Delicious when candied!

I love candied Witches Butter! Fantastic with a raspberry syrup.

Looks like an oysterling, maybe?

I eat them.... They are listed as edible in my guides here in the Pacific Northwest, just need to be thoroughly cooked to break down some unpleasant compounds.

Wow! Are the ones on the light vs shaded side of the trunk the same species?

Tastes like nothing when eatten fresh off the log. You're not missing much but great for grossing out friends.

I agree that there are often ambiguities that are problematic! I used to worry more about a lot of these things, but have come realize that I'm not gonna die on any of these hills. There are no perfect or easy answers to any of this... which is why I still have a job 😁

Intuitively, restricting the sample to the patients with full data does not always remove issues of bias from selecting the complete case samples.

Is the objective to directly compare the models or to assess the relationship between predictor and outcome? If the former, then I agree that is problematic, but if the latter (which is where I mostly see this), then I'd qualitatively evaluate whether the association still holds.

Yes, and I feel like that is pretty standard. The unadjusted model is considered primary with adjusted as auxiliary. There are assumptions regarding reasons for missingness which can be problematic. But outside of that, I'm not sure I see a problem with this out of hand.

This seems usually due to missingness. I tend to think it fine to do complete case analysis but taking care to ensure missingness is completely at random. Seems pretty standard to me...

Location?

I get the point, but honestly, I'm not even sure linear regression should be called regression... seems to be a mistake where regression to the mean was conflated with the process of fitting models... Words and their meanings change..

I think we basically reviewed on these criteria anyway so it's helpful to formalize it. I'm curious whether we will still need to make specific statements regarding rigor of prior research etc.

Which is why we use it when talking to some Docs! Meet them where they're at! 😜

I'm not sure that this is strictly the case? They are tests for equality of distributions and could reject for other alternatives, e.g. differential dispersion, though I agree that they are not are not very good at it.

Totally irrelevant to this discussion, but what is the pronunciation of cauchit? Looks like cow-sh*t...

Do parasitic worms count? Or maybe that's a question for RFK...