Bernie Sanders speaks to supporters on the night of his upset victory in Michigan. (Reuters/Carlo Allegri)

Pollsters aren’t the only ones who incorrectly projected a Hillary Clinton victory in Tuesday’s Democratic presidential primary in Michigan. Journalists also made clear that they expected Clinton to beat Bernie Sanders — if not through explicit predictions, then through stories about the strength of her African American base in the state and the political wisdom of her making the Flint water crisis a campaign issue so early and often.

A Clinton win made so much sense!

“Ever since Flint’s water crisis became a national story at the beginning of the year, Hillary Clinton has done everything in her power to own the issue in the Democratic primary, to Bernie Sanders’s detriment,” the New Republic’s Rebecca Leber wrote in February.

The Los Angeles Times’ Cathleen Decker wrote on Sunday that “African American voters steadied Hillary Clinton's campaign in the South and now are poised to propel her forward in a corridor of Northern industrial states where voting kicks off with Tuesday's Michigan primary.”

Leber and Decker and all the other journalists who penned similar sentences weren’t exactly wrong about Flint or about black voters. Clinton was indisputably out in front on the toxic water issue, and while Sanders fared slightly better than expected among African Americans, Clinton still won that slice of the electorate by 40 points on Tuesday.

The flaw in pre-primary reporting on Michigan was that it perhaps underemphasized Sanders’s advantages in the state — most notably his popularity among voters who believe international trade agreements have cost the region jobs. And while Clinton was generally considered the winner of Sunday’s debate in Flint, she handed Sanders a ready-made attack with her misleading claim that the Vermont senator opposed the auto industry bailout in 2008 — a low blow in a state where the carmaking business is vital.

It's hard to know for sure which factors contributed to Sanders's upset, but in exit polls, nearly 6 in 10 voters said trade took away American jobs, and about 60 percent of people who said that backed Sanders.

In any case, it turned out that Sanders’s pluses were more than enough to counteract Clinton’s.

Sanders supporters, convinced that the press is biased against their candidate, might assume that reporters deliberately ignored the underdog’s strengths in Michigan. The truth is probably less insidious but nevertheless deserves criticism.

It goes back to polling. Pre-primary surveys showed Clinton leading, on average, by 21.4 points. That’s massive. With journalists knowing (or thinking they knew) that the former secretary of state was dominating in Michigan, it was only natural for them to look for the logical reasons that supported the conclusion.

Remember in school, when you peeked at the answer key in the back of your math workbook and then made sure that your problem-solving method — no matter how faulty — produced the correct result? (You know you did that.) Michigan reporting was kind of like that. The media worked backward, starting with the “correct” answer — a certain Clinton victory — and then finding the rationale to fit.

The trouble is that, every now and then, there’s a misprint in the answer key, just like polls are sometimes way off. That’s why it’s important in math and journalism to set assumptions aside and follow the evidence wherever it leads.

Without the influence of polling, a Sanders win in Michigan still might have been difficult to forecast. The final result was extremely close, and the case for Clinton was valid and made complete sense. But a more open-minded approach to covering the race — one that put more stock in on-the-ground accounts of Sanders's support, perhaps, and went beyond the black-voters-mean-Clinton-wins narrative — probably would have pointed to something other than a blowout.