Date Published: May 2, 2019
Publisher: Public Library of Science
Author(s): Katrin Hussinger, Maikel Pellens, Giuseppe Sartori.
Increasing complexity and multidisciplinarity make collaboration essential for modern science. This, however, raises the question of how to assign accountability for scientific misconduct among larger teams of authors. Biomedical societies and science associations have put forward various sets of guidelines. Some state that all authors are jointly accountable for the integrity of the work. Others stipulate that authors are only accountable for their own contribution. Alternatively, there are guarantor type models that assign accountability to a single author. We contribute to this debate by analyzing the outcomes of 80 scientific misconduct investigations of biomedical scholars conducted by the U.S. Office of Research Integrity (ORI). We show that the position of authors on the byline of 184 publications involved in misconduct cases correlates with responsibility for the misconduct. Based on a series of binary regression models, we show that first authors are 38% more likely to be responsible for scientific misconduct than authors listed in the middle of the byline (p<0.01). Corresponding authors are 14% more likely (p<0.05). These findings suggest that a guarantor-like model where first authors are ex-ante accountable for misconduct is highly likely to not miss catching the author responsible, while not afflicting too many bystanders.
Over the course of the 20th century, science evolved from a norm of single authorship to collaboration. The average number of co-authors per publication grew from almost zero to between 2 and 7 at the end of the century, depending on the discipline . 99.4% of science and engineering subfields exhibited an increase in average team sizes between 1955 and 2000, while team-based publications receive more citations than single-authored publications in 97.7% of subfields . Growing complexity and multidisciplinarity as well as harsh performance evaluation policies render teamwork indispensable [1–4]. This raises normative questions about the distribution of credit and accountability among team members [5,6]. As more collaborators take part in a project, it becomes more difficult to negotiate who should and should not receive credit, and who should be held accountable for individual contributions and the integrity of the final work.
The analysis shows that scientific misconduct is not equally likely to arise from all members of an author team. First authors, corresponding authors, and to some extent senior authors are much more likely than middle authors to be found responsible for scientific misconduct in Findings of Research Misconduct by the U.S. Office of Research Integrity. Authorship policies that impose accountability for the integrity of all research findings on the shoulders of all coauthors miss this important fact. Our findings make intuitive sense. Those who have most to gain from the work in terms of scientific credit arguably have the highest incentives to commit scientific misconduct. While middle authors will do equal harm to their careers if found guilty of scientific misconduct, they reap much less of the reward since they are not lead author of the projects.