The Bentley University accounting professor whose retraction we first reported on in November 2012 fabricated the data behind two papers, a university investigation has concluded.
James E. Hunton, who resigned in December 2012:
engaged in research misconduct by fabricating the data underlying Fraud Brainstorming and Tone at the Top…
according to a statement from Bentley ethics officer Judith A. Malone that was first reported by The Boston Globe.
“Fraud Brainstorming” is short for “A Field Experiment Comparing the Outcomes of Three Fraud Brainstorming Procedures: Nominal Group, Round Robin, and Open Discussion,” which was retracted in 2012 from Accounting Review. “Tone at the Top” refers to “The Relationship between Perceived Tone at the Top and Earnings Quality,” a 2011 paper in Contemporary Accounting Research whose results had been questioned by the journal editor just after the Accounting Review retraction but which has not been retracted.
Hunton told Retraction Watch in 2012 that he was hampered by confidentiality agreements he was forced to sign by the firms whose data he was using in his research. The university report makes it clear he repeated this claim:
Dr. Hunton repeatedly told Bentley, his co-authors and journal editors that strict confidentiality agreements prevented him from sharing with them the original data or the identities of the CPA firm (the supposed source of the data reported in Fraud Brainstorming) and the consulting/training firm (the supposed source of the data reported in Tone at the Top). He claimed that disclosure of either the data or the identities of the firms would result in him being subject to lawsuits, to the loss of his CPA license, and to a loss of confidence in the field and thus access to further research opportunities.
But after investigators contacted “any major accounting firm that might reasonably have supplied the data reported in Fraud Brainstorming, and any training or consulting firm that might reasonably have supplied the data reported in Tone at the Top,”
None was able to find any evidence that it had a research relationship with Dr. Hunton or any evidence that it had been the source of the data. No training or consulting firms (the supposed sources of the data in the Tone at the Top paper) were contacted because none could be identified in Dr. Hunton’s files.
The existence of confidentiality agreements seems highly unlikely. As the report notes:
The investigation recovered from Dr. Hunton’s computer copies of multiple versions of confidentiality agreements purportedly governing these two articles. These agreements – none of which indicated that they had been executed – all revealed unusual redactions, contradictory dates, and – most damaging of all – evidence that the documents had been revised after allegations were raised to make the documents’ prohibitions of disclosure more stringent.1 One of the firms, whose name appeared in one version of a confidentiality agreement, reviewed the document at Bentley’s request and reported that it bore no resemblance to any document that it ever used with research partners.
Hunton also seems to have destroyed evidence, which the university report said “supports the conclusion that he engaged in research misconduct:”
Despite having been cautioned on numerous occasions to retain all relevant documents concerning these papers, Bentley discovered after Dr. Hunton resigned that his office had been completely cleaned out of all physical files, and that his laptop had been wiped clean of all of his electronic files. Statements by Dr. Hunton found among his emails stored on Bentley’s servers strongly suggest that he had begun the process of removing or destroying files sometime in the fall of 2012 – prior to the time that Bentley had been informed of any allegations but after The Accounting Review had approached Dr. Hunton directly about the validity of the data in Fraud Brainstorming.
The investigation required, therefore, that Bentley engage a forensic consulting firm and a research assistant to analyze the hard drive of his laptop computer, to undertake recovery efforts, and to analyze and review the results of those recovery efforts. The consultant’s forensic analysis and recovery efforts revealed that Dr. Hunton’s laptop once housed a great deal of electronic data, and revealed no evidence to suggest that the data had been inadvertently lost. To the contrary, the forensic analysis discovered that an “eraser” utility had been loaded onto the laptop during the fall of 2012. This analysis also revealed that a second utility, which permits users to manipulate the creation dates in certain metadata fields in certain kinds of files, had also been loaded onto the laptop during the fall of 2012. Further analysis determined that the eraser utility had been run multiple times during that fall in an attempt (only partially successful) to delete all of the files on the laptop – including the metadata manipulation utility.
There is no evidence that Hunton’s co-authors — one of whose names was signed, probably by Hunton, to a comment on our original post explaining the retraction — knew anything about the fabrication, let alone took part in it, according to the report. The university will now ask Contemporary Accounting Research to retract “The Relationship between Perceived Tone at the Top and Earnings Quality.”
The report concludes that “The whole body of Dr. Hunton’s extensive research while a faculty member at Bentley University must now be considered suspect:”
He published approximately 50 papers while at Bentley, and there is reason to believe that many of those papers involved data that were provided by Dr. Hunton alone, as in the case of the Fraud Brainstorming and Tone at the Top research. Bentley University wrote to all of the authors of the additional papers identified by the confidential reporter to ask if they could personally verify the source of the data. Of the authors who responded, some said they were confident the data were valid, but none was able to produce original data or to identify with certainty the source of the data. Instead, and like in the case of Fraud Brainstorming and Tone at the Top, Dr. Hunton appears to have supplied the data allegedly obtained through his contacts within the accounting field, and provided the co-authors only summary spreadsheets of the data.
Ten of Hunton’s papers have been cited more than 20 times, according to Thomson Scientific’s Web of Knowledge.
Update, 12:30 p.m. Eastern, 7/22/14: We asked Dan Stone, an accounting professor at the University of Kentucky who has co-authored work with Hunton and who left a comment about the case on Retraction Watch, for his thoughts on the findings. He responded:
The Bentley investigation of Jim Hunton is a model of professionalism and competence. That it reveals that I was wrong about Jim; that, in fact, he did fabricate data is surely less important than is the clarity that this investigation provides. Although I have never been contacted in relation to this investigation, I commit to fully cooperating with it. Our community of scholars must now begin the important work of recovering from, and assessing the further damage caused by, Jim’s actions. One small benefit that has accrued from Jim’s misconduct is that our community is better prepared to prevent and catch future cases of scholarly misconduct.
http://retractionwatch.com/2014/07/21/accounting-professor-faked-data-for-two-studies-destroyed-evidence-university-report/ goes to a 404 error page
For some reason, it was posted with a different date in the URL. Here’s the right one: http://retractionwatch.com/2014/07/22/accounting-professor-faked-data-for-two-studies-destroyed-evidence-university-report/ Thanks.
It’s the Stapel-Forster Syndrome all over again. When the provenance of the data relies entirely on the word of one person.
It is a bit ironic, that a noted and highly regarded and award winning professor of accounting, computer IT and fraud can’t even get the job done, of erasing a hard drive (after buying software for that purpose).
https://www.bentley.edu/files/Hunton%20report%20July21.pdf
Harry Markopolos (who ever you are) should be given a medal for your hard work (see the original RW post).
The poster Anonymous (11 January 2013) was very prophetic indeed.
Yeah. Always physically destroy the drive. That means opening it up and grinding the platters. Nothing less will work. And of course e-mail and anything posted to the interwebs can never be removed.
Interesting. Hard to say whether he had any “expertise” beyond first hand experience of committing (poorly executed) fraud!
He would have been better to transfer all files on computer to external hard drive, go buy a new hard drive for his laptop and install with a backdated start date. Then transfer good (cleaned up) files to the new hard drive on laptop. Would have been a little less obvious, and the “data trail” would have been lost forever. But given his education/expertise he should have known that.
Bentley deserves a great deal of credit doing what appears to be a very extensive investigation (not only time but money) and making this investigation public.
Laptops get stolen all the time. Theft, damage, and spontaneous hard drive failure — the latter becoming rarer as disks get more reliable and solid-state drives become more popular, but it still happens — are the most common causes of data loss. (Almost no data is destroyed by viruses, but that meme will never die.) So by far the easiest way to do this would have been to “disappear” his laptop. The school is hardly going to get a warrant to search his cousin’s garage. (Amusingly, Diederik Stapel’s autobiography describes his laptop actually being stolen from his car, some time _before_ he was unmasked.)
I think Bentley University deserves credit for the way they’ve handled this situation. Many a university would have shied away from publicly concluding misconduct through fabrication in the absence of hard and direct evidence (which can usually only come in the form of a full confession), especially when a complaint concerns a well-established scientist. So kudos to them for doing a difficult job thoroughly and then publishing a summary that doesn’t flinch from addressing the matter squarely.
What should happen to the co-authors of these papers? Do they have no responsibility for what occurred? If they were not privy to the data, how could they have participated in writing the paper?
As I understand it, in more than one field the common practice is that the PI collects the data and analyses it. Using this analysis, several coauthors join in writing the paper, or they may have only been involved in the experiment design, or their names were added for more obscure reasons. The raw data is not usually shared with coauthors.
I would say that all authors share responsibility for all facets of the publication. In my opinion, coauthors should insist on receiving a copy of all the data used for preparing an article. And perhaps it is sufficient punishment to have their names attached to the retraction.
I don’t know what fields you are referring to, but most hard sciences rely on students and lab technicians to get data (the PI is in an office the whole time). And the many studies I participated in while a poor undergrad also had grad students running the show, so this doesn’t seem unique to studies that don’t involve human subjects.
The circumstances are important. If the experiment was emailed out to participants, then maybe only the PI would have been involved. But if the experiment was administered in person, I can’t imagine one person being able to handle running every experiment (e.g. subjects might be in several rooms). Is this how research in accounting usually works? Do people recruit help from grad students or colleagues when collecting data? Could one person handle everything?
Of course, the experimental setup might make a single person handling the data fishy, or not. I’m not sure whether that’s applicable here. I am sure that none of my collaborators want me emailing them terabytes of simulation outputs for them to peruse themselves; they’d much rather just see the plots. (And I don’t really want them emailing me terabytes of observational data, either).
@Andrew. Your example strongly supports my claim that the circumstances (what you call “experimental setup”) are crucial in determining the reasonableness of a single author handling all aspects of the data. In your example, it would be reasonable for a single person to handle terabytes of data.
I suppose there are some circumstances that would lead to more questions than answers; that might be the case with Hunton. I don’t know, but an investigation into all his work seems warranted.
@Skeptical Scalpel: Dan Stone, who has co-authored work other than the two papers most under question, has responded to Retraction Watch.
The response of Dan Stone (quote from an update at RW, 12:30 p.m. Eastern, 7/22/14):
“The Bentley investigation of Jim Hunton is a model of professionalism and competence. That it reveals that I was wrong about Jim; that, in fact, he did fabricate data is surely less important than is the clarity that this investigation provides. Although I have never been contacted in relation to this investigation, I commit to fully cooperating with it. Our community of scholars must now begin the important work of recovering from, and assessing the further damage caused by, Jim’s actions. One small benefit that has accrued from Jim’s misconduct is that our community is better prepared to prevent and catch future cases of scholarly misconduct.”
I am wondering if co-author Anna Gold ( http://www.feweb.vu.nl/nl/afdelingen-en-instituten/accounting/scientific-staff/gold/index.asp ) is also willing to comment on the statement attributed to her and Hunton of 27 November 2012 (= response #3, http://retractionwatch.com/2012/11/20/accounting-fraud-paper-retracted-for-misstatement/ ).
Hey all, I was a coauthor on one of Hunton’s papers that was independently validated because we got the data e-mailed from the participating financial firm directly. I just wanted to chime in that it is pretty rare these days (and has been for years) to have hard copies of data at all, much less to keep them for years. For my dissertation I administered hardcopies to fewer than 100 participants, and I carted the boxes of data around with me on two interstate moves and then finally threw them out. It was literally taking up so much space that I could no longer fit them in my office cabinets or closets at home. And that was just ONE experiment. Most researchers, at least in business, have converted over to computer or internet administration of studies (I was doing this back in 2006 as a Ph.D. student and still do today, for the most part), and most universities have some kind of software or online platform for administering studies. Thus, to say that “no hardcopies could be located” really doesn’t mean anything at all. For better or worse, that’s the world we live in now. One of the best ways to judge the validity of a paper is through replication. I know a number of Hunton’s results have been replicated, but it’s something we definitely need to do more of in business research. One other somewhat funny observation – on the paper I worked with him on, where I did most of the data analysis, he had done an initial analysis and concluded that “guys the results just aren’t there…too bad.” The funny thing is, the results were there just as we had predicted and very significantly so, and he had apparently done the analysis incorrectly. This isn’t rocket science, it’s a simple 2 x 2 ANOVA. At the time I remember thinking, gosh, how did this guy get so far without even being able to do a standard ANOVA correctly? I guess it’s easier to know the results are there when you fake the data…
“One of the best ways to judge the validity of a paper is through replication.”
That’s a good way to judge whether the claimed effects do in fact exist. The question of whether the data were made up is entirely orthogonal to that.
If I were going to make up some data, I would try to make up data that are realistic, i.e. data that would be replicable if anyone really did the study.
However that would still be fraud. It would be exactly the same degree of fraud as if I made up some unrealistic data.
They’re two entirely separate issues.
wouldn’t that require the author to know ahead of time what the result would be?
Nick, thanks for providing your insight in this matter. There has been quite a bit of speculation regarding Hunton’s other publications and those papers in which he helped provide data so you can help clear the air here. So you affirm that someone besides Hunton first received the participant data? Is this also true for the hardcopy data that you threw away?
Is the paper you refer to as the data being independently validated one of the JAR papers retracted in May? If the data was not the issue, what was? I see at the byu ranking website under Hunton’s record that the number of retractions is now six.
“The American Accounting Association has retracted 25 articles, and one section of one article.” http://aaahq.org/Portals/0/documents/Website-FinalListofRetractedArticles-6-25-15.pdf