iiinova — where your digital self keeps evolving

Reposted by Jesper N. Wulff

Julia M. Rohrer @dingdingpeng.the100.ci · 15/09/2026

The question that’s dominating the field is not a causality question. It’s really, how can one sell a causal relationship while crafting the language so that there’s plausible deniability when somebody criticizes anything.

56612

Reposted by Jesper N. Wulff

Brian Nosek @briannosek.bsky.social · 12/09/2026

The cover-up is always worse than the crime? Harvard's latest filing says that Gino manipulated a file to cover-up fabrication, hid the laptop used to do it during the discovery process, and did more [redacted] destruction of evidence after being compelled by the court to share the laptop.

38130

Reposted by Jesper N. Wulff

Julia M. Rohrer @dingdingpeng.the100.ci · 05/09/2026

Fun little game in which a human life is randomly drawn for you. I ended up a girl, born in the Yangtze Basin in 670 CE. Died in infancy at 6 months 🥲 anyhumanever.com

anyhumanever.com

Any Human Ever

One life, drawn at random from all who have ever lived.

185321

Reposted by Jesper N. Wulff

Data Colada @datacolada.bsky.social · 03/07/2026

The Cover-Up in Gino v Harvard datacolada.org/136

12412

Jesper N. Wulff @jnwulff.bsky.social · 02/09/2026

The gift that keeps on giving

000

Reposted by Jesper N. Wulff

Daniel Lakens @lakens.bsky.social · 04/08/2026

We compute a number that should be higher or lower than a criterion. If the effect size is larger than a critical value, it the t-value is larger, or a p-value is smaller, you can say there is something because the long run probability that you are fooling yourself is small enough.

121

Reposted by Jesper N. Wulff

Daniel Lakens @lakens.bsky.social · 03/08/2026

In the last 2 years I (and our lovely collaborators) have been rather obsessively working on Metacheck - a tool to automatically check best practices in papers. In this RIOTS summerschool talk, I walk you through what the tool can do, and future plans www.youtube.com/watch?v=OWd2...

youtube.com

Metacheck: automatically check papers for best practices - Daniel Lakens

Can automated tools help researchers catch problems before their work is published? In this talk from the King’s Open Research Summer School 2026, Professor Daniël Lakens (Eindhoven University of…

12613

Reposted by Jesper N. Wulff

Daniel Lakens @lakens.bsky.social · 01/08/2026

I was not reviewer 2. But I am so annoyed by Americans pretending the whole open science movement is funded by John Arnold. His investments are absolutely dwarfed by what the European Union and European science funders have invested in Open science. Only incredibly biased people would ignore this.

36514

Reposted by Jesper N. Wulff

Daniel Benneworth-Gray @danielgray.com · 13/07/2026

Sam Neill on tackling depression and imposter syndrome when you’re between jobs.

64107543684

Reposted by Jesper N. Wulff

Daniel Lakens @lakens.bsky.social · 25/05/2026

New blog post: Evaluating Dr. Cuddy’s Claim that the Debunking of Power Posing is a Myth. daniellakens.blogspot.com/2026/05/eval... On an AI generated description of a non-existent study, incorrectly citing findings from studies, and the importance of scientific criticism.

daniellakens.blogspot.com

Evaluating Dr. Cuddy’s Claim that the Debunking of Power Posing is a Myth

In this blog post I will analyse the arguments that Dr. Amy Cuddy provided in a blog post “The "Power Posing Was Debunked" Myth: What the Re...

1517870

Reposted by Jesper N. Wulff

Richard D. Morey @richarddmorey.bsky.social · 25/05/2026

One of the arguments put to me re: scientific reform is essentially that it will sort itself (poor science is ignored) but of course things only get sorted if scientists are *constantly* doing the difficult work of evaluating competing claims, so "grey" science doesn't spread.

2336

Jesper N. Wulff @jnwulff.bsky.social · 21/05/2026

"If I hated to be the bearer of bad news, I wouldn’t have specialized in research methods."

1245

Reposted by Jesper N. Wulff

Led By Donkeys @ledbydonkeys.org · 17/05/2026

Immigration make Britain brilliant

510139214210

Reposted by Jesper N. Wulff

Stephen Wild @stephenjwild.bsky.social · 07/05/2026

Love it

An incorrect formula for standard deviation.

in pseudo code:

sum(sqrt((sum(x-i)^2)/n))

1302

Jesper N. Wulff @jnwulff.bsky.social · 01/04/2026

In this enormous meta-research project we put claims in social-science papers to the test and the results from the project have now been published in Nature.

091

Reposted by Jesper N. Wulff

Greg Pak @gregpak.net · 09/03/2026

i’ve watched this three times. VERY satisfying.

11711127

Reposted by Jesper N. Wulff

Richard McElreath 🐈‍⬛ @rmcelreath.bsky.social · 08/03/2026

Last week, I mentioned this in passing in a workshop: In 1938 Enrico Fermi won a Nobel Prize for discovering two new elements of the periodic table. Lise Meitner shortly showed that Fermi was mistaken and instead had produced known lighter elements by fission. She did not win a Nobel prize.

Portrait of Lise Meitner taken in 1928. She is smoking a cigarette and looking impatient to get back to her experiments.

316251

Reposted by Jesper N. Wulff

Ben Collins @bencollins.bsky.social · 10/02/2026

Listen to this. Not a penny more for this. Abolish and prosecute anyone who had anything to do with it.

230152615818

Reposted by Jesper N. Wulff

Richard McElreath 🐈‍⬛ @rmcelreath.bsky.social · 02/02/2026

Wow this scoring chaos seems to be an extreme case of what I say about many ad hoc analyses: no derivation of method from a clear scientific theory, no assessment of statistical properties, and decades pass before someone notices. This happens in biology too, so let’s not pick on psychology only

23612

Reposted by Jesper N. Wulff

samuel mehr @mehr.nz · 05/11/2023

who did this

428078

Reposted by Jesper N. Wulff

Ben Williamson @benpatrickwill.bsky.social · 19/12/2025

(Also, icyi, here are the 42 papers citing our non-existent paper which includes a "meta-analysis" - often called the evidence "gold standard" - of "LLM effects" in education 🤮 scholar.google.com.vn/scholar?star...)

scholar.google.com.vn

1070891

Reposted by Jesper N. Wulff

Ian Hussey @ianhussey.mmmdata.io · 12/12/2025

Keeping this at hand in case I need to point to it and tap

510234

Reposted by Jesper N. Wulff

Andrew Heiss @andrew.heiss.phd · 09/12/2025

Some closing thoughts for my students this semester on LLMs and learning #rstats datavizf25.classes.andrewheiss.com/news/2025-12...

Will you incorporate LLMs and AI prompting into the course in the future?
No.

Why won’t you incorporate LLMs and AI prompting into the course?
These tools are useful for coding (see this for my personal take on this).

However, they’re only useful if you know what you’re doing first. If you skip the learning-the-process-of-writing-code step and just copy/paste output from ChatGPT, you will not learn. You cannot learn. You cannot improve. You will not understand the code.

In that post, it warns that you cannot use it as a beginner:

…to use Databot effectively and safely, you still need the skills of a data scientist: background and domain knowledge, data analysis expertise, and coding ability.

There is no LLM-based shortcut to those skills. You cannot LLM your way into domain knowledge, data analysis expertise, or coding ability.

The only way to gain domain knowledge, data analysis expertise, and coding ability is to struggle. To get errors. To google those errors. To look over the documentation. To copy/paste your own code and adapt it for different purposes. To explore messy datasets. To struggle to clean those datasets. To spend an hour looking for a missing comma.

This isn’t a form of programming hazing, like “I had to walk to school uphill both ways in the snow and now you must too.” It’s the actual process of learning and growing and developing and improving. You’ve gotta struggle.

This Tumblr post puts it well (it’s about art specifically, but it applies to coding and data analysis too):

Contrary to popular belief the biggest beginner’s roadblock to art isn’t even technical skill it’s frustration tolerance, especially in the age of social media. It hurts and the frustration is endless but you must build the frustration tolerance equivalent to a roach’s capacity to survive a nuclear explosion. That’s how you build on the technical skill. Throw that “won’t even start because I’m afraid it won’t be perfect” shit out the window. Just do it. Just start. Good luck. (The original post has disappeared, but here’s a reblog.)

It’s hard, but struggling is the only way to learn anything.

You might not enjoy code as much as Williams does (or I do), but there’s still value in maintaining codings skills as you improve and learn more. You don’t want your skills to atrophy.

As I discuss here, when I do use LLMs for coding-related tasks, I purposely throw as much friction into the process as possible:

To avoid falling into over-reliance on LLM-assisted code help, I add as much friction into my workflow as possible. I only use GitHub Copilot and Claude in the browser, not through the chat sidebar in Positron or Visual Studio Code. I treat the code it generates like random answers from StackOverflow or blog posts and generally rewrite it completely. I disable the inline LLM-based auto complete in text editors. For routine tasks like generating {roxygen2} documentation scaffolding for functions, I use the {chores} package, which requires a bunch of pointing and clicking to use.

Even though I use Positron, I purposely do not use either Positron Assistant or Databot. I have them disabled.

So in the end, for pedagogical reasons, I don’t foresee me incorporating LLMs into this class. I’m pedagogically opposed to it. I’m facing all sorts of external pressure to do it, but I’m resisting.

You’ve got to learn first.

14336103

Reposted by Jesper N. Wulff

Julia M. Rohrer @dingdingpeng.the100.ci · 07/12/2025

“No, we’ve all accepted that the goal is to publish anything and everything and see what sticks. More students. More “collaborations”. So many papers that no human can possibly be paying very much attention to any of it. And we all know where the incentives lie.”

0236

Reposted by Jesper N. Wulff

Erik Angner @erikangner.com · 27/11/2025

"Runctitiononal features"? "Medical fymblal"? "1 Tol Line storee"? This gets worse the longer you look at it. But it's got to be good, because it was published in Nature Scientific Reports last week: www.nature.com/articles/s41... h/t @asa.tsbalans.se

Infographic with AI slop published in Nature Scientific Reports

2012302737

Reposted by Jesper N. Wulff

Andrew Heiss @andrew.heiss.phd · 21/11/2025

hahahaha just got an email from someone who was using Claude to generate a boilerplate #QuartoPub document and the LLM *used my name* as the author. The computers are literally trying to be me now 😂🤣🙃🫠

A Quarto document in RStudio with the author field set to "Andrew Heiss"

1533529

Reposted by Jesper N. Wulff

ElieNYC @elienyc.bsky.social · 19/11/2025

I know that at this point it's a subplot in the Epstein files drama, but I feel compelled to point out, once again, that Larry Summers HAS NO BUSINESS teaching students at ANY university ever again! My latest cries into the abyss, in @thenation.com www.thenation.com/article/soci...

thenation.com

Why Is Larry Summers Still Employed?

The revelations about the economist’s attempts to pressure a women into a “relationship”—with guidance from Jeffrey Epstein—should finally disqualify him from teaching students.

692535596

Reposted by Jesper N. Wulff

Retraction Watch @retractionwatch.com · 16/11/2025

A paper critiquing post-publication peer review has numerous made-up references, including a @nature.com article falsely attributed to our Ivan Oransky. link.springer.com/article/10.1...

pubpeer.com

PubPeer - An expert criticism on post-publication peer review platform...

There are comments on PubPeer for publication: An expert criticism on post-publication peer review platforms: the case of pubpeer (2025)

15829

Reposted by Jesper N. Wulff

Aaron Parnas @aaronparnas.bsky.social · 16/11/2025

NEW: Epstein survivors release the most powerful PSA I have ever seen. Make this go viral so every member of the House of Representatives sees it.

13005945736317

Reposted by Jesper N. Wulff

Cursor @cursortue.bsky.social · 13/11/2025

TU/e has gained a new research centre: META/e. Daniël Lakens and Krist Vaesen were among the founders of this knowledge hub for metascience—research aimed at improving the practice of science itself. “We want to be a home for every researcher who occasionally wonders: what are we even doing?”

cursor.tue.nl

Knowledge centre META/e: home for those improving science

02613

Reposted by Jesper N. Wulff

Daniel Lakens @lakens.bsky.social · 28/09/2025

New blog post on Gelman's recent claim that Type S and M errors are intended as a 'rhetorical tool', and if I was wrong to believe they were recommended more routinely in our recent preprint criticizing the idea of Type S and M errors. daniellakens.blogspot.com/2025/09/type...

daniellakens.blogspot.com

Type S and M errors as a “rhetorical tool”

We recently posted a preprint criticizing the idea of Type S and M errors ( https://osf.io/2phzb_v1 ). From our abstract: “While these conce...

0107

Reposted by Jesper N. Wulff

Joachim Baumann @joachimbaumann.bsky.social · 12/09/2025

🚨 New paper alert 🚨 Using LLMs as data annotators, you can produce any scientific result you want. We call this **LLM Hacking**. Paper: arxiv.org/pdf/2509.08825

$We present our new preprint titled "Large Language Model Hacking: Quantifying the Hidden Risks of Using LLMs for Text Annotation". We quantify LLM hacking risk through systematic replication of 37 diverse computational social science annotation tasks. For these tasks, we use a combined set of 2,361 realistic hypotheses that researchers might test using these annotations. Then, we collect 13 million LLM annotations across plausible LLM configurations. These annotations feed into 1.4 million regressions testing the hypotheses. For a hypothesis with no true effect (ground truth $p > 0.05$), different LLM configurations yield conflicting conclusions. Checkmarks indicate correct statistical conclusions matching ground truth; crosses indicate LLM hacking -- incorrect conclusions due to annotation errors. Across all experiments, LLM hacking occurs in 31-50\% of cases even with highly capable models. Since minor configuration changes can flip scientific conclusions, from correct to incorrect, LLM hacking can be exploited to present anything as statistically significant.$

6303106

Reposted by Jesper N. Wulff

Marieke van Vugt @mvugt.bsky.social · 09/09/2025

"OpenAI is making “small steps that are good, but I don’t think we’re anywhere near where we need to be”, says Mark Steyvers, a cognitive science and AI researcher at UC Irvine. “It’s not frequent enough that GPT says ‘I don’t know’.”" www.nature.com/articles/d41...

nature.com

Can researchers stop AI making up citations?

OpenAI’s GPT-5 hallucinates less than previous models do, but cutting hallucination completely might prove impossible.

053

Reposted by Jesper N. Wulff

Paul Hünermund @p-hunermund.com · 30/08/2025

➡️ Deadline approaching—only one month left to send in your papers and presentation proposals for #CDSM2025! 🚨 𝗖𝗮𝗹𝗹 𝗳𝗼𝗿 𝗣𝗮𝗽𝗲𝗿𝘀: 𝗖𝗮𝘂𝘀𝗮𝗹 𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 𝗠𝗲𝗲𝘁𝗶𝗻𝗴 𝟮𝟬𝟮𝟱 🚨 📅 𝗡𝗼𝘃 𝟭𝟮–𝟭𝟯, 𝟮𝟬𝟮𝟱 (𝗩𝗶𝗿𝘁𝘂𝗮𝗹) 📥 Submission Deadline: 𝗦𝗲𝗽𝘁 𝟯𝟬, 𝟮𝟬𝟮𝟱

02021

Reposted by Jesper N. Wulff

Julia M. Rohrer @dingdingpeng.the100.ci · 25/08/2025

Ever stared at a table of regression coefficients & wondered what you're doing with your life? Very excited to share this gentle introduction to another way of making sense of statistical models (w @vincentab.bsky.social) Preprint: doi.org/10.31234/osf... Website: j-rohrer.github.io/marginal-psy...

Models as Prediction Machines: How to Convert Confusing Coefficients into Clear Quantities

Abstract
Psychological researchers usually make sense of regression models by interpreting coefficient estimates directly. This works well enough for simple linear models, but is more challenging for more complex models with, for example, categorical variables, interactions, non-linearities, and hierarchical structures. Here, we introduce an alternative approach to making sense of statistical models. The central idea is to abstract away from the mechanics of estimation, and to treat models as “counterfactual prediction machines,” which are subsequently queried to estimate quantities and conduct tests that matter substantively. This workflow is model-agnostic; it can be applied in a consistent fashion to draw causal or descriptive inference from a wide range of models. We illustrate how to implement this workflow with the marginaleffects package, which supports over 100 different classes of models in R and Python, and present two worked examples. These examples show how the workflow can be applied across designs (e.g., observational study, randomized experiment) to answer different research questions (e.g., associations, causal effects, effect heterogeneity) while facing various challenges (e.g., controlling for confounders in a flexible manner, modelling ordinal outcomes, and interpreting non-linear models).

Figure illustrating model predictions. On the X-axis the predictor, annual gross income in Euro. On the Y-axis the outcome, predicted life satisfaction. A solid line marks the curve of predictions on which individual data points are marked as model-implied outcomes at incomes of interest. Comparing two such predictions gives us a comparison. We can also fit a tangent to the line of predictions, which illustrates the slope at any given point of the curve.

A figure illustrating various ways to include age as a predictor in a model. On the x-axis age (predictor), on the y-axis the outcome (model-implied importance of friends, including confidence intervals).

Illustrated are
1. age as a categorical predictor, resultings in the predictions bouncing around a lot with wide confidence intervals
2. age as a linear predictor, which forces a straight line through the data points that has a very tight confidence band and
3. age splines, which lies somewhere in between as it smoothly follows the data but has more uncertainty than the straight line.

461001285

Reposted by Jesper N. Wulff

Theiss Bendixen @theissbendixen.bsky.social · 25/08/2025

"Being Bayesian in a Frequentist World" New post on "Bayesian dynamic borrowing" in R 📚 Link 👇

2102

Reposted by Jesper N. Wulff

Daniel Lakens @lakens.bsky.social · 25/08/2025

If you are preparing your bachelor statistics course and would like to add optional material for students to better understand statistics on a conceptual level (see topics in the screenshot) my free textbook provides a state of the art overview. lakens.github.io/statistical_...

320865

Reposted by Jesper N. Wulff

Dr. Casey Fiesler @cfiesler.bsky.social · 23/08/2025

My video about how LLMs are not search engines has led to many, MANY comments telling me that I should be using Perplexity. Some insisting that Perplexity does not hallucinate. Out of a list of 26 papers it just provided me (in "Research" mode) 4 were real. FOUR. 85% hallucination rate.

281488465

Reposted by Jesper N. Wulff

Saloni @scientificdiscovery.dev · 17/08/2025

TIL the original paper describing CRISPR, by Francisco Mojica, was rejected by 4 journals and took 2 years to be published

629877

Reposted by Jesper N. Wulff

Carl T. Bergstrom @carlbergstrom.com · 16/08/2025

Just in case there was any doubt, ChatGPT 5.0 still makes up completely random citations that don't exist and should not be used for literature search.

1. David Ackerly (UC Berkeley)

While his most-cited work is on leaf size and SLA, he also wrote explicitly about plasticity in leaf traits, including shape, in the context of ecological strategies.

Example: Ackerly (1997), “Allocation, leaf display, and growth in fluctuating light environments: A comparative study of deciduous and evergreen species” (Oecologia). This emphasizes how plasticity in leaf traits mediates adaptation to light.

26957264

Reposted by Jesper N. Wulff

Carlisle Rainey 👨‍💻📊📚 @carlislerainey.bsky.social · 12/08/2025

‼️Cool new paper‼️ Finds that journal data policies in psychology boost sharing statements to ~100%, but only about half of datasets are complete, understandable, reusable. Open: open.lnu.se/index.php/me...

1164

Reposted by Jesper N. Wulff

Zeta Of 1 @zetaof1.bsky.social · 04/08/2025

5. Most frequentist methods are just *fine* and there's no need to always go full luxury bayesian in every application.

1266