Bar charts with y-axes that don't begin at zero can visually exaggerate effect sizes, and in turn lead to unjustified or erroneous judgments. However, advice for whether or not to truncate the y-axis can be equivocal for other visualization types, and there is little existing empirical work on how axis truncation impacts judgments. In this paper we present examples of visualizations where this y-axis truncation can be beneficial as well as harmful, depending on the communicative and analytical intent. We also present the results of a series of crowd-sourced experiments in which we examine how y-axis truncation impacts subjective effect size across visualization types, and we explore alternative designs that more directly alert viewers to this truncation. We find that the subjective impact of axis truncation is persistent across visualizations designs, even for viewers that are aware of the presence of truncated axes. We therefore advise against ironclad rules about when y-axes are appropriate, but ask designers to consider the scale of the meaningful effect sizes and variation they intend to communicate, regardless of the visual encoding.
View on arXiv