US scientists are significantly more likely to publish fake research than scientists from elsewhere, according to a bold statement published in a BMJ press release. The press release is about a paper called 'Retractions in the scientific literature: do authors deliberately commit research fraud?' in the Journal of Medical Ethics. How did he arrive at that conclusion? Language and apparently poor understanding of statistics.
The author is Grant Steen, president of Medical Communications Consultants, a company that sells medical writing services. That will be ironically important farther down, so keep it in mind. He searched the PubMed database for every scientific research paper that had been withdrawn between 2000 and 2010.
A total of 788 papers had been retracted, around three quarters because of a serious error (545) while the rest of the retractions were attributed to fraud (data plagiarism, data fabrication, or data falsification). 5.8% of papers, a significant number, did not have reasons listed for the retraction. The highest number of retracted papers were written by U.S. first authors (260), accounting for a third of the total. One in three of those was attributed to fraud.(1)
Fair enough, but that does not mean that a US scientist is more likely to engage in data fraud than a researcher from another country, despite the strange assertion. It simply means that PubMed is dominated by US researchers and had more submissions and retractions.
And he found fakes were more likely to appear in leading publications with a high "impact factor", which is the common measure of how often research is cited in other peer reviewed journals. As you can imagine, high impact factor journals like Nature had to be alarmed over this news. Normalizing for the dominance of the US in PubMed, Nature found that "US researchers have a lower fraud and retraction rate than authors affiliated with China, India, and South Korea". Again, not what the claims are.
So given the less than clear nature of the data, how did Steen determine "These results suggest that American scientists are significantly more prone to engage in data fabrication or falsification than scientists from other countries"? Simply because the US had the highest total.
Which country's scientists are most likely to produce frauds, as his paper suggests? He doesn't say that nor does the data yet he chooses language that makes it appear that way.
This is a guy who writes for a living so either he did it intentionally or he isn't very good at his job.
Other interesting factoids, devoid of hype:
NOTE:
(1) The UK, India, Japan, and China each had more than 40 papers withdrawn during the decade. Asian nations, including South Korea, accounted for 30% of retractions. Of these, one in four was attributed to fraud.
Citation: G Steen, 'Retractions in the scientific literature: do authors deliberately commit research fraud?', J. Med. Ethics, 2010; doi: 10.1136/jme.2010.038125
The author is Grant Steen, president of Medical Communications Consultants, a company that sells medical writing services. That will be ironically important farther down, so keep it in mind. He searched the PubMed database for every scientific research paper that had been withdrawn between 2000 and 2010.
A total of 788 papers had been retracted, around three quarters because of a serious error (545) while the rest of the retractions were attributed to fraud (data plagiarism, data fabrication, or data falsification). 5.8% of papers, a significant number, did not have reasons listed for the retraction. The highest number of retracted papers were written by U.S. first authors (260), accounting for a third of the total. One in three of those was attributed to fraud.(1)
Fair enough, but that does not mean that a US scientist is more likely to engage in data fraud than a researcher from another country, despite the strange assertion. It simply means that PubMed is dominated by US researchers and had more submissions and retractions.
And he found fakes were more likely to appear in leading publications with a high "impact factor", which is the common measure of how often research is cited in other peer reviewed journals. As you can imagine, high impact factor journals like Nature had to be alarmed over this news. Normalizing for the dominance of the US in PubMed, Nature found that "US researchers have a lower fraud and retraction rate than authors affiliated with China, India, and South Korea". Again, not what the claims are.
So given the less than clear nature of the data, how did Steen determine "These results suggest that American scientists are significantly more prone to engage in data fabrication or falsification than scientists from other countries"? Simply because the US had the highest total.
Which country's scientists are most likely to produce frauds, as his paper suggests? He doesn't say that nor does the data yet he chooses language that makes it appear that way.
This is a guy who writes for a living so either he did it intentionally or he isn't very good at his job.
Other interesting factoids, devoid of hype:
- More than half (53%) of the faked research papers had been written by a first author who was a "repeat offender." This was the case in only one in five (18%) of the erroneous papers.
- The average number of authors on all retracted papers was three, but some had 10 or more. Faked research papers were significantly more likely to have multiple authors.
- Each first author who was a repeat fraudster had an average of six co-authors, each of whom had had another three retractions.
NOTE:
(1) The UK, India, Japan, and China each had more than 40 papers withdrawn during the decade. Asian nations, including South Korea, accounted for 30% of retractions. Of these, one in four was attributed to fraud.
Citation: G Steen, 'Retractions in the scientific literature: do authors deliberately commit research fraud?', J. Med. Ethics, 2010; doi: 10.1136/jme.2010.038125
...but that does not mean that a US scientist is more likely to engage in data fraud than a researcher from another country, despite the strange assertion....
I never could understand to what degree people are sloppy with statistics and correlations. There must be a profound lack of knowledge of modal logic and elementary scientific methods and philosophy out there!
You have provided me with several linkable articles lately! Thanks! ;-)