By Camille Fairbourn (Michigan State University)
Information
My poster will display the relationship between homework achievement and exam performance in a large introductory statistics course in the semesters immediately before and since the launch of Chat GPT. (Spoiler: It's not good.) Michigan State University is a large, land-grant research university. Our introductory course is taught to first- and second-year students from a variety of majors in a large lecture + small recitation format. Homework is completed online using problems we have written and coded using WeBWorK and exams are delivered on paper and in person. Formative assessment (homework) questions and summative assessment (exam) questions have been carefully written to align to learning objectives and course learning activities. Data in the poster shows the scatter plot of homework achievement (graded homework problems comprising 25% of the course grade) and final course percentage from anonymized student scores over several semesters, including before 2020, after 2020 but before generative AI tools were released, and each semester since then. The data show a progressive change in student scores and course outcomes, providing participants an opportunity to reflect on the use of online homework as a student learning tool in an era where having a computer do your classwork is easier than ever.