I caused an entire BOOK to be retracted. Reported 6 months ago. Swift decision by
@SpringerNature
. Also reported several books by the same authors to
@ElsevierConnect
, did heard a beep.
Editor: You didn't stop for a second and think "honey is amazing for Alzheimer's?!". ....Anyway, lots of back and forth between the authors and sleuths here
I matched a neuroscience PI at Harvard yesterday, on a dating app. Of course I immediately said I am one of those sleuths. Of course he immediately unmatched me. I laughed so hard. Empowering doesn't come close to describe it.
To friends who followed me from India and Iran this week, thank you for your support. My home country (China) is also plagued with fraudulent publications, and I share your frustration. The cause of the issue is very complicated, but together, we can start to make changes. 🙏
Weekend conversation on integrity: We need to address "toxic pressure in the lab" in order reduce fraud. Many cases of fraud were probably the result of desperate people working under duress. Implicit messages from PIs on what results are desirable are very harmful. Your turn.
I caused an entire BOOK to be retracted. Reported 6 months ago. Swift decision by
@SpringerNature
. Also reported several books by the same authors to
@ElsevierConnect
, did heard a beep.
This is not about a specific country. China (where I came from) has its huge share of F papers. Ultimately, the editors and publishers are the gatekeepers. The responsibility is not negotiable, given profits are involved. Let's keep the conversation civil.
Editor: You didn't stop for a second and think "honey is amazing for Alzheimer's?!". ....Anyway, lots of back and forth between the authors and sleuths here
After a behavioral experiment today (my actual job), I held a mouse in my hand and thought about the number of compromised studies that used mice, and how many little lives were lived for nothing, I choked up. Team🐭🐁
Never mind AI, I am not sure you even need a science degree to see it (and I didn't mark everything). Nevertheless it is in Eur J. Medicinal Chemistry (2023).
Marking is messy because same images were differently cropped and used multiple times. I didn't mark all, otherwise too messy. Nanjing Univ is one of top 10 universities in China. And this is in the Nature journal with all-Chinese editors.
When you are too lazy to report each paper to each EIC, you dump the "monthly collection" to the Integrity team at
@ElsevierConnect
and hope for miracles.
This sleuth-in-training needs your opinions: 1) What could be the incentives to publish a lot of papers that are rarely cited? 2) Should we worry about such papers?
I accidentally caused a Retraction "The corresponding author contacted the editorial office with a request to correct Figure 3A. However, image analysis performed by the editorial office revealed the revised Figure 3A still contained image duplication..." PMID 38608700
Error bars are present, but VERY VERY hard to see. This seems to be a Nature journal that's under an all -Chinese editorial team. Not sure how if there is a mechanism to prevent favoritism. I asked to see raw data on Pubpeer.
I communicated with the EIC of J. Chemical Neuroanatomy (Elsevier) many many times about this. He kept promising actions and even asked me what tools I use, and I told him. A random check today on a recent paper indicates a lack of improvement.
Sleuth pointed out the pasted part in A, author volunteered to correct B, incorrectly, saying .... "I am not very skilled at putting these figures together....". Now we just don't know.
@lonnibesancon
.
The "Survival of the Fittest" mentality is the source of more than a couple of vile things that plagued our culture. Now it is also plaguing the science community. About time to be woke on this one.
@SocialImpurity
@mumumouse2
Alternatively, it seems like you are putting the personal responsibility on the people who are being honest and not producing flashy results because of it, while suggesting they don't belong. That attitude is benefitting the fraudsters.
Just as I was about to conclude that Elsevier doesn't do anything about anything, here comes evidence of some effort. Appreciated it! Reported last May.
The black line has several segments that seem to be identical repeats or very similar. See green, yellow, and blue boxes. I most likely missed a lot, but am better than AI on this one haha. pubmed 35759866
It is shocking that mice can find the platform in a 120cm pool on day 1 in about 10s on average, with two reaching almost immediately. Notwithstanding a visual cue was present. Yes yes their tracking software "blinked" too long, I am sure that's the reason.
@VVoikar
How many times is too many times to re-use one's control data without acknowledging it in later papers? Yellow arrows point to data identical in 3 papers.
This email from an Elsevier publisher is a bit perplexing. This is after I raised the concern that an Elsevier journal has many "problematic" papers. They are just implementing AI software, after I have been telling them about the problem for two years? The"Prof" is the new EIC.
Fourth from this group. I can't think of how adjusting for baseline, age, and BMI could lead to identical error bars on all measures between two separate groups.
1) Different mouse models, 2) re-use not mentioned, 2) blots were partial in the 2012 (earlier) study, 4) supposed to show 4 lanes in the 2012 study and 8 lanes in the 2015 study?
*This is a scrutiny/discussion*. This study shows an astonishing effect of 30-day fish oil in repairing axons and drastically reducing APP levels in rats. It seems very very amazing. PMID 20635852
@jsmoliga
@mumumouse2
medicine is actually rife with these types of papers. it makes it extremely difficult to determine what is a real effect/a finding worth considering. at least in bio or other fields you can validate with little consequence, with medicine a bad study can have deadly consequences
Retracted a year after it was flagged. That is pretty fast.
@SpringerNature
is doing a lot better than Elsevier on this front, in my experience dealing with integrity folks in both.
I am perfectly happy to believe the same SDs (0.13) in most columns is a mistake, but it's kind of cavalier that the authors didn't bother to reply or anything since I flagged it in January. pubmed: 36899414
A certain level of data anomaly needs to be considered as "hard evidence" in the future. If we can only/primarily rely on on image evidence, then what happens when people start to launder their images with AI tools which *will* become more and more available?
This data anomaly was listed as one of the reasons to RETRACT this paper!
@SpringerNature
is very progressive in terms of heeding (traditionally ignored) data anomaly evidence!
2nd and 3rd month MAS (a clinical scale 0-4) data appear identical. The cerebrolysin group has the same SD in months 1,2,and 3. This was used to shown clinical efficacy of the drug, so maybe a little scrutiny is not too rude. pubmed 38585517
Part 1: The Erratum was published after I pointed out the red circle in A. Supposedly the correct image was made from the Original blots. I can't seem to match most of it. And the retina and PRE blots seem same in the correction.
@mumumouse2
That is quite a lot! There are 8000 that were retracted from Hindawi Special Issues, which were mainly done through its current publisher Wiley, without specific contributions of volunteers, so 40/2000 is a lot.
Thank you so much for your contributions. Much appreciated ❤️
Saw this tweet on someone's phone in the elevator after work. Oh good God, the days when I'm only known to the Columbia community as the "woman who runs the mouse behavior core" may be numbered.
I agree that mistakes happen and MANY image issues were caused by mistakes. But a good sleuth (that I hope to become someday) differentiates mistakes and non-mistakes. Here is an example.
Since there are more people willing to talk about error bars, let's revisit this PNAS paper (pubmed: 35353605) where I have questions about missing error bars in B and D. N=3.