Post a banger that isn’t in English youtu.be/OG2VEI0yPW8

This is so nicely said: It's the authors' responsibility to tell the reader how they might interpret an association.

Sometimes I hear, "we need to frame our research as associational otherwise the MDs reading it will not interpret it correctly." It took me a while to realize how bonkers this is. It's saying some important primary users of the knowledge created are unable to understand what they're reading.

DAGs describe discrete snapshots. Most causal mechanisms of interest in science on humans and other living things operate continuously. The two may not correspond. Is it time to panic? wildetruth.substack.com/p/can-we-bel...

Cool paper from @keling-wang.com showing that 35% of randomly selected diabetes guideline recommendations use stronger causal language than the papers they cite. Such "causal jumps" are not necessarily unwarranted but, at best, lack of transparency in how causal conclusions were arrived at.

'to treat peer review as a throughput problem is to misunderstand what is at stake. Review is not simply a production stage in the research pipeline; it is one of the few remaining spaces where the scientific community talks to itself.' 1/3

This also goes for scientific conferences in the USA: just, no.

I almost appreciate how Trump has exposed how vacant every single high profile conservative belief was: free speech, states’ rights, letting the market decide, NATO, the right to bear arms, democracy, all of it. It’s been an empty, malicious political project for decades and decades now

Really hard to look at this and all the other examples of bravery and kindness from ordinary people in this moment not think also of the tremendous cowardice exhibited over the past year by some of the richest, most well-protected people in business, media and politics

1. This short thread is about the issue that dominates almost every aspect of politics, and causes or exacerbates most of our problems: the extreme wealth of a small number of people. Here’s the amazing thing: almost the entire political class aligns with the ultra-rich against the rest. 🧵1/10

If you are a Canadian, and you travel to or support the United States in any way, you are a traitor.

Posting for no reason: european-alternatives.eu

My apologies to people named Al because I keep reading your name as A.I. now.

It's funny when people write "these results suggest that exposure and outcome are associated" when they clearly found they were associated. What they mean, I think, is the results suggest the possibility of causation.

*researcher is unable to answer and is hurled into the abyss*

Are you or a loved one LGBTQ+? Or does your work serve LGBTQ+ communities? Take this Survey from The Cancer Network and share how this past year has impacted you! I took it and am looking forward to seeing how the data can and will be used to advocate for our community.

Having trouble with time zero when using healthcare databases to emulate a #TargetTrial? See our review of procedures to align eligibility and treatment assignment in observational emulations. We use 3 target trials of increasing complexity and provide a decision diagram www.bmj.com/content/392/...

Who's ever seen LLM output say, "we don't know"? And yet so often that is the best response. There's so much that isn't known in medicine and really even official resources could be a bit clearer about that. www.theguardian.com/technology/2...

It's super easy to exercise your choice as a consumer. And if more people did, these companies would be more receptive to change. I switched from Spotify to Qobuz last year and am loving it. They also pay artists more than other streaming services.

*Determinant* is a synonym for cause. But it seems implausible for your voting behaviour to cause your mortality risk (beyond large scale macro changes to the government in control). More likely it's a complex multidimensional marker of your overall circumstances which in turn predicts your risk.

Here's a question for the causal inference people: What is your take on causality without counterfactuals and/or causal inference without counterfactuals?

I reckon CI does two good things to a field, neither of which are philosophical or even technical: first, it obliges it to admit it actually wants causal knowledge. And second, demands it says what that knowledge amounts to in concrete interventional terms (or admit that it doesn't quite know).

Books I read this year that were particularly good I will never do a top ten list you can't make me rank art.

‪It has a name now 😜 Many thanks to Ken for agreeing to put his good name to my...artwork. The image is in the public domain (CC 0), but citations to the linked documents are warmly welcomed. ✅ zenodo.org/records/1808... ✅ pubmed.ncbi.nlm.nih.gov/24452418/

Poverty is a policy choice. Concentrated wealth is a policy choice. Inequality is a policy choice. None of it is natural or inevitable. Remember: We have the power to build a system that serves the many, not the powerful few.

I don't think the claim is puritanical, to be honest. X is a digital platform built on network effects, so every user who posts there creates a negative externality by amplifying its owner, who is openly hostile to democratic rule and the Western liberal order.

One of the major problems with epidemiologic research is that we focus so much on the statistics that many (most?) researchers think that "doing research" means running the regression model and not the actual research part, which is everything described in this historian's post. 1/n

These tools would be useful if they help the researcher achieve equal or better quality result faster or with fewer resources. But we have no idea whether they do that. No idea. And little idea of other potentially negative impacts of these tools like eroding our ability to think…

suspect a big reason why many academics and others who work in areas where getting facts RIGHT is key are disinterested in using LLMs for research: they’ve tried it, they keep noticing major errors in output, and they conclude that having to verify all that doesn’t actually save them time.

Me, in many talks: Use causal language for this type of thing and use associational language for this other type of thing. Someone, inevitably: Why are you trying to ban the word association? Me: I literally just said "use associational language...".

As an editor I often see authors fall into a trap of stressing some fancy statistical technique for estimating causal effects. Often neglected are the necessary assumptions to map the data to the estimand. @pausalz.bsky.social and I wrote about it here www.bmj.com/content/391/... #EpiSky

🚨Job alert! We're hiring experienced programmer to analyze national Medicaid claims data, along w/ colleague Ari Ne'eman @hsph.harvard.edu. Details & application instructions: careers.harvard.edu/job/statisti... @societyforepi.bsky.social @popassocamerica.bsky.social @ashecon.bsky.social @apha.org

I wish we were as skeptical that we've found artificial intelligence as we are skeptical that we’ve found alien intelligence.

Yeah. And before this, half of reviewers were submitting trivial, short reviews that either said "looks good to me" or "this is excrement". This is why reviews should be part of the public record for any paper. www.nature.com/articles/d41...

A lot of people seem to be under the impression that causal inference mak you more certain about causation. It sometimes can but often it just shows you all the ways your study could be spectacularly wrong.

Such an important piece highlighting the impacts of recent events on trans and nb researchers. We are at risk of losing lot of incredible talent from the academic pipeline. The ripple effects will be felt for decades.

Rewatched some Star Trek tonight. Still wild how a show about warp drives and weird nebulae ends up teaching the cleanest lessons about being human. Every episode says the same thing in its own way: ‘Be curious. Be brave. …Try not to fire phasers unless you really have to.”

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. 🤔

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?

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.

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

Recently I heard someone say “we ran an RCT based on our associational studies and unfortunately found null results.” How did we get in a place where we don't see the problem in using associational evidence (which cannot even say the word cause) to inform RCTs.

If you have a manuscript ready to submit in mid-December do you: a) Submit in December (when editors are bombarded with manuscripts) b) Hold off until mid-January to let the editors have a damn holiday already

If the world is as complex as advocates for complexity research say (and it may be), then everything is hopelessly confounded and we can't estimate the parameters that you would plug into a complex model.

When fixed effects aren't fixed... Important work by @dlmillimet.bsky.social and Marc Bellemare that should influence econometric practice. onlinelibrary.wiley.com/doi/10.1111/...

this is one of my favourite observations about sample size calculations. (afaik first articulated by Miettinen in 1985)

What's it called when you're waiting for the bus and there are two different buses that could take you home? A doubly ro-busstop.

Should you include people in your study who have no opportunity to be exposed? Like someone with an absolute contraindication? Turns out the answer is...it depends. There are conditions under which excluding them can actually increase bias through bias amplification.

Great perspective from #CDSM2025: causal inference is 'what-if' analysis. You don’t have to get everything perfect—or how dare you use the c-word. What matters is laying out your assumptions transparently and showing us what happens when they're violated.

🥳 Registration for abstracts for EuroCIM 2026 (Oxford) is now OPEN and the deadline for submissions is 9 January 2026: eurocim.org/oxford-2026/... 👉 Theme? “Causal inference in health, economic and social science” 👉 When? April 14-17 👉 Where? Oxford 👉 Register? eurocim.org/oxford-2026/...