Seeing How the Sausage is Made: Data Storytelling as Means and Method in a Computer Science Writing Course

As data corpus-driven tools and technologies increasingly push users to passively search for an answer, rather than search to understand, we believe that technical and computing disciplinary writing courses have a duty to teach the process of responsible data storytelling. While students can grasp that generative AI makes mistakes, hallucinates, and perpetuates bias, they can need help understanding the antecedent causes of those difficulties. All algorithmically driven decision-making or recommending software have in common a large data set that has been labeled, either by users or by the system itself. The origins of that data and the reasonable applications/deductions and conclusions possible for any given dataset have everything to do with why some tools help and some tools perpetuate harms. By starting at the very beginning and asking students to make sense of data, students can more easily see how purpose and audience impact analysis of any given collection of data. Once those opportunities for rhetorical choice making are known, students become ready to understand the connection between data and complex A.I. systems and some of the ways that bias and other kinds of harm can result if designers are not careful. Combining instruction in a technical coding environment with basic data literacy lessons such as ‘the seven data stories,’ [14] we developed and delivered a three-week writing unit designed around responsible data exploration and storytelling. In this experience report, we provide the assignment we used, and the scaffolded activities we employed to bring students through the process, remarking on what worked well and what we want to improve. We provide attendees with a link to an R-based notebook with a walk-through lesson on data exploration commands, and the rubric used to assess students’ texts, notebooks with code and commentary and results, all existing in a referential context. We provide the survey results of students’ perception of learning from this activity. Early findings demonstrate that students internalized lessons about the non-objective nature of data analysis and of specific responsible data storytelling practices required by anyone seeking to ethically represent answers within and limitations of any dataset.