Pie Chart Explained: When to Use It, How to Read It and How to Report It
Pie charts are among the most popular and most criticised charts in data visualisation. Used well, they communicate "this category is about a quarter of the whole" at a glance. Used badly, they hide differences, crowd the page with slices, and tempt readers to compare angles that humans judge poorly. This guide explains when a pie chart is a defensible choice, how the numbers behind it are computed, how to read one critically, and how to report it.
What is a pie chart?
A pie chart divides a circle into slices whose areas, arc lengths and angles are proportional to the share each category contributes to a total. Because the slices must add to a whole, a pie chart only suits data where the categories are mutually exclusive and exhaustive, such as the percentage of respondents choosing each answer to a single-choice question. A doughnut chart is a pie chart with the centre removed and carries the same strengths and weaknesses.
When should you use a pie chart?
- Showing a small number of categories, ideally five or fewer, that sum to 100%.
- Emphasising one dominant share (for example, "more than half of respondents chose online").
- Communicating a rough composition to a general audience where exact comparison is not the goal.
Pie chart versus bar chart. Experimental work on graphical perception shows that people judge position along a common scale more accurately than angle or area (Cleveland & McGill, 1984). When the reader needs to compare similar-sized categories, or when there are many categories, a bar plot is the safer choice. Spence (2005) reviews the origins and usage of the pie chart and the evidence on how well it communicates.
The mathematical foundation
For category i with count ni out of a total N, the proportion, percentage and slice angle are:
pi = ni / N, % = 100 × pi, anglei = 360° × pi
The proportions must sum to 1 and the angles to 360°. If a category's percentage is rounded for display, the displayed percentages may add to 99% or 101%; say so in a note or adjust the rounding transparently.
To test whether the observed shares differ from an expected distribution, a chi-square goodness-of-fit test (Pearson, 1900) is commonly used:
χ² = Σ (Oi − Ei)² / Ei, df = k − 1
where O are observed counts, E expected counts and k the number of categories. The pie chart shows the shares; the test addresses whether they depart from an expectation.
Assumptions and data requirements
- Parts of a whole. Categories must be mutually exclusive and add to a meaningful total. Multiple-response questions (where one person can choose several options) do not sum to 100% and do not suit a pie chart.
- Non-negative values. Slices cannot represent negative numbers.
- Few categories. Beyond about five or six, slices become thin and labels collide.
- Clear ordering. Start the largest slice at 12 o'clock and proceed clockwise in decreasing order, unless the categories have a natural order.
- Sufficient sample size. Percentages from tiny samples are unstable. Always report N.
Original research example (illustrative)
Background. A library surveys students on where they prefer to study.
Objective and question. What share of students prefer each study location, and are preferences evenly spread?
Data. N = 200 students each choosing one location. The values below are invented for teaching.
Step 1: Compute proportions and angles
| Location | Count | Proportion | Percent | Angle |
|---|---|---|---|---|
| Online / virtual spaces | 84 | .42 | 42% | 151.2° |
| Library | 52 | .26 | 26% | 93.6° |
| Home | 40 | .20 | 20% | 72.0° |
| Café | 24 | .12 | 12% | 43.2° |
| Total | 200 | 1.00 | 100% | 360.0° |
Step 2: Check the arithmetic
84 + 52 + 40 + 24 = 200. Proportions: 84/200 = .42, 52/200 = .26, 40/200 = .20, 24/200 = .12, which sum to 1.00. Angles: .42 × 360 = 151.2, .26 × 360 = 93.6, .20 × 360 = 72.0, .12 × 360 = 43.2, which sum to 360.0.
Step 3: Test against an even split (optional)
If preferences were equal, each location would have E = 200 / 4 = 50 students. Then χ² = (84 − 50)²/50 + (52 − 50)²/50 + (40 − 50)²/50 + (24 − 50)²/50 = 23.12 + 0.08 + 2.00 + 13.52 = 38.72, with df = 3. This far exceeds the 5% critical value of 7.81, so p < .001: preferences are not evenly spread.
Step 4: Interpret
Online spaces are the most popular choice (42%) and cafés the least (12%). The pie chart makes the dominant category obvious, but note that the library (26%) and home (20%) slices are hard to rank by eye, which is why the percentages should be printed on or beside the slices. If the aim were to compare these closely, a bar chart would do the job more clearly. The chi-square result says the distribution is uneven; it does not say which categories differ from each other, and an expected even split may not be a sensible null in the first place.
How to make a pie chart with the StatFAS Pie Chart Maker
Open statfas.com/pie-chart-maker.html.
- Prepare your data. One column of category labels and one of counts (or raw responses, if the tool tabulates them).
- Load the data and map columns.
- Set options. Typical choices are labels, percentage display, colours, start angle, doughnut style and title.
- Review the figure. Check that slices are ordered sensibly, percentages are readable and colours are distinguishable.
- Download. Export at a resolution suitable for publication.
statfas-pie-chart-maker-interface.pngAlt text: "StatFAS pie chart maker interface before data entry."
Caption: The Pie Chart Maker on StatFAS.com.
statfas-pie-chart-example-data.png · Alt: "Study location counts loaded in the StatFAS pie chart maker."Screenshot 3:
statfas-pie-chart-settings.png · Alt: "Label and colour settings in the StatFAS pie chart maker."Screenshot 4:
statfas-pie-chart-result.png · Alt: "Finished pie chart with percentage labels." Use only genuine captures of the tool.Reading a pie chart
- Slice size. Read the printed percentages rather than estimating angles. Mistake: ranking slices of similar size by eye.
- Total. Check what the whole represents and its size. Mistake: ignoring N, so that 50% of 8 people looks as convincing as 50% of 800.
- Category definitions. A large "Other" slice hides information.
- Colour. Colours should be distinguishable and not imply an order that does not exist.
- Comparisons across charts. Comparing slices across two separate pies is difficult; use a grouped or stacked bar chart.
APA 7 reporting example (based on the illustrative data)
Figure 1 shows students' preferred study location (N = 200). Online spaces were the most frequently chosen option (42%, n = 84), followed by the library (26%, n = 52), home (20%, n = 40) and cafés (12%, n = 24). The distribution differed from an even split, χ²(3, N = 200) = 38.72, p < .001.
Essential elements: a numbered figure with a descriptive caption, the total N, counts as well as percentages, and the statistical test if you make an inferential claim. In APA style, figures should be understandable without the surrounding text, so state in the caption what the whole represents.
Common mistakes and troubleshooting
- Too many slices. Group small categories into "Other" only if the grouped share remains small, or switch to a bar chart.
- Percentages that do not add to 100. Explain rounding, or show one decimal place.
- Using a pie for non-additive data. Multiple-response data or averages are not parts of a whole.
- 3-D and exploded slices. Perspective changes the apparent size of slices and misleads readers.
- Alphabetical ordering. Order by size so the pattern is visible.
- Omitting N. Always state the total.
- Comparing time points with several pies. Use a line graph or stacked bar instead.
Limitations and alternatives
A pie chart shows one distribution at one moment and cannot show uncertainty. For precise comparison, use a bar plot. For change over time, use a line graph. For two categorical variables at once, consider a grouped bar chart or a contingency table with a chi-square test. For simple tabulations, the descriptive statistics calculator can help.
Frequently asked questions
When should I use a pie chart?
When you show about five or fewer mutually exclusive categories that form a whole, and the message is a rough proportion.
How is the angle of a pie slice calculated?
Divide the category count by the total and multiply by 360°.
Is a bar chart better than a pie chart?
For precise comparison, usually. People judge length along a shared baseline better than angle or area.
Can a pie chart show negative values?
No. Slices represent non-negative shares of a total.
Conclusion
A pie chart earns its place when there are few categories, they form a whole, and the message is the rough size of one or two shares. Print the percentages, state the total, and switch to a bar chart when precise comparison matters. When you are ready to draw one from your own data, the StatFAS Pie Chart Maker provides a browser-based route, and the exported figure should be checked by you before publication.
References
- Cleveland, W. S., & McGill, R. (1984). Graphical perception: Theory, experimentation, and application to the development of graphical methods. Journal of the American Statistical Association, 79(387), 531–554.
- Pearson, K. (1900). On the criterion that a given system of deviations from the probable in the case of a correlated system of variables is such that it can be reasonably supposed to have arisen from random sampling. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 50(302), 157–175.
- Spence, I. (2005). No humble pie: The origins and usage of a statistical chart. Journal of Educational and Behavioral Statistics, 30(4), 353–368.
- Tufte, E. R. (2001). The visual display of quantitative information (2nd ed.). Graphics Press.