One of the most memorable polls of the last year, and among the most encouraging for Trump, was the NYT’s battleground survey from last November. Trump trailed various Democratic primary candidates big in national polling even at the time, but at the end of the day national polls don’t matter. States matter.

And only a handful of states, at that.

Look past the national polls, the Times wrote in November, and you’ll see that Trump is well positioned to pull the same trick he pulled in 2016. Biden was his toughest match-up among the Democratic contenders, but even against Biden he managed to make the battlegrounds a coin flip.

That was eight months, one pandemic, and a series of nationwide anti-racism protests ago. You already know what the latest NYT national poll looks like, but Nate Cohn went digging through that data to see what the margins look like in the battlegrounds specifically. How far has Trump slipped since November?

A lot, it turns out:

Note that the November survey was a poll of likely voters whereas the new one’s a poll of registered voters. We can safely knock a few points off of Biden’s margin in each state here. But not six points. And certainly not 11.

Cohn sees the same problem in the battlegrounds as we saw in the national data yesterday: The white vote is in play, and you can’t win as a Republican when that’s the case. Specifically, Biden’s up 21 points among white college grads; in the November poll he led them by just 10. Trump’s also lost ground among his core base of whites without a degree, sliding from a 24-point lead in November to a 16-point lead now. In fact, notes Cohn, Trump trails outright among whites in all three Rust Belt battlegrounds. In 2016 he won that demographic in those states by nearly 10 points.

So, ~all of Biden's gains relative to Clinton in the battlegrounds are because of white voters, both college and noncollege whites. Interesting in the context of the protests and Trump's attempts to appeal to white identity. https://t.co/Zv1BMiIDBK

— Nate Silver (@NateSilver538) June 25, 2020

Despite everything that’s happened in America since November 2019, Biden’s margins among blacks and Latinos in the swing states have barely budged. All of the erosion for Trump is among whites.

The pandemic and the protests — or rather, his tone-deaf response to each — seem to be what’s killing him:

A majority of voters, 63 percent, say they would rather back a presidential candidate who focuses on the cause of protests, even when the protests go too far, while just 31 percent say they would prefer to support a candidate who says we need to be tough on demonstrations that go too far.

Despite double-digit unemployment, 55 percent of voters in these six states say the federal government’s priority should be to limit the spread of the coronavirus, even if it hurts the economy, while just 35 percent say the federal government’s priority should be to restart the economy. Even the newly unemployed, who would seem to have the most to gain from a reopened economy, say stopping the coronavirus should be the government’s priority.

Those are attitudes you’d expect among the white college grads who are defecting to Biden. They’re a bit more progressive on racial issues but probably also a bit more reluctant to reopen since they have financial reserves to weather a lockdown that the working class doesn’t. Trump’s on the wrong side of both issues from them and it’s showing. Combine that with the fact that suburbanites were already trending blue since the 2018 midterms due to skepticism about Trump and he may have lost those votes for good. His only option at this point might be to try to run up the score to a staggering degree among blue-collar whites. And I think he’s trying, with his threats of military force against looters and his focus on getting the economy moving.

But it’s not working right now, for whatever reason. Maybe even working-class whites are more eager to see the virus contained than they are to get back to work.

The most noteworthy detail from the head-to-head-numbers above is that Trump is polling at 41 percent or worse in every state. His national job approval right now is 42.5 percent. That means we’re seeing the same thing here that we saw in yesterday’s NYT poll, namely, some small share of people who think he’s doing a good enough job are nonetheless inclined to support Biden, which means Trump might need a job approval of something like 48 percent on Election Day to make another electoral-college win likely. He’s never touched 48 in the RCP average; in fact, the only time he’s touched 47 was briefly at the end of March, when political leaders across the spectrum got a bounce in their approval amid public anxiety about the erupting pandemic.

How much room does he have to grow his support in the battlegrounds before November? Not much, if you believe the Times’s sample:

Over all, 55 percent of registered voters in the battleground states said there was at least “some chance” they would support Mr. Biden in the election, including 12 percent of Republicans, 11 percent of voters who backed Mr. Trump in 2016, and 44 percent of the Republican-tilting undecided voters.

As for Mr. Trump, 55 percent of registered voters in the battlegrounds said there was “not really any chance” they would vote for him this November.

Again, that’s registered voters, not likely voters. But if, say, 50 percent of likelies are a hard no four months out from the election, it’s easy to see how ugly this might get. The UVA “Crystal Ball” has 268 electoral votes in Biden’s column right now if you include states that are leaning Democratic in polling (which means Michigan and Pennsylvania). That puts him one state away from victory with Wisconsin, Arizona, North Carolina, and Florida all in play (and Ohio, Iowa, Georgia, and Texas not looking rock-solid at the moment). If Biden won all six battlegrounds polled by the Times plus all the Hillary states from 2016 he’d have 333 electoral votes, more than Trump won four years ago.

POTUS is annoyed at the polling news this week:

https://twitter.com/realDonaldTrump/status/1276173394053156865

He likes spinning discouraging polling as a “suppression” tactic, which is part of his conspiratorial outlook, but I’ve seen Biden fans on social media spin them in an oddly similar way — not as a “suppression” tactic, of course, but as a complacency risk. Righties want to believe that a Biden +10 survey is aimed at convincing them not to bother voting, which is silly when we’re still more than four months out from the election. But paranoid lefties and Never Trumpers, spooked by Trump’s victory in 2016, fear that Democrats will be the ones inadvertently persuaded not to bother voting (or donating or phone-banking or organizing) in the mistaken belief that this election is in the bag. Forgive the profanity, but this tweet by Trump-hater Joe Walsh captures it succinctly:

To every American who wants Donald Trump to lose in November: IGNORE THE FUCKING POLLS!

— Joe Walsh (@WalshFreedom) June 25, 2020

It’s a “suppression poll” for Republicans and a “complacency poll” for Dems and their bedfellows. Sheesh.

One last note about the Times data. It would help Trump greatly if lefties were holding Biden’s victory over Bernie Sanders, and his general squishy centrism, against him. But they aren’t so far. He’s winning 93 percent of Democrats in the Times battleground poll and his margin over Trump in the 18-29 group has more than doubled since November, from +17 to +38. (Which isn’t completely surprising since Bernie dropped out over that period.) Maybe Biden will say or do something between now and November to alienate them, but Biden himself may be irrelevant to the equation here: More than half his voters, 55 percent, say they’re voting against Trump, not “for” Biden. It’s a Trump referendum election, plain and simple. At least for now.