Avoid truncating the value axis in bar charts

Keep the bar-chart value axis untruncated so bar length reflects the actual magnitude difference. For example, show both bars on the full value scale rather than raising the axis minimum to make a small difference look much larger.

Bar length is read against the displayed axis range. When the axis range is shortened, the same data produces a much stronger visual gap, so readers overstate how much larger one value is than another.

Mechanism: Truncating the value axis exaggerates the perceived difference while leaving the underlying numbers unchanged, so the visual message becomes stronger than the data.

Evidence: In the bar-chart experiment, truncated-axis charts produced much higher “how much bigger” judgments than full-axis charts (mean 2.77 vs. 1.45 on a 5-point scale; p = 0.0003), even though the chart still showed the actual numbers (Pandey et al., 2015).

• User Goal: Judge how much one category exceeds another.
• Task: Compare magnitude, not just direction.
• Data: A small set of quantitative category values.
• Chart Setting: A bar chart with a visible value axis and a message-level comparison.
• Audience: Readers who will interpret the chart visually rather than recalculate from the numbers.
• Success Criterion: The perceived size of the difference matches the data instead of an exaggerated visual gap.

Break it when: The graphic is not an axis-based bar chart. Why: This guideline targets the exaggeration created by changing the axis range that bar length is read against, and the paper tested that mechanism with bar charts.

Sacrifice: You give up some visual drama. Risk: A small real difference may look modest. Mitigation: Keep the full axis and let the chart communicate the actual size of the difference rather than amplifying it with the scale.

Mistake: Show the exact values on the bars but still truncate the value axis. Why it fails: The study presented accurate numbers on the chart and readers still gave exaggerated magnitude judgments.

Failure Sign: The bars look far more different than the numeric values suggest. Quick Check: Inspect the displayed minimum on the value axis; if the scale starts above the natural baseline used in the control-style version, the chart uses the distortion tested here. Stronger Test: Compare the current chart to a full-axis version and ask a reviewer the same “how much bigger” question; if the truncated version yields a stronger answer, keep the full-axis version.

• Reset the bar-chart value axis to the full baseline instead of an enlarged segment.
• Redraw the same bars against the unaltered axis range.
• Remove the truncated version rather than trying to offset it with printed numbers alone.