Collaborative problem solving (CPS) is considered a necessary skill for students and workers in the 21st century as the advent of technology requires more and more people to frequently work in teams. In the current study, we employed theoretically-grounded data mining techniques to identify four profiles of collaborative problem solvers interacting with an online electronics task. The profiles were created based on 11 theoretically-grounded CPS skills defined a priori. The resulting four profiles correlated in expected directions with in-task performance and had interesting relationships with external measures associated with prior knowledge and CPS skills. These results inform and partially replicate findings from our previous research using a similar approach on a smaller dataset. Implications and comparisons between the two studies will be discussed. [For the full proceedings, see ED607784.]