Retraction Watch

Tracking retractions as a window into the scientific process

Inquiry finds no evidence of misconduct in high-profile Science paper flagged by allegations

with 14 comments

scienceAn expert group at Uppsala University has recommended not proceeding with a full investigation into allegations of misconduct in a high-profile Science paper showing how human pollution may be harming fish.

The June paper — which caught the media’s attention for suggesting fish larvae are eating small particles of plastic rather than their natural prey — became the focus of scrutiny soon after it was published when a group of researchers raised allegations of misconduct. Earlier this year, Science told us it was considering issuing an Expression of Concern (EOC) for the paper, and Uppsala said it was conducting an inquiry, the first step in determining whether to launch a formal investigation.

The expert group who conducted the preliminary investigation has ultimately recommended against an investigation of the paper, according to an Uppsala spokesperson:

Their conclusion is that there is no evidence of research misconduct and that there are insufficient reasons for a full investigation.

The spokesperson added that the university has passed the matter onto the Central Ethical Review Board, which is also preparing a report:

Uppsala University will make a decision on this issue when both reports are available to make an overall assessment.

You can read the expert group’s report in full here. It details a response to each of the accusers’ specific complaints about the paper, noting when they believe the authors — Oona M. Lönnstedt, a postdoctoral research fellow at the Department of Ecology and Genetics, and her supervisor, Peter Eklöv — had provided sufficient responses.

It concludes:

The investigators note that [the complainants] appear to have a very strong desire that the article be examined for research misconduct, but that the large majority of their objections come within the ambit of normal scholarly discussion, which could have been conducted directly with the authors of the article. The most serious accusations, which could potentially indicate misconduct, concern assertions that the experiments were not executed as asserted, and that Lönnstedt and Eklöv thus fabricated their data. The investigators have not found any evidence that this was the case.

However, the investigators note inadequate documentation of the research on the part of the accused, with necessary documentation only stored on one computer (which was subsequently stolen, as confirmed by the report of the theft to the police), and with a lack of back-up storage at Uppsala University. However, this cannot be judged to be a sign of any research misconduct.

Update 9/20/16 8 p.m. eastern: We’ve received a statement from Eklöv, who told us:

We are of course happy for the result from the investigating group where we were freed from all accusations. Although  we have always been confident about our own innocence it is very stressful to be attacked by a group with the only purpose of discrediting our research.

Update 9/21/16 9:10 a.m. eastern: We’ve received a statement from Science about the paper:

The Science editorial staff awaits Uppsala University’s assessment of the validity of findings in the study by O.M. Lönnstedt and P. Eklöv. Once the Science editorial staff has sufficient information from Uppsala University, they can work to determine the most appropriate next steps.

Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.

Written by Alison McCook

September 20th, 2016 at 2:15 pm

Comments
  • Kenneth P. Price, Ph.D. September 20, 2016 at 2:45 pm

    Perhaps journals -especially the most prestigious and influential- need to start requiring submitters of manuscripts to include with their submissions all documentation and raw data, whether in hard copy or on a flash drive, so their data can be checked if desired and other scientists can see, if they so desire, if their results are reproducible. Certainly, if no documentation is available (“the dog ate my homework”), the manuscript ought be rejected outright. If retractionwatch has proven anything, it is that many, many researchers and their articles cannot be trusted. That has serious implications for society, not just for the researchers’ careers. Especially since science today is becoming so politicized, on the one hand, and “scientific” data are being used to support political causes (e.g. banning firearms used for self defense; keeping opiates out of the hands of chronic pain patients; sun cycles have nothing to do with global warming or freezing – or is it climate change) on the other hand, it behooves the scientific community to police itself, adhere to honesty in conducting research and humility in drawing conclusions, especially if the authors are hawking drugs or making recommendations regarding physical health (how many eggs may we eat nowadays?) or societal issues. Ivan “the Terrible” is a national treasure. I salute you and your work exposing fraud and incompetence and encouraging honesty in research. (Full disclosure: I was an Asst. Prof. in Psych. for 3 years, published over 2 dozen journal articles and books, some of which are still cited, over 30 years after their publication, was an NIH site visitor and reviewer for a half-dozen or so journals, before leaving for independent practice.)

    • Brian September 20, 2016 at 4:36 pm

      The discovery of the Higgs boson used about 200 petabytes of data – that’s $12 million dollars worth of flash drives you need to ship to Nature’s offices. Probably a little over five hundred years time to download if you’ve got a decent connection (though I expect here Moore’s law will save you a lot of that).

      • Hard cases make bad law September 20, 2016 at 5:28 pm

        The fact that the odd exception may have to be made is no reason not to move towards robust data sharing.

        • Brian September 21, 2016 at 8:54 am

          It’s a somewhat extreme case, but not terribly extreme. My average publication has a few Tb of data – less than the LHC, to be sure, but it’d be beyond my own resources to make all of it pre-emptively available. Especially when the hassle and expense make it a useful tool for Creationists (or whatever the relevant political group is for you) to harass and undermine you.

      • Kenneth P. Price, Ph.D. September 20, 2016 at 10:24 pm

        Let’s give the alleged finders of the alleged God particle a pass. As it turns out, I have a set of five books that contain all of the God particle’s information that he/she wishes to pass on to us, in any case. I think those data (let’s use the Greek and call them, say, oh, idunno, the Pentateuch) will fit on one large flash drive. I like the second comment to my post. I suspect the data from the vast majority of research articles -certainly in the social sciences- will fit on a 128gig flash drive, probably a cheap 8 gig. Maybe 2 gig. Actually, most social science manuscripts shouldn’t be submitted for publication. It’s not for nothing that the median number of publications of Psychology Ph.D.’s is 1. The mean and probably mode is/are 0 if I remember correctly. BTW, Brian, something that today takes 500 years to download, will probably take 10 sec. or less in 20 years. Maybe fewer. Maybe a LOT fewer if we do the download on one of those planets circling Alpha Centauri. I hear the weather is temperate there and gravity less weighty. BTW, thanks for your comment. It sure made me think about retracting my suggestion. Jes sayin’…

      • Dave Fernig September 21, 2016 at 3:06 am

        These data are available – CERN has a data archive, since storage costs are negligible compared to the costs of an experiment. Similarly, atom coordinates for molecular structures, nucleic acid sequences, protein mass spectrometry and more are all stored somewhere, but not at the journal. A couple of clicks and you are there. Moreover, I would be very surprised if Swedish funders did not require a Data Management Plan, which requires the data to be maintained for at least 5 years.

        • Brian September 21, 2016 at 9:02 am

          This is simply not true. LHC only records events that set off a trigger. Most of their data is disgarded immediately. Roughly 0.00001% of the data taken at the LHC is actually recorded (see, e.g., http://www.lhc-closer.es/taking_a_closer_look_at_lhc/0.lhc_data_analysis )

          After that, it’s still not all available to just anybody. About half of it is in theory publicly accessible, but of course you can’t actually spend ~250 years downloading it. You’ll be dead first, as will your graduate students.

  • D Cameron September 21, 2016 at 2:30 am

    I am reminded of an exchange from the great Elia Kazan film “On The Waterfront”…

    “Sure. We got records.
    Where are they?
    We was robbed last night.”

  • Dave Fernig September 21, 2016 at 3:07 am

    One has to laugh at the statement in the report “On the other hand, the amount of space available in these prestigious journals is so limited that it is often impossible to include all details.”
    So prestigious that it doesn’t meet the bar for science set in high school.

    • TL September 21, 2016 at 4:54 am

      Even funnier is “it has to be assumed that Science has rigid and thorough review processes,
      including statistics experts who do not appear to have had any objections to the statistics published”. If it passed through peer-review in Science, it has to be accurate!

  • J. Näslund September 22, 2016 at 4:02 pm

    The report states: “All necessary raw data has been freely available to readers for some time, in supplementary data published by Science”.

    However, I think the committee did not really do their work appropriately, because a lot of data is still missing, from what I can see in the supplements. The activity measurements are repeated measures, reported as change in activity (clearly seen in Fig. 2). However, there are no repeated measures in the data provided in the supplements – only one set of data, which I assume are the either the “Before treatment” or the “After treatment” measurements…

  • Karin Johansson September 23, 2016 at 5:14 am

    It seems like a very unfortunate coincidence the raw data of criticized work gets stolen…

  • oliver September 24, 2016 at 5:19 am

    The-dog-ate-my-homework defense does work for SCIENCE?

  • Jtom September 24, 2016 at 4:06 pm

    Regardless of an investigation, if a researcher manages to lose supporting documentation, the paper is deserving retraction. Let them reproduce the research, obtain new data, and resubmit their finding.

    With flash drives being absurdly cheap, and cloud storage essentially free, there is no excuse for not backing up ones research. When I wrote research papers in private industry, I had multiple backups of every revision I made. Why would someone take such a reckless risk of losing months of work? A simple lightning storm, drop, or coffee spill could have wiped out all that work.

  • Post a comment

    Threaded commenting powered by interconnect/it code.