I hope you are having a good and relaxed day. replicationindex.com/2025/12/25/b...

The Robust Bayesian Meta-Analysis package got updated with additional vignettes explaining how to perform Bayesian model-averaged publication bias-adjusted - multilevel meta-analysis (cran.r-project.org/web/packages...) - multilevel meta-regression (cran.r-project.org/web/packages...)

Evidence of publication bias at Christmas

2025 Reflections A famous saying is "A crisis is an opportunity" And every opportunity attracts opportunists. And many opportunists have used the replication crisis to sell their solutions. Now we see what really makes a difference and what was opportunistic BS.

This important meta-science article should not be hidden behind a paywall.

I think there is an argument to be made that Gelman has not made any positive contribution to meta-science. I mean Type-S error. Cut Type-I error in half and call it an important advance in inferential statistics? LOL

However, it beats PMS-SEM

Somebody never plaid Monopoly.

The young kids may not remember the attack on personality psychology is that all of the correlations are just a measure of semantic similarity anyways. And internal consistency is just four semantic similar items, alpha > .8. The real correlations are harder to predict.

Fortunately, I don't have to read Jordan Peterson anymore and can just ask AI to summarize his bullshit for me. Apparently, he sided with Trump against Harvard. Here is my response. Fuck you Nazi JP (but written more nicely thanks to AI). replicationindex.com/2025/12/16/a...

So, 2025 was another bad year. When was the last good one?

I'm very happy to continue my work at the Human-Technology Interaction group as a Full Professor of Metascience. Working at this department has been a great source of joy over the last 15 years.

This is going to be a treasure trove for science of science

"Unconscionable practices in unconscious bias research" replicationindex.com/2025/12/09/a... replicationindex.com/2025/12/05/w...

We have made public #openscience materials from the Münster Center for Open Science, specifically flyers and cheatsheets on #OpenAccess, #OpenData, and #Preregistration. Flyers are already on Zenodo, Cheatsheets will follow: zenodo.org/communities/muecos/ and lukasroeseler.github.io/MueCOS-Infom...

At the regional level, the sexual-orientation IAT predicted rate of same-sex couples better than self-report (Snyder, J. S., & Henry, 2023), but only in 1 specification out of 42 multiverses. In 13 the difference was significant in the other direction. replicationindex.com/2025/12/09/a...

Are you an experimental psychologist? Other sciences understand the benefits of drawing causal inferences from data that are not randomized controlled studies. These inferences are not conclusions, but good ones make some causal claims more plausible than others. Even experiments are not fool prove.

@ulrichschimmack.bsky.social , you inspired a new blog post that I think you will like! :) daniellakens.blogspot.com/2025/12/dogm...

Cultivate resistance against fascism, indeed. Fuck Trump www.theguardian.com/us-news/2025...

if this helps to get rid of network models, so be it, but the real problem is that pretty pictures of correlations are not science.

Although no formal test has been done, a thesis student of mine showed my equivalence testing papers (cited 4k times since 2017) reduce misreporting of non-significant results. journals.sagepub.com/doi/full/10....

A blind spot in implicit bias research. I thought it is nicer to attribute @jordanaxt.bsky.social and his 12 co-authors mistake to #WillfulIncompetence (an inability to see problems) rather than #WillfulIgnorance (a motivated bias to ignore them). replicationindex.com/2025/12/05/w...

Zen Philosophy of Science: Science is supposed to be self-correcting. Scientists are invested in their ideas, publication, etc. If science corrects itself and no scientists' feelings get hurt, did it actually happen?

25 years of research to see whether the race IAT predict racially biased behaviors without any success. Maybe it is time to study what actually predicts racial biases in the real world! For White researchers to use the social relevance of racism to advance careers is racist itself.

This an open invitation for @jordanaxt.bsky.social to participate in an adversarial collaboration to model these data without confirmation bias. I see no evidence that the race IAT measures anything implicit that is not reflected in self-ratings. replicationindex.com/2025/12/02/i...

Scientific utopia. @briannosek.bsky.social responds to my criticism of the IAT (LOL) Sometimes silence says more than a thousand words. replicationindex.com/2025/12/02/i...

Political ideology distorts interpretation of science. Yes, there are genetic influences on behavior. No, that does not mean that some people are better than others. Genetic variability exists because nature thought it is fine (it did not hurt evo. fitness). Eugenics is man-made dogma.

How many failed attempts to show that the IAT measures something of importance do we need? replicationindex.com/2025/12/02/i...

What have we learned? 1. All measures measure a single construct. No evidence that implicit measures like the IAT measure anything distinct (Schimmack, 2020). 2. Self-report measures are much more valid than implicit measures like the IAT. 3. Time to tell Project Implicit visitors about this.

A better fitting model shows no discriminant validity for any set of measures (implicit, explicit, behavioral). A single attitude best measured with self-ratings. Prejudice is real, people know it, and report it. Time to study racism, not laboratory tasks.

"From a more optimistic lens, these results could be viewed as encouraging for the predictive validity and utility of indirect measures. " LOL where are the adversarial voices here? EPT and AMP: over 90% of the variance is ERROR So, they cannot have predictive validity. IAT .57 * .16 * .62 = .06!

I think this is a much more informative figure to share. 1. Implicit to disposition (not behavior) b = .05 to .27 2. No convergent validity for Implict measures (eval. prime = .08, AMP = .25) 3. Implicit factor = IAT without random error.

For the record. I was never asked to participate in this adversarial collaboration. Maybe I am too adversarial for collaboration, but I am happy that this study was conducted. I also like the title, no implication that IAT measures implicit constructs. Maybe Project Implicit will take note.

We just preprinted a huge meta-meta-analysis examining the effects of exercise on cognition, memory, and executive function In short - 2239 effect sizes - extreme between-study heterogeneity - extensive publication bias - some subgroup/exercise-specific effects More below (doi.org/10.31234/osf...)

Thanksgiving is over. No more fake smiles left? Enjoy sone open and honest truth.

A positive response to the validation crisis. Systematically reviewing psychometric properties; here measures of prejudice. osf.io/preprints/ps...

"it is easier to spot a splinter in somebody else's eye: replicationindex.com/2025/11/29/b...

Not sure who cares, but Project Implicit continues to give people useless feedback without any information about measurement error and the invalidity of IATs. Apparently, Harvard does not have to follow rules of ethical use of psychometric tests. replicationindex.com/2025/11/28/w...

Watch this video before you do your Black Friday shopping on Amazon.

Against Agreeabless www.theguardian.com/lifeandstyle...

Preregistrations without Code do not Prevent P-Hacking: You can increase your chances for a significant finding in the absence of real effects even with correlations and t test despite having preregistered your hypothesis (e.g., simply changing arguments in the functions). doi.org/10.31222/osf...

#AcademicSky #PsycSciSky #Metascience How psychologists handle sampling, measurement, & statistical test assumptions

Á position for an über-psychologist in Germany.

A device looks impressive, but we forget they simplify reality. Eye-tracking shows where people look, but not what they think. AI can summarize medical symptoms, but it cannot read uncertainty or context. The danger is not “bad technology” but blind trust in its outputs. #SciWri #metasci

come on we already went through the replication crisis

Is there any debate about the stupidity of p-values in economics. Why not CI?

What scientists are you. I mean some are better than others. Interested in the good science that is not corrupted by dark research funding money and 10% funding rates.

JPSP: don't let methodological rigor get in the way of a good story. This one was falsified by a blog post that was published years before the article came out. replicationindex.com/2025/11/18/p...

This would never happen in psychology because everything is significant even without industry sponsers.

How do you teach students about the replication crisis. With examples from blood glucose, estrogen, to nasal spray oxytocin studies. replicationindex.com/2025/11/18/p...