I'm talking specifically about "long division" in Europe... and have heard this factoid from multiple mathematicians including Nasim Taleb. So I expect there is a basis for the claim...
He has bit of a habit exaggerating and elaborating on stories for the sake of making tasty reading. There was a story in Black Swan about two hospitals and a simple statistical riddle that he claimed professional statisticians would get wrong. It wasn't even a trick question, most people would get it right. I found this hard to believe but I was still sort of surprised when the source spoke of a much harder statistics question that was surveyed at a psychologists convention, not statisticians. That was the first (and the last) source I checked on him. I'd have checked a few more but I don't have the book any more.
But now I guess I can add this one to the list of examples.
However, if we are talking about the number of people being able to accurately do 1234 / 56 = 22 r 2. That was regularly taught in the Middle Ages with plenty of examples of such calculations being available.
That’s wildly inaccurate. Arithmetic was very much a subject at several schools with the addition, subtraction, multiplication, and division all being taught. https://en.m.wikipedia.org/wiki/Quadrivium
Long division being difficult in Roman numerals is one of the reasons for the switch.