FWIW, I would summarize the substantive point as "Given behavior B from agent A where B is differentially characteristic of group G and trait T is also differentially characteristic of group G, don't infer that A has T."

To which I would respond "Nah, go ahead and infer it, but be aware that you might be wrong and keep your confidence levels as well-calibrated as you can."

That said, I might be missing the OP's intended point.