On Saturday, we highlighted a great two-part series by Joseph Neff of the News & Observer diving into the story of “Stefan Franzen, a chemistry professor at North Carolina State University who has been trying unsuccessfully to correct the scientific record.” Today, that series became a three-part series, with a new story revealing that an investigation by the National Science Foundation (NSF) Office of the Inspector General (OIG) had found “reckless” falsification in the work in question.
One of the key papers in the controversy was published by Lina A. Gugliotti, Daniel L. Feldheim, and Bruce E. Eaton in Science in 2004 and cited 125 times, according to Thomson Scientific’s Web of Knowledge. (Eaton is now at the University of Colorado at Boulder.) In 2011, Nature reported on the situation:
In 2009, Terri Lomax, the NCSU’s vice-chancellor for research, drafted a letter to Science recommending that the journal retract the paper or contact the authors for an erratum. The letter was not sent, however, and some months later, Lomax sent a milder version that simply noted the misstatements and suggested that the editors assess whether these had been corrected in the record.
A spokeswoman for Science says that, after reading the letter, editors contacted Eaton and Feldheim, who replied that their 2008 response to Franzen contained the necessary correction. Aware that the inspector-general of the NSF was still investigating, Science decided not to take further action.
That investigation would appear to be complete, although its recommendations have not been approved by top NSF OIG. According to a September 2013 report to Congress published today by the N&O, however, NSF investigators
…determined two faculty members and a graduate student at a North Carolina university recklessly omitted experimental details and overstated their experimental results in a published article, to an extent that constituted falsification.
To be clear: The NSF OIG report is anonymized, so even though the circumstances match, the N&O couldn’t confirm that it was the Eaton case. A university spokesperson told the newspaper, however:
The OIG report appears to be the Franzen-Eaton-Feldheim case, and it does appear that the OIG has found research misconduct and recommended sanctions, but we cannot definitively confirm that.
According to the report to Congress, the university did request a retraction:
The university’s investigation concluded that at least one of the faculty members had falsified but had done so carelessly, which did not constitute research misconduct. Nevertheless, the university requested that the authors retract the article. When the authors disregarded that request, the university sent the request directly to the journal – which did not retract the article.
And now, the NSF OIG — pending final approval — recommends the same thing:
We concluded that collectively the coauthors recklessly falsified their work in the original article. We recommended that NSF require retraction of the article and three years of certifications and assurance for each author, and bar each author as an NSF reviewer, advisor, or consultant for three years.
Eaton wrote what can only be called a remarkable letter to Neff in response to a request to interview him. Excerpts:
Every year just before the holidays professor Franzen launches some attack of some sort. Numerous institutions and government agencies have had scientific experts review the allegations. Professor Franzen has not been satisfied by their conclusions and so he has come to you and the popular press.
“I do not care about this area of science at all,” Eaton wrote, citing a 2010 paper he co-authored as “the definitive article that explains everything:”
I did care at one time, but Franzen’s relentless attacks have killed the program. He has won and you can write an article to congratulate him on stifling innovative thought and creativity. Furthermore, you can encourage him to write threatening letters to my students as he has done where he promises to ruin their careers. Great guy, you should work closely with him and get inside his brain. I am sure you will be fascinated to learn how it works.
The N&O reports that
Franzen said he had warned graduate students of falsified data and scientific problems. He said he never threatened anyone’s career.
Find some time to read Neff’s whole series here.
If I understand it, this is a case where the authors could not produce the data that supports the claims in their paper, and such data was not found by the OIG in the original notebooks and other records. Apparently the data that does exist contradicts the claims.
This seems different to what we usually see here, and a bit more serious.
DanZ
interesting… didn’t deny fabrication/falsification
This is a fascinating case, esp since I’m in the area- and I’m also a chemist. Just got a notification about a talk Franzen’s going to give on research integrity at Duke. It’s tempting to actually go…….
Peer review is suffering as journals and universities seem more dedicated to defending their reputations instead of promoting the truth and science we can trust. I have been trying to get a paper retracted that failed to report that butterfly populations in a natural habitat were thriving, in order to blame global warming for the extinction of neighboring butterflies a few yards away that had just colonized the newly logged area and new food plant but then went extinct. By only reporting half of the data the authors created a misleading illusion, but the editors for the American Meteorological Society justified the half truths. read How the American Meteorological Society Justified Publishing Half-Truths http://landscapesandcycles.net/American_Meterological_Society_half-truth.html
Fascinating!
“NCSU concluded, “that at least one of the faculty members had falsified but had done so carelessly, which did not constitute research misconduct.”
NSF concluded, “that collectively the coauthors recklessly falsified their work in the original article.”
Apparently, a VERY FINE LINE between careless and reckless.
My prediction: a lot of falsifiers will adapt the careless defense.
I actually think that this is a line, but what are the criteria that NSF used that the university did not, and what are the distinctions between careless and reckless?
The problem is that we often see these things lined up as if there was a continuum of guilt like this:
innocent – inadvertent – careless – reckless – fraud
It often seems (and I am not specifically talking about this case) that investigators think there was intentional fraud, but are afraid to say it, and instead use ‘careless’ as a dodge to avoid upsetting people.
I think instead these things each have their own qualities. If there are irregularities an investigation should only say ‘careless’ if there is reason to use that specific word. ‘Careless’ is not a junior form of ‘fraud’.
Dan , I agree with your “continuum of guilt.”
Reckless, is included in ORI’s definition of research misconduct, and apparently also NSF’s. Good for them. Reckless leads to sanctions.
Can we agree that careless is not research misconduct? or fraud?
My problem is that knowing what it’s not does not tell what careless is – a careless mistake – copying down the wrong number in a Table, for example.
How is careless distinguished from reckless?
And a caveat: Innocent, inadvertent and careless, errors nevertheless certainly should be corrected in the scientific literature.
With apologies to Yes “That’s another of those irregular verbs, isn’t it? I give confidential press briefings; you leak; he’s being charged under section 2A of the Official Secrets Act.” Minister.
I make inadvertant mistakes, you are careless, he committed fraud
A few issues here warrant highlighting.
1. The NSF investigation apparently ran for four years or more. Why did it take so long?
2. The University says they asked Science to retract the article but Science says there was only a mild suggestion to look into the issue. One of these accounts can’t be true.
3. After receiving the mild letter, Science was satisfied with the authors’ response. There is mention of a “necessary correction” in some letter but no corrigendum to the original article appears to have been published.
If you have read NSF OIG’s Semi-Annual Reports to Congress [ http://www.nsf.gov/oig/pubs.jsp ] over the past 25 years, you may have noticed that routinely over 90% of the stories of NSF findings (63 of 70 in the past 5 years) and over 80% of the stories on NSF OIG recommended findings (71 of 86 in past five years) involve plagiarism, copying, failure to cite, theft of intellectual property, and such. In contrast, ORI has over those same 25 years reported 90-95% of its cases and findings involved falsification and/or fabrication. So, to answer your question about why it took NSF OIG over four years to investigate and recommend a finding in this case, a complex evidentiary and contested case of falsification of research is relatively rare for NSF OIG, and such cases often do take such federal offices many years to investigate and resolve, and to try to negotiate a settlement (often in order to avoid a lengthy and costly debarment hearing).
After a quick skim of the articles here, it seems that this is the sort of thing that could be settled with a good old fashioned video camera and some referees. Gather a couple neutral scientists in a room, order some fresh chemicals, let the original scientists do their thing and see if the Pd really falls out of solution or not. Record it and BOOM. Since both sides seem to claim they are legit, make an old fashioned science dual.
Since this finding has merited a science paper +, it’s clearly worth finding out if it works… Science should offer said dual and retract if it’s declined given the allegations and the results of the investigation. Also, I would enjoy watching this video.
Great idea! Actually, a version of this was done. I did a laboratory demonstration for the investigation committee. I prepared the solutions in sealed vials in a glove box (as the authors say one should do). Then I used a Hamilton gas tight syringe to add THF solution containing 400 uM Pd2(dba)3 to the distilled H2O solution (this is the control). If you use a 5%, 10% or even 20% THF to water percentage a black precipitate forms immediately. The picture of this is in the JACS paper in 2007 we published. I urged the committee to ask those other authors to show how they could do this. The committee did not ask to see it.
Despite this demonstration, there is a strange contradictory evidence item in the committee report. Evidence item #39 is claimed to be an experiment by one of the committee members (sic) that purports to show the Pd2(dba)3 can be dissolved in 5% THF. When I asked the Research Integrity Officer to see the evidence, I was told that there was no “evidence”. There is just the statement by one member of the committee that it is possible to do this. Based on this statement by the committee member, the committee decided that there was no falsification of the solvent conditions. What was reported was fine. This evidence item probably delayed the outcome of this investigation by 6 years. We need a better process for adjudication of these cases. I agree wholeheartedly with the idea of a scientific duel. Just not a duel to the death please! We should keep this in perspective.
It seems you could get a paper out of #39. Since it is a point of controversy, it should be easy to get it published, and would be simple to do.
Good point. Please see my laboratory experiments for undergraduate students published in the Journal of Chemical Education (Franzen, S. “Understanding organic solubility in mixed solvents” J. Chem. Ed., 2011, 88, 619-623). However, the part about it being easy to publish is not quite on point. No papers of this type are “easy to publish”. Correcting the record is not part of the sphere of interest of any of the mainstream journals. If you want to correct the record you need to do something original (novel) and then the have the correction be an incidental part of the whole work.
Thanks for the link to the Bad Science articles – well written and fascinating. I would lean towards the original errors likely being due to sloppiness (poor supervision, the grad student doing what the profs should have been doing or at least checking once the interesting results were in). Otherwise, why would the authors have asked Franzen to join them or run the electron microscopy tests? Seems like pretty dumb things to do if the original results were deliberately fabricated. But there does seem to have been an extensive effort at a cover up. More importantly, the authors have had a decade in which to prove that their process works and this does not seem to have been done. The best test is always replication.
That reminds me of the following paper:
http://cen.acs.org/articles/90/web/2012/03/Nano-Nuisance-Palladium-Source.html
“Surprisingly, it was found that commercially available samples of Pd2(dba)3 may readily contain up to 40% of Pd nanoparticles in a wide range of sizes (10–200 nm). “
Yes. We reported that too in our J. Am. Chem. Soc. paper (Franzen S.; Cerruti, M.; Leonard, D.; Duscher, G. “The Role of Selection Pressure in RNA-mediated Evolutionary Materials Synthesis” J. Am. Chem. Soc. 2007, 129, 15340-15346). We did not report a yield (and 40% sounds a bit high), but clearly Pd2(dba)3 spontaneously loses Pd atoms, which combine to form small spheroidal nanoparticles in the size range from 10-200 nm.