We redesigned the upper-level chemical engineering laboratory sequence at Stanford University to shift the focus from learning various analytic techniques, to having students learn to make the same decisions that an expert experimentalist makes in the laboratory. Each course in the two-course sequence had different levels of structure and available agency to scaffold the decision-making process for students in a way that would help them practice making the relevant decisions. In the first lab course, students were given targeted feedback as they made a limited set of experimentation decisions, such as deciding how much data to collect or how to control variables, for a predefined experimental question. In the second lab course, students were given the opportunity to make almost all of the same decisions an expert experimentalist would conducting their research, including choosing a research question. We found that students tended to make more decisions per week in the second lab course than in the first, partly because some decisions, such as whether an experiment was feasible, were made for the students in the first lab. In the projects in the second lab course, we found substantial variations in the temporal patterns of student decision-making between groups, but the variation in decision-making did not seem to be correlated with the type of research question students chose to explore. Student testimonials suggest that the first lab course laid the groundwork for students to be successful in the more open-ended projects in the second course. These results motivate future research that systematically varies the amount of agency afforded to students to understand the role of other variables, such as relevant course preparation or prior research experiences, in learning to make expert-level decisions during experimentation. © 2021 Institution of Chemical Engineers