A hobbyhorse of mine. I think that one great source of confusion in causal analysis is that causality itself is not a statistical concept, but it so happens that a lot of our statistical toolkit(s) are necessary to help us empirically analyze causal questions.

this rings to me very close to scientists hoping if they do everything by the book the nature will confess its secrets without a doubt. there is no book that can conveniently remove all uncertainty. we are never promised answers. human relationships are way more precarious. so many unknowns,

There is no "science-wide replication crisis" because every word in that phrase carries unfounded assumptions. 🧵 1. There is no evidence that progress has generally stalled in the sciences. Also, no one has actually tried to estimate replicability across fields.

I was just reminded of this thread and given this new "Science: A New Golden Age" report, I think it's time to re-up it. We should all stop attributing a number of different observations, symptoms, grievances, and vibes about science to a replication crisis. That would be the responsible conduct.

Berna Devezer@devezer.bsky.social · last yr.

There is no "science-wide replication crisis" because every word in that phrase carries unfounded assumptions. 🧵 1. There is no evidence that progress has generally stalled in the sciences. Also, no one has actually tried to estimate replicability across fields.

One thing I think that think conversation is often missing (which Frank covers in this post, albeit in different language) is that we need to define characteristics for which we want our sample to be "representative" of the target population and then *explicitly state how we achieve it.*

Frank Harrell@f2harrell.bsky.social · 3mo ago

Good to be annoyed. It's only a minor of clinicians who understand that what's important is representativeness on interacting factors that we forgot to interact with treatment: www.fharrell.com/post/ia/

New blog post: "Conviviality in computational science" blog.khinsen.net/posts/2026/0... "Conviviality matters for science ... if you want to derive knowledge from your work, you need to know exactly what you are doing, and that includes a detailed understanding of your tools." 🧪 #metascience

> Conviviality can only happen if a majority of a community adopts it as a value. What small groups of people can do, however, is develop and use convivial tools at their modest scale, to [...] provide a model that others can learn from if they want to. — @khinsen.net

Konrad Hinsen@khinsen.net · 3mo ago

New blog post: "Conviviality in computational science" blog.khinsen.net/posts/2026/0... "Conviviality matters for science ... if you want to derive knowledge from your work, you need to know exactly what you are doing, and that includes a detailed understanding of your tools." 🧪 #metascience

Clear points about why AI-generated summaries are problematic despite claims by some that they're supposedly less problematic than other genAI use. Can I summarize this long text for you? NOPE, NOPE, NOPE

Maarten Paulusse@maartenp.bsky.social · 7mo ago

Tired of AI hype posts? You might like my sober assessment of whether AI-generated summaries are suitable for studying and research. Spoiler alert, they are not. The text is primarily aimed at students and researchers, but has much broader relevance. So share freely!: www.tue.nl/en/our-unive...

People always say "tell a story" but then don't explain how to do so! How is that helping? If we knew how to tell a story we wouldn't need the advice in the first place. 1/

Stephen Turner@stephenturner.us · 3mo ago

How to give a bad talk www.nature.com/articles/s41...

How about a “do better statistics” movement? Or “be clear about what you did and saw” manifesto? Or “share your fuckin phage libraries” declaration? Or, critique people’s data on PubPeer? Proviso: I don’t have any familiarity with psychology, where I guess this fooferaw started.

@devezer.bsky.social has the hands-down best perspective on this “reproducibility crisis” people like to bray about. Statistical rigor, data access, transparent experimental design, research material sharing have always been crucial and sometimes problematic. The “crisis” is a fake moral panic.

pasted in text, all spacing changes, inconsistently throughout the doc. manually updating headings one by one since document-wide styles breaks spacing. added a numbered list and the numbers are in a different font than text. wtf do people know that normal software doesn't act this nuts???

ATTENTION EVERYONE! I sewed Nova a bucket hat🥹🧵🪡

At long last, our meta-analysis covering close to 70 studies conducted over the last 25 years on phonetic convergence between speakers in speech is out in the Journal of Phonetics authors.elsevier.com/c/1nEdMLix~5... @laboratoire-parole-et-langage @ilcb.bsky.social @montclair-state-university

Testu zaharretan laburdurak ohikoak dira, latinez bereziki. Euskaraz, haatik, zailagoak dira aurkitzen. Horra bi adibide, #Munarritz-en (#Goñibar) agertu berri den testu honetan (XVIII.-XIX. m): ▶️ lezaq > Lezake ▶️ gurutceareq > Gurutzearekin #euskara

Wonderful news! Journal of Research on Research is launched. "if you are a researcher who researches any aspect of research, then this is the journal and community for you." Worth reading the whole editorial for a transparent and thoughtful account of this amazing community-driven effort.

Research on Research Association@roration.bsky.social · 4mo ago

We're excited to launch J·ROR, the Journal of Research on Research. A new open-access home for research on how research is funded, organised, conducted, communicated, and evaluated. Our first editorial: www.tandfonline.com/doi/full/10.... #ResearchOnResearch #MetaScience #OpenScience #STS

Don’t really know what the Times says, but I’m pretty sure one could do an amazing sociolinguistics experiment on this. Shame I don’t have neither the time nor the money to do it myself

Hagen Blix@hagenblix.bsky.social · 4mo ago

Lovely instance of a collective process of creating abstract meaning from a concrete meaning. Humans and language are such fascinating things. The video from the article also does a great job explaining it! www.tiktok.com/@wordsatwork...

"Fake citations have turned into a nightmare for research librarians, who by some estimates are wasting up to 15 percent of their work hours responding to requests for nonexistent records that ChatGPT or Google Gemini alluded to." open access link: archive.ph/qt59c 2/🧵

AI Is Inventing Academic Papers That Don't Exist -- And They're Being Cited in Real Journals

Academic articles from authors using large language model are creating an ecosystem of fake research that threatens human knowledge itself.

rollingstone.com

the metascience feed has two modes: 1. i saw some research i didn't like so it's p-hacked 2. there's this shady research practice hence the replication crisis the more these terms have been adopted, the more their meaning got diluted to ultimately stand for nothing but content-free critical vibes

If you're a #QuartoPub or #RStats user, you might find it hard to find information about how those work with #GitLab. So here's a blog post showing you a few different ways to deploy Quarto documents with GitLab Pages! (Featuring penguins 🐧obviously ) Link: nrennie.rbind.io/blog/deploy-...

Deploying Quarto documents with GitLab – Nicola Rennie

GitLab is a common alternative to GitHub, and you can publish Quarto documents that are hosted via GitLab fairly easily. This blog post documents how to do it.

nrennie.rbind.io

many dysfunctions of academic publishing get exposed by way or llm use, and i keep hoping we'll realize how poorly we've been doing so many things, but we end up focusing on llms as the root cause. i've read so many unrealistic, utopic statements about how scientific practice allegedly works.

I have been using Whisper for the last couple of years, and it has gotten sensibly worse in the last months (perhaps a year?)

Jed Brown@jedbrown.org · 8mo ago

This is a study of Whisper, the LLM-based "transcription" software behind the products used by many medical providers. It gives the appearance of "fluency", but routinely goes entirely off-script, fabricating extensive passages and relations. Human scribes don't do this. arxiv.org/abs/2402.08021

Okay, @glossapsycholx.bsky.social, we need to talk…

You can now watch Jeremy's excellent talk on the term "association" and what it's even supposed to be referring to on YouTube. Must watch if you're doing associational research! www.youtube.com/watch?v=IKdC...

J Labrecque | Association. You keep using that word. I do not think it means what you think it means

YouTube video by the Causal Inference Interest Group (CIIG)

youtube.com

Julia M. Rohrer@dingdingpeng.the100.ci · 5mo ago

Jeremy right now

When I was still a neurocognitive psychologist it always seemed absurd when finding something in the brain was presented as if it validatwd the realness of the behavioral finding. Where do these people think behavior comes from? I was thought we settled the dualism/monism debate a while back...

Kevin Mitchell@wiringthebrain.bsky.social · 5mo ago

Activation of some brain area does not constitute an explanation of a psychological phenomenon - certainly not a better explanation or the real explanation

No one should ever use this b.s., especially not in the social sciences. The fact that we’re constantly inundated by this stuff is atrocious for the future of research.

Adam Kucharski@adamjkucharski.bsky.social · 5mo ago

"Based on the dataset you shared, US and UK responses differ mainly in tone, intensity, and wording style, even though they express similar emotional states" New post on what happened when I got Copilot to do some data analysis: kucharski.substack.com/p/real-signa...

Now on LingMethodsHub, @rpuggaardrode.bsky.social's tutorial on multitaper spectra in R!

Generating and analyzing multitaper spectra in R – LingMethodsHub

This is a tutorial showing how to generate multitaper spectra in R and how to compute spectral moments and DCT coefficients from multitaper spectra.

lingmethodshub.github.io

Do you know a good framework to document data analysis choices? Ex: dropping obs, recoding a var, tuning pars, model + test statistics, etc. Ideally human-readable so people don't have to read my entire codebase. Low-maintenance+ machine-readable for bonus points! #RStats #PyData

3 main take aways: (1) Scientific inference cannot be automated or proceduralized. (2) Not heeding the warning in (1) will necessarily limit our ability for scientific discovery and understanding. (3) The only way not to limit scientific discovery is to allow for unbounded pluralism. /end

How does one even react to such daily savagery