Start the y-axis at zero when values are read from the baseline

Start the y-axis at zero when readers judge values by the distance from the x-axis. For example, in a bar chart or other baseline-read view, extend the axis to zero instead of truncating it above the data range.

Readers form the gist of a chart before they read axis labels. When the baseline is moved upward, the visible distances between marks no longer match the numeric differences in the data.

Mechanism: A zero baseline keeps apparent size and apparent difference closer to the real values readers are trying to compare. This reduces exaggerated ratios that arise from truncated axes.

Evidence: The paper shows that non-zero y-axes can make small differences look much larger than they are, and it notes that people seldom read axis labels before drawing conclusions from what they see at a glance (Szafir, 2018).

• User Goal: Compare magnitudes or differences quickly.
• Task: Judge relative size from mark height or position measured from the x-axis.
• Data: Quantitative values where raw magnitude matters.
• Chart Setting: A chart uses an axis and readers are expected to interpret it at a glance.
• Audience: Readers who will infer differences visually before reading labels closely.
• Success Criterion: Apparent ratios and differences match the actual data more closely.

Break it when: The chart is a line graph and the analysis cares about variation rather than magnitude. Why: The source notes this as a debated case because a zero baseline can make small variations hard to notice when variation is the quantity of interest.

Sacrifice: You give up vertical space that could magnify small changes. Risk: Important small-scale variation can become hard to see. Mitigation: Compute and visualize change relative to a baseline, or visualize rate of growth, instead of truncating the raw axis.

Mistake: Truncating the y-axis and relying on labels to prevent misreading. Why it fails: Readers often see the apparent ratio first and never correct it with the labels.

Failure Sign: A modest numeric change looks like a dramatic visual jump. Quick Check: Compare the visible ratio of mark heights to the numeric ratio in the data. Stronger Test: Ask whether the chart suggests a much larger difference when viewed briefly than the numbers actually support.

• Extend the y-axis to zero.
• Replot the same values with the zero baseline and compare the apparent ratio.
• If the story is about change from a baseline, compute that change and plot it directly.
• If the story is about growth or decline, plot the growth rate instead of truncating the raw values.