Evaluating Visualization Techniques for Categorical Data: Addressing Line Width Illusions in Parallel Sets and Hammock Plots

This paper investigates the efficacy of different visualization techniques for categorical data, specifically parallel sets and hammock plots, in the context of line width illusions. It presents evidence that both parallel sets and hammock plots are susceptible to perceptual distortions, which we term the “line width illusion” and its reverse. To address these issues, we introduce the common angle plot, an alternative visualization method that avoids these illusions and offers more accurate data representation. Our study, involving user surveys, demonstrates a significant preference for common angle plots over traditional methods, highlighting their effectiveness in providing clearer visual comparisons. The paper concludes that while straight-line approaches in visualization can be mathematically sound, they often fail to align with intuitive human perception, thus undermining the accuracy of data interpretation. The common angle plot emerges as a robust solution to these challenges, offering a more reliable alternative for categorical data visualization.