Fading the Training Wheels: Chart Taxonomies as Instructional Scaffolding in Data Visualization Education

Authors

Hannah Yanhua Zong (Purdue University ), Nabin Khanal (Purdue University), Yingjie Victor Chen (Purdue University)

Abstract

Chart taxonomies are widely used in introductory data visualization education to support chart selection. Their pedagogical value, however, remains debated: while they may reduce cognitive load, they have also been criticized for encouraging rigid, template-driven reasoning. This study reframes that debate through instructional scaffolding and examines whether chart taxonomies can function as temporary supports that improve performance without creating strong scaffold dependence. 29 students in an introductory data visualization course completed a baseline assessment, taxonomy-supported practice, and a delayed post-test without taxonomy access. Chart-selection accuracy increased from 67.2% at baseline to 78.3% at post-test, with a significant phase effect, F(2, 56) = 7.10, p = .002. Decision time decreased, while confidence and perceived workload remained relatively stable. Analysis of written justifications from 22 students showed a modest, nonsignificant increase in reasoning quality. Exploratory analysis of complex and creativity-oriented post-test tasks showed that students frequently selected multiple charts, combined chart types across taxonomy categories, and produced largely defensible solutions with substantive reasoning. Although these tasks had no pre-test counterparts, they provide descriptive evidence of flexible taxonomy use in more complex contexts. Overall, the findings are consistent with chart taxonomies functioning as fading scaffolds rather than rigid templates.