Don't posit a theory that is only interesting because it implies X causes variation in Y, test for a statistical relationship between X and Y, but then say you are not interpreting causally. Of course you are interpreting it causally. That was the whole point of what you just did.

Mechanically, what this means is that it’s assuming fairly smooth interactions between covariates. So if you want flexible models of interaction you have to build them in yourself. Which is, after all, what you’d expect with BART.

This is definitely interesting. It’s probably worth noting in high dimensions that mixed sobolev spaces are much smoother than usual sobolev spaces. So think of this kernel as closer to a squared exponential than a Matern-1. But still rough enough that the paths aren’t analytic.

Noah Greifer@noahgreifer.bsky.social · 2d ago

Fascinating and clear paper by @corymccartan.com and @melodyyhuang.bsky.social, greatly enhancing our understanding of how Bayesian Additive Regression Trees (BART) works and why it is so effective. A must-read for my fellow BART enthusiasts. #statssky #causalinference

Random tree features are also implemented in my 'bases' software! corymccartan.com/bases/refere...

Bayesian Additive Regression Tree (BART) features — b_bart

Generates random features from a BART prior on symmetric trees. Equivalently, the features are the interaction of a small number of indicator functions. The number of interacted indicators is the dept...

corymccartan.com

Cory McCartan@corymccartan.com · 2d ago

New WP w/@melodyyhuang.bsky.social studying the success of BART models, which regularly win causal inference competitions! We argue that BART should be thought of as a random features approximation to a limiting GP. This view helps understand BART & apply it in more places arxiv.org/abs/2607.28844

Fascinating and clear paper by @corymccartan.com and @melodyyhuang.bsky.social, greatly enhancing our understanding of how Bayesian Additive Regression Trees (BART) works and why it is so effective. A must-read for my fellow BART enthusiasts. #statssky #causalinference

ArXiv Paperboy (Stat.ME+Econ.EM)@paperposterbot.bsky.social · 3d ago

arXiv📈🤖 Seeing the Forest for the Trees: The Gaussian Process Limit of BART By McCartan, Huang

New WP w/@melodyyhuang.bsky.social studying the success of BART models, which regularly win causal inference competitions! We argue that BART should be thought of as a random features approximation to a limiting GP. This view helps understand BART & apply it in more places arxiv.org/abs/2607.28844

arXiv📈🤖 Seeing the Forest for the Trees: The Gaussian Process Limit of BART By McCartan, Huang

Dems strongly favored to take the House in our new election forecast. Senate just better than a tossup: Iowa: 41% Dem Texas: 48 Alaska: 51 Ohio: 53 Maine: 68 Michigan: 69 North Carolina: 84 Georgia: 94 www.gelliottmorris.com/p/democrats-...

Democrats strongly favored to take the House, Senate tilts Democratic in our new election forecast (crosspost)

Our brand-new election forecast for the 2026 congressional midterms puts Democrats at above even odds to take control of both chambers

gelliottmorris.com

New working paper with @corymccartan.com on the Gaussian Process limit of BART! (Bayesian Additive Regression Trees... not the Bay Area Rapid Transit sadly) arxiv.org/abs/2607.28844

Seeing the Forest for the Trees: The Gaussian Process Limit of BART

Bayesian Additive Regression Trees (BART) have shown state-of-the-art performance in both prediction and causal inference problems. Previous theoretical work has attempted to explain BART's superior p...

arxiv.org

#OpenAccess from @apsrjournal.bsky.social - Redistricting Reforms Reduce Gerrymandering by Constraining Partisan Actors - https://cup.org/4pNwXA7 - @corymccartan.com, @chriskenny.bsky.social, @simko.bsky.social, EMMA EBOWE, MICHAEL Y. ZHAO & KOSUKE IMAI #FirstView

Out today in the APSR: our paper studying which gerrymandering reforms work! Long story short, do what Michigan does! Slightly longer story: we project these different complex reforms onto a 1-dimensional "partisan leeway" axis, and do continuous-trt DiDiD there. doi.org/10.1017/S000...

If there is a practical difference then people should just talk in simple terms about what is to be done. The dictionary definition stuff is largely exhausting, pedantic navel gazing

this elite mindset is tedious in the extreme and desperately unaware of how little correspondence it has with empirical mass opinion (including the actual 'middle of the road' voter in the wild), but it would also be less of a problem for the rest of us under proportional representation

Luke O'Neil@lukeoneil47.bsky.social · 2w ago

👍 archive.is/YLulE#select...

Thrilled to share that I received POLMETH's Statistical Software Award for 2026 for the #rstats package geomander. It's a performant toolkit for working with spatial data, especially election and demographic data. Learn more about the package at christophertkenny.com/geomander/

Geographic Tools for Studying Gerrymandering

A compilation of tools to complete common tasks for studying gerrymandering. This focuses on the geographic tool side of common problems, such as linking different levels of spatial units or estimatin...

christophertkenny.com

Peacetime politics involves less physical danger than war, but it has some dangers in common: both require what Clausewitz calls “moral courage” by which he means the willingness to accept risks and moral responsibility. This is what I saw lacking on that panel. 1/🧵

Liberal Currents@liberalcurrents.com · 2w ago

"I had only one reaction to the panelists and their cheerleaders: these gutless losers will never win. If they have their way, we will, at best, remain in this crisis for a generation longer, and at worst we will simply lose the war." www.liberalcurrents.com/tit-for-tat-...

Here's something I post from time to time. My answer to a reader who asked me: what could journalists do NOW to break with some of their more corrosive habits.

Interesting read. I think both can be true: (a) the public reads more than ever, and (b) standardized tests, phones (TM), & pandemic have hindered college students' ability to engage with longer works. Academics see (b) and are receptive to hearing that reading has cratered everywhere

I have just resigned from the board of "Statistics and Computing", along with 17 other Associate Editors. This was a difficult decision: this great journal has published many tremendous articles under the leadership of EiC Ajay Jasra, and before him David Hand, Gilles Celeux, and Mark Girolami.

If there’s an issue Democrats could run on and win in a landslide, it’s anti-corruption. But first the Dem establishment needs to deal with its own corruption problem. It’s not enough to be right about Trump’s corruption—they also have to be credible.

Had a great time presenting Shiro Kuriwaki's and my new EI methods to EPSS in Belfast on Saturday! Our R package implementing all these methods (as well as partial ID bounds) is also now on CRAN! corymccartan.com/seine/

New R package on CRAN! ONNX is a runtime & file format for ML models. 'onnxr' lets you load & run models in about 2 lines of code! E.g. image detection running in ms from a pretrained model. Perfect for embeddings, smaller models, etc. And lots of .onnx available online! corymccartan.com/onnxr/

This take may ruffle some feathers. Not only do political parties expel members all the time, but the ability to do so is absolutely crucial for maintaining political cohesion and protecting against demagogic capture. Like would anyone on this platform object if the Democrats expelled Fetterman?

Few people are putting this so plainly. This is what is happening. Miss me with partisanship, as if partisanship isn't a vessel for achieving policy goals — including racial ones

So I urge you to look at the folks who will inevitable call for us to privatize social security in the wake of this report and check for the overlap with the folks who brought us the surging inequality and a sluggish economic recovery that got us here. 4/4

This is truly one of the worst things the court has done — both in terms of legitimacy and in terms of the law — since I have been covering the court.

Chris Geidner@chrisgeidner.bsky.social · 2mo ago

BREAKING: In an unsigned, per curiam opinion on the shadow docket, the SCOTUS conservatives allow Alabama to use a congressional map held repeatedly by a lower court to have been enacted with discriminatory intent.

Roberts court: "the mere fact of racially polarized voting is not relevant to proving racially polarized voting patterns"

Our VoteShield team is seeking a civic-minded Data Scientist to analyze voter registration data and support election officials nationwide to maintain free and fair elections. 💰$84.3K–$92K 📍Remote (US-Based) Bring your Python skills to protect free and fair elections: 🔗 https://protdem.org/3S1RFQ1

Another clear centering of Trump’s emotions from a few days ago: the number of headlines and articles that referred to E. Jean Carroll that referred to her as an “enemy” or “adversary” rather than “victim.”

Larry Glickman@larryglickman.bsky.social · 2mo ago

Yet another article centering Trump’s emotions. We learn here that he is ”angry,” “irate” and that he issued a “scathing rebuke” of the decision. www.nytimes.com/2026/05/30/a...

Newly updated simple congressional model (tinyurl.com/cmchousemodel), now with LA map & recalibrated election model fit only to 2026-2024 data! Ds have lost 4 seats on avg to redistricting, but the loss grows to 5.6 seats in a D+6 environment. Ds 98% to win house in a D+6 environment, though

A small 🧵 about frustrations and opportunities w/ open-source research software for the social sciences: Came across a new paper in a top methods journal, explaining/demo'ing a new R package. Sounded v. cool, so was excited to dig in. What I found was largely a mess...