Abstract

Scientific research has evolved into a highly organized collaborative process, making Scientific Research Collaboration Networks (SRCNs) an important perspective for understanding collaboration patterns and their formation mechanisms. Existing studies mainly focus on individual dimensions of proximity, with limited attention to the combined effects of multiple proximities, and their findings are largely derived from the natural sciences. Using co-authorship data from the Chinese public management discipline indexed in the Web of Science database, this study employs the Exponential Random Graph Model (ERGM) to examine the effects of geographical, cognitive, institutional, and social proximities on the formation and evolution of research collaboration networks, as well as their interaction effects. The results show that cognitive proximity consistently serves as the primary driver of collaboration formation. The effect of institutional proximity strengthens over time and becomes the dominant factor in the mature stage of network development. Geographical proximity mainly facilitates collaboration in the early stage, while social proximity has a relatively limited influence throughout the evolution process. Interaction effects among different proximity dimensions are significant only during the rapid expansion stage and remain generally modest. These findings deepen the understanding of the dynamic mechanisms underlying research collaboration network evolution and provide theoretical implications for optimizing research collaboration and research management policies.