By Tiffany Reyes-Denis (University of Illinois Urbana-Champaign) and V.N. Vimal Rao (University of Illinois Urbana-Champaign) and Amos Jeng (University of Michigan)
Information
The rapid integration of AI in educational contexts has strongly influenced how teachers teach and how students learn, yet little is known about how these systems shape students’ statistical thinking and learning, especially in statistics, where students increasingly rely more on software to solve statistical problems. This study rigorously examines how constrained-AI environments embedded in a pedagogical activity influences students’ statistical thinking and reasoning. Twelve students (10 female) from a large Midwestern university and with varying levels of prior knowledge in statistics (i.e., low, medium, and high prior knowledge) participated in task-based interviews to solve a statistical task using R, either with (n=6) or without AI (n=6). Specifically, students were tasked to provide a recommendation based on given data, first by following step-by-step guidelines to create a graph in R, and then independently. Audio and video recordings of the interview, students’ sketches, and researcher memos were analyzed in a inductive thematic analysis including the following phases: 1) preliminary coding (to familiarize with data), 2) descriptive and process coding across data sources (Saldaña, 2016), 3) resulting codes were developed into categories for within-participant analysis, 4) the resulting categories were developed into themes for within-group analysis, and 5) the themes were used for across-group analysis. We additionally employed peer debriefing (Lincoln & Guba,1985) throughout all phases. Students were recruited from a large R1 university. Students had completed a general education introductory-level statistics course prior to participating in this study. The course is a large-enrollment course with approximately 600 students per section, taught in an auditorium or theater. The course employs a quasi-flipped curriculum with interactivity in lectures focused on group problem solving of check-in questions and worked-examples. The activity and task developed for this study were designed to align with this curriculum, but is intended for use in any educational setting, in-person or remote, guided or unguided.
Link: https://mediaspace.illinois.edu/media/t/1_n5sphtlm