What are the best practices for using pie charts?

Pie charts are useful for showing the breakdown of a total, but not for showing distributions (i.e., how values are distributed across a range). If the point of a chart is to show how values are distributed, consider chart types such as histograms or strip plots instead.

Pie charts are used full for showing one part out of a whole. Generally, any more pie "pieces" reduce the message. In terms of increasing pieces, there are diminishing returns. If used appropriately, they can be very effective.

To sum up: 1. A pie chart is used for parts-to- whole, showing composition or distribution of a variable 2. A complete pie chart (360 degree) represents total composition (100%)

A pie chart is a circular statistical graphic, which is divided into slices to illustrate numerical proportions. In a pie chart, the arc length of each slice, and consequently its central angle and area, is proportional to the quantity it represents. This type of chart is widely used in various fields for its simplicity and visual immediacy in showing the composition of a dataset.

It is hard to define any case where a pie chart is the best option. In 99.9% of cases, a bar or column chart is far more effective. The main reason being it’s too easy to be deceived with data visualized in pie charts. Best to avoid pie charts altogether.

As humans, we perceive things visually in the following order: 1. Order 2. Color 3. Size 4. Shape I avoid using pie charts because they present a lot of challenges, especially for user accessibility. In relation to the list of characteristics above, it's very often hard to look at a pie and determine whether the segments are already in order. Also, the colors are different to distinguish between the slices, which means that we can't use colors to emphasize certain segments or data points. Distinguishing between them also comes down to distinguishing between shapes rather than size, which is conversely more difficult as well.

A pie chart can work well to show a part-to-whole relationship, which a bar chart won’t do. One needs to make sure though that the categories are sorted, and only 1-2 are highlighted. If there are many small categories, they can be grouped together as they won’t be readable on a pie in any case.

Pie charts are usually a poor visualisation choice. Nevertheless they may be useful when you want to compare two quantities and show a big difference between the two. If you have 95-5% and want to underline the difference a pie chart may be a good option.

Agreed with all points mentioned by all contributors in this section. Basically, don't use pie chart. Else, it will look more like pizza, with colourful toppings.

A pie chart is most effective when you need to show the proportion or percentage of a whole. It's ideal for displaying data that are categorically distinct and represent parts of a whole, like market share, survey responses, or budget allocations. It works best with a limited number of categories, typically no more than five to six slices, to keep it readable and comprehensible.

Just don't. Don't use pie charts. Please. A provocative response, I know but highly effective! There are lots of better charts to use, and we could even challenge ourselves to create new ones!

How would one even show negative values on a pie chart? 🤯 In theory, I agree with the notes above. In practice, it will depend on your use case and design choices, so there’s no right answer for all. I’ll also add that a lot of non-data people love pies—something to keep in mind.

Agreed with the points mentioned by all contributors in this section. Basically, avoid pie chart at all time. Bar or column chart is a better option.

Avoid a pie chart when you use more than two pieces (one part and one whole). Increasing the number of pieces reduces the effectiveness and has diminishing returns. Simple pies are best. Donut charts that can show the one part of a whole, and a key metric (often a percentage) in the middle are the most effective.

Avoid using a pie chart when you have a large number of categories, as the slices become too thin and difficult to compare. It's also less effective for showing trends over time or relationships between data points. Additionally, if the data points are of similar values or require precise comparison, bar or line charts are more suitable.

I'd dispute many of these formatting guidelines: - Pie slices should be sorted so that those of *greatest interest* are first (i.e., start at the "12 o'clock" position and go clockwise). Often, this means sorting from largest value to smallest, but sometimes it could be smallest to largest, or some other sort order. - Using shades (i.e., darker and lighter shades of the same hue) is generally a bad idea because audiences perceive darker shades as representing greater quantities, even though that's not the case in a pie chart. - Labelling slices directly is almost always more desirable than using a legend, since it removes the need to memorize colors. If very high precision is required, add value (%, $, etc.) labels.

When designing a pie chart, ensure that it is simple and uncluttered. Start the largest slice at the 12 o'clock position and order the slices by size. Use contrasting colors for different slices for better differentiation. Label each slice clearly, either directly on the chart or in an accompanying legend. Avoid using too many decorative elements that might distract from the data.

I give it the 3 second rule. If I can't determine the main point of the visual within 3 seconds, then it's not effective. That is why making the main takeaway as clear as possible is the best solution.

Consider examining the labeling of each pie slice to ensure clarity and precision in conveying specific data points. Evaluate the color choices to guarantee accessibility for all viewers, particularly those with color vision deficiencies. Furthermore, assess whether a pie chart is the most appropriate visualisation for your data compared to alternative options like bar graphs or scatter plots. Lastly, scrutinize the source and reliability of the data presented in the pie chart to enhance the overall trustworthiness of your analysis.

I'll just say it... Sometimes a pie chart is okay... Honestly, if the data set is large but can be grouped effectively without creating a thousand segments then go for it! If not, this isn't the viz you're looking for!

To evaluate a pie chart, check whether it represents the proportion of each category. Ensure that the chart is not overloaded with information, and the categories are distinct and easily comparable. The colors and labels should aid in understanding, not complicate it. The data should be accurate, and the chart should give a true representation of the underlying numbers.

The easiest way to improve on a pie chart is by literally, and physically, chosing any other form of visualisation. Limited use only

For those prone to using pie charts, bar charts are a great option to try out and most of the time I find they work for the visualization purposes that I need them to work for!

To improve a pie chart, focus on clarity and simplicity. Use a consistent color scheme and limit the number of slices. Consider using a sorted pie chart, where slices are ordered from largest to smallest. For better readability, you might convert very small slices into a separate “other” category. Interactive elements like tooltips can be added for digital versions to provide more information on hover.

There is a reason why the most popular memes featuring charts use pie charts: they are almost universally understood. Indeed, there are many instances where a pie chart may not be the most appropriate choice (such as when there are more than four or five categories, or when the differences between categories are not significant). Nonetheless, if your intended audience is the general public and your data can be effectively illustrated using a pie or donut chart, what's the harm?