Scientists are commonly portrayed as paragons of rationality, subsisting solely on fact-based evidence and hard data; what goes on behind laboratory doors is in large part a mystery to the general public. In the past several years, however, there have been several notable cases of scientific fraud exposed: in 2006, it was revealed that Hwang Woo Suk fabricated data in papers claiming that he had managed to extract stem cells from cloned embryos. At around the same time it was discovered that Norwegian researcher Jon Sudbo made up a study where he claimed that anti-inflammatory drugs reduced the risk of oral cancer.
Just last month, the Office of Research Integrity (the watchdog of the scientific community) found former Harvard psychology professor Marc Hauser guilty of research misconduct and fabrication of published data, although it is still unclear whether the fabrications were the result of intentional, outright fraud. In all of these instances the news made national headlines and shocked the general public, shattering the illusion that scientists are immune to human frailty. These isolated cases, however, are not representative of the state of science. For the most part, scientists are still pretty trustworthy, but maybe not as much as originally believed.
An article released in the early edition of the Proceedings of the National Academy of Sciences Monday found that scientific misconduct, not error, was the principle cause of article retractions. What was most concerning about the author’s conclusions, however, was the finding that the rate of articles retracted as a result of fraud has increased tenfold since the late 1970s. While these findings aren’t necessarily harbingers of doom for the future of science, they do underscore several issues which need to be addressed.
The authors of the paper vetted the PubMed database and identified 2,047 retracted articles. Each article was then classified based on the reason for retraction, with the most innocent being retraction as a result of error and an assortment of other miscellaneous reasons (authorship dispute, etc.). The more egregious circumstances fell under the umbrella of ‘scientific misconduct’, namely confirmed and suspected fraud, plagiarism, or duplicate publication.
Previous studies of retracted articles have found that the majority of retractions were the result of error; however, after re-analyzing the data the authors concluded that the majority of retractions were actually due to scientific misconduct. The discrepancy arose from the obscure and often misleading ‘retraction notices’ which accompanied some articles. The authors cited an instance where they found that the retraction notice accompanying an article originally published in the Journal of Cell Biology stated that the “lack of FOXO1a expression reported in figure 1 is not correct”, while a report issued from the Office of Research Integrity found that the author had knowingly and intentionally falsified the results presented. In total, 158 of 742 retractions originally chalked up to error were reclassified as fraud.
After carefully reviewing the reasons for retraction, the authors found that 67.4% of retractions were a consequence of scientific misconduct. Articles retracted because of error made up only 21.3% of the total retracted articles, while fraud or suspected fraud accounted for 43.4% of all retracted publications, plagiarism for 9.8%, and duplicate publication for 14.2%.
They also found that the percentage of published articles retracted as a result of fraud or suspected fraud per year has increased tenfold since the first retraction in 1977, up from approximately .001% to .01%. These numbers may not be staggering, but it has been suggested that the actual occurrence of scientific misconduct is most likely much higher. In a 2009 survey it was found that nearly 2% of scientists reported having fabricated data at least once, and 14% knew of fabrication committed by their colleagues (as my fiancé, who works on the policy side of energy policy, read this, his jaw dropped: 2%? Really?). A paper published in January reported that of 2,782 doctors and scientists surveyed in the UK, 13% claimed that they had witnessed or had knowledge of scientific misconduct.
That’s not to say that fraud has become rampant; 2,047 retractions out of the 25 million articles published on PubMed isn’t such a bad track record, after all. Although we haven’t yet reached crisis levels, the number of articles retracted annually has been steadily rising, and it is likely that several variables are contributing to the increase. The establishment of the Office of Scientific Integrity (now the Office of Research Integrity) and the passage of the Whistleblower Protection Act made it clear that fraud would not be tolerated and made it safer for researchers to come forward and voice their concerns about observed scientific misconduct. Plagiarism is more readily detected as the use of plagiarism detection software is becoming increasingly popular. The rise in the number of retractions due to scientific misconduct, therefore, can in part be attributed to more sophisticated methods of detection, and a more widespread effort to maintain scientific integrity.
While these efforts are positive attempts to maintain scientific integrity, more stringent measures need to be applied to ensure suspect data is not published or, when fraud is uncovered, that punitive action is taken to serve as a warning to other scientists. Many retracted articles are cited years after their retraction, as their status is not made clear or is willfully ignored by the authors who cite them. In some cases, when concerns are raised about suspect publications a more thorough investigation never takes place- or such concerns are never even voiced at all.
The current state of science and the pressure to publish, however, cannot be ignored. Scientists, especially those at top-tier research institutions, are often under relentless pressure to publish, publish often, and publish well (in journals with the highest impact factor), working around the clock to produce results which can be assembled into publications. When applying for grants, productivity is often an important part of the funding considerations- and if its not published, it never happened. Research in the US is also facing an uncertain future in light of looming cuts in federal research funding. If ‘budget sequestration’ proceeds as scheduled, federal R&D funds will lose $57.5 billion dollars over the next five years, beginning January 2nd of next year.
All of these variables keep many researchers in a constant state of high-stress, as they worry where the money to continue their work will come from and scramble to gather publication-worthy data, although sometimes the data just isn’t compliant. Months of work can sometimes yield unusable results, rendering all of that effort fruitless. Scientists are constantly laboring in a pressure cooker, and in extreme cases the pressure clouds their moral judgement and scientific integrity. While the environment is not an excuse for fraud, it is an unfortunate consequence of the constant need to be on the cutting edge of research.
The reluctance of many laboratories to collaborate and the constant need to publish stifles the creativity of many scientists. Scientists need to get past the myopia which is often developed after years of working at the bench and take the time to explore and be creative without fear of their research being deemed inadequate, simply because it has failed to yield many publications. Instead, researchers should be rewarded based on the soundness of their work and its significance. Unfortunately, however, many of the reviewers who ultimately decide who gets funded or which articles are publication-worthy are scientists themselves, and if there is one thing scientists lack, it's time.
But I digress. For most of these issues there is no quick fix; first and foremost, however, fraud needs to be addressed and handled more severely, and it needs to be made clear in no uncertain terms that it has no place in science. A massive overhaul of the scientific environment is also unlikely, but it at least needs to be recognized that at present there are many major flaws. So what's a scientist to do? It's not likely that we will find the answers to all of these problems anytime soon, but that doesn't mean we should stop looking.
Ferric C. Fang, R. Grant Steen, and Arturo Casadevall. Misconduct accounts for the majority of retracted scientific publications. PNAS 2012 ; published ahead of print October 1, 2012,doi:10.1073/pnas.1212247109
Fanelli D (2009) How Many Scientists Fabricate and Falsify Research? A Systematic Review and Meta-Analysis of Survey Data. PLoS ONE 4(5): e5738.doi:10.1371/journal.pone.0005738
Tavare A. Scientific misconduct is worryingly prevalent in the UK, shows BMJ survey.British Medical Journal 2012; 344: e377
Just last month, the Office of Research Integrity (the watchdog of the scientific community) found former Harvard psychology professor Marc Hauser guilty of research misconduct and fabrication of published data, although it is still unclear whether the fabrications were the result of intentional, outright fraud. In all of these instances the news made national headlines and shocked the general public, shattering the illusion that scientists are immune to human frailty. These isolated cases, however, are not representative of the state of science. For the most part, scientists are still pretty trustworthy, but maybe not as much as originally believed.
An article released in the early edition of the Proceedings of the National Academy of Sciences Monday found that scientific misconduct, not error, was the principle cause of article retractions. What was most concerning about the author’s conclusions, however, was the finding that the rate of articles retracted as a result of fraud has increased tenfold since the late 1970s. While these findings aren’t necessarily harbingers of doom for the future of science, they do underscore several issues which need to be addressed.
The authors of the paper vetted the PubMed database and identified 2,047 retracted articles. Each article was then classified based on the reason for retraction, with the most innocent being retraction as a result of error and an assortment of other miscellaneous reasons (authorship dispute, etc.). The more egregious circumstances fell under the umbrella of ‘scientific misconduct’, namely confirmed and suspected fraud, plagiarism, or duplicate publication.
Previous studies of retracted articles have found that the majority of retractions were the result of error; however, after re-analyzing the data the authors concluded that the majority of retractions were actually due to scientific misconduct. The discrepancy arose from the obscure and often misleading ‘retraction notices’ which accompanied some articles. The authors cited an instance where they found that the retraction notice accompanying an article originally published in the Journal of Cell Biology stated that the “lack of FOXO1a expression reported in figure 1 is not correct”, while a report issued from the Office of Research Integrity found that the author had knowingly and intentionally falsified the results presented. In total, 158 of 742 retractions originally chalked up to error were reclassified as fraud.
After carefully reviewing the reasons for retraction, the authors found that 67.4% of retractions were a consequence of scientific misconduct. Articles retracted because of error made up only 21.3% of the total retracted articles, while fraud or suspected fraud accounted for 43.4% of all retracted publications, plagiarism for 9.8%, and duplicate publication for 14.2%.
They also found that the percentage of published articles retracted as a result of fraud or suspected fraud per year has increased tenfold since the first retraction in 1977, up from approximately .001% to .01%. These numbers may not be staggering, but it has been suggested that the actual occurrence of scientific misconduct is most likely much higher. In a 2009 survey it was found that nearly 2% of scientists reported having fabricated data at least once, and 14% knew of fabrication committed by their colleagues (as my fiancé, who works on the policy side of energy policy, read this, his jaw dropped: 2%? Really?). A paper published in January reported that of 2,782 doctors and scientists surveyed in the UK, 13% claimed that they had witnessed or had knowledge of scientific misconduct.
That’s not to say that fraud has become rampant; 2,047 retractions out of the 25 million articles published on PubMed isn’t such a bad track record, after all. Although we haven’t yet reached crisis levels, the number of articles retracted annually has been steadily rising, and it is likely that several variables are contributing to the increase. The establishment of the Office of Scientific Integrity (now the Office of Research Integrity) and the passage of the Whistleblower Protection Act made it clear that fraud would not be tolerated and made it safer for researchers to come forward and voice their concerns about observed scientific misconduct. Plagiarism is more readily detected as the use of plagiarism detection software is becoming increasingly popular. The rise in the number of retractions due to scientific misconduct, therefore, can in part be attributed to more sophisticated methods of detection, and a more widespread effort to maintain scientific integrity.
While these efforts are positive attempts to maintain scientific integrity, more stringent measures need to be applied to ensure suspect data is not published or, when fraud is uncovered, that punitive action is taken to serve as a warning to other scientists. Many retracted articles are cited years after their retraction, as their status is not made clear or is willfully ignored by the authors who cite them. In some cases, when concerns are raised about suspect publications a more thorough investigation never takes place- or such concerns are never even voiced at all.
The current state of science and the pressure to publish, however, cannot be ignored. Scientists, especially those at top-tier research institutions, are often under relentless pressure to publish, publish often, and publish well (in journals with the highest impact factor), working around the clock to produce results which can be assembled into publications. When applying for grants, productivity is often an important part of the funding considerations- and if its not published, it never happened. Research in the US is also facing an uncertain future in light of looming cuts in federal research funding. If ‘budget sequestration’ proceeds as scheduled, federal R&D funds will lose $57.5 billion dollars over the next five years, beginning January 2nd of next year.
All of these variables keep many researchers in a constant state of high-stress, as they worry where the money to continue their work will come from and scramble to gather publication-worthy data, although sometimes the data just isn’t compliant. Months of work can sometimes yield unusable results, rendering all of that effort fruitless. Scientists are constantly laboring in a pressure cooker, and in extreme cases the pressure clouds their moral judgement and scientific integrity. While the environment is not an excuse for fraud, it is an unfortunate consequence of the constant need to be on the cutting edge of research.
The reluctance of many laboratories to collaborate and the constant need to publish stifles the creativity of many scientists. Scientists need to get past the myopia which is often developed after years of working at the bench and take the time to explore and be creative without fear of their research being deemed inadequate, simply because it has failed to yield many publications. Instead, researchers should be rewarded based on the soundness of their work and its significance. Unfortunately, however, many of the reviewers who ultimately decide who gets funded or which articles are publication-worthy are scientists themselves, and if there is one thing scientists lack, it's time.
But I digress. For most of these issues there is no quick fix; first and foremost, however, fraud needs to be addressed and handled more severely, and it needs to be made clear in no uncertain terms that it has no place in science. A massive overhaul of the scientific environment is also unlikely, but it at least needs to be recognized that at present there are many major flaws. So what's a scientist to do? It's not likely that we will find the answers to all of these problems anytime soon, but that doesn't mean we should stop looking.
Ferric C. Fang, R. Grant Steen, and Arturo Casadevall. Misconduct accounts for the majority of retracted scientific publications. PNAS 2012 ; published ahead of print October 1, 2012,doi:10.1073/pnas.1212247109
Fanelli D (2009) How Many Scientists Fabricate and Falsify Research? A Systematic Review and Meta-Analysis of Survey Data. PLoS ONE 4(5): e5738.doi:10.1371/journal.pone.0005738
Tavare A. Scientific misconduct is worryingly prevalent in the UK, shows BMJ survey.British Medical Journal 2012; 344: e377
