Use stacked bar charts instead of stacked area charts for negative correlation judgments
For association judgments, prefer stacked bar charts over stacked area charts on negatively correlated quantitative series to improve fidelity and address weaker discrimination of correlation strength for people comparing opposite-moving values.
- purpose:select
- basis:empirical
- task:relate
- chart:bar:use
- chart:area:avoid
- quality:fidelity
- lever:chart-family
- operator:association
advice
Choose stacked bars over stacked areas for negative relationships
Use a stacked bar chart instead of a stacked area chart when viewers need to judge the strength of a negative correlation. For example, if two quantitative series move in opposite directions and you are choosing between these two stacked forms, render them as stacked bars rather than stacked areas.
reason
Why the stacked bar performs better here
The stacked forms were not perceptually interchangeable. For negative correlations, stacked bars supported more precise judgments than stacked areas.
Mechanism: The stacked bar layout gave viewers a more discriminable pattern for comparing stronger versus weaker negative relationships than the stacked area form.
Evidence: For negatively correlated data, stacked bar charts significantly outperformed stacked area charts in correlation-judgment precision, and both were modeled with Weber-style fits after filtering unreliable conditions (Harrison et al., 2014).
Notes: The paper also reports that viewers appeared to use different visual cues in stacked bars than in stacked areas.
context
Use when the sign is negative and the chart family is still open
- User Goal: Help viewers compare which negative relationship is stronger.
- Task: Judge association strength, not exact values.
- Data: Two quantitative series with negative correlation.
- Chart Setting: A stacked display is desired and the choice is between stacked bars and stacked areas.
- Success Criterion: Viewers can more reliably rank stronger and weaker negative correlations.
exceptions
Do not extend this contrast to unsupported sign conditions
Break it when: The relationship to be judged is positive rather than negative. Why: Positive stacked-area and positive stacked-bar conditions frequently hit the study’s chance boundary, so this contrast was not supported for positive correlations.
costs
Costs of changing the stacked form
Sacrifice: This recommendation only covers the negative-correlation case. Risk: Treating it as a general rule for all stacked relationships can overreach the evidence. Mitigation: Check the correlation sign before deciding between stacked bars and stacked areas.
mistakes
Common failure mode in this comparison
Mistake: Assume stacked bars and stacked areas are perceptually equivalent because their overall shapes look similar. Why it fails: The study found a significant performance difference between them for negative correlations.
check
How to test the choice
Failure Sign: Reviewers struggle to tell which of two negative stacked-area relationships is more strongly correlated. Quick Check: Render the same negative relationship once as stacked bars and once as stacked areas, then ask which version makes the stronger correlation easier to identify. Stronger Test: Use repeated paired comparisons of slightly different negative correlations and keep the chart type that produces more consistent correct choices.
fix
What to change
- Replace the stacked area chart with a stacked bar chart for the negative-correlation view.
- Re-run the comparison on the same data after the swap instead of assuming the stacked forms behave the same.
- If positive and negative relationships both appear in the project, choose chart families for those cases separately.