team of once-trusted data wizards predicted Hillary Clinton had a greater than 99% chance of winning the Michigan primary. At 10 PM last night, Decision Desk HQ, generally a trusted source of coverage, tweeted out their projection in favor of Clinton while Bernie Sanders was in the process of winning.
Clearly they screwed up bad, which got us wondering if this would constitute a history-breaking screw-up. Turns out, yes.
When the early returns showed a commanding Sanders lead, Silver hedged his expectations in a sort of Rubio-esque fashion, then noted that a Clinton loss in Michigan would "count as among the greatest polling errors in primary history."
And it turns out that the 1984 primary is notable here for several reasons. For one thing, it was the first year in which the superdelegate rule went into effect in the Democratic party. And somewhat familiarly, the race featured an initially dismissed campaign that gained a solid following among young voters dissatisfied with the establishment-ties of presumed nominee, former-Vice President Walter Mondale. Mondale, in turn, accused the self-described long-shot Gary Hart of lacking substance in his policy positions, famously asking him in a debate, "Where's the beef?"
Once the caucusing kicked off in Iowa, Mondale won with 48.9% of the vote (Hillary took 49.9 in February). Then in New Hampshire, Hart managed to pull off his shocking upset, winning the primary thanks to a grassroots canvassing effort that few pollsters had accounted for. As Silver (correctly) notes above, Hart surged to victory despite polls predicting Mondale would win by 17.1 percentage points.
A New York Times article the following day noted the "startling upset that damaged the aura of invincibility with which Walter F. Mondale began the campaign year."
And that's probably just as far as Sanders supporters want this comparison to go. Mondale finished with a razor-thin lead in votes and delegates, Hart wound up winning the most states, and Rev. Jesse Jackson's 20% of delegates kept either from a majority. A contested convention was averted, however, thanks to Mondale's support from superdelegates. This remains the first and only time that superdelegates were decisive in a convention.
But take heart #Berners, because nothing makes much sense in this election cycle, and it's safe to assume that historical parallels have as much predictive power as public polling—that is, none. All we have now are the facts: Bernie Sanders won Michigan and the pollsters have never been more wrong.