Insert a gap between compared segments in a single stacked bar

Insert a visible gap between bar segments that readers must compare within one stack. For example, separate the target segments with a gap or other spacing cue instead of leaving them immediately adjacent when viewers must compare one segment’s height to another.

Touching segments inside one stack invite a part-to-whole reading that pulls estimates downward. Separating the target segments weakens that whole-bar cue and reduces the bias.

Mechanism: A gap shifts the read from “part of one whole” toward “two lengths to compare,” which lowers the negative bias seen with adjacent segments.

Evidence: In divided-bar comparisons, immediately adjacent target segments produced high error and strong negative bias, while separating the compared segments reduced error and attenuated the bias; the review includes this paper as evidence on bar-chart variant performance (Talbot et al., 2014; Zeng & Battle, 2023).

• User Goal: Compare two parts inside the same whole exactly.
• Task: Judge one stacked segment relative to another segment in the same bar.
• Data: Quantitative parts shown inside a single stacked or divided bar.
• Chart Setting: The comparison happens within one stack, and the reader could confuse part-to-part with part-to-whole.
• Audience: Viewers making quick visual estimates.
• Success Criterion: Lower bias and lower absolute error for the target comparison.

Break it when: The compared values are separate bars in a simple bar chart. Why: In simple bar charts, adding separation makes comparison harder rather than easier.

Sacrifice: Some of the immediate continuous-whole appearance of the stack.
Risk: If the whole composition is the only message, extra spacing can weaken that part-to-whole cue.
Mitigation: Use the gap when part-to-part comparison is the priority.

Mistake: Leaving the compared segments touching because they belong to the same whole. Why it fails: Adjacency encourages a part-to-whole interpretation and biases the comparison.

Failure Sign: Readers systematically underestimate the smaller segment when the two target segments touch.
Quick Check: Compare an adjacent version with a gapped version and inspect whether the target ratio looks less like a part-of-total judgment.
Stronger Test: Run an A/B estimate task on adjacent versus separated segments and compare bias as well as absolute error.

• Add a small gap between the compared segments inside the stack.
• If a gap is not possible, avoid making the touching within-stack segment comparison the primary readout.
• When layout can change, separate the target segments instead of keeping them immediately adjacent.