Silver is no Republican, another reason to weigh what he wrote. He starts off

Earlier this week, John Burn-Murdoch of the Financial Times posted a thread that purported to show substantial losses for Democrats among non-white voters, which he termed a “racial realignment”. If you’re an election data junkie, you’ve probably seen it; it’s been viewed more than 7 million times on Twitter.

This was followed by the graphic that “kicked if off.” Silver says he’s not a fan of digging into poll crosstabs.  Why? “Because of the small sample sizes and difficulties in reaching certain underrepresented groups, you can always find something “wrong” with them and use that to dismiss polling results you don’t like.”

Fair enough. But…

However, the Adam Carlson has been performing an invaluable service by aggregating the results of different polls together, which at least solves the sample size problem. And he’s finding that Joe Biden’s share of the vote has dropped dramatically among Black and Hispanic voters as compared with an average reliable estimates of the 2020 vote:

This is more difficult for the usual suspect to dismiss. Moreover

As you can see, Biden’s margin against Donald Trump has basically not moved an inch among white voters; he’s losing them by 12 percentage points, as he did in 2020. However, Biden is now only winning Hispanics by 7 percentage points — down from 24 points in 2020 — and Black voters by “only” 55 points, as compared with 83 points in 2020.

Silver looks at two areas where ”non-white voters are plentiful. One is somewhere I’ve never been to, Starr County in South Texas, and the other is the place where I live, New York City.” Silver writes

I’ve charted these as the total number of votes rather than just the vote margin, because that’s really what tells the story. Biden received about as many votes in Starr County as Hillary Clinton did in 2016, or as Barack Obama did in 2008 or 2012. But Trump surged from receiving 2218 votes in 2016 to 8247 votes, almost four times as many, in 2020. I’ve rarely seen anything like that, especially in the contemporary American political landscape where partisan preferences tend to be relatively stable. Turnout was much higher in Starr County in 2020 — but those new voters came out overwhelmingly for Trump, contradicting the longstanding belief that Democrats benefit from higher turnout among minority groups.

The same sort of thing happened in New York City where Democrats won by lopsided margins but not nearly as large.

Silver concludes with this ominous warning for Democrats:

Without winning huge majorities of Black voters, and solid majorities of Hispanics and Asian Americans, Democrats’ electoral math doesn’t add up to a majority.

 

Let’s keep it to that, for now. Although Burn-Murdoch’s theory is plausible — that Democrats’ increasing progressivism and generational turnover is the root of the problem — that’s something that deserves a longer analysis. What he’s seeing in the data shouldn’t be dismissed as some kind of outlier, however. It’s been replicated in poll after poll, and it has become increasingly apparent in election results, too.