Student-student interactions are foundational for active learning, but are seldom studied at classroom-wide scales. I present early results from a social network analysis of detailed student interaction data throughout a term of introductory physics. Network data are gathered via regular surveys and are differentiated by the content and setting of the interaction. The resulting collection of weekly network layers shows a complex set of evolving and interlocking discussion clusters among the 187 students in the course. The analysis casts this data as a multiplex network object, a framework where distinct layers of links are preserved among a common set of student nodes. It then becomes possible to study this dense object along temporal scales (time-aggregated vs. evolving), for content focus (e.g. problem solving, concept discussion), or for the context of student collaborations (in-class vs. outside, teacher assigned vs. self-selected). I focus on the content-based network layers, what they show about student groupings during the semester, and how different measures of network position correlate with students' course outcomes. This analysis framework opens possibilities for studying a class-wide view of how different learning environments and curriculum changes play out among student interactions.
