Scientific synthesis is the process by which researchers bring together heterogeneous data and knowledge sets in ways that yield novel insights or explanations. Over the last 20 years, there has been a rapid emergence of new research facilities around the world dedicated to scientific synthesis. Little is known, however, about team and individual factors related to success in synthesis teams. Evaluators of synthetic research groups face a daunting challenge of measuring not only the productivity of group members, but also the antecedents to successful production. This paper will demonstrate the use of hierarchical modeling methods to evaluate the factors leading to productivity of 20 interdisciplinary synthetic research groups that took place over six years at The National Institute for Mathematical and Biological Synthesis (NIMBioS).
NIMBioS is funded through a National Science Foundation award and located on the University of Tennessee campus. The institute draws a diverse cadre of researchers from around the world to take part in interdisciplinary synthetic research groups to find creative solutions to pressing problems from animal disease to wildfire control. The subjects of this presentation are Working Groups, which are assemblages of 10-14 researchers who focus on major well-defined scientific questions at the interface between biology and mathematics. The groups typically meet 2-4 times over a 2-year period and vary in their composition, size, and activities.
A known observation of scientific productivity is that is tends to be unequally distributed, with relatively few researchers being responsible for the vast majority of publications. This is also the case for NIMBioS, where 50% of the 44 working groups have produced the entirety of the working group publications for the center. In addition, only 35% of the working group participants are coauthors on these papers. A need exists, therefore, to evaluate the group and individual level factors that predict scientific productivity within these research groups. This paper attempts to evaluate these factors using multilevel modeling methods where the traits of individual research group participants (e.g. gender, ethnicity, discipline area) are modeled within group-level factors (e.g. number of meetings, group size, group composition) as determinants of Working Group-related journal article production.
The results of this study will have implications for designing and evaluating synthetic interdisciplinary research teams, although the methods used could be replicated to evaluate many different types of team science groups.
Preliminary results of 22 working groups with 353 participants
- Each 5% increase in number of underrepresented minorities in a group resulted in a 20% overall greater incidence of co-authorship within the group
- Being a group organizer resulted in a 52% increase in incidence of co-authoring a paper in the group
- Attending more face-to-face meetings resulted in greater incidence of co-authoring a paper in the group
- Being a social scientist decreased incidence rate of co-authorship within the group by 46%
- The % of female organizers moderates the expected co-authorship count for female participants. For groups with three organizers, the inclusion of one female organizer increased the incidence rate of co-authorship for females in the group by 92%