commentary on the 1%

I just read this article highlight in Nature about the distribution of publication output that is so misleading I feel compelled to write about it. The finding of the original article by Ioannidis is that only 1% of researchers publishing between 1996 and 2011 published every year. The implication/spin given to this finding is that the top 1% of scientists are contributing disproportionately (41%) of the scientific output — a ludicrous misinterpretation. The fraction is so low simply because most people who publish happen to be graduate students and post-docs (just pick any scientific paper and go through the author affiliations) and very few of those are in the game for 15 years. For example, lots of the authors in the denominator in this list were graduate students on papers published between 2000 and 2011, very few of whom were full time scientists in 1996 and thus immediately eliminated from the publish-every-year numerator. This 15 year time period is long enough to cover plenty of productive, publish every year grad students who became publish every year post docs who became publish every year junior PIs between 1997 and 2011 and still don’t make it into the numerator which requires they published in 1996 before their scientific career began.

At the very least the denomenator could have been out of researchers that started publishing in 1996, what fraction continued to publish every year there-after. This of course would still be rather unuseful statistic, since the amount of hours required to lead a project (typically denoted by first authorship, also typically carried out by a graduate student or post-doc) is vastly in excess of the amount of hours required to be senior author. Not to diminish the critical role of senior authors in funding, often designing, and usually mentoring a project. I doubt most PIs with a lab of N people believe they contribute N times more to science than their average group member, though if all group members include the PI on their publications the PIs publication rate is N times higher.

This entry was posted in Blog. Bookmark the permalink.