When three reviewers (1 human, 2 AI) spot the same fatal flaw, why did the authors not see it or ask AI for a pre-submission review?
replicationindex.com/2026/03/05/d...
Did hindsight bias replicate? 😃
Oversight Democrats have been investigating the handling of serious allegations by a survivor against Donald Trump when she was a minor. We have now confirmed that some of the documents related to this survivor appear to be missing from the files.
Here’s what we know ⬇️
Meta-psychology is the scientific study of psychological scientists' behavior. They like z-curve as much, as rats liked a Skinner box. Too bad. Now you know how it feels to be a subject and not a participant.
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.
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/...
The replication crisis is only over when psychology journals stop publishing such biased articles that blame replication failures on replicators and not the use of p-hacking in original studies. Pathetic.
replicationindex.com/2026/02/13/r...
Is this finally the baseball quote, where somebody compared their 1/3 success rate to a 33% batting average in baseball?
I think it's best for everyone to understand that the unified class project of billionaires right now is to do to white collar workers what globalization and neoliberalism did to blue collar workers.
A Federal judge in Oregon ruled that ICE can NOT use chemicals on protestors for exercising their 1st amendment rights.
youtu.be/gvL9ckyrDTM?...
I am not a doctor.
I am a doctor of philosophy.
That just means I doctored my dissertation.
😃😁😇
I want to laugh at it but sadly it’s true.
Next level simulation work. The simulation is still running and you can check the results as they accumulate. Z-curve produces 95% CI with 95% coverage even with k = 50 sig. results across many realistic scenarios.
replicationindex.com/2026/02/03/c...
“Unconscionable,” Jon thought as he found an email address online for the lead prosecutor, Joseph Dernbach, who was named in the story. Peering through metal-rimmed glasses, Jon opened Gmail on his computer monitor. “Mr. Dernbach, don’t play Russian roulette with H’s life,” he wrote. “Err on the side of caution. There’s a reason the US government along with many other governments don’t recognise the Taliban. Apply principles of common sense and decency.” That was it. In five minutes, Jon said, he finished the note, signed his first and last name, pressed send and hoped his plea would make a difference. Five hours and one minute later, Jon was watching TV with his wife when an email popped up in his inbox. He noticed it on his phone. “Google,” the message read, “has received legal process from a Law Enforcement authority compelling the release of information related to your Google Account.”
A retiree wrote this email to a DHS attorney. Within five hours, DHS demanded Google turn over records for his account.
A Kafkaesque form of domestic surveillance, intimidating Americans for lawful speech.
New from us at the Post: www.washingtonpost.com/investigatio... @johnwoodrowcox.bsky.social
The hidden power of anonymous peer reviewers in legacy journals with closed peer-review. Maybe just publish in open, scientific journals like Meta-Psychology.
replicationindex.com/2026/02/03/h...
Applied Social Psychology LOL
Based on the amazing FReD replication explorer, I finally have some evidence for the claim that the observed replication rate is lower in social than in disciplines like cognitive psychology
forrt.org/FReD-apps/ex...
1/
Didn't Bem describe his PSI experiments as rhetorical devices?
daniellakens.blogspot.com/2025/09/type...
Another AI gem:
1. Articles optimize for novelty; books optimize for understanding
Maybe journal culture and performance indicators like H-index killed social science. Tower of bable. Every article starts with a new framework, no integration possible. Yes, I am writing a book.
old wine in new bottles, but if you haven't seen this preregistered p-hacked 9 study article yet, take a look
It even comes with a response by the authors, no we did not p-hack, but they did.
replicationindex.com/2026/01/18/p...
How do you respond to the concerns about z-curve that z-curve is a selection model and should not be used when there is no selection bias???
replicationindex.com/2026/01/27/z...
Ian Austin, an Army veteran, was arrested for protesting ICE in Minneapolis. But he’s continued protesting. Senior reporter @julialurie.bsky.social spoke to him.
“When they say, ‘Why would you be out here?’ How the fuck could I not be out here?” he says. “My nation is under attack."
This was a really great discussion with people pushing the boundaries of scientific resesrch in their ManyMany projects. Really the cutting edge of scientific knowledge generation!
Me: Erik, you made a mistake in your simulation. Sometimes there were no false positives to detect missing non-significant tests of true H0
Erik: I’m not interested in more pointless discussions with you. I can do without your insults and misrepresentations.
replicationindex.com/2026/01/24/o...
For context. Social psych has spread the claim "we are not alone but we are doing the mosr go improve" but their practices wete the worst. Not the same in other fields
The problem is not AI. The problem is human ab/users.
SPSS did not phack either
The existing reproducibility crisis is exacerbated by publication pressure. Lack of incentives for replication studies & funding difficulties make science vulnerable. AI could worsen this by enabling rapid generation of plausible but flawed papers. #Reproducibility 3/6
Some are better than others, but a blanket statement is not warranted. With reasonable set of studies, The random effects selection model implemented in weightr should always be used and results reported. Also @fbartos.bsky.social RoBMA.
but how do you know whether effect sizes are large without checking publication bias, ego depletion went from .6 to .06.
