Guidelines
Suggest edit

Use a pie chart instead of a treemap for few-slice part-to-whole comparisons

For exact part-to-whole comparison, prefer a pie chart over a treemap on few-category single-level share displays to improve fidelity and address higher percentage-estimation error for readers judging one segment at a time.

  • purpose:select
  • basis:empirical
  • chart:pie-donut:use
  • chart:treemap:avoid
  • quality:fidelity:use
  • lever:chart-family
  • operator:part-whole
  • group-cardinality:few
  • reading-mode:exact

advice

Replace the chart family

Choose a pie chart instead of a treemap when the reader must estimate the share of one segment in a small part-to-whole display. For example, replace a single-level treemap of five shares with a pie chart when the task is to read one category’s percentage of the whole.

reason

Why the pie chart works better here

The tested treemap made percentage judgments less precise than the tested pie chart in a few-segment part-to-whole display. This matters when the chart’s job is to let readers estimate one segment’s share rather than browse a packed rectangular layout.

Mechanism: The pie chart gave readers a more accurate visual basis for estimating one segment’s fraction of the whole than the treemap in the tested five-segment displays.

Evidence: In the extracted experimental ranking for accuracy on the study’s part-to-whole comparison task, the pie chart ranked above the treemap, and the recorded significance pairs show the pie chart as significantly more accurate than the treemap (Zeng & Battle, 2023; Kosara, 2019).

Notes: The evidence is about accuracy, not a broad claim that pie charts are best for every part-to-whole design.

context

Use when this situation is true

  • User Goal: Estimate how much of the whole a selected segment represents.
  • Task: Read or compare segment percentages in a part-to-whole display.
  • Data: One-level part-to-whole data with a few segments; the tested case used five segments.
  • Chart Setting: Static single-view chart used for direct percentage reading.
  • Audience: Readers answering direct percentage questions about a chosen segment.
  • Success Criterion: Lower percentage-estimation error.

exceptions

Do not use when this condition breaks

Break it when: The display is not a few-segment, single-level part-to-whole chart or the task is not estimating a segment’s share of the whole. Why: The evidence only covers five-segment part-to-whole judgments in this specific comparison.

costs

Costs of replacing the treemap

Sacrifice: You give up the treemap’s rectangular packing and any value it has as a space-filling layout. Risk: A pie chart is only supported here against a treemap, not as the best option against every other part-to-whole chart. Mitigation: Use this replacement only for the tested use case: a few segments and direct percentage reading.

mistakes

Common mistake in this comparison

Mistake: Keeping a treemap as the default chart for a few-category part-to-whole readout. Why it fails: In the tested five-segment displays, the treemap produced higher percentage-estimation error than the pie chart.

check

How to check the decision

Failure Sign: Reviewers give less accurate percentage estimates from the treemap than from an equivalent pie chart. Quick Check: Build matched pie and treemap versions with the same few shares and ask reviewers to estimate the same target segment in both versions. Stronger Test: Run a timed A/B test on the same target-segment percentage questions and compare estimation error between the pie chart and treemap.

fix

What to change

  • Replace the single-level treemap with a pie chart when the display shows only a few shares of one whole.
  • Keep the same segment set across both versions so the revision changes chart family rather than content.
  • Re-test the target percentage questions on the pie chart and keep the version with lower estimation error.

References

Kosara, R. (2019). The Impact of Distribution and Chart Type on Part-to-Whole Comparisons. EuroVis 2019 - Short Papers, 5 pages. https://doi.org/10.2312/EVS.20191162
Zeng, Z., & Battle, L. (2023). A Review and Collation of Graphical Perception Knowledge for Visualization Recommendation. Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, 1–16. https://doi.org/10.1145/3544548.3581349