Judging a pie slice's angle is a coarser task than judging a bar's length or position

What it is

A pie chart maps each part's share of a whole to a central angle, also producing corresponding arc length and area. Angle judgment is usually less suitable for subtle magnitude comparison than position or length on a common scale. Pies therefore communicate conspicuous part-to-whole structure more reliably than exact differences or strict ranks. “Imprecisely” is relative to task and graphic conditions: it does not make every angle indiscriminable or guarantee that every bar design wins every task.

Why it happens

Wedges do not give every part a common origin. Readers may combine angle, arc, and area cues while moving attention among separated wedges, making similar shares hard to compare consistently. A donut preserves arc proportion but removes the center and wedge tips, shifting reliance toward arc length or outer contour; inner radius, ring width, and chart size affect the available cues. Geometrically proportional angles prevent encoding distortion but do not eliminate perceptual error.

Studying it

Cleveland and McGill's graphical-perception experiments compared position, length, angle, and other elementary tasks using proportion-estimation error. Heer and Bostock later replicated and extended selected conditions through crowdsourcing. The evidence places angle behind common-scale position for typical magnitude judgments, not at one constant rank across all data, tasks, and readers. Pie studies should compare pie, donut, bar, and dot versions on largest-part identification, pairwise comparison, share estimation, and part-to-whole understanding while varying separation, rotation, labels, delivered size, and ring width; report error distributions, time, and confidence.

Where it stops holding

Circular structure may help when the task is to identify a clearly dominant part, verify a total, or communicate that parts compose one whole. If parts overlap, the denominator is unclear, or displayed values do not exhaust the whole, the pie is semantically wrong before perceptual precision is considered. Direct values support exact lookup without improving comparison among unlabeled wedges. A donut center can hold a total or state, but cannot replace each part's name, value, and denominator.

Applying it

• Confirm mutually exclusive categories, one denominator, and a complete whole before considering a pie or donut; otherwise use independent magnitude marks.
• Separate dominant-part detection and composition overview from exact comparison. Test pies for the former and common-scale bars or dots for the latter.
• Label categories and values locally and provide a keyboard-accessible table, total, denominator, order, and text summary. Colour, angle, and hover cannot be the sole access route.
• Compare largest-part, pairwise-difference, and share tasks on target proportions and delivered sizes. Select by acceptable error rather than a fixed angular threshold.

Related

• Same group: U2.07.2 Past a few slices, pies become unreadable · U2.07.3 Stacked bars are usually the better substitute
• Adjacent: U1.03.2 Angle judgments are markedly less accurate than length under equal conditions · U2.02.1 Bar charts suit magnitude comparison across categories
• Search terms: pie chart · angle judgment · graphical perception · donut chart · part to whole