Comparing the Efficacy of Static and Dynamic Graph Types in Communicating Complex Statistical Relationships

Graphs are useful tools to communicate meaningful patterns in data, but their efficacy varies considerably based on the figure’s construction and presentation medium. Specifically, a digital format figure can be dynamic, allowing the reader to manipulate it and little is known about the efficacy of dynamic figures. This present study compared how effectively static and dynamic graphical formats convey relationship information, and in particular variable interactions. Undergraduates (N = 128, 56% female, M(age) = 18.9) were given a brief tutorial on main effects and interactions in data and then answered 48 multiple-choice questions about specific graphs. Each question involved one of four figure types and one of four relationship types (main effect only, interaction only, main effect and interaction, or no relationship), with relationship types and graphical formats fully crossed. Multilevel logistic regression analysis revealed that participants were fairly accurate at detecting main effects and null relationships but struggled with interaction effects. Additionally, the static 3D graph lowered performance for detecting main effects, although this negative effect disappeared when participants were allowed to rotate the 3D graph. These results suggest that dynamic figures in digital publications are a potential tool to effectively communicate data, but they are not a panacea. Undergraduates continued to struggle with more complicated relationships (e.g., interactions) regardless of graph type. Future studies will need to examine more experienced populations and additional dynamic graph formats, especially ones tailored for demonstrating interactions (e.g., profiler plots).