Donut Chart

Donut chart is a pie chart with a center opening. The center space can carry a summary label, such as total sample size, making it useful when one key number belongs alongside the composition display. The opening does not improve comparison accuracy between slices.

Figure purpose

Donut chart serves the same purpose as pie chart: showing how a total divides into categories. The structural difference, a hollow center, is a layout feature, not a perceptual improvement. Research shows that the hole size does not meaningfully change reading accuracy, and that placing a label in the center does not harm performance [6].

Choose donut over pie when the center space adds useful information. Choose pie when a full circle is cleaner for the message.

When to use

Use donut chart under the same conditions as pie chart:

  • Five or fewer categories
  • Part-to-whole composition is the main message
  • Differences between slices are large enough to read from arc length alone

Additionally, donut is preferable to pie when:

  • You want to display a total n, a primary percentage, or another key summary value in the center opening

Do not choose donut because it looks more modern, or because you believe it corrects pie chart's comparison limitations; it does not [5, 6].

Licklider can render the donut and calculate shares, but it does not determine automatically whether your category count is low enough for easy reading, whether nearby slice sizes are still interpretable for your audience, or whether the real question is precise comparison rather than part-to-whole composition.

That limitation matters because a donut chart can look polished while still being a poor fit for the scientific question. If readers need to compare similar values accurately, the center opening does not solve that problem.

How donut differs from pie chart in Licklider

Donut chart uses the same data contract and the same rendering engine as pie chart. The center opening is set at 50% of the radius and is not currently configurable.

In the current product, the center label can display Total n, Total value, Largest category, or custom text. Slice labels are also configurable, with the same options as pie chart: Name + %, Name + value, Name + value + %, % only, and Value only.

These are layout and labeling features, not accuracy features. The fixed 50% opening creates space for a center label, but it does not make adjacent slices easier to compare than they would be in a pie chart.

Required columns

Same as Pie Chart: one categorical column and one numeric value column. Licklider aggregates by summing values within each label before rendering.

This aggregation is a display convenience. It does not determine whether summing within each label is scientifically the right summary for your dataset, nor does it distinguish automatically between one true whole and values that should really be compared as separate groups or conditions.

Example

Appropriate use

A flow cytometry panel shows the proportion of cell populations: CD4+ T cells (38%), CD8+ T cells (29%), B cells (21%), NK cells (12%). Four populations, clear shares. A donut chart with total cell count in the center would communicate both the composition and the sample size in one figure.

Not a reason to use donut

"The pie chart slices are too similar in size; maybe the donut will be easier to read." The hole does not resolve this. If slices are too similar to distinguish in a pie, they are equally indistinguishable in a donut. Use a bar chart instead.

Alternative figures

  • Use Pie Chart when no center label is needed and a full circle is cleaner.
  • Use Stacked Bar Chart when you need to compare composition across several categories at once.
  • Use a bar chart or Strip Plot for accurate category comparison.
  • Use Group Comparison figures when the question is whether groups differ.

Design Rationale & References

See Part to Whole - Design Rationale & References for the perceptual research basis for Licklider's approach to pie-style figures, including evidence on donut chart accuracy [5, 6].

The key design rule is simple: choose donut only when the center opening adds useful context, not because it is expected to fix pie-chart comparison limits. Licklider therefore treats donut as a layout variant of pie rather than as a perceptually stronger chart type.

Two relevant references are:

  1. Skau, D., Harrison, L., & Kosara, R. (2016). Arcs, angles, or areas: Individual data encodings in pie and donut charts. Computer Graphics Forum, 35(3), 121-130. https://doi.org/10.1111/cgf.12888
  2. Kosara, R., Skau, D., & Burch, M. (2019). There is no spoon: The effects of hole size on the perception of donut charts. Computer Graphics Forum, 38(3), 183-194. https://doi.org/10.1111/cgf.13694

These references support the specific Licklider claim made on this page: the center opening can be useful for labeling, but donut charts do not meaningfully improve slice-comparison accuracy over pie charts.

See also