As we close in on the second-most-important election of our lives, we’re being barraged with polls and forecasts and punditry predicting big wins for Democrats. While the Senate is a heavy lift for Democrats because of a tough map, it can feel like a blue wave that will sweep the party into power in the House and in a bunch of state legislative races is inevitable. But given the stakes, it’s really important to tune all of that out and work as hard as possible to avoid disaster next week. Here’s why.

Make no mistake, the GOP has a solid chance of holding everything next Tuesday. Not great, but solid. As Princeton number-cruncher Sam Wang points out, the average polling error in midterm contests is significantly larger than that in presidential years. The polls could easily be off by three or four percentage points. At present, Democrats enjoy an advantage on the generic congressional ballot of around eight points. If they were to overperform that number by three, it’d be a blowout win; if they underperform by the same margin, Republicans would get their “red tide.”

It’s difficult to overstate how catastrophic that outcome would be. Not only would it deliver a demoralizing blow to the entire left, it would also validate Trump’s toxic demagoguery and Republicans’ wildly dishonest campaign claims about everything from health care to that caravan of refugees that’s still almost 1,000 miles from our border. It would send yet another a message to right-wing extremists of all stripes that their odious worldviews are in fact mainstream. And we could kiss any meaningful oversight goodbye as Trump and his cronies would continue to enjoy impunity by congressional majority.

The 2016 presidential contest has stark lessons for the reality we face today. Hillary Clinton’s presidency was seen as all but inevitable, even though all the forecasting models gave Trump a real chance of winning. FiveThirtyEight’s final run gave him a 29 percent likelihood; The New York Times’ more bearish model still had the probability of his victory at 15 percent. Those odds aren’t actually that low—things that have a 15 percent chance of happening occur all the time. It’s not like buying a Powerball ticket. But as any meteorologist will tell you, most people simply do not know how to interpret statistical likelihoods. As Nate Silver has repeatedly attempted to pound into people’s heads, the media didn’t cover that race as if Trump had a realistic chance of becoming president, even though he did. If they had, it’s possible—perhaps likely—that he wouldn’t be in the White House today. Although the mainstream coverage of both candidates was overwhelmingly negative, relatively little of it focused on Trump’s myriad scandals. With Clinton, it was the exact opposite. The fact that he was seen as a clownish figure who had no chance of becoming president had to play into those editorial decisions.

It’s likely that a consequential handful of voters also took a Democratic win for granted in 2016. A post-election poll of 100,000 registered voters found that those who decided to stay home skewed more Democratic than the ones who voted. According to CNN forecaster Harry Enten, that difference probably cost Clinton the election.

The midterms feature even more uncertainty than the 2016 election. As of this writing, FiveThirtyEight’s forecasting model gives Republicans similar odds of keeping their House majority, losing 61 or more seats or losing fewer than 20. At 18 percent, Democrats’ likelihood of winning the Senate is the same as Trump’s chances of winning the presidency were in late October of 2016.

There’s some anecdotal evidence suggesting that lots of irregular voters and first-time voters have been lining up to cast their ballots during this early-voting period. If people who don’t usually vote in midterms are in fact turning out this year, polls of likely voters may be missing them. And those irregular voters are likely Democrats—Democrats typically stay home during midterm contests at a significantly higher rate than Republicans. Put all of that together, and it’s entirely possible that polls are systematically undercounting Democratic enthusiasm in a big way, and we’re actually heading for a blue tsunami.

At the same time, polls measure voters’ intent, nor whether their registration has been improperly purged or they have an acceptable form of identification. Voters across the country are at risk of not being able to cast ballots, thanks to voter-suppression efforts. An election official in Dallas County, Texas, told ProPublica that the amount of voter intimidation and harassment at the polls “is the worst she’s seen in decades.” Polls of the North Dakota Senate race don’t tell us anything about whether Native Americans will be able to vote, and the same is true of the Georgia governor’s race, where secretary of state and gubernatorial candidate Brian Kemp is using every method available to him to steal that seat. Whether or not it’s a result of foul play, voting machines in Texas appear to be switching votes from Beto O’Rourke to Ted Cruz.

So find a campaign and knock on some doors. Throw a few bucks your candidates’ way if you can afford it. You can also sign up to help get out the vote at thelastweekend.org, a coalition of Swing Left and 22 other progressive groups. MoveOn even lets you do this work from your computer in the comfort of your own home.

Next week’s election hinges on whether the backlash against Trump is great enough to overcome the GOP’s relentless efforts to keep turnout down. The polls and the pundits may point to an approaching victory, but we need to tune out all the noise and work as if lives are in the balance, because they are.