Happy 206th 🎂 to Florence Nightingale b #OTD 1820 (d 13 Aug 1910). Not only founder of modern nursing, she was a renowned statistician & dataviz pioneer. First woman member of the Royal Statistical Society 1859 and ASA honorary member 1875. 1/5🧵
Chat Control's final trilogue: June 29. The "win" (no mandatory scanning of encrypted messages) hides two quieter losses: risk-mitigation scanning + mandatory age verification. Encryption survives. Anonymity doesn't.
Paper update: wildetruth.substack.com/p/encryption...
New methods for estimating effects of mass screenings for low-prevalence problems account for time better than ever. But also raise open questions about mechanisms. Like, is colonoscopist quality a facet of accuracy heterogeneity or strategic behavior?
wildetruth.substack.com/p/when-bette...
NEW PAPER 🚨
"Causal Graphs for Conditional Parallel Trends" with @henripf.bsky.social
It connects causal graphs and the modern Diff-in-Diff literature by introducing Δ-SWIGs as a graphical tool to reason about controls in DiD settings under standard additively separability assumptions.
Thread 🧵
Background on the June 29 endgame — EDRi's explainer, good on the age-verification threat: edri.org/our-work/cha...
One thing that just keeps giving is @vincentab.bsky.social's marginaleffects package (see marginaleffects.com). It's just a complete game changer, moving from "trying to discern what model coefficients tell you after careful recoding" to "you know what, just compare these slopes, thx"
Validation and transportability are different problems. Some mass screenings may be unvalidatable in principle.
h/t @p-hunermund.com for the convo + Mooij/Forré/Magliacane pointers I'm still chasing.
wildetruth.substack.com/p/validate-me
American Statistical Association History of Statistics
Nice critique of mass screening for low-prevalence problems in one of their most popular incarnations, colon cancer screening: www.sensible-med.com/p/the-hubris...
95%CI [last Tuesday, heat death of the universe]
Also, the singularity identifies off the assumption that nothing in the DGP changes when you fully automate the DGP. Bold consistency assumption!
I really enjoyed writing this one because marginaleffects just makes so much sense! It's a bit like with causal graphs -- a great tool to make sense of stuff, so why not teach it early on to make life a bit easier?
