Daniel Lakens
@lakens
Metascience, statistics, psychology, philosophy of science. Eindhoven University of Technology, The Netherlands. Omnia probate. 🇪🇺
In our recent paper criticizing the idea of Type S and M errors, we discuss the idea that 'the null is never true' and explain why that idea is silly (why not doubt that an effect of 1 × 10−32 can be true?) and scientifically unfalsifiable. journals.sagepub.com/doi/full/10....
In Norway, staying in this lovely farm (the elderly home of my friend and collaborator Peder Isager) for a week with the lab and friends, to talk about research waste, Meehl's corroboration index, agent based models, and Metacheck. Good food, great weather, and amazing people ❤️
Campbell, 1988, explaining why he disliked working in Chicago and almost considered leaving academia (which would have been a huge loss) due to the highly competitive environment and the idea you should write only brilliant papers.
These discussion led to the code of conduct for research integrity. The one in my country is very clear. If you want to discuss this, join the open conversations about the next edition. For now, you signed yours and you have to follow it.
We launched our beta version 0.0.1 of Metacheck today! After our second year of work on the tool, we now have validated modules, hugely expanded functionality, optional (or local) LLM use, checks inside code files, and much more! Try it out in R now: www.scienceverse.org/metacheck/
Vlasta Sikimić is opening the conference on AI and Peer Review at Eindhoven University of Technology. It will be a very interesting day, no doubt full of spirited discussions!
Excellent news! Nature is expanding Registered Reports to all the fields in which they publish! A great result by all those proposing this strong solution to publication bias and selective reporting (and getting expert feedback before data collection)! www.nature.com/articles/d41...
Very happy to have Metin Bulus over for the next 11 months to work on sample size considerations for multi-site collaborative research projects! And of course @debruine.bsky.social who is on a Metacheck programming craze to get the last bugs sorted in the next 2 weeks!
At the Vrije Universiteit Amsterdam to join today's session on the national dialogue on academic freedom.
New interactive simulation in my online textbook illustrating publication bias in meta-analyses, and some techniques to model bias-adjusted effect sizes. None of these work perfectly. We need an unbiased literature to draw valid inferences. lakens.github.io/statistical_...
I also created a version of the 'guess the correlation' game. Students always enjoyed it, but the original website (with fun retro design) started to require a google account login. So I created my own version. Play it here: lakens.github.io/statistical_...
Updated my textbook with a cleaner lay-out, better accessibility, and eight new interactive visualizations made completely in html (such as this illustration of the positive predictive value). lakens.github.io/statistical_...
Many Labs 5 was such a cool project. Authors who argued that a study from the Reproducibility Project: Psychology was 'too different', leading to a null result, could propose their own replication. Mostly null results again. Blaming auxiliaries is easy. Replication is hard.
Lovely to see everyone coding away during the Paul Meehl Graduate School workshop on Cognitive Computational models paulmeehlschool.github.io/2026-03-10-c..., with visitors from Poland, Czech Republic, and Germany. Thanks to Leendert van Maanen and Dominik Bachmann for teaching it!
There is! Although this picture is the old set up. Lisa, I now have my own office, and the lab has a very luxurious huge space with 6 desks and some additional tables. You will see it next time you come over!
Standing in line for Record Store Day in Rotterdam with some good coffee. The wait time (1.5 hours!) was well spent giving feedback on a paper that I am very excited about.
"comprehensive study pre-registrations or Registered Reports, which ensure peer reviewing of the study protocol and acceptance in principle before data collection and analysis, are a readily implementable measure that should be considered essential." www.frontiersin.org/journals/neu...
I am sure, because here is the abstract of a 2024 talk by him (not too far away in time from the prereg of Study 1). And I do not think it is that farfetched to assume people discuss if they should preregister studies with collaborators, but sure, they might not have (for all 3 studies).
It is very funny to see that authors who in 2019 argued that "Preregistration is redundant, at best", are now co-authoring papers where their collaborators want to preregister all studies. I guess they weren't even able to even convince their direct co-authors!
Lovely day in Eindhoven, and not just because of the weather! I had Maciej Behnke over as visitor for a few days to talk science and pick up an old collaboration! And @crist14n.bsky.social had Jenny Murphy, Simon Nolte, and Robert Rein over to talk computational reproducibility in Sports science.
198 effect sizes in ego depletion resesrch showed an effect size of d=0.62. Preregistered large replications (including some by original authors) yielded an effect size of 0. No one has been able to offer any other explanation for this huge research waste than massive p-hacking.
Approximately 40% of all studies preregistered on the Open Science Framework are never shared publicly. That’s a lot. Researchers give several reason, and null results and bad planning (or lack of priority) are the main reasons. journals.sagepub.com/doi/10.1177/...
In standard scientific reports in psych 96% of first mentioned main hypotheses are supported. In Registered Reports, this is 46%. The 96% is clearly biased (given true H1 rate and power). Lack of transparency means we do not know the true baserate. journals.sagepub.com/doi/10.1177/...
Babbage, 1830, discussing the problem that scientists selectively report findings that they want to be true. Confirmation bias is a strong human tendency. This is why we need to design science in a way that prevents conformation bias from leading us away from the truth.
Very pleased with these new hex stickers, for the #PSE8 conference, but also for our TOSTER package, Metacheck, and my Improving Your Statistical Inferences book and course :)
The Metacheck team (www.scienceverse.org/metacheck/) and some interested collaborators are hackatonning away on how to create and validate new metacheck modules to automatically detect information in scientific papers. #PSE8
We are starting the 3rd day of #PSE8, which is completely dedicated to hackathons. People can join one in the morning, and one in the afternoon, and collaboratively work on projects, or develop new ideas for future projects!