Work Enthusiasm or Authorship Abuse? (Part 1)
(February 27th, 2015) How many research publications can a scientist claim to author in a year? Can we really be the author of one paper a week, every week, for years on end? Jeremy Garwood looks at authorship abuse in biomedical science.
In the latest issue of Lab Times, we looked at authorship contributions in multi-author research publications, how these can be abused, and the proposed use of ‘digital badges’ as a partial solution.
But it has been recently pointed out that Lab Times may itself be promoting unethical authorship practices through its regular feature on ‘Publication Analysis’ that lists the most highly-cited scientists in an area of research. In each issue of the magazine, citation statistics are presented for a subject area in medical and life sciences. These statistics reflect the number of times a research publication has been cited by subsequent research articles. This is taken to be a crude measure of research success because being cited by other researchers suggests that they have read the original paper and taken its research results and ideas into account when performing their own studies.
As such, Lab Times presents the total number of citations for a given subject over a given time period (currently the years 2007-2013) and ranks them according to who are the 30 ‘most cited authors’. In addition to presenting the total number of citations per author, it lists the total number of research articles that were cited. For example, the ‘winner’ in Physiology had 375 articles, closely followed by the ‘winner’ in Gastroenterology/Hepatology on 371 articles. With so many papers, it is hardly surprising to learn that their citation counts were also impressive: 8,848 and 25,767, respectively.
However, the question has been posed as to how a scientist becomes a genuine author of so many scientific papers. The senior Cambridge biologist, Peter Lawrence, has spent years condemning the ways in which senior scientists have abused their authority to steal credit from juniors (e.g. ‘Rank Injustice’, Nature 2002, ‘The politics of publication’, Nature 2003 and ‘The Mismeasurement of Science’, Current Biology 2007).
He recently asked Lab Times how it was possible for these senior scientists to co-author so many articles. For example, the top person in Physiology who had published 375 articles in 2005-2011 - “This is more than one paper a week. To me it shows how corrupt authorship is. It takes me about 1-2 years to write a paper, the work often takes longer. Now maybe I am unusually slow, but one a week, winter or summer holidays and conference trips etc. etc. not included?! It’s just a fiddle. How can all those people in that league table know what has been done in their name? If someone in their team is busy cheating, how would they know, how then can they take the credit for papers/discoveries?”
As he has previously noted in ‘Rank Injustice’, “everyone has their views on the contentious topic of authorship. Mine are that the person who is most responsible for the scientific findings and conclusions should be the first author, write the paper, settle any differences take responsibility for its contents (good and bad). Yet in reality, the PI (principal investigator) takes the lead (and his/her name occupies the final position in the author list), rarely having done the experiments or sometimes even written the paper - providing many opportunities for muddles and worse.” He points at the example of Robert Gallo “who spent much of the mid-1980s travelling (promoting his disputed claims to the discovery of the AIDS virus, yet still managed to author up to 90 papers per year!”
As discussed in the recent LT article, increasingly strict guidelines are being called for to clarify who should be an author on a research paper. For example, the ICMJE guidelines insist that insufficient claims to authorship should be relegated to a contributor’s role, i.e. mentioned in the Acknowledgements section. This includes scientists who have only performed a role of general supervision (or administration), or those who have provided a technique that was routine, or an antibody or other reagent that has already been published, or who only obtained funding. This last point is delicate, since without funding the research might never have been performed, but does this mean that, for example, the funding agency should be a co-author rather than appear in the Acknowledgements?
How much does an active researcher publish on average?
There is effectively a huge difference between researchers who publish 1-2 papers a year and 1-2 papers a week! The most active full-time research period is usually considered to be the post-doctoral years when junior researchers have learnt enough from their PhD studies to do independent research, do not have administrative responsibilities, may not have family ‘distractions’ (children, etc.), and are motivated to work hard by the hope that they can do enough to obtain a stable employment position.
Two recent articles have studied the association between publication success and the chances that a post-doc will succeed in becoming a principal investigator (‘PI’). ‘Predicting Publication Success for Biologists’, looks at the numbers of research papers published by successful academics in biological and environmental sciences at 35 universities worldwide. Their measure of academic success was the number of peer-reviewed papers a researcher had published in the 10 years after they had obtained their PhD. They also considered the prestige of the university at which they obtained their PhD (as determined by the Academic Ranking of World Universities). During the first 10 years after their PhD award, they found that graduates of top-100-ranked universities averaged 26.1 papers, whereas graduates of universities ranked outside the top 100 published an average of 23.4 papers.
If we extrapolate from this sample of fairly successful scientists, it appears that during their most active period as post-doctoral researchers, they were averaging around 2.3-2.6 papers a year. At this rate, a senior scientist with a large lab group would need some 20 to 22.6 active post-docs at their command to attain an average of 52 papers a year, or one a week (and that’s assuming they consistently find results worth publishing).
However, the second paper, ‘Publication metrics and success on the academic job market’, finds that average publication rates for post-docs are lower. They used a machine-learning approach to predict who would become a PI, based on data from over 25,000 scientists in the PubMed database of research publications. They claim that their results should hold for biomedical scientists from all fields. The graph in their Figure 1D provides a measure of ‘publication rate’ (publications per year per author) versus the time since the start of their careers (showing the first 10 years) for authors who subsequently became Pls and authors who subsequently left academia. For the authors that left academia, the rate varies between 0.6 and 0.8 papers/year, while for the authors that succeeded in becoming Pls, the rate is higher, but not spectacular, at between 0.8 and 1.2 papers per year over their first 10 years of research.
These figures suggest that for a senior scientist to attain publication rates of one paper a week over a 6 year period, would require them to have even more post-docs at their command – 43.3 to 65 of those with a chance of becoming a PI, and 65 to 86.7 post-docs who are coming around to the realisation that they’re not going to make it in academia.
But, of course, most research laboratories also have research students doing their PhDs. Yet, even when present in larger numbers than post-docs (since they cost considerably less to employ), they also have lower publication rates. So, is the secret of high publishing senior scientists that they have extremely large labs, filled with, say, 20 hard-working postdocs and 50 PhD students? Or are they abusing their position of authority?
... to be continued next week