VFF: Deceptive Graphs: Quantifying the Truncation Effect in Bar Graphs & Exploring Interventions
We approach the question of potentially deceptive visualizations using the tools of experimental psychology. In a series of experiments, we quantify the effects of truncating the y-axis of bar graphs, which remains a common practice, despite being frowned upon by data visualization experts. We focus on relative judgments of quantity ("How different are A and B?"), rather than objective measures ("What is the value of X?"). We'll discuss our experimental design, results to date (spoiler: truncation matters!), and variations on the study, including studies recruiting PhD students in the quantitative fields and the humanities. I'll discuss our working definition of a "misleading graph," and how this might work might fit into existing work on misinformation. I'd love to solicit discussion and feedback on this paradigm from a non-psychological perspective.