Use vertically stacked small multiples instead of superposition for mean and range comparison

Choose a vertically stacked small-multiple layout when readers must decide which bar set has the larger mean or wider range. For example, place the two bar charts one above the other and avoid a single superposed view that overlays both series in the same space.

Set-level comparisons depend on seeing each bar set as its own unit. A stacked layout keeps the two sets separate while preserving direct vertical alignment, whereas superposition forces readers to separate overlapping groups before comparing them.

Mechanism: Vertical separation supports whole-set comparison of the bar sets, and the stacked arrangement lets readers compare corresponding bar lengths by slicing downward between the two charts.

Evidence: In two crowdsourced staircase experiments, vertically stacked bar charts produced the most precise judgments for largest-mean and widest-range comparisons, while superposed charts produced the least precise judgments for both tasks (Jardine et al., 2020).

Notes: The paper also reports that mean judgments were consistent with global proxies such as mean length and bar centroids, while range judgments were consistent with focal within-set delta proxies.

• User Goal: Decide which of two bar sets has the larger mean or the wider range.
• Task: Set-to-set comparison rather than item-to-item change detection.
• Data: Two quantitative series shown as multiple bars per set.
• Chart Setting: Paired horizontal bar charts where the layout choice is between separated views and an overlaid view.
• Audience: Readers making a brief visual judgment.
• Success Criterion: Small differences in mean or range remain discriminable.

Break it when: The reader must find which individual item changed the most between the two series. Why: The paper reports that item-to-item biggest-delta judgments were better supported by animated or superposed arrangements than by separated stacked views.

Sacrifice: You give up the compactness of a single overlaid chart. Risk: If applied to an item-to-item change task, the separated layout can weaken the direct local cues that highlight the largest change. Mitigation: Reserve the stacked layout for whole-set mean or range comparison.

Mistake: Overlay the two bar sets in one shared plotting space to save space. Why it fails: Superposition was the least precise arrangement for deciding which set had the larger mean or the wider range.

Failure Sign: Reviewers hesitate, inspect overlaps, or lose track of which whole set is larger. Quick Check: Build stacked and superposed versions of the same two bar sets and ask for a brief larger-mean or wider-range judgment; prefer the stacked version if it yields surer answers. Stronger Test: Run a short timed A/B test and compare correctness on the same comparison task across the two layouts.

• Split the superposed bar chart into two separate bar charts.
• Place one chart directly above the other so corresponding bars align vertically.
• Remove the superposed version when the decision is about the larger mean or wider range.