Nate Cohn recently published an article in the Times somewhat oddly titled "Is Biden Gaining Older Voters, and Losing Young Ones?" I say oddly titled since his data show a huge swing to Biden among older voters relative to Clinton's 2016 performance but very little change relative to Clinton among young voters. So the implication of a trade-off does not seem to follow.
Anyway, the most newsworthy part of the article to me was his claim of no pro-Democratic movement among white noncollege voters relative to 2016. He has Biden's current deficit among these voters at an average of 29 points, identical to his 2016 point of comparison. Is this believable?
Sure, it is possible but I do have serious reservations about this finding. First of all, his methodology seems idiosyncratic, using only "high quality RDD live interviewer polls" for his averages and not even all of them them (at least by any reasonable definition of high quality). And his point of comparison for 2016 is the same type of polls conducted after the third Presidential debate in that campaign. Huh? That doesn't seem like the obvious point of comparison to me.
Second, while his chosen group of polls may indeed average out at this point to a 29 point Biden white noncollege deficit, there is other data out there. For example, the Nationscape data (70,000 cases since the beginning of the year and counting) has Biden's white noncollege deficit at 15, a huge swing relative to the 31 point Democratic deficit from the actual 2016 election. On the other hand, the Nationscape data show Biden winning white college voters but by about the same amount as Clinton in the 2016 election.
Third, we have a raft of new polls from key swing states that do seem to show superior performance for Biden relative to Clinton's 2016 election performance among this group. I'll use here the just-released Fox polls from Michigan, Pennsylvania and Florida (all presumably Cohn-approved since they are high quality, RDD, live-interviewer polls).
Start with Michigan. The poll has Biden ahead 49-41. His white noncollege deficit: 8 points. Clnton's 2016 deficit: 21 points. Note especially Biden's performance among white noncollege women--dead even. Note also that Biden and Clinton do about the same among white college voters.
Then look at Pennsylvania. The poll has Biden ahead 50-42. His white noncollege deficit: 15 points. Clnton's 2016 deficit: 29 points. And Biden's performance among white noncollege women is a mere one point deficit. Biden also does better among white college voters than Clinton, but not by as much.
Finally, look at Florida, The poll has Biden ahead 46-43. His white noncollege deficit: 24 points. Clnton's 2016 deficit: 30 points (Quinnipiac--Cohn-approved!--also has an FL poll out and they have basically the same result). Here again Biden and Clinton do about the same among white college voters.
So am I sure these data are telling the right story and Cohn's the wrong one? No, there is always room for argument--and new data!--on these matters. And I always applaud an effort to keep a focus on the white noncollege vote as an area of Democratic vulnerability. But, if I had to put money on it, I'd say the white noncollege swing--at least at this point in time--is real.