This paper studies productivity and preferences in scientific research. Collaboration is increasingly important for innovation in science, and other domains, but we have limited understanding of the factors researchers use to choose their collaborators and the projects they work on. Here, we use a model of strategic network formation and a recently developed econometric method to examine this question in the context of economics researchers. We learn that research teams with more collaborators tend to produce papers with higher impact, without increasing individual costs of communication and coordination. This suggests the trend toward larger research teams in economics will continue.