Patti Frazer Lock & Robin Lock (St. Lawrence University); Kari Lock Morgan (Pennsylvania State University)
Abstract
Numerous sessions at previous USCOTS have explored using simulation-based inference methods for introducing students to the key ideas of statistical inference – usually with randomization tests and bootstrap confidence intervals for one or two sample problems involving means or proportions. In this breakout we look at ways to extend these simple ideas to more advanced situations, such as testing relationships in two-way tables, means for more than two groups, or regression models. The common structure of randomization tests makes these procedures more accessible to introductory students and provides an intuitive setting for bridging to chi-square and F-distributions.