Jeremy Labrecque

Yes. You just have to make sure you haven’t induced any biases by conditioning on that subgroup.

Or could conceivably be pregnancy but not birth (loss or given away to adoption). there are many potential meanings of the effect of birth.

Yeah, that is the tough question! Even if you ask a question only about the effect of birth, you can’t have a birth without also having been pregnant (well, you could adopt) so pregnancy is necessarily part of the treatment. And the control arm could either be no birth or pregnancy

Shhhh! Don't anger the gods of perinatal research!

I think there are potentially some issues though with the precise research question and with competing events (infertility, losses) but my own eyes are rolling as I write this so I'll just keep them to myself.

Ignore the competing events of perinatal epidemiology at your peril!

Particularly because I've been in rooms where people say "let's plug that estimate into our model." When I tell them you can't because it's from a predictive model they say "but surely having some number and being able to run our model is better than having no number."

Yeah, my feeling as well. I think they might be useful to find dynamics that we might not find otherwise but I'm really skeptical about their ability to estimate magnitudes of causal effects.

I hope you're successful in getting people to think about their controls!

My suggestion was going to be to simulate the DAG to check. But even better is to post it on bluesky and wait for people like @stephenjwild.bsky.social to do the simulation for you!

Or am I missing something here?

nice work on the paper. But for this one, adjusting for the parent shouldn’t adjust (even partially) for the descendent. The inverse is true: adjusting for the descendent partially adjusts for the parent. We would be in big trouble if we could never adjust for variables that influenced mediators!

I’m just an astronomy nerd so I was pumped about this one

I don’t get this one. Why is adjusting for judge characteristics a problem on this DAG? And why (or how) is judge prejudice being conditioned on?

Did not think I could enjoy M estimator more until you added in astronomy

I like this! But what are the identification assumptions when the research question is an association?

How often do you see in an intro section the authors making a clear distinction bw causal papers they cite and observational ones? All seem to be magically causal. 🤔

There‘a just no way to get any shred of evidence for a causal effect (even if that evidence is just something that makes you slightly more likely to do a second study) without inputing some kind of causal assumption.

But maybe I’m totally out to lunch!

The way I think about this is it only works if you can argue that if X does not cause Y you would have been less likely to observe an association. Which is an argument about the potential absence of biases which is, therefore, a causal argument!

I get most of mine while running or cycling. I had a bad knee injury for a while and I was worried I wouldn't be to run and therefore not have any new ideas.

I both approve and am extremely envious!

Something something Dahlphi process

I particularly like jokes in a peer review where I’m forced to be really critical because I hope it conveys a message like “look, i‘m being tough here but it’s nothing personal. You still deserve a chuckle.”

We don’t just assume new treatments will work, we test them thoroughly.

Why do we think we can just throw AI at a problem and it will just work?

And more jokes during the peer review process. I always like to include one or two

People rarely check the positivity assumption but I’d say 5% of the time it shows something strange is going on that we wouldn’t have caught otherwise.

Will this be my chance to buy you a beer?

I also feel better inside when DAGs flow from left to right.

I could almost tell you the day AI was incorporated into autocorrect. The change was that drastic. Started automatically changing the word “the” to tge. (and right now it autocorrected to thé even though I’m using an English keyboard

"Unusually insightful" is a perfect way to put it.

Thank you for making me aware that this exists