The structure of collaborative relationships impacts research productivity.  Studies of co-authorship have shown varied outcomes and quality based on aspects including size and composition of research team size, location of research team members, institutional types, among other factors. In the United States, research capacity and productivity also vary greatly across the vast set of academic research institutions.  Because the role of universities in the production of science in the United States is well recognized as significant to the overall research capacity of the nation (Geiger, 2004; Jaffe, 1989; Hall et. al, 2003), considerable federal investments have addressed capacity discrepancies with the goal of enhancing the ability of less productive institutions (and ultimately regions) to contribute to the research capacity of the nation. The Experimental Program to Stimulate Competitive Research (EPSCoR) was created in 1980 to address these discrepancies and build capacity in less resourced states. Since its inception, the EPSCoR program has grown from a set of five states, to the current set of twenty-seven states.  The program’s goals have been expanded to include “systemic change” in state science and technology environments, contribution to state technology-based economic development, and enhancement of human resources in R&D. The largest of the current set of NSF-EPSCoR awards involve multi-disciplinary, multi-institutional projects over five years, and are typically funded at $20 million for that period. The expectation is that the research teams develop in such a way as to not only further the science specific to the project, but importantly to build capacity of faculty and institutions across the state. Ultimately, these projects build networks of academic researchers, students, and external stakeholder that engage in various dimensions of the research.

We ask, how do the network ties of academic scientists funded by EPSCoR matter for publication production?  While NSF EPSCoR has made significant investments in state science, there is remarkably little literature addressing its outcomes or processes. Most recently, retrospective evaluations have been conducted by the National Academies of Science, and the Science and Technology Policy Institute to address programmatic outcomes overall. Other works have addressed the larger policy issues surrounding EPSCoR (Lambright, 2000; Feller 2000), and state-wide capacity issues specific to EPSCoR (Melkers and Wu, 2009).  Yet, the structure and goals of EPSCoR present a rich environment for studying programmatic effects and enables more generalizable research on the evolution of research collaborative networks and their impacts on productivity. This paper examines the structure and evolution of EPSCoR research collaborations, and their resulting outcomes, in a recent set of funded EPSCoR projects.  Data include survey-based social network data that capture the structure and resources of the collaborative ties of scientists in these projects, and bibliometric data for project-base publications. Analysis examines how measures of network structure (eg. centrality, density, constraint) and relational characteristics (eg. disciplinary homophily, strength of ties) predict important near term outcomes (eg. grants submissions, grant awards, publications and citations).