The problem is it takes a snapshot and people immediately compare it with what they assume was the case years ago. It's invariably worse now than before. What we remember from years before isn't a random sample--those who read such surveys and write such stories are typically educated, and remember hanging out with educated people.
The debunking was to look at prior surveys that asked the same or equivalent questions. In other words, compare not data against an assumed past but against the real past. Usually the results are better now than before, but that's not much of a story. Teachers may care, but nobody else: Neither politicians nor advocates get any advantage from saying, "We're doing better but have a long ways to go." It's nuanced. It's better just to say, "A quarter of our kids ___________". Those educated get no ego boost from knowing that they're less special than they used to be, and those they look down are decreasing in number.
Yes, the numbers are better for most developed countries than for the US. But still, a survey last year had people not knowing what the Earth's orbital period was. "Year." The problem was vocabulary, of course, but the article I read billed it as not understanding the definition of "year"--not the definition of "orbital period."
= new reply since forum marked as read