Keep bar aspect ratios near square for memory-based value reading

Keep bar aspect ratios near square when readers must remember bar-top positions after the chart disappears. For example, in a bar chart avoid very wide bars that are later recalled too high and very tall bars that are later recalled too low; square bars showed no systematic bias.

Bar-top position was not remembered independently of the bar’s shape. Memory for the mark was pulled toward a prototypical square, so wide bars were recalled as higher and tall bars were recalled as lower.

Mechanism: A bar’s incidental width:height ratio changes the remembered top position even when position is the intended encoding. Keeping the mark closer to square reduces this shape-driven pull and preserves more accurate value recall.

Evidence: Across three experiments, wide bars were overestimated, tall bars were underestimated, square bars were not systematically biased, and the same aspect-ratio pattern remained when area was controlled or varied, showing that aspect ratio rather than area drove the bias (Ceja et al., 2021).

Notes: Tall bars also produced larger absolute error and variability than square or wide bars.

• User Goal: Remember or report an exact bar value after a short delay.
• Task: Reproduce or compare bar-top positions from a previous view.
• Data: Quantitative values encoded by vertical position in bars.
• Chart Setting: The original bar will not stay visible during the judgment, and the layout can make bars very wide or very tall.
• Audience: Viewers making memory-based readings.
• Success Criterion: Minimize systematic overestimation of wide bars and underestimation of tall bars.

Break it when: The original bar remains simultaneously visible during the comparison or tracing response. Why: The tested wide-versus-tall memory-bias pattern did not persist under direct concurrent perception.

Sacrifice: This rule limits freedom to stretch bars into very wide or very tall shapes as a layout choice.
Risk: Applying the same correction outside bar charts assumes the same effect without direct test.
Mitigation: Use this rule directly for bar charts and treat other mark types as separate review cases.

Mistake: Changing bar area while leaving the bars distinctly wide or tall. Why it fails: Area did not account for the bias pattern; wide bars still skewed high and tall bars still skewed low.

Failure Sign: Bars are visibly elongated and the reader must recall them from a prior view.
Quick Check: Inspect the width:height ratios of the bars; if they are clearly wide or clearly tall, expect directional recall bias.
Stronger Test: Hide the chart briefly and ask reviewers to reproduce a bar’s top position; systematic positive error on wide bars or negative error on tall bars indicates the problem.

• Reduce extreme width:height ratios so the bars are closer to square.
• Resize the chart region or reallocate panel space when the current layout creates very wide or very tall bars.
• Avoid using elongated bars for tasks that require exact value recall after the chart disappears.
• If exact comparison must happen across views, keep a visible reference bar onscreen instead of relying only on memory.