|
Abstract:
|
Hypothesis testing provides a substantial framework for the statistical evaluation of evidence. This classical framework has not changed much over time even though our approach to teaching statistics continues to evolve. The classical framework require students to use a considerable amount of technical language and students often must accept at face value many of the subtle details of classical inference, e.g. cannot accept Ho. Some believe the recent advances in the use of simulation / randomization-based methods have alleviated most of these issues. However, the use of such methods requires the understanding of not one, but two frameworks for doing inference. Furthermore, important concepts in one framework are unimportant in the other, e.g. standard error is necessary for classical inference, but not needed in randomization tests. In this talk, I will discuss approaches we use in our teaching to streamline the simulation-based and classical frameworks of inference.
|
