Axis truncation in bar charts is widely criticized as misleading, often based on ratio judgments or subjective ratings (i.e., Likert scale comparisons). This perspective, however, overly relies on these tasks and lacks nuance. We conducted three experiments to examine the effects of truncation across seven bar chart tasks. Our results show that truncation increases error for ratio calculations but improves accuracy or speed for tasks such as filtering and value retrieval. We further find that the magnitude of these effects depends on the degree of truncation and that direct data labeling substantially mitigates the negative effects of truncation in our experimental setting. These findings add nuance to bar chart truncation and invite discussion around the inherent "deceptiveness" of design elements.
ACM CHI Conference on Human Factors in Computing Systems