Pie / Donut Chart

The pie chart is the most restricted figure type in the Publishable Figs library. Human perception of angles and areas is poor —we systematically misjudge the relative sizes of similarly-sized pie slices. Pie charts are only clearly appropriate when there are 竕､ 5 categories with substantially different proportions. For most scientific use cases, a bar chart is more readable and more accurate.

STEP 1 —When to Use

A pie chart is appropriate when: there are 竕､ 5 categories; the proportions differ substantially (no two slices are within 5% of each other); the message is "part of a whole" and the absolute sizes do not matter; each slice represents a distinctly different and mutually exclusive category with a simple label.

Licklider blocks pie chart generation when: more than 7 categories are present; any slice is less than 3% (invisible at print size); two or more slices are within 5% of each other (visual differentiation fails); the data are not truly compositional (do not sum to 100%).

[Image placeholder: Left —pie chart with 6 similar-sized slices (around 14-18% each). Almost impossible to rank without looking at the labels. Right —the same data as a horizontal bar chart. Immediately clear which category is largest and which is smallest. Demonstrates why bar charts are preferred for similar-sized categories.]

Similar-sized slices in a pie chart (left) cannot be visually ranked. The same data as a bar chart (right) is immediately interpretable.

STEP 2 —Journal Requirement

Each slice must be labeled with its percentage value directly on or next to the slice (not relying solely on the legend for value identification).
n (total observations represented) must be stated in the legend or figure.
Category labels must be clearly associated with slices (not relying on color alone —accessibility requirement).
If used for sample composition (demographics), the data source must be referenced.

STEP 3 —Licklider Implementation

Input

Category names and proportional values (percentages or counts that sum to 100%)
Chart type: pie (default) or donut (with optional center label)
Color palette: colorblind-safe (default), journal palette, or custom

Guard Checks

Category count: > 7 xblocked, stacked bar suggested
Minimum slice: < 3% xwarning, "Other" category merge suggested
Similar slices: any two within 5% xwarning with bar chart alternative
Non-compositional data: values do not sum to ~100% xblocked

Output

Pie or donut with percentage labels on each slice
Legend with category names and color swatches
n total labeled in legend or center of donut
Colorblind-accessible palette (default: ColorBrewer qualitative, max 7 colors)

[Image placeholder: Licklider donut chart with 4 categories (30%, 28%, 25%, 17%). Percentage labels on each slice. Category labels on slice via leader lines. Center: "n=240". Colorblind-safe 4-color palette. Legend below. Guard check result shown: "4 categories, min slice 17%, all clear —pie chart appropriate."]

Licklider donut chart: percentage labels, leader lines, n in center, colorblind-safe palette, guard check passed.

STEP 4 —Draft Output (Draft / Needs review)

Donut chart showing the distribution of [variable] across [N] categories
(n = 240 total observations). Percentage values are labeled on each segment.
Categories: [A] = 30%, [B] = 28%, [C] = 25%, [D] = 17%. Colors are
from the ColorBrewer qualitative palette and are distinguishable under
common forms of color vision deficiency.