Scientific expertise is manifested through extensive cycles of making and acting on decisions. To learn the processes and practices of science, therefore, students must have practice with scientific decision making. We argue that this can only happen if students are afforded agency: the opportunity to make decisions to pursue a goal. In this study, we compared two different introductory physics labs through the lens of structuring and supporting student agency. We explore both the role of decision making agency (students are afforded opportunities to make decisions about their lab investigations) and epistemic agency (students are afforded opportunities to generate knowledge). We found that the two labs differed in both the amount and type of structure provided to students: one lab (the intervention condition), with more overall structure, cued student decision making; the other lab (control condition), with less overall structure, made decisions for students. Students in the intervention condition documented, on average, ten times more decisions during their investigations, including more unprompted decisions. In contrast, in labs with less available epistemic agency (i.e., labs where they had to verify or demonstrate canonical models or constants), students in the intervention condition documented fewer decisions than in labs with more available epistemic agency. We associate the improved decision making with students taking up personal agency, based on significant shifts in students' tone as they described their experimental methods (from passive, third-person narratives to active, first-person narratives). Students in the intervention condition also made higher-quality decisions in line with scientific practice. The study speaks to the important nuances between structure and agency in instructional labs and tests the generalizability of previous research on lab instruction to a two-year college population.