By Madeline Hunt (University of Illinois)
Information
Applied statistics courses often acquaint students with a common repertoire of standard data visualizations, such as histograms, barplots, boxplots, and scatterplots. But do students leave these courses with the understanding needed to interpret the complex, unconventional data visualizations that they encounter in the media? Furthermore, does their own prior knowledge and biases affect how they interpret data visualizations involving critical, contemporary issues? We are currently developing an assessment of students’ data visualization literacy spanning a variety of representations and contemporary contexts. After identifying 10 visualizations that fit our criteria, we completed two rounds of think-aloud interviews with students from a large introductory statistics course at a public university. Using our findings from 16 student interviews, we explored common ways of reasoning that students expressed and the shared challenges that many encountered. One theme from the data is students’ tendency to believe that a graph supports a claim when the data needed to gauge that claim are not actually contained in the graph. We also found that students commonly struggled with rates, percentages, and identifying the unit of observation in many visualizations. These results highlight aspects of visual data literacy that we believe are needed in the modern statistics and data science curriculum. Our findings provide a roadmap for how statistics instructors can use authentic data visualizations to help prepare their students for lives as data literate citizens.