Over the course of 2017, SurveyMonkey conducted 605,172 interviews of Americans. A quirk of statistical analysis is that the precision of poll results from a survey of 605,000 people vs. only 1,000 people is small; the former has essentially no margin of error, but the latter has a margin of error of only three points. This is why most pollsters don’t bother polling hundreds of thousands of people. Why spend the money when your estimate is good with far fewer people?
Those 605,172 interviews, though, were conducted over the course of the year, meaning that they offered an interesting look at how Americans’ views may have changed during the presidency of Donald Trump. So the Atlantic’s Ron Brownstein asked SurveyMonkey to pull out how Trump’s approval rating had shifted among various demographic groups.
That analysis, Brownstein wrote, showed “Trump losing ground over his tumultuous first year not only with the younger voters and white-collar whites who have always been skeptical of him, but also with the blue-collar whites central to his coalition.” Most of those who disliked Trump at the outset of his presidency still disliked him, although there was one interesting quirk: About twice as many black men as women approved of Trump’s job performance, meaning that “black men are one of the few groups for which Trump’s 2017 average approval rating significantly exceeds his 2016 vote share.” Given that the vote share among black men was so low — 13 percent, per exit polls — that’s not a tough figure to beat in terms of approval rating. But, of course, vote share and approval rating measure two different things and, therefore, should be compared directly only if you’re paying close attention to those differences.
Or if you’re a media outlet eager to find good news for Trump. And so, on Sunday, Breitbart’s Neil Munro compared them directly and bizarrely.
“Donald Trump’s support among blacks has doubled since 2016, amid racism claims,” the headline for Munro’s piece read. It cites the numbers that Brownstein noted and declares that 23 percent approval from black men and 11 percent approval from black women “averages out to 17 percent, or twice the 8 percent score he was given in the 2016 exit polls.”
This is not how this works. You can’t just “average” the values from men and women, because they may not make up an even portion of the population. Nor can you then compare that average — which evaluates how people view Trump’s performance as president — to how those groups voted — an evaluation of whom people liked better as a candidate, Trump or Hillary Clinton. Trump’s overall approval rating is consistently lower than his vote share because a lot of people liked him a bit better than Clinton as a candidate but think he’s doing a bad job in office. Brownstein isolated the figure among black men because it was unusual in going the other direction.
Any objective observer who is at all familiar with poll numbers would see Munro’s article for what it is. Then there’s Fox News’s Brian Kilmeade.
On Tuesday morning’s “Fox and Friends,” the hosts were discussing a survey showing that most 2017 coverage of Trump’s presidency was negative. Kilmeade interjected with some good news.
“Believe it or not,” he said, “through all this negative coverage, they did a survey of 600,000 people about how black America views this president. His numbers have actually doubled in approval. It’s still low, it’s around 25 percent, but it’s doubled since the election.”
Okay. So. First of all, they didn’t do a survey of 600,000 black Americans. Second of all, Kilmeade clearly thinks that saying “600,000” adds heft to the results, which, as we noted above, it doesn’t. Third, Trump’s approval numbers haven’t doubled, for the reasons above — and then some.
Approval numbers necessarily start only when a president takes office; after all, how are you going to evaluate the job performance of someone who doesn’t yet hold a job? Gallup has asked Americans their views of Trump’s job as president since his first week in office, allowing us to compare approval ratings among black Americans from the earliest point to the most recently available ratings (through the end of 2017).
Trump’s approval among black Americans fell nine points from January to December. Rather than doubling, his approval rating among those Americans was actually more than cut in half, dropping from 15 percent to 6 percent.
Kilmeade had an explanation for why black Americans suddenly loved Trump.
“They’re seeing something, maybe in their pocketbooks or their job opportunities,” he said, “because we know that unemployment for African Americans or black Americans is under 7 percent for the first time ever.”
That part about the low rate is true — but it’s very misleading. It’s not as if black unemployment was 18 percent under Barack Obama and, as soon as Trump took office, it plummeted. Black unemployment fell fairly consistently from 2010 on, as did the rates for whites and Hispanics.
From January to December 2017, the unemployment rate among black Americans fell 1 percentage point. During the same period in 2016, it fell the same amount. In 2015, it fell 1.9 points. The previous year, it fell 1.5 points. The year before that, it fell 1.8 points.
Did black Americans suddenly credit Trump with this slide? Well, if they did, they still were also increasingly likely to disapprove of his job performance, according to Gallup.
On Twitter, Munro claimed that the media’s focus on Trump’s recent racially iffy comments were meant to distract from his theoretical great numbers among black Americans. Of course, the opposite is true: Munro’s article is an attempt to distract from critiques of Trump.
The president seized on it. After spending the weekend at his private resort in Florida, Trump indulged in a little “executive time” on Tuesday morning, meaning that he was watching Fox and tweeting. He clearly lit up when he heard Kilmeade’s depiction of his popularity among black Americans, tweeting:
After washing through Breitbart and then Fox, Trump somehow missed the nuanced context from Brownstein’s original aside.
Bonus irony: After Kilmeade’s tangent from the study showing that coverage of Trump was negative, host Steve Doocy brought the “Fox and Friends” team back to the subject at hand.
“When you look at their bar graphs” of coverage over 2017 (which were actually line graphs), Doocy said, “I think you could say easily: not fair and balanced. Should be closer to 50, but instead it’s closer to 90 and 10.”
Yes, half of the coverage of Trump should be negative, as it is on “Fox and Friends.”