What it is
A doughnut chart shows the parts of a whole as arcs around a ring. Chart Buffet writes the total in the centre, so the chart reads as “this whole, made of these parts”.
Removing the centre takes away the angle at the middle, so readers judge slices by arc length and area instead. Research suggests they do this about as well as with a full pie (Skau & Kosara 2016).
✓ Use it when
- Two to six parts that add up to a meaningful whole.
- You want to show the total alongside its parts.
- A dashboard or report where several doughnuts show single headline shares.
✗ Avoid it when
- Many or similar-sized parts — use bars.
- Comparing compositions across groups — use a 100% stacked chart.
Tips for a better chart
- Keep the ring fairly thick: a very thin ring makes small slices disappear.
- Use the centre for one number, not a paragraph.
- Label the arcs directly with percentages.
The maths
Slice i spans the angle θi = 360° · vi/Σ v; its arc on the outer edge has length r · θi (in radians), and its area is ½ θi (R² − r²) for outer radius R and hole radius r. All three remain proportional to the share, so removing the centre loses no information.
Skau and Kosara tested arcs, angles and areas separately and found that people read pie and doughnut charts mainly by arc length and area, not by the central angle (Skau & Kosara 2016).
A little history
The doughnut is a twentieth-century variant of Playfair’s pie (Playfair 1801), common in business graphics and spreadsheet software. Its reputation improved in 2016, when experiments showed it is read no worse than a full pie (Skau & Kosara 2016).
Questions
Can I change the size of the hole?
Yes. In “Marks and labels”, move the “Hole size” slider.
References
- Skau, D. & Kosara, R. (2016). Arcs, angles, or areas: individual data encodings in pie and donut charts. Computer Graphics Forum 35(3), 121–130.
- Playfair, W. (1801). The Statistical Breviary. London: J. Wallis (printed by T. Bensley).
- Spence, I. & Lewandowsky, S. (1991). Displaying proportions and percentages. Applied Cognitive Psychology 5(1), 61–77.