This comes with a caveat, but even so it may be the best batch of polls Trump has had all summer. Two weeks ago Change Research found Biden leading comfortably in every swing state except North Carolina:

Two weeks later Change sees the race as a jump ball, with Trump having gained meaningfully in every battleground — again, except North Carolina:

His approval across all six battleground states has risen two points in the last two weeks. The same goes for his handling of the pandemic. He’s still underwater in both metrics, at 47/53 overall approval in the battlegrounds and 45/55 on COVID-19, but even modest improvement is potentially significant. Maybe his semi-new tone on fighting the virus is paying off. He restarted the daily briefing, canceled the GOP convention, and became a little more forceful about wearing masks … before he started demanding anew that governors reopen and babbling about hydroxychloroquine again. Conceivably his recent spate of “safety first” messaging is bringing some wayward voters back to him.

There’s another possibility, that the outburst of violence in Portland, Oakland, and other cities has begun to erode some of Biden’s support. I wrote about that last night in the context of Minnesota, where Morning Consult recently detected a surprisingly close race. The “defund the police” push there may be leading undecideds towards Trump. Change Research didn’t poll about crime today but I noticed that Biden felt obliged to issue this statement last night:

I wonder what Democratic internal polling is showing right now about Trump’s “law and order” message making inroads with the electorate.

Like I said, though, there are caveats. Here’s one: New data from Morning Consult suggesting that Trump’s message isn’t having a big impact on its intended audience.

Suburbanites are the “swing” demographic in that group, one that normally favors the GOP in elections but has been drifting leftward recently. Highlighting the violence in Portland by deploying DHS is Trump’s way of trying to lure them back. He doesn’t have majority support for it here — but that’s not necessarily inconsistent with the improvement in battlegrounds being seen by Change Research. Maybe suburbanites in swing states are more receptive to his message about crime than suburbanites in bluer states. Even if they aren’t, it’s possible that “law and order” has netted him more suburbanite votes than it’s cost him. Those who support the feds’ presence in Portland may be less decided about their vote this fall than those who oppose it.

The other caveat is that some of Change Research’s numbers are outlier-ish. Not all of them: Their toplines in North Carolina, Arizona, and Wisconsin closely match the RCP polling averages in those states. But in Florida they have Biden by three while RCP has him up 7.2 points; in Michigan it’s Biden by four versus Biden by eight in RCP; and in Pennsylvania Change Research has Biden by two while RCP sees Biden ahead by 6.2 points. That’s significant because winning FL, PA, and MI while holding the other 2016 Hillary states would easily be enough for Biden to win the election. If RCP’s average is more accurate, Trump’s still in big trouble.

We’ll see what the next round of polling says. Florida, Arizona, and Texas are especially important to watch right now because the COVID outbreaks there seem to be easing — noticeably so in Arizona, somewhat less dramatically in Texas, and less dramatically still in Florida at the moment. It could be that having cases abate will ease voters’ hard feelings towards Trump about the state of the pandemic. That may be all he needs to make Arizona and Florida competitive again.