Do Innocent Errors Cause Most Retractions?

Contrary to previous studies, a new publication finds that most retractions from scholarly literature are not due to misconduct.  

By | November 2, 2012

Wikimedia, VmenkovIn a sweeping study that examined nearly 4,500 papers retracted from scholarly literature between 1928 and 2011, researchers found that most retractions were not a result of misconduct, and fewer than half were due to questionable data or interpretations. Instead, the study suggested, many retractions are a consequence of publishing misconduct or errors, such as plagiarism, duplicate publication, authorship issues, and unspecified copyright issues.

The results, published last week (October 24) in PLOS ONE, were based on information collected from retraction notices, however, critics say such notices may not always be accurate.

It’s a “methodological weakness,” microbiologist Ferric Fang of the University of Washington told Retraction Watch. “The information given in retraction notices was taken at face value,” said Fang, who led a similar study last month that found that most retractions were, in fact, due to misconduct and that many retractions notices were incomplete. Because “no attempt was made to independently verify the accuracy of the statements made in the notices,” he added, the new study’s findings—that 47 percent of paper retractions were due to publishing misconduct or errors, while only 20 percent were due to research misconduct and 42 percent were due to questionable data or analysis—may not be accurate. (The total totals more than 100 percent because some retraction notices listed multiple reasons for the retraction.)

Moreover, Fang also pointed out that the new study used an unusual classification system for retractions. “Specifically, the classification of data falsification or fabrication, plagiarism, and intentional duplicate publication as forms of ‘author error’ is confusing, as most studies have characterized these practices as misconduct, as opposed to error,” Fang noted. 

Despite the difference, the new study did echo previous findings that the percentage of retractions has been increasing in the past decade, though they still represent a small fraction of the overall number of publications. The study also found that “fifteen prolific individuals accounted for more than half of all retractions due to alleged research misconduct, and strongly influenced all retraction characteristics,” the authors wrote.

Add a Comment

Avatar of: You



Sign In with your LabX Media Group Passport to leave a comment

Not a member? Register Now!

LabX Media Group Passport Logo


Avatar of: Paul Stein

Paul Stein

Posts: 229

November 2, 2012

Last time I looked, "plagiarism, duplicate publication, authorship issues, and unspecified copyright issues" were all forms of misconduct.  All of these are willful actions and are just plain wrong.  Has the dictionary definition of misconduct changed?

Avatar of: jeenious


Posts: 45

November 2, 2012

It's encouraging to read that results of a serious study indicated most retractions of articles published for peer review are not due to dishonesty.

Some sixty years ago, as a high school student needing summer employment, I had the wonderful good fortune to be hired to assist in gathering samples for studies by an entomologist (Harold Newsome, PhD, Louisiana State Agricultural Extension Service, L.S.U. professor).

I was at an impressionable age and was much influenced by Dr. Newsome's personal committment to the highest standards of accuracy attainable in the gathering and recording of samples of infestations following a number of then-experimental cotton poisons. (We even experimented with organic mercury, as well as experimental products named Aldrin and Dildrin (sp?), and compare to the then standard cotton dusting compounds sprayed on specially sequestered stands of cotton. We literally pulled up plants, for one experiment involving treated versus control thrips-targeting substances, placed each leaf on light-sensitive paper, to be measured for total comparative leaf area to be done by PhD candidate students during the following winter.)

One hundred-plus-degree fahrenheit day, as I counted and recorded bowl weavil damaged versus undamaged cotton bowls, a sleek car with a sleekly dressed (and obviously unaccustomed to sweating this much) came rolling up behind Dr. Newsom's truck. The sleek young man introduced himself as a "representative" of a particular chemical company and -- glancing furtively at me and another high school-age sample taker, kept urging Dr. Newsome to get into his sleek, air conditioned car to "discuss some business."

Dr. Newsome, to his great credit, and to my impressionable admiration said, "No. If you want to talk, we'll talk right here."

In a nutshell, the young man offered Dr. Newsome a very large amount of money to "endorse" his company's product.

Dr. Newsome told the representative to submit an inquiry in writing, to the Agricultural Extension Service, requesting the statistical results of research on the product in question, and the published results would be provided to him at no charge.

Then, he told the young man we had much work to do, and the "business" between them was, at that juncture, "over".

Since that day I have known many who engaged in scientific research of one kind or another. Some I know for a fact could be persuaded to slant results, and some I know for a fact have been careless in their sample-taking. Some, like Dr. Newsome, however, were rigorously insistent upon accuracy and statistical discipline and doing science responsibly.

Whether or not fudging one's stats is "dishonest" may be a matter of personal subjectivity and standards. Some say, "Everybody does it."

Well, by golly, not EVERYBODY cheats. And not EVERYBODY sells out. And not EVERYBODY rationalizes that dishonesty in science research allows for a little rationalizing, a little sloppy carelessness, or a little fudging is within the bounds of scientific honesty.

Avatar of: Alan Price

Alan Price

Posts: 4

November 3, 2012

Paul Stein's comment is absolutely correct -- the authors' definition of "research misconduct" in this paper excluded from "research misconduct" acts of plagiarism (listed instead as "publication misconduct") and other types of misconduct that are uniformaly considered to be misconduct in science.  

Thus, at least half of their "publishing misconduct" cases (796 for plagiarism alone) should be moved in Figure 3 to the "research misconduct" side, and likely many of the 915 "distrusted data" cases should be moved to "research misconduct" as well -- so that the total of research misconduct would be over 2,300 (of the total 4,232 retactions studied) -- well over 50% then being "reseach misconduct."

The authors classified the cause for retractions based on the language published by the authors of the retraction.  In my 17 years of experience as the cheif research fraud investigator for ORI, it was sometimes very difficult for ORI to get the guilty party or the lab chief or the journal editor to wait for the public ORI misconduct finding and ORI's demand for retraction, before they proceeded themselves with a retraction of the paper -- which often did not mention any misconduct, nor name the gulity party, instead just claiming that the results were "not reporducible." 

I did a similar research study when I was Associate Director in the Office of Research Integrity (ORI), working with Dr. Mary Scheetz of ORI and Dr. Sheldon Kotzin of NIH National Library of Medicine (Medlline). We submitted an abstract for Drummond Rennie’s 2005 Peer Review Congress (but it was not accepted for presentation):   “We noticed a high correlation between the retraction of publications involving United States Public Health Service (USPHS) support as listed in MEDLINE® and the existence of a related research misconduct case involving one of the authors of the retracted publications. We describe an analysis of that data, and related data involving other cases known to ORI staff or made public as involving allegations of misconduct. Approximately 64% to 72% of such PHS-related retracted publications were found to be associated with allegations and/or findings of research misconduct known to ORI, over the years that such retractions have been indexed in MEDLINE. We speculate whether a significant fraction of the other 28% of the retracted papers may also have involved PHS-related research misconduct, rather than being retracted because of errors or other scientific judgment reasons. We furthermore encourage editors to examine seriously requests for retraction, to ensure that those involving scientific misconduct will identify the person/author responsible and exonerate the other authors. . . .   “Over the past three decades there have been 572 retracted publications listed in MEDLINE, an average of about 20 publications retracted per year, about 30% of which involved PHS support. Of the 175 retracted publications citing in MEDLINE with PHS support, 114 (64%) were known by ORI staff to have involved cases in which research misconduct was alleged (some cases led inquires but no investigation, but others led to investigations, most of which found research misconduct for the authors/papers cited). When the 43 other retracted publications known to ORI staff to have been involved in such allegations or investigations related to PHS-appropriated funds are added, then of all of the 218 such retracted publications, 156 (72%) were known by ORI staff to have been related to research misconduct cases.”   Alan Price, P.R.I.C.E.

Popular Now

  1. So You’ve Been Mistaken as a White Nationalist
  2. Opinion: We Need a Replacement for Beall’s List
  3. Trump Releases Science Spending Priorities for FY2019
  4. Seeding the Gut Microbiome Prevents Sepsis in Infants