Republican presidential candidate businessman Donald Trump speaks during a campaign pep rally, Friday, Aug. 21, 2015, in Mobile, Ala. (AP Photo/Brynn Anderson)

Donald Trump did a tweet, as Donald Trump does, focused on a new poll that shows him scooping up a staggering one-fourth of the black vote. Here's the tweet Donald did:

The link in that goes to a site called "The American Mirror," which I will admit to having never heard of prior to clicking the link. But as with Trump's often-cited poll about how well he's doing with Nevada Hispanics, it's worth tapping the brakes a bit on this one, too.

The American Mirror post basically pulls out a bit of data from a recent SurveyUSA poll that shows Trump getting 25 percent of the black vote against Hillary Clinton. That compares favorably with past Republican candidates in general elections; George W. Bush's 11 percent of the black vote in 2004 was a recent high. This is more than twice that!

And now the caveats:

Caveat 1: Donald Trump and Hillary Clinton have not completed a general election campaign

This is a variant on the old "early polls don't mean much" argument. This sub-point from an early poll doesn't matter either.

It's just more than 100 respondents who identified themselves as black -- one-quarter of whom said they'd vote for Trump over Clinton in a year and two months. There's a lot of campaign before then -- even assuming one or both of those people gets his or her party's nomination.

Caveat 2: Trump-Clinton results are all over the map

Other recent polls have also asked about Clinton and Trump. The percentage of support from black voters varies widely. (The other polls we looked at are from CNN/ORC, conducted July 22-25, Fox News, August 11-13, and Quinnipiac University, August 20-25.) On all of the graphs that follow, the Democratic share of the vote is shown in blue; the Republican, in red.


In Quinnipiac's poll, Trump gets 3 percent of the black vote. In Fox's, 10 percent. In CNN/ORC's, Trump gets 24 percent of the vote -- among all non-white voters (Hispanics, for instance, lean Democratic but not nearly as much as African Americans). That poll is also about a month older than SurveyUSA's. SurveyUSA is something of an outlier on this question, showing much more support for Trump among black voters than Quinnipiac or CNN.

Notice too that SurveyUSA has a much higher percentage of people who said that they didn't know how they would vote -- in other words, there's a lot of empty space between the bars showing support for Trump and Clinton and 100 percent of respondents. This is likely a function of the methodology used by SurveyUSA.

Caveat 3: All the other Democrats do about as well against Trump

There's nothing special about this Trump-Clinton matchup. Trump gets about the same percentage of the black vote against every Democrat in the survey -- between 21 and 26 percent.

Trump also gets more support from black voters against Bernie Sanders, for example.


Caveat 4: Clinton does about the same with black voters against other Republicans

The flip side of that is also true. SurveyUSA only asked about Democrats against Trump, but in the other three polls, we can see how Clinton-Trump compares to, say, Clinton-Bush.


She gets basically the same level of support from black voters in each contest.

Caveat 5: All of which means that most of this is partisanship

Combining caveats 3 and 4 suggests that the split is largely a function of partisanship: black voters prefer Democrats over Trump at consistent levels and Clinton over Republicans at consistent levels -- whatever those levels happen to be.

We can see that more clearly when looking at Sanders versus Republicans and Democrats versus Trump. Sanders fares the same against Trump and Bush in Quinnipiac's poll. He and Clinton fare the same against Trump in SurveyUSA's.


In short? SurveyUSA's poll probably is mostly showing partisanship by black voters, and the specific numbers reflected in that group are outliers from past findings, perhaps due to methodology.

In less short? Trump is cherry-picking, in order to make the case that his rhetoric isn't hurting him and the party with non-white voters. Maybe, over the long run, it won't. But this one poll doesn't show that by itself.

We do not expect Trump to include that nuanced point in corrected future tweets.