How Is Biden Doing in the Battleground States?

Sanders Endorsement Day is a good time to check in on the state of the race to beat Trump. Coincidentally, today also saw the publication of a big piece by Nate Cohn on the subject of how Biden is faring the battleground states. Despite Biden leading in enough states to win (Michigan, Pennsylvania and Wisconsin plus Arizona--see graphic below), according to his aggregated data, he mostly emphasizes that Biden's lead is fragile in these key states, compared to more his robust lead nationally.

This is fair I think, the race is still tight when looked at on a state by state level; we do live in a world where it's quite conceivable that Trump could lose the popular vote but carry the electoral college. Cohn also emphasizes the key role that weakness among white noncollege voters for Biden could be a driver of such an outcome. Again, fair enough.

I do have some questions about his analysis though. He suggests a 4-5 point difference between national and key battleground state results when, in 2016, the difference, at least in MI-PA-Wi, was on the order of 2.3-2.8 points. Also, while Cohn's national aggregate result gives Biden pretty much the same lead that RCP's average does, his state results are more different:

(Cohn margin then RCP margin)

MI 4 vs 4
AZ 4 vs 4
PA 2 vs 4
WI 1 vs 3
FL -1 vs .4

Cohn also has Biden's nationwide deficit among white noncollege voters at -29, though other polls have it considerably smaller, including the Nationscape data I cited awhile back.

National: -=31 Clinton 2016, -16 Biden 2020, +15 Biden difference

Arizona: -27 Clinton, -20 Biden, +7 Biden
Florida: -30 Clinton, -20 Biden, +10 Biden
Michigan -21 Clinton, -7 Biden, +14 Biden
Minnesota, -21 Clinton, -5 Biden, +16 Biden
North Carolina: -51 Clinton, -39 Biden, +12 Biden
Ohio: -32 Clinton, -16 Biden, +16 Biden
Pennsylvania: -29 Clinton, -17 Biden, +12 Biden
Wisconsin: -19 Clinton, --5 Biden, +14 Biden

None of this means that Cohn is wrong; in fact, I think his overall caution about reasoning from Biden's nationwide results to electoral college results is entirely reasonable. Still, his results are just one take and may have something to do with his decisions about what data to include (only live enterviewer polls) and other methodological matters. So read and digest but keep in mind those limitations.

https://www.nytimes.com/…/ups…/polling-2020-biden-trump.html