Thread by @WhitneyEpi

I’m so excited! Past couple years, I’ve wanted to sit down & structurally* deconstuct Roland Fryer’s 2019 paper on racial difs in police use of force to use as a teaching example** re: #colliderbias. Alas life got in the way...1/
HT @l_farland
*analyze “structure” in a #DAG

https://twitter.com/StableMarkets/status/1276276257790066694

#ColliderBias^ is rampant in US disparities research that uses admin data like hospital claims, police records. Almost all non-random samples will differentially select by race & other imp factors
^a selection bias with a particular structure in #causalinference #DAGs
2/

Side note: Dr. Chanelle Howe (@BrownUniversity) and I explore a different type of selection bias in this @EpidemiologyLWW paper

ncbi.nlm.nih.gov/pmc/articles/P…

Structural selection bias* is EVERYwhere in health disparities, research, y’all! 3/

*To understand the “structural” in structural selection bias, read this paper (@_MiguelHernan, S. Hernandez-Diez, J. Robins). Then read it again. Like every few months. 4/
scholar.google.com/scholar?hl=en&…

Anyway, I kept meaning to read this paper but never got around to it. But now some #epitwitter-conversant poli sci folks seem to have done it!

**even put fryer paper in syllabus in my social epi course in 2019 to make myself read it but it got pushed out by something else
5/

So far, I’ve just read the @FiveThirtyEight article by @laurabronner (see 👆🏾thread). I like it bc there are graphics. Selection bias is hard, y’all. It’s not intuitive. That’s why I love the structural approach. But graphics help a lot too! 6/

Even many smart experienced #epitwitter don’t get structural selection bias and how it messes with interpretations of race/ethnicity and gender and immigrant and #LGBTQ coefficients in epidemiology. So im always looking for better ways to explain 7/

Question for y’all: Do you find the graphical representation in the @laurabronner @FiveThirtyEight article helpful? Illuminating of some aspect of selection bias DAGs don’t communicate well? How do you teach these concepts? 8/

Also I’ll ❤️ @laurabronner article forever bc of footnote 2: “The term got its name because if you’re drawing a sloppy diagram of the directions in which you think causal effects go, the arrow heads pointing to one of the boxes in the diagram will collide. I wish I was kidding.”

I still want to go back and read the Fryer et al. Commenters are citing sensitivity analyses they did that further restrict on type of encounter. Further restriction is a seductive strategy that sometimes causes its own selection bias. 10/

I want to #DAG it out to decide if these sensitivity analyses really address the causal question of interest 11/

That’s often what our research comes down to. Yes, you can get an answer from a statistical model. But does that model result actually answer the question you say you’re addressing? 12/