i thoroughly deconstructed this during the primaries, and i agree that the primary factor was age....in the north.
in the south, the fact is that you actually didn't see a lot of difference in the results, based on race - that sanders lost badly to white people, too. it was widely reported that sanders badly lost to black voters in these states, but the media forgot to mention that he lost by roughly the same margins to white people, too.
the conclusion that i drew was actually that the south is just a conservative place, and would prefer to vote for the more right-leaning candidate, regardless of gender or race. further, one might forget that this was true in 2008 as well, when obama ran to the right of clinton in the primaries; in 2008, it was clinton that supported universal health care, for example, and obama that didn't. obama ran as a kind of market-driven pro-business libertarian. or, at least, that was the impression i got from him, from day one. clinton was a communist that wanted universal health care, gender equality and the village to raise your kids.
so, you're going to see these surveys and studies come up and they're going to say "x% of black people think...", and they're going to all make the same fundamental error: they're going to treat the black vote in the united states as a homogeneous whole that is determined by ancestry (because the negro is preconditioned, of course), rather than a fragmented collection of geographically-specific interests.
if you look at the data carefully, what you're actually going to find out is that black people in new york aren't that different than white people in new york, and that black people in atlanta aren't that different than white people there, either - and that if you want predictive surveying, you should be looking at geography, not at identifiable characteristics.
but, i shouldn't have to make the argument too forcefully, because this is the same mistake everybody made in 2016: they looked at national polls, then got confused by the electoral college. the fact that hillary won the popular vote nationwide did not help her win an election, and a study that says that sanders is competitive for black voters across the country doesn't mean he's competitive for black voters in the carolinas.
it is far too early to expect polls to be predictive. but, if you're going to have these discussions, you need to look at local polls, and not national ones.
https://www.vox.com/2019/3/7/18216899/bernie-sanders-bro-base-polling-2020-president