By Meg Ellingwood
Information
Writing is an important part of statistics education, both as it helps students consolidate their statistical knowledge and as a skill in communicating results. However, while written reports are often used as assessment tools, how students actually write in statistics courses has been underexplored. Recent work has shown that asking students to think about the structure of their writing improves outcomes, so we further explore differences in writing structure over the course of statistics education. First, we leverage a large dataset of student reports collected from different levels of statistics and data science courses to build a classifier that labels sentences by rhetorical function (e.g., method definition, interpretation of results, etc.). This enables us to study the rhetorical structure of student writing at scale. Then, we explore ways to compare writing at the introductory level with writing from upper-level courses based on how they present information and construct statistical arguments. We investigate the development of statistical writing as students gain familiarity with the genre and learn to convey information in ways that match reader expectations. Understanding this development process could help instructors give constructive feedback to promote growth in writing skills, in addition to growth in statistical thinking.