Fostering data literacy by engaging in data cleaning

The increasing societal relevance of data-driven technologies highlights the importance of fostering data literacy in education. One important part is data cleaning, which plays a crucial role in data- driven technologies and offers authentic opportunities to foster data literacy through critical engagement with real-world data. Despite its mathematical richness, data cleaning – particularly outlier detection – remains underrepresented in school curricula and educational research. This paper presents a design-based research project focusing on the mathematical foundations of outlier detection methods. Using the four-level approach by Hußmann and Prediger (2016), we specify and structure the mathematical topic of boxplots for outlier detection. We explore how these concepts can be meaningfully embedded in intended learning trajectories to promote students’ understanding of variability, robustness, and the impact of assumptions. The material is based on real datasets and aims to support critical reflection on data-driven decision-making.