Annals of derp: Good news for John McCain

Ratifying the 15th Amendment, 1870. Via United States Embassy to Nigeria.

A fairly comical example of some willful derp from the National Review, with that conservative penchant for looking through every bend in the chicken entrails for the signs that are in some sense or another "good for the right", in this case a way of arguing that there's something about a YouGov poll that you can see in a dream as an indication that there is a universe where black people might vote Republican:

The White Vote

By Ramesh Ponnuru — October 23, 2015

Great Scott Alexander post:

[T]he [polls with racial breakdowns that] we have suggest that Donald Trump’s supporters are about as diverse as any other Republican’s and maybe more so. . . . 

On the other hand, there is a candidate whom the media narrative fits like a glove. A candidate who may win [a] primary among whites, but loses in a landslide among minorities. A candidate whose black support is almost an entire order of magnitude lower than his white support. . . .

The what vote? Great Scott! It's a blog post arguing that Donald Trump isn't a magnet for angry racist white voters but if he is then Bernie Sanders is too only just kidding:

Vox says that Donald Trump practices the “politics of white insecurity”. US News says that Trump shows “the rising power of the white vote”. Salon wants to tell you “eight reasons why white America falls for demagogues like Donald Trump”. The Week says Donald Trump represents “the rise of white identity politics”. The National Journal says Trump is creating problems by “preaching to a shrinking white electorate”. Read enough of these articles, and you might start to get the feeling that Donald Trump’s supporters are disproportionately white.

Not directly I wouldn't, because that's not what these discussions are about. They are about what Trump says, not who is listening, which is not necessarily the same thing. I wouldn't be surprised if a vast majority of his supporters were white, but I don't know it from Vox and Salon.

You would be wrong.

Really?

Well, probably. Data are sketchy.

Ah, that's more like it. And then we're off to a YouGov poll from last August where 25% of the black voters who identified as Republican chose Trump as their favorite candidate, and 28% of the Latinos. Sounds like a lot!

Well, one black voter, because there were only five African Americans who identified as Republicans in the sample, out of between 200 and 210 black respondents altogether, or well under 3% of the total (one out of five was 25% instead of 20% because of statistical weighting). And about eight Latinos, out of 30, or 14% Republicans of a total sample of 210 Latinos. Trump was, two months ago, the choice of less than 0.5% of African Americans, and 3.8% of Hispanics, or less than the margin of error. The raw numbers of black and Latino Republicans are too small to get meaningful statistics.

Compared to maybe 100 (weighted 24%) of the 416 white Republicans in the survey. Trump support is disproportionately white, if only because all Republicans are disproportionately white, but nevertheless. It's possible to say that "Donald Trump's voters are as diverse as any other Republican's, maybe more so," because that's how low the bar is set. Very, very low, as Trump would say.

But then there's the candidate described at top, who could win a primary among white voters but seems to have no black support, and who might that be?

That candidate is Bernie Sanders.

Indeed, it's true as Alexander claims that Bernie Sanders only had 4% of the self-proclaimed black Democrats in that preference poll, that's around 6 out of 140 ( plus 13% of 77 Latinos), and an order of magnitude more than Martin O'Malley got (1%). But that's not because they don't like Bernie! It's because they had a candidate that they liked best, Hillary Clinton (64%), and a choice that they liked a lot (the now definitely out Biden). As a second choice, Sanders got 11% of the black vote.

But the most important numbers in the poll, if you want to contrast Trump's vs. Sanders's minority appeal, aren't in these horserace charts at all, but in the favorability ratings. There, black voters liked Sanders more decisively (35/22) than white voters did (40/34), and disliked Trump overwhelmingly more (23/60), although whites certainly hated Trump too (43/51). That one black dude who told the pollsters he'd vote for Trump was quite the outlier. (For comparison purposes, blacks disfavor Carson 29/34, Cruz, 23/45, Fiorina 19/36, Rubio 21/44, and JEB! 21/56. )

And then for the "sketchy data" with which he tries to make his point, Alexander needs to go back two months. But it just so happens there is some non-sketchy poll-aggregate data from right now. Wonder why Alexander didn't think of looking at that?

Sanders mostly wins very clearly over Trump in the head-to-head, in fact, in up-to-date polling, as you probably know. If it happens, it won't be with a lack of black votes: in the Quinnipiac version, for instance, Sanders beats Trump by 75 to 15% of the African American vote (where he beats Carson 65 to 26%, and Bush 82 to 9%). And Trump's favorables among black voters have gone way south from the drowning point they were at two months ago: 7/85 (Sanders is 40/10 positive, too many still haven't heard of him).

There is no unwillingness in the black community to vote for Sanders, if that's what's required. There's a legitimate worry, as almost always, about minority turnout, if our dear party fails to represent everybody the way it's supposed to do, the Democrats' chronic illness, and I do think that's a seriously bigger worry if Clinton is not the nominee (which will certainly influence my own primary vote, I don't know how much). But if they make it into the booth, they'll be just fine, Ramesh, thanks so much for your concern.