In 2016, Sen. Bernie Sanders (I-Vt.) notched a shocking win in the Michigan primary in March despite polls showing him trailing by double digits. It was labeled as one of the worst polling errors in political history, and it gave Sanders renewed hope after a rough Super Tuesday.

On Tuesday, Sanders needs lightning to strike twice in the Wolverine State. Unfortunately for him, it’s not at all clear how that might happen.

Both an EPIC-MRA poll and a Monmouth University poll on Monday showed former vice president Joe Biden building a double-digit lead on Sanders. Biden led 51 to 27 percent in the former survey and 51 to 36 percent in the latter.

The problem for Sanders lies in the cross tabs. In 2016, the biggest reason for the polling miss was that pollsters badly underestimated a surge in the youth vote. Part of this was because there wasn’t a recent competitive primary in Michigan that they could use to model the presumed electorate. (Before 2016, the last contest was in 2008, but that year Barack Obama and Hillary Clinton didn’t formally compete in the state because it violated party rules by holding its primary too early.) While polls estimated that voters under 30 would make up about 1 in 10 votes, they were actually 1 in 5. Given that Sanders took about 80 percent of those votes, it was a huge windfall and delivered him a narrow win over Clinton.

In 2020, pollsters have that recent primary to model the electorate and adjust accordingly. Even if they still get the electorate somewhat wrong, it’s hard to see how Sanders could pull off another massive upset.

Sanders is underperforming his 2016 numbers pretty much across the board. That year, he won voters between 18 and 44 years old by 33 points. The two polls Monday show him leading among 18-to-49-year-olds (a slightly different group) by 16 points in one poll and 11 in the other.

He’s also doing worse with older voters. Exit polls in 2016 showed him losing voters 45 and older 62 to 37 percent. Now, the EPIC-MRA poll has him trailing 65 to 14 percent among those 50 and older, and Monmouth has him trailing by more than 2 to 1.

Again, these are slightly different age groups, but they tell the same tale. Sanders does significantly worse than in 2016 in both the younger and the older groups.

Another key group Sanders won in 2016 was white voters. He offset Clinton’s strength among black voters by carrying whites 56 to 42 percent. This time he actually trails among white voters by 14 points in the Monmouth poll and an astounding 30 points in the EPIC-MRA poll.

Another key finding from the EPIC-MRA poll is this, via the Detroit Free Press: “Among the 37% of respondents who decided in the last few days, Biden led overwhelmingly, 75%-18%, over Sanders.”

If late-breaking voters go for Biden by anywhere close to such a margin, and the findings on young and white voters are anywhere close to accurate, it’s not clear what kind of a polling error might rescue Sanders’s chances in Michigan. Pollsters aren’t going to miss another youth surge by that much again, and even the demographics Sanders would be counting on to mount a comeback don’t seem to be delivering for him.

The bigger question seems to be whether Sanders can keep it close and avoid a delegate drubbing in Michigan, which with 125 delegates is the biggest delegate prize of the six states voting Tuesday. He won it in 2016, and it still didn’t fundamentally change the race that lay ahead. So it would be tough to argue that a loss was anything less than a disappointment.