With Upneet Cheema and Beth Chance (Cal Poly, San Luis Obispo)
The primary focus for this project is to explore the impacts of different ways to present confidence intervals to introductory students. Students have been engrained to plug numbers into a formula and stopping there. Statistics is usually their first exposure to interpreting their results and outcomes. Our main emphasis is to see whether one approach or sequence could instill a more conceptual understanding of confidence intervals for the introductory student.
For our study, we recruited the instructor of two sections of the same class. At the beginning of the two-day experiment, we first gave a demographic assessment that asked students about their background in statistics. Both sections introduced the conceptual of confidence intervals in different ways: one class began with learning about confidence intervals with the bootstrapping method; the other class used a more common z-interval introduction. At the end of the class session, both classes were presented with a post-test (five multiple-choice items from CAOS and other sources). On the second day, both sections worked through a similar activity using the other approach, but in a different context. At the end the second day, both classes were presented with a second post-test.
Our experiment consisted of introducing multiple new and old methods of confidence intervals for a single proportion such as bootstrapping and the usual one-sample z method. In analyzing the data, we will look for differences in initial understanding after day 1, growth in understanding after day 2, and whether the day 2 performances differed between the two sections. We will also compare midterm performance after additional instruction.