Went to look up textbook results after getting the nagging feeling that an ML paper was reinventing classical ideas, and found this gem: "Not reading to the end of Le Cam's papers became not uncommon in later years. His ideas have been regularly rediscovered." At least they're in good company.

From twitter: A short thread: It amazes me how many crucial ideas underlying now-popular semiparametrics (aka doubly robust parameter/functional estimation / TMLE / double/debiased/orthogonal ML etc etc) were first proposed many decades ago. I think this is widely under-appreciated!

"Randomized trials should be used to answer any causal question that can be so studied... But the reality is that observational methods are used everyday to answer pressing causal questions that cannot be studied in randomized trials." - Jamie Robins, 2002 tinyurl.com/4yuxfxes tinyurl.com/zncp39mr

What's the best paper you read this year?

Thank you Alec for leading this project, I learned a lot! This paper has a very useful study of what contrasts are feasible in situations with many treatments and positivity violations, including necessary assumptions and efficient one-step estimators. Check it out!

New-ish paper alert! arxiv.org/abs/2410.13522   We tackle the challenge of comparing multiple treatments when some subjects have zero prob. of receiving certain treatments. Eg, provider profiling: comparing hospitals (the “treatments”) for patient outcomes. Positivity violations are everywhere.

Found slides by Ankur Moitra (presented at a TCS For All event) on "How to do theoretical research." Full of great advice! My favourite: "Find the easiest problem you can't solve. The more embarrassing, the better!" Slides: drive.google.com/file/d/15VaT... TCS For all: sigact.org/tcsforall/

@bonv.bsky.social presented this at NYU this week -- terrific work with an excellent presentation (no surprise there)! I found the connections to higher-order estimators and the orthogonalizing property of the U-stat kernel fascinating&illuminating.

Should we use structure-agnostic (arxiv.org/abs/2305.04116) or smooth (arxiv.org/pdf/1512.02174) models for causal inference? Why not both? Here we propose novel hybrid smooth+agnostic model, give minimax rates, & new optimal methods arxiv.org/pdf/2405.08525 -> fast rates under weaker conditions

I see renewed discussion on #statsky about the interpretation of confidence intervals. I will leave here this quote from Larry Wasserman's All of Statistics, which I love. Controlling one's lifetime proportion of studies with an interval that does not contain the parameter is surely desirable!

Reminder/plug: my graduate-level monograph on "Topics and Techniques in Distribution Testing" (FnT Comm. and Inf Theory, 2022). 📖 ccanonne.github.io/survey-topic... [Latest draft+exercise solns, free] 📗 nowpublishers.com/article/Deta... [Official pub] 📝 github.com/ccanonne/sur... [LaTeX source]

🔥🔥🔥 from Chris Adams's "Learning Microeconometrics with R:"

"There’s no way you can just sit down & do a `big thing', or at least I can’t. So I just went back to doing lots of little things, & hoping that some of them will turn out okay. Statistics is a wonderfully forgiving field... all you have to do is get an idea & keep at it." - Brad Efron #statsquotes

Kandiros, Pipis, Daskalakis, and Harshaw have a really Interesting new arxiv preprint on "conflict graph designs" for interference/spillovers: arxiv.org/abs/2411.10908 For GATE estimation the improvement is very significant and I'm optimistic/excited about how the ideas will impact the literature..!

The European Causal Inference Meeting 2025 is coming to Ghent! ✨ Share your work with experts across the globe – abstract submission for oral & poster presentations is now open! eurocim.org/abstracts.html

New paper! arxiv.org/pdf/2411.14285 Led by amazing postdoc Alex Levis: www.awlevis.com/about/ We show causal effects of new "soft" interventions are less sensitive to unmeasured confounding & study which effects are *least* sensitive to confounding -> makes new connections to optimal transport

As my first post on this platform, allow me to advertise the RL theory lecture notes I have been developing with Sasha Rakhlin: arxiv.org/abs/2312.16730 (shameless repost of my pinned tweet)

What’s the best way to structure a quantitative methods sequence? Our current take is roughly: 1) Probability/Inference/Regression 2) Causal Inference 3) Model based inference (MLE/Bayes) 4) Machine Learning

In this paper we consider incremental effects of continuous exposures: arxiv.org/abs/2409.11967 i.e., soft interventions on cts treatments like dose, duration, frequency it turns out exponential tilts preserve all nice properties of incremental effects with binary trt (arxiv.org/abs/1704.00211)

This work was led by Alex Levis - an amazingly talented postdoc, who I've been lucky to work with on a surprisingly wide variety of really interesting causal inference problems www.awlevis.com/about/

This paper was so fun: arxiv.org/pdf/2301.121... We give new methods for estimating bounds on avg treatment effects - trt is confounded, but an instrument is available. Super common in practice The bounds are non-smooth, so std efficiency theory isn't applicable Lots of useful nuggets throughout!

Very excited about this paper! arxiv.org/abs/2305.04116 We study if one can improve popular semiparametric / doubly robust / DML causal effect estimators - w/o adding structural assumptions... Short answer: nope! Turns out these methods are minimax optimal here www.ehkennedy.com/uploads/5/8/...

Wow things seem to actually be taking off here…

there are *surprisingly many* open problems when it comes to theory/methods in causal inference check out this talk by Siva Balakrishnan for an excellent & comprehensive summary of the state of the art www.youtube.com/live/Mnum0Ox... www.stat.cmu.edu/~siva/

CMU Stats & Data Science is hiring! www.cmu.edu/dietrich/sta... forms.stat.ufl.edu/statistics-j...

Short story - the ideas behind “causal ML” and “double machine learning” go back at least 40 years Here is an estimator from a 1982 textbook that today would be called double machine learning or something similar

From twitter: A short thread: It amazes me how many crucial ideas underlying now-popular semiparametrics (aka doubly robust parameter/functional estimation / TMLE / double/debiased/orthogonal ML etc etc) were first proposed many decades ago. I think this is widely under-appreciated!

Awesome. This interview is also a gem www.youtube.com/watch?v=IIO2...

Some amazing quotes by Herbert Robbins here: jiayinggu.weebly.com/uploads/3/8/... "Why does it take so long? Why haven't I done ten times as much as I have? Why do I bother over & over again trying the wrong way when the right way was staring me in the face all the time? I don't know."

Larry Wasserman’s talk on “Problems with Bayesian causal inference” youtu.be/sZyyaNdvfto?...

One of my pastimes is to watch videos about writing fiction. Many things that work when writing fiction also work when writing academic papers! I think this video on editing is quite useful: youtu.be/WLAmilJx3Us?.... E.g., accept that your manuscript will evolve a lot over time!

Check out this great paper by Mateo Rubio (scholar.google.com/citations?us...), rigorously estimating causal effects of the "cycle of violence" Superb example of how to tell a story including average effects, heterogeneous effects, & sensitivity analysis (i.e., relaxing assumptions abt confounding)

Excellent news! blog.arxiv.org/2023/10/20/a...

CMU Stats & Data Science is hiring! apply.interfolio.com/134121 www.cmu.edu/dietrich/sta...

I have some line in my first set of slides like "Not every economics research paper is estimating a causal quantity. But, the implication or takeaway of papers is (almost) always a causal one." I should just quote Wasserman!

"There are two types of statisticians: those who do causal inference and those who lie about it." - Larry Wasserman #statsquotes www.jstor.org/stable/26699...

"Statisticians are engaged in an exhausting but exhilarating struggle with the biggest challenge that philosophy makes to science: how do we translate information into knowledge?" - Stephen Senn #statsquotes

Econometrics Thread (#EconSky) Today, I will talk very briefly about a few recent methodological papers that I think are super useful to applied researchers. Basically, below, you will find some new tools that may help you to answer relevant empirical questions. +

Matteo Bonvini made an R package for our “proportion of unmeasured confounding” sensitivity approach here: github.com/matteobonvin... Paper: arxiv.org/abs/1912.02793

"Sensitivity analyses can remain agnostic about [causal] structure. This is one reason they are useful; they adjudicate debates bc people can agree on validity w/out reaching full agreement on what constitutes plausible causal knowledge." - Aronow & Savje #statsquotes arxiv.org/pdf/2003.116...

Model-Agnostic Covariate-Assisted Inference on Partially Identified Causal Effects http://arxiv.org/abs/2310.08115 Many causal estimands are only partially identifiable since they depend on the unobservable joint distribution between potential outcomes. Stratification on pretreatment covariates c 📈🤖

this new Interactive Causal Learning Conference looks pretty great: interactivecausallearning.com/2023/#speakers can't go wrong with people like: Judith Lok, Eli Ben-Michael, Roshni Sahoo, Mats Stensrud, Mark van der Laan, Linbo Wang, etc etc. bet their talks will be 🔥🔥🔥

This paper was so fun: arxiv.org/pdf/2301.121... We give new methods for estimating bounds on avg treatment effects - trt is confounded, but an instrument is available. Super common in practice The bounds are non-smooth, so std efficiency theory isn't applicable Lots of useful nuggets throughout!

"Don't worry about the overall importance of the problem; work on it if it looks interesting. I think there's a sufficient correlation between interest and importance." - David Blackwell #statsquotes en.wikipedia.org/wiki/David_B...

Cool thread! MTE/LIV is pretty neat - very natural & advantageous w/ cts IVs I worked on semiparametric methods for this effect here: arxiv.org/pdf/1607.025... Also interesting statistically - a ratio of derivatives of partially averaged regression functions, leading to some quirky theory/methods

minimax & mountaineering causal inference is like inca climbers in 1400s - some methods have nice error guarantees, but we don't know if better errors / higher summits are possible very surprising how little is known about optimality in causal - eg ATE in smooth model w/ unknown covariate density

The Harvard Data Science Initiative has launched its application for the 2024–2025 Postdoctoral Fellowship. Share with anyone who is a good fit; and have them reach out if their interest aligns with mine (e.g., causal inference, fairness, model validation, anomaly detection, GenAI, etc.).