iiinova — where your digital self keeps evolving

Note: An earlier version of this thread described the problems with the ML techniques as severe, but this was misleading. Only without the described countermeasures, these might have resulted in severe problems.

001

The full rationale, including the review and author response, are available on PsyArXiv: osf.io/hxnum The study materials and all new materials generated as part of the review are on OSF osf.io/y3hxg/ and Github github.com/FlorianParge.... Original article available here: doi.org/10.1177/0956...

OSF

100

This shows how replications can function as a safety net not just for random fluctuation, but also for problems with methods that were not known at the time.

110

Best practices evolve and especially in machine learning, avoiding overfitting can be difficult. Simple, easy to explain techniques like splitting the sample and straightforward replications are easy by comparison and will remain a best practice.

By doing training/test set splits and conducting a straightforward validation of their results in a second sample, the authors reined in the bias.

100

However, he also found that the machine learning techniques used, state-of-the-art at the time, could lead to inflated performance estimates. This could have led to bias had the authors not conducted self-replication.

Florian found some minor transcription errors and improvable documentation practices (unsurprising given the age of the work and the studies involved).

100

We are grateful to Florian Pargent for his in-depth review and reanalysis and to @datingdecisions.bsky.social @pauleastwick.bsky.social @elijfinkel.bsky.social for their willingness to have their impactful work scrutinised for errors.

110

New report: Joel, Eastwick, & Finkel (2017) “Is Romantic Desire Predictable? Machine Learning Applied to Initial Romantic Attraction”. Based on the review by @florianpargent.bsky.social, we find Minor Errors that do not affect the core conclusions of the manuscript. osf.io/hxnum

ERROR @error.reviews · 23/01/2026

OSF

1188

This is a follow-up on bsky.app/profile/erro...

010

ERROR @error.reviews · 23/01/2026

News from scientific self-correction: Authors pushing to get errors in their papers corrected. Lukas continues to be a role model for how scientists should handle post-publication peer review.

195

ERROR @error.reviews · 22/01/2026

Improving scientific practice can seem daunting. In this fantastic talk (and thread below), Julia Rohrer shares practical ways to communicate methodological insights to a wider audience of researchers.

12310

ERROR @error.reviews · 18/12/2025

At ERROR, we cannot compete with million-dollar bounties for whistleblowers. But it is great to see sleuthing work rewarded, and institutions admitting when their researchers engaged in misconduct.

0137

ERROR @error.reviews · 17/12/2025

Post-publication peer review is at it best when it's thoughtful, scrupulous, steeped in detail – and challenges key claims of the paper. @janhove.bsky.social's discussion of a recent paper on multilingualism exemplifies this.

0189

ERROR @error.reviews · 08/12/2025

Metascientists step up as role models for a healthy error culture in science. Here is a great case where an author and a critical reader collaborated to set the record straight.

0133

ERROR @error.reviews · 02/12/2025

Voluntary retraction remains a key way to put scientific self-correction into practice. Zhu and Holmes (2024) did the right thing when they realized that some of their results were based on a coding error. Original: psycnet.apa.org/fulltext/202... With retraction: psycnet.apa.org/fulltext/202...

psycnet.apa.org

APA PsycNet

0167

ERROR @error.reviews · 02/12/2025

Many error remain to be found in clinical trials. Patients deserve reliable results. Kudos to these authors for their persistent work to correct the record.

0135

ERROR @error.reviews · 24/11/2025

Congratulations to @simine.com for winning the Einstein Foundation Individual Award! 🎉 A well-deserved recognition for her seminal efforts to improve scientific rigor, which includes instituting detailed checks for errors and computational reproducibility at Psychological Science.

0216

Reposted by ERROR

Ian Hussey @ianhussey.mmmdata.io · 14/11/2025

I think this is an overly pessimistic take from the @bmj.com. Sharing data does not inherently increase trust, rather it enables verification which allows for trust calibration. This example is a win. Serious issues were rapidly detected that would not have been without mandatory data sharing.

35214

ERROR @error.reviews · 14/11/2025

Synchronous Robustness Reports could explore implications of different analytical choices – but they could still suffer from bias. Hardwicke argues that preregistration is crucial to prevent it. @tomhardwicke.bsky.social

089

ERROR @error.reviews · 14/11/2025

Are methodological and causal inference errors creating a false impression that the gut microbiome causes autism? In this strong analysis, Mitchell, Dahly, and Bishop question the evidence. They show that triangulation in science requires multiple robust lines of research.

01812

ERROR @error.reviews · 30/07/2025

New Nature podcast episode about ERROR and the Perspectives on Scientific Error workshop!

162

Reposted by ERROR

Andrew Gelman et al. @statmodeling.bsky.social · 13/07/2025

“We pay experts to examine important and influential scientific publications for errors ... We expect most published research to contain some errors ... our reward system pays bonuses to both authors and reviewers even when minor errors are found ..." statmodeling.stat.columbia.edu/2025/07/13/e...

statmodeling.stat.columbia.edu

The ERROR project: “We pay experts to examine important and influential scientific publications for errors . . . We expect most published research to contain at least some errors . . . our reward sy...

1259

Reposted by ERROR

Ian Hussey @ianhussey.mmmdata.io · 15/02/2025

EU legislation requiring clothes be reused and recycled may be based on a numerical error in a 2017 NGO report where $460 billion was added instead of subtracted. www.frontiersin.org/journals/sus...

frontiersin.org

Frontiers | The Emperor’s old clothes: a critical review of circular fashion in gray literature

094

Reposted by ERROR

Ruben C. Arslan @ruben.the100.ci · 20/01/2025

My first time acting as a recommender for the ERROR project. Huge props to the authors @erichehman.bsky.social et al. for exposing themselves to this detailed post-pub peer review and to Will Lowe @conjugateprior.org for his detailed review and reproduction.

13010

Reposted by ERROR

Anne Scheel @annemscheel.bsky.social · 20/01/2025

Really amazingly heartening to see that academic discourse can also be like this

0387

Reposted by ERROR

Jamie Cummins @jamiecummins.bsky.social · 20/01/2025

One of our main goals with @error.reviews is to shift scientific culture to identify, acknowledge, and correct errors in the literature. Eric’s response to the recent report on his work is a perfect example of embodying these values. Kudos to the authorship team and reviewer Will Lowe.

0287

Reposted by ERROR

JK Flake @jkflake.bsky.social · 20/01/2025

It is an honor for anyone to care enough about your work to check it, so thanks dudes. A mistake I suggest others not make is failing to ensure that all the data, from download, to merge, to analysis is available and each step, not just the final analysis, is reproducible.

15816

Reposted by ERROR

Richard McElreath 🐈‍⬛ @rmcelreath.bsky.social · 21/01/2025

This sounds like the same error I blogged about a few years ago, the common error of trying to control for population (or body size or many etc) by dividing the outcome variable by it. Props to the authors for seeking review and taking the issue seriously. Role models for us all.

elevanth.org

There Are No Magic Outcome Variables

I'm very sorry but I am going to write about statistics and causal inference again. I'd much rather be doing science. I'll make it brief. I was reading a preprint about the relationship between kinsh...

510120

Reposted by ERROR

Lisa DeBruine @debruine.bsky.social · 21/01/2025

A ⭐️⭐️⭐️⭐️⭐️ model response to finding problems in your published research. If you’ve published more than a few papers, you too have almost certainly made at least one mistake, so we should normalise addressing them like this! Great to see @error.reviews approaching error detection so sensitively

0539

Reposted by ERROR

Sebastian Kuhn @sebastiankuhn.bsky.social · 20/01/2025

THIS is how science works. A great and honorable reaction - especially in these times of resistance to facts and smug self-righteousness. Thanks to @erichehman.bsky.social & @error.reviews!

0144

Reposted by ERROR

Eric Hehman @erichehman.bsky.social · 20/01/2025

We've had a paper reviewed by ERROR, which has concluded we had a "major error that affects a core conclusion" Few comments on the conclusion and the process in general 1/n

420954

The full rationale, including the review and author response, are available on PsyArXiv: osf.io/preprints/ps.... The study materials and all new materials generated as part of the review are on OSF: osf.io/abyqs/. Original article available here: journals.sagepub.com/doi/10.1177/....

OSF | Sign in

170

We are very grateful to @erichehman.bsky.social and colleagues for their participation in this project. This serves to help improve our culture around scientific error, and is a testament to their commitment to scientific values of transparency and verification.

151

A reformulated regression model that accounts for base rates more effectively finds no significant relationship between Black/White Racial Prejudice IAT scores and police killings of Black people. We recommend the authors seek a correction.

100

The error review found that Hehman et al. (2018) did contain a major error that affects a core conclusion. Specifically, the use of a difference of proportions variable as the regression outcome did not effectively adjust for the Black population base rate.

Hehman et al (2018)'s paper is on a highly relevant topic of great public interest. It is also highly cited, including by textbooks, news outlets, and policy white papers. Therefore, errors in this article matter.

130

We are grateful to Will Lowe for his diligent examination of the article and its materials, and to @erichehman.bsky.social, @jkflake.bsky.social, & @jimmycalanchini.bsky.social for their willingness to have their impactful work examined for errors.

161

ERROR is a bug bounty program for science to systematically detect and report errors in academic publications

New report! Hehman et al. (2018) "Disproportionate Use of Lethal Force in Policing Is Associated With Regional Racial Biases". Based on the review by @conjugateprior.org, we find a Major Error that affects a core conclusion. We recommend the authors seek a correction. error.reviews/reviews/hehm...

error.reviews

Hehman et al. (2018)

14624

Reposted by ERROR

Jamie Cummins @jamiecummins.bsky.social · 10/01/2025

ERROR is our scientific bug-bounty program. We pay reviewers and authors of impactful published articles to do a deep dive on results, data, code, etc to check for errors. Our second report of 2025 (and my first acting as Recommender for the review) is out now!

0136

ERROR @error.reviews · 10/01/2025

The full rationale, including the review and author response, are available on PsyArXiv: osf.io/preprints/os.... The study materials and all new materials generated as part of the review are on OSF: osf.io/8kuxp/. Original article available here: www.nature.com/articles/s41...

ERROR @error.reviews · 10/01/2025

OSF

010

We are once again grateful to both the original authorship team, and R Chris Fraley, for their openness to the pursuit and correction of errors in an open and honest atmosphere. Their participation is a testament to their commitment to transparency and openness in science.