RESEARCH COLLABORATION WITHIN A DEPARTMENT OF PEDIATRICS: A SOCIAL NETWORK ANALYSIS OF COAUTHORSHIP PATTERNS

Abstract BACKGROUND Research is a collaborative undertaking. Through collaborators, researchers have access to expertise, experience, and resources which may result in increased research productivity. Social network analysis is a set of techniques that focuses primarily on the patterns and characteristics of relationships among individuals. OBJECTIVES To describe the social network structure —the extent and patterns of collaboration among members — of a department of paediatrics and identify prominent individuals and divisions. DESIGN/METHODS We conducted a social network analysis of coauthorship. We included faculty members in a single department of paediatrics with at least 1 publication. We excluded those with a clinical appointment. We used PubMed to identify publications and Web of Science to obtain the total citations for each publication. RESULTS We included 99 faculty who authored 3 939 publications. The median (Q1, Q3) number of publications per faculty member was 12 (5, 39), ranging from 1 to 478. 83 (80%) of the faculty have coauthored a publication with another faculty member; the median (Q1, Q3) number of collaborators per faculty member was 3 (2, 8) and ranged from 0 to 21. 450 (11%) of publications included more than one faculty member as a coauthor. In the network diagram, 80 (81%) of faculty members were connected by coauthorship to a single large cluster. Neither the number of publications (increase in odds 1.0, 95% CI 1.0–1.1; p = 0.16) or h-index (increase in odds 1.0, 95% CI 1.0-1.0; p = 0.74) was associated with increased odds of a faculty member collaborating with another faculty member. Factors associated with increased odds of any two faculty members collaborating were: being from the same division (increase in odds 5.0, 95% CI 3.9–6.3; p<0.001) and both coauthoring a publication with a common faculty member (increase in odds 4.8, 95% CI 3.8–6.2; p<0.001). Being of different genders or differences in number of publications or h-index was not associated with changes in the odds of collaboration. CONCLUSION Social network analysis of coauthorship can provide insight into the social structure and research collaboration of an academic department. This structure should be considered in efforts to improve collaboration and research productivity.