Cette année, les bilans financiers annuels de Nintendo et Sony tombent le même jour, donc c'est parti pour un fil qui cumul les deux.

Killer paper intro, from this 2021 paper on arxiv: arxiv.org/pdf/2106.11285

If you drew an arrow ... you have a causal model. If you are calling some variables IV and others DV ... you have a causal model.

This is consistent with earlier psychometric work that suggests 5-7 is the best response scale options, but good to see that the finding holds up in contemporary research. Also, good to see that labeling scales whether anchored or not has little impact on findings. academic.oup.com/ijpor/articl...

Fontana thinking his 6000+ experiments may not have been enough to feel confident in his conclusions in the 18th century while we expect our singular experiments and their standalone replications to do wonders some 250 years later... (excerpt from the intro of Jutta Schickore's About Experiment)

Heureux d'assister à un moment particulier dans le numérique : de plus en plus souvent, des outils libres (pour PC, ou accessibles sur notre intranet) sont bien supérieurs aux équivalents sous Windows (sans être plus compliqués, ni plus moches). (1/4) #OpenSource #LogicielsLibres #FreeSoftware

Evidence of unreliable data and poor data provenance in published clinical prediction models and clinical practice 📣 My first PhD study is available on MedRxiv: www.medrxiv.org/content/10.6...

Edward L. Deci, 83, Dies; Found Self-Determination as a Key to Happiness www.nytimes.com/2026/02/26/s...

Someone should make a Simpson's meme for maximum relevance. In the meantime:

Nature retracted a paper that claimed ChatGPT had a positive impact on student learning. it's not the only paper making this claim that has this problem www.404media.co/nature-retra...

🚬 Si tu voulais vraiment arrêter, tu y arriverais ! Dans « Communiquer en santé mentale : repères pour de nouvelles stratégies et pratiques », Emeline Chauchard nous parle de la complexité du processus de changement. communication-santementale.fr/changer-faci...

My final piece in a 4-part series on vitamin D supplementation. I think that the evidence as a whole shows quite strongly that vitamin D pills are probably a waste of time for virtually everyone. Science! gidmk.substack.com/p/theres-no-...

I'm a cognitive scientist with an interest in epistemic vigilance, and this essay that's been going around gave me pause. I don't think it's straightforward to apply the concept of epistemic vigilance to interactions with LLMs, as this essay does. 🧵/ sbgeoaiphd.github.io/rotating_the...

After learning about collider bias, reading this gave me the second-worst existential crisis about learning anything causal from data *even when you can run experiments*. Really a must-read if you’re not already familiar with this stuff.

Yeah, as psychologists, we are doomed to face a bunch of existential crises bc of what we are (not) taught (even at the grad. level). My fav. example of what Julia so eloquently talks about here is what I call the Stewie Griffin graph, see e.g. p. 230 here czasopisma.uwm.edu.pl/index.php/pp... 🕳️

I have written a piece articulating what I believe is a reasonably new position on AI, or at least one extremely underrepresented in the discourse, and I hope people will read it. The full piece is behind the New Scientist paywall, but I will share some snippets www.newscientist.com/article/2516...

This recent RCT of an "AI stethoscope" claims the technology "shows promise" for diagnosing cardiovascular conditions. It does not. It is a textbook example of the risks of conducting unprincipled 'per protocol analyses'. Once again, peer review at a major medical journal has failed. 🧵 1/

A must-read for anyone interested in causal inference

Science sleuths share their common-sense tips for sniffing out fishy articles go.nature.com/4ceKgVF

Study in Nature that just came out shows a 51 PERCENT drop in Dementia post Shingrix vaccine. That is earth shattering. 19% drop was reported from the older+ less effective vaccine given up to 2019. Whew. So glad I got mine already!

#rstats Here's a useful guide to creating publication-ready #ggplot figures to journal specifications, which is often quite fiddly. jaquent.github.io/2026/02/crea...

New preprint from me argues that p-hacking inflates Type I error rates in some philosophies of significance testing but not in others. osf.io/preprints/ps...

FDA grants breakthrough device designation for schizophrenia/bipolar bloodtests, based on below paper with n=30 per disorder, "using 'leave-one-out' analyses, indicating that the models should be predictive when applied to independent data cohorts." Ouch. pmc.ncbi.nlm.nih.gov/articles/PMC...

Coincidentally, my second paper ever (where I’m first author) was just published yesterday (!) and relates to the difficulties with pre-post differences. bsky.app/profile/vilg...

Like I said 😌 bsky.app/profile/ding...

Find someone who believes in you like researchers believe in pre-post differences as valid treatment effect estimates.

P.S. Pre-post differences are *not* valid treatment effect estimates. Why? Here's a post by @statsepi.bsky.social: statsepi.substack.com/p/one-simple..., here's a post by me: www.the100.ci/2025/01/22/r... >

@matthewfacciani.bsky.social I'm skeptical of this. First, the essay is just speculation, with no evidence that 'plain language abstracts' have any impact. My own experience suggests that this approach will waste resources. A few considerations: 1. Primary scientific publications are inherently […]

Something related:

Great blog by @deevybee.bsky.social on how often PubPeer entries are preceding highly cited retracted articles. Fully agree with Prof. Bishop's description of PubPeer. I am glad that this platform exists because it is a non-annoying way to point out errors/problems.

Pendant que l'Assemblée Nationale se faisait un remake Lacombe Lucien, je suis allé voir Nuremberg. Magistral.

@authentisci.bsky.social is the most exciting tech I've seen in months. Problem: People are unsure of the reliability of science news. Solution: Users click through a browser extension to nominate articles for review by relevant, active experts. authentisci.com/dashboard

An analysis of PubPeer comments on highly-cited retracted articles buff.ly/pWXAj9R

Hate to say it but I believe most, if not all, of my followers can be replaced by AI in 12-18 months

This is just nuts. In a well-powered randomised trial, giving immunotherapy in the morning rather than afternoon/evening produced ludicrously big improvements in progression-free survival www.nature.com/articles/s41... via www.science.org/content/blog...

A while back, I wrote a thing. If you like experiments and causal inference, you should read it:

I'd like to re-up this paper we published last year bec I believe it makes a fundamental contribution to theoretical metascience but it is woefully underappreciated. We address a key challenge in estimating the reproducibility of a result: The distance of a replication study from the original. 1/n

👉🏼 PHQ-9 (ou 2/8) est un outil très utilisé pour évaluer la santé mentale. Quoiqu'omniprésent, il mesure des choses différentes selon les répondants. Pourquoi ? Parce qu’on prête peu d’attention à un point capital : la compréhension des questions par les répondants. jamanetwork.com/journals/jam...

Et si un seul mauvais choix pouvait tout arrêter ? Un “jeu du risque d’extinction” teste comment on décide quand tout peut s’éteindre. psychopapers.kessel.media/posts/pst_3e...

👉🏼 Mé/désinformation : la santé est un des principaux domaines touchés. La santé mentale, devenue sujet d'influence, n'est pas en reste. 👉🏼 Un édito d'Aude Caria dans « Communiquer en santé mentale : repères pour de nouvelles stratégies et pratiques ». communication-santementale.fr/2026/04/21/i...

Where do parents find the time to parent? Less sleep, work and screens. Amazing chart feat. in @alphaville.ft.com Further Reading. homeeconomics.substack.com/p/where-do-p...

New blog post about the age-period-cohort identification problem! In which, for the first time ever, I ask "What's the mechanism?" and also suggest that sometimes you may actually *not* be interested in causal inference. www.the100.ci/2026/02/13/o...

There are 80 title/abstract hits for (nurses health study AND coffee) on pubmed, the first of these from 1998.

what is the charge. what crime has been done

…reproducibility problems are widespread. Only one in three authors (35%) share code, and even when code is available, only every second main result (51%) is numerically reproducible.

Do researchers share their code upon request? Does running their orginal code on the original data produce the original results? We provide evidence in a new Royal Society Open Science publication. Studying more than 1,000 articles which use data from the European Social Survey, we find that... 🧵