Philosophers of science and social scientists have argued that diverse perspectives, methods, and background assumptions are critical to the progress of science. One way to achieve such diversity is to ensure that a scientific community is made up of individuals from diverse personal backgrounds. In many scientific disciplines, though, minority groups are underrepresented. In some cases minority members further segregate into sub-fields, thus decreasing the effective diversity of research collaborations. In this paper, we employ agent-based, game theoretic models to investigate various types of initiatives aimed at improving the diversity of collaborative groups. This formal framework provides a platform to discuss the potential efficacy of these various proposals. As we point out, though, such proposals may have unintended negative consequences.