And no peer reviewer dares question the integrity of the data or researcher, it's taken as gospel.
So while many might know about, nobody dares say a thing.
And no peer reviewer dares question the integrity of the data or researcher, it's taken as gospel.
So while many might know about, nobody dares say a thing.
> "The agency’s reply, which Schrag shared with Science, noted that complaints deemed credible will go to the Department of Health and Human Services Office of Research Integrity (ORI) for review. That agency could then instruct grantee universities to investigate prior to a final ORI review, a process that can take years and remains confidential absent an official misconduct finding. To Science, NIH said it takes research misconduct seriously, but otherwise declined to comment."
https://www.science.org/content/article/potential-fabricatio...
There is simply no willingness on the part of federal funding agencies to investigate their own grantees after they receive credible evidence of fraud.
Maybe, but it's more likely that compiling solid evidence for fraud is just far too resource intensive and there is no reward for replication--especially negative replication.
And committing the fraud is far too tempting--you got your needed publication for tenure, and you're mostly safe as long as nobody accidentally makes your paper the cornerstone for something that becomes famous.
As for peers, you can spend your time moving your own ideas forward, or you can take a detour to prove one particular important researcher's ideas wrong. It's pretty clear which is going to be more beneficial to your career.
As the article points out, this was triggered by some short sellers looking to make a buck--not anybody in the field itself.
We were always questioning each other’s results. Especially when something seemed too good to be true. “Integrity” can take many forms, and it’s surprisingly easy to fool yourself when you’re doing work in the field. So the default assumption was that we were all fooling ourselves, not each other, unless proven otherwise.
However, in other areas where you have "real world" experiments, you don't even expect the experiments to replicate - two clinical trials on different sets of patients won't necessarily yield the same results, and different results when repeating a biological experiment does not necessarily imply fraud; we know that in this domain (unlike ML) we sometimes do have unknown confounders that experiments don't control for.
I'm not in that particular field of neuroscience (albeit Alzheimer's patients do come in to the clinic...because where else are they going to go?). But when I was in the research world, a big factor was who you knew, and who you could cozy up to, in terms of getting grants, papers published in top journals, etc. Who you ate lunch w/ in big conferences. That being said, fortunately, there was an element of recognition of skill/ability and good work, as well as a decent amount of (sometimes irrelevant) challenge/response and general peacock-ing in terms of the peer review...but it was a reminder that science is above all, a human endeavor and not immune to humanity's sins.