By Catalina Medina (University of California, Irvine)
Information
There is a general consensus among the statistics education community that statistics should be taught with real data, and some educators go further by recommending the real data context should be interesting or relevant to students. There are various suggestions about how to choose relevant data, but defining relevancy is difficult and may vary by student. Some instructors try to engage students with a data context by allowing the student to pick the data context. We hypothesized that autonomy and multiple data contexts may lead students to be more engaged with the material, and we investigated this through a quasi-experiment. In one section of a course students were allowed to choose their weekly homework assignment from three similar instructor-provided options that varied by the data context, while the other course section did not receive a choice. This investigation was implemented at a large public institution in two sections, totaling 180 active students, of an undergraduate introductory probability and statistics course. We sought to learn if given the choice between similar assignments that differ by data context (1) do students have more positive perceptions of the value of statistics, (2) do students achieve higher grades relative to students not given choice, (3) what characteristics of the data contexts led students to choose one assignment over the other options? Question (1) was measured with a survey inspired from established surveys, (2) was assessed with a generalized estimating equation, and (3) was qualitatively assessed from identified common themes in student provided reasonings.