A recent Politico article talks about how Trump might defeat Clinton. The article’s lede focuses on potential support among black voters:

If Donald Trump becomes the next president of the United States, there will be plenty of surprises along the way. One of the biggest will be the help he gets from black voters.

According to Republican pollsters and Trump’s allies, the GOP poll-leader — who has been dogged by accusations of racism, most recently for tweeting out a chart that exaggerated the share of murders committed by blacks — is poised to out-perform with this demographic group in a general-election matchup with Hillary Clinton.

However, although it quotes pollsters, the article doesn’t link to any actual polls that show Trump’s support from black voters. Nor do those pollsters mention any poll numbers that would support the contention that black voters support Trump.

Instead, the article says that Frank Luntz has talked to some black people who say they voted for Obama and that they would consider voting for Trump this time. But it is mum on any sort of quantification: what percentage of former Obama voters might feel that way, for example, and/or how many of them also said they would consider voting for the other Republicans. Was that question even asked?

There are not actually very many polls that attempt to measure the amount of Democratic and/or black support for the various GOP candidates, but there are a few. When I studied the numbers from one of those polls recently, this is what I found:

If you look closely at questions 22-25 in that poll, which was taken January 4-7…you will see that Trump does slightly worse against Hillary among Democrats and among black voters than the other leading GOP candidates do…and these are typical of results I’ve seen in earlier polls.

Against Hillary, Cruz gets 5% of the black vote, but Rubio gets 9% of the black vote…Bush gets 6% of the black vote against her. And Trump? 4% of the black vote. Again—except perhaps for Rubio—they all are very similar, but Trump does slightly worse.

Unlike that poll from the first week of January, the most recent nationwide poll available for the general election lacks a breakdown that measures black support for each GOP candidate. But it does have head-to-head numbers for the GOP lead candidates against Clinton (questions 33-35). I’ve listed the Clinton figure first in each case:

Clinton/Trump 51/41
Clinton/Rubio 47/46
Clinton/Cruz 49/45

This is consistent with every national poll I’ve seen during the last few months. Sometimes all the Republicans win in a certain poll, sometimes they all lose (except Rubio, who consistently beats Clinton), and sometimes it is mixed. But Trump virtually always does the worst of the leading Republican candidates against Clinton, and also against Sanders (see this, for example).

So, how to square this with the oft-repeated claim that Trump does well with black voters?

Aside from anecdotal evidence of black voters here and there who have declared support for him, there are polls such as this one from late December, conducted by Clout Research and reported in World Net Daily with the headline: “Minorities line up behind…Donald Trump: new poll shows billionaire solidifying support.” First of all, it’s easy to miss if you read the article quickly, but the poll concerns the Republican primary rather than the general election. Here’s an excerpt from the article:

…[A] new poll, which still has Trump leading the race, shows 40 percent of blacks are lining up behind Trump, as are 45 percent of Hispanics, and even nearly 19 percent of Asians.

Blacks and Hispanics, in fact, even support Trump at a higher level than whites.

The results are from a new WND/Clout poll by Clout Research, a national opinion research firm in Columbus, Ohio. The telephone survey of registered voters was taken Dec. 18-27.

The poll was reported by many writers (this one, for example, who doesn’t mention—and may not even perceive—that the poll refers to the primaries rather than the general). The poll questioned 365 Republicans about which candidate they would support in the GOP primary, and there was a smattering of Democrats (19) and Independents (61) as well. Among all the respondents there were 10 black voters, and the breakdown among them was as follows: 5 for Carson, 4 for Trump, and 1 for Cruz.

That’s how Trump got his “40% of black voters” in the poll—it was actually 40% of the 10 black voters asked who they would vote for in the Republican primary. You don’t have to be a pollster to recognize that this is a meaningless figure because the number of black voters polled was so very low.

The only poll I’ve been able to locate so far that shows Trump pulling a significant percentage of black voters in a general election is this rather curious one from September 6 (which was an awfully long time ago). I call it “curious” for several reasons.

It was conducted by SurveyUSA, which uses automated questions and touch-tone responses rather than live interviews. It never revealed the number of each group within the sample (at least, I could find nothing about it), so we have no way of knowing how many black voters were actually surveyed—and as you could see from the Clout survey I just described, this can be important. And SurveyUSA only asked about a general election between Trump and each Democrat, so there was no way to compare his results with those of the other GOP candidates, because there were no such results. Without all this information, the poll tells us little or nothing about other Republican candidates and their share of the black voters in a head-to-head general election—only Trump, who is reported to draw 25% of the black vote.

This sort of figure is such an extreme outlier—no other poll has found anything remotely like that figure—and the poll’s methodology, and lack of questions about any GOP candidates other than Trump, was unusual enough that, even if the poll were current (which it is not) it would be hard to fathom its meaning. And yet I suspect that some of the claims of black support are based on this poll and the Clout poll (with the four black Trump voters), because I could find no other polls indicating anything about the black vote that would cheer the Trump camp at all.

When the SurveyUSA poll was published, there were a number of articles such as this one that appeared in The Daily Beast. It stated:

Last week, Survey USA released an eye-catching poll showing how Donald Trump would fare in head-to-head matchups against potential Democratic nominees Hillary Clinton, Bernie Sanders, Joe Biden, and Al Gore. The most shocking result was not that he beat all four of the candidates. What made everyone’s jaw hit the floor is that he received more than 20 percent support among African-American voters in every matchup.

However, the article not only neglected to mention any of the odd things about the poll’s methodology, but the writer added, “…why isn’t Ben Carson, the only African-American presidential candidate—who according to a recent Monmouth poll is tied with Trump in Iowa with 23 percent, receiving a similar level of support among African-American voters?”

Why, indeed. Well, actually, that’s an easy one to answer, since the pollsters never asked respondents about a Carson/Clinton matchup, or anything about Carson at all—nor about any other GOP candidate except for Trump. So it would be impossible to know about their black support as measured by this poll, and to compare it to Trump’s.

All we really seem to know so far (other than anecdotal evidence) about the black vote and the Republican candidates, including Trump, is that more conventional polling has consistently indicated a small percentage of black support for all of them in a general election, with Rubio polling best with this demographic at about 9% in the latest poll that bothered to measure it.

[Neo-neocon is a writer with degrees in law and family therapy, who blogs at neo-neocon.]