Collaborative problem solving (CPS) is a critical competency in a variety of contexts, including the workplace, school, and home. However, only recently have assessment and curriculum reformers begun to focus to a greater extent on the acquisition and development of CPS skill. One of the major challenges in psychometric modeling of CPS is collecting large-scale data on teams and processes. In this study, we explore the use of agent-based modeling (ABM) to model the CPS process, test the sensitivity of outcomes to different population characteristics, and generate simulated data that can provide a novel means by which to refine and develop psychometric models. Methods of adapting trait-based stochastic processes to a specific task are described, and preliminary results are presented.