By Jonathan Brown, University of Minnesota
Statistics courses have been developed where simulation methods are a major part of course content, changing numerous aspects of the student experience, including how students solve statistical tasks. Eight students from an undergraduate introductory statistics course that taught only simulation-based inference engaged in guided task-based interviews to explore problem-solving cycles in students’ approaches to the tasks. Participants attempted to solve four tasks from multiple contexts using randomization tests and bootstrapping procedures. Interview transcripts were qualitatively coded for problem-solving phases that emerged. Results suggest that statistical problem-solving with simulations may require students to navigate a cycle of phases specific to simulation methods, which corroborates and expands upon other models of statistical problem-solving. A typical cycle arose composed of six phases: Plan, Model, Simulate, Evaluate, Conclusions, and Context. Examples are presented to highlight variations observed in participant success, the navigation of phases, and cycle complexity across participants and tasks.