Data Reasoning With Self-Authored Visualization in the MVP (Mathematizing, Visualizing, and Power) Project

• In Event: Community, Design, and Data Reasoning in Youth Data Visualization Practices

Abstract

In mathematics, reasoning is to draw logical conclusions based on evidence and assumptions, which is a key skill for effective mathematics learning (Shaughnessy et al., 2009). Although reasoning has been studied in mathematics education, how students engage in data reasoning–drawing insights, making inferences, and explaining the rationale behind the data they have–is an emerging topic in data science education. In this “big data” era, students need to be more prepared to become data literate (National Academies of Sciences, 2023), how to develop the skill of data reasoning is worth exploring.
We hypothesize that authoring data visualizations enables students to engage in a variety of other data reasoning: make sense of quantities, relate variables, and justify inferences (Rubel et al., 2021). When students have the opportunity to create their own data visualizations, their self-authored data visualizations can facilitate better connections between their data reasoning and the topics they are familiar with. In this study, we explored students’ data reasoning skills after they created and exhibited their data visualizations.
To examine student’s data reasoning, we operationalize Rubin's (2020) aspects of working with data–context, variability, aggregate, inference, and visualization. Specifically, context refers to the use of relevant context to explain the data; variability means to know the frequency of different values in the dataset; aggregate depicts the understanding of the relationship between a single data point and the whole dataset; inference stands for those data-based predictions; visualization is the practice to turn numeric data into imagery representations. Based on this framework, we deductively coded post-program interviews with six participants about the data they collected and the data visualizations they authored. Each one-hour interview was with two participants who were paired in the data visualization program.
By analyzing our coding results, we had three findings: 1) As we saw that the codes of data variability, inference, and visualization co-occurred a couple of times, it was plausible that students relied on their data visualization to understand their dataset through creating the mapping between their data and the data representations (e.g., one student talked about his data visualization created by his painted soccer ball when asked to describe his dataset); 2) we realized that the code variability frequently appeared, so we assumed that students paid special attention to the variability in their dataset to draw insights from it (e.g., one student talked about the mainstream music by comparing the counts of music genres that people liked); 3) “informal statistical inference,” which was defined in Makar and Rubin (2018), seemed essential for data reasoning as this practice allowed students to connect the context of their dataset with reasonable predictions that were based on and even beyond the data they had. We argue that better understanding students’ data reasoning can help us develop more tailored curricula, emphasizing the linkages between the self-authored data visualizations and the reasoning of the behind dataset, to teach students essential data skills in the era of “big data.”

Author

• Yilang Zhao, University at Buffalo - SUNY