To help researchers understand potential issues one can encounter when conducting propensity matching studies in complex settings, this paper describes methodological complications faced when studying schools using deeper learning practices to improve college and career readiness. The study uses data from high schools located in six districts across two states: California and New York. The study used a quasi-experimental matched design, in which each network school was matched to a non-network school within the same (or neighboring) school district. Results from the "Study of Deeper Learning" illustrated the tensions between the realities associated with conducting research based on matching methods and the simplicity of textbook study design and analysis. In this paper, different areas were examined in which the nature of the intervention being studied and the constraints and challenges of working with districts and schools affected design and analysis decisions and the potential threats to the validity of research findings. Tables are appended.