In a study of group work and physics, multiple theoretical lenses were used as analytical strategy to explore university students’ learning possibilities and patterns of interaction when talking physics in small groups together. The main empirical data was video recordings of students discussing and solving problems in Newtonian mechanics. Due to the complexity and wealth of details in the video data the analytical process was developed into three different levels for investigating what happened in the students’ interaction: phenomenography and variation theory, positioning theory, and finally, conversation analysis. Every analytical perspective was chosen for both its descriptive and explanatory capacities in relation to the students’ learning possibilities and patterns of interaction. We propose that these three analytical perspectives are complementary rather than commensurable, because they are not referring to the same data, even though the data they reference is derived from the same social setting, situation and event. Consequently, ontological mismatch is not crucial since the different perspectives do not address the same issues - they do not even exist in the same theoretical space, and relationships between theoretical spaces may instead be explored post-analysis in a research synthesis. This paper addresses both benefits and drawbacks with this analytical strategy.