Accepted at Graphics Interface 2026

Abstract: Although Virtual Reality (VR) is increasingly used for information work, there is a limited understanding of how data should be encoded in immersive visualizations to support accurate interpretation. It remains unclear whether established two-dimensional graphical perception guidelines generalize to VR settings with different viewing conditions. To address this gap, we conducted two immersive experiments in which participants performed ratio-estimation tasks, with mean absolute error as the metric. In Experiment 1, we replicated a foundational graphical perception study on a virtual tabletop display and found that accuracy followed the established ranking of visual variables, with the lowest mean absolute error for bar charts, followed by length and area encodings. In Experiment 2, we examined the effects of chart orientation and viewing distance on a virtual wall display, extending an established large-display paradigm to VR, and found that horizontal bar charts produced higher error and greater between-participant variability than vertical bar charts, with accuracy degrading more sharply with viewing distance for horizontal than vertical orientations. These findings indicate that while classic perceptual rankings largely carry over to VR, viewing conditions interact with chart design in ways that affect accuracy, informing recommendations for optimal encoding of 2D visualizations for immersive environments.