Yesterday, Aaron Blake of the Washington Post's "The Fix" posted a thoughtful piece on misleading polling, especially that of Morning Consult's — an outfit, I have remarked of late, that "I have had little trust in," but "perhaps there's some hope for them." Boy was I wrong.
Blake looks at its recent poll …
that supposedly proves … Trump’s net approval rating [has sunk] in every battleground state…. It’s overwrought…. Those shifts look pretty damning. He has lost 18 points off his old net approval rating … in New Hampshire, 20 points in Wisconsin, 18 in Michigan, 18 in Nevada, 26 in Arizona and even 23 in Florida….
What could account for such shifts? The answer: because [they're] being compared with the very beginning of his presidency, which was a high point of his presidency…. Pretty much every president is popular upon inauguration, meaning that if you compare their later approval ratings to that, you’re likely to find some sort of regression…. The biggest problem is that this creates the appearance of change, when Trump’s numbers have shown very little of that. He has had some of the most remarkably static approval ratings for a president, in fact, and there is little reason to believe that has changed much in recent weeks, months or really since the start of 2018….
A better comparison than to Trump’s inauguration would be … before he got the inaugural bump. His favorable rating on Election Day 2016 was minus-29 in Wisconsin, minus-20 in Michigan and minus-14 in Pennsylvania, according to exit polls — all worse than his new approval numbers in those states….
Trump may be in trouble, but this is hardly adding to the data that demonstrate that. And when you see polls suggesting some kind of big shift in Trump’s numbers, you should always be wary.
My most enjoyable class when I was an undergrad student double-majoring in political science was on public opinion and polling. My enjoyment of these topics perseveres, so forgive me for my rather intense concentration on political polling.
But this stuff is important, it seems to me, no matter how often a columnist or television commentator insists that it's far too early to be responsibly polling the presidential race. I disagree, obviously, because early, aggregate polling establishes a kind of baseline from which we can monitor trends. One poll here and another there may not mean much, but when 20 polls suggest that Trump is in really big trouble, then odds are, he's in really big trouble.
Thus even though Morning Consult appears not to know how to poll, if 20 other polls lean to its findings more professionally, then we have something we can pretty much rely on — and keep our eyes on.