In fact that makes sense lol

If I’m not mistaken, B acts a precision parameter. No backdoors through B, bc arrows originate from B into D and Y (no arrow into B). I don’t think you need B, but you’d get more precise estimates w/o introducing bias. Also, the problem of colliders is conditioning on them, not their presence

To put it in a different way, I worry about the infinite number of assumptions we make, without any comment or even awareness, before we discuss anything. Esp. because many or most of us would disagree with many of those assumptions. Yet, we take them with us

100%!!! I’m currently teaching an intro to Linguistics, and I have this constant feeling that we behave as if methodological constructs (required to a great degree) were not constructs but ontological realities, and, as such, they inadvertently become the (only) targets of our enquiries

I’ve been thinking so much about this lately. Especially relevant for linguistics

A tool that is impossible to educate people about has a tool problem.

A

LOL I read something earlier today that made me think exactly that