STUDENTS' UNDERSTANDING OF RANDOMIZATION-BASED INFERENCE

INTRODUCTION Before modern computing power allowed for rapid simulations, introductory statistics courses necessarily relied on methods like z tests and t tests to introduce the core logic of inference; today, a growing number of statistics educators (e.g., Cobb, 2007) are proposing that these traditional methods be replaced or supplemented with randomization-based tests. Because randomization tests more directly model the randomness inherent in the study design, some believe that these tests can help students develop a deeper conceptual understanding of statistical significance and p-values. In this study, we explored whether conceptual understanding of p-values could be improved by exposure to randomization-based inference, even in students familiar with more formal methods.