I'm not high-confidence about this heuristic because 1) the "class label" is a bit subjective and 2) I don't have a massive sample size, but attitudes toward typing is one of the best filters for engineer quality I've come across.
(No, I haven't put this in practice in interviews, and I don't ask any related question. It's just a correlation whose strength I've found really striking)
It's part of a broader class of heuristic I've been noticing recently (again, still a very preliminary, low-confidence model): things like "lack of reliable, quality devtools" and "inability to customize interfaces" bother highly-productive engineers more than less-productive ones, which is what you'd expect given that inefficient interfaces to their work has a far greater cost to the former. Along these same lines, if (eg) you're already pretty bad at understanding complex systems by reading code, you're much less likely to notice or care about how much more difficult dynamic typing makes this.
I should be clear: this isn't a general-purpose argument against dynamic typing, and I'm not suggesting that the situation is as simple as "dynamic typing bad, static good". They both have their strengths and weaknesses, which make each appropriate for different situations; I ended up picking Python for a company whose tech org I was responsible for building, and as much as I personally find engineering in Python to be horribly unpleasant, it was the right choice for the company given the environment it was in, and I would make the same choice in the same situation.
I don't think my theory above is particularly novel, and I understand the self-serving bias that it aligns with, which is part of why it's still such a low-confidence model. But I've been unable to wrap my head around the number of nonsensical defenses of dynamic typing I've seen over the last few years, and this model is the first to offer an explanation: if you can't tap into the benefits of static typing, of course the accessibility and flexibility of dynamic typing is going to seem like it far outweighs the foofy, theoretical benefits that static typing brings to stability and comprehensibility.