• Megan Johanson

Pie for Those that Feel Down

This month’s Storytelling with Data challenge is to find a good use for a pie chart. This is not my favorite chart type, but I do believe it can tell a story when done right.

When to Make Pie (Charts)

Pie charts have become infamous in the data viz community and have largely been replaced with other visualization types. However, there are still some situations where a pie chart can convey data insights better than a data table without all the challenges pie charts were known for in the past.


This depends on 1) the data set and 2) the design of the chart.


Start with an Appropriate Data Set

The data set should have only 2 or 3 categories or be able to be grouped into 2 or 3 categories. By doing so you can reduce the issue of a pie chart becoming cluttered with too many pieces.


Two pie charts: one with seven slices and one with three slices.

In addition, it’s ideal if the slices align with easily recognizable angles such as 25%, 33%, 50%, etc. Generally, people are not good at estimating angles, especially when there are many similarly sized slices of the pie.


Two pie charts, both have three slices. One pie chart has two slices that are approximately 25% and the other chart has two slices that are much smaller, approximately 8 or 10 percent each.


Mix in Strategic Design

To make the most of a pie chart you want to use color intentionally and label the slices directly.


Using color intentionally could look like identifying the pie slice that is the most important to you, putting that in a bold color, and putting the other slice(s) in a gray. The colored slice will draw the reader’s attention and the gray slice(s) will slide into the background.


Two pie charts that each have two slices, one is 26% and one is 74%. In one chart the bigger slice is purple and the smaller slice is blue. In the other chart the larger slice is light gray and the smaller slice is blue.


Label the slices directly by putting the data label and percentage within the pie, if they fit, or right next to the relevant slice. This eliminates the need for a legend, which just adds an unnecessary mental load for people as they scan between the legend and the chart to understand what each slice represents.


Two pie charts that both have a 26% slice and a 74% slice. One chart has a legend and no percentages labelled. The other chart has the category labels and percentages directly on top of the relevant slices.


Serve with Quarantine Blues

I used data from the most recent Household Pulse Survey (Week 12) conducted by the United States Census Bureau. The surveys were sent to a sample of Americans each week for 12 weeks asking about their food security, employment status, healthcare access, housing status, home-based education activities, stimulus payment usage, and mental health.


I focused on the mental health question: “Over the last 7 days, how frequently have you felt down, depressed, or hopeless?”


Image of a subset of the original Census data file for the question of how often respondents felt down, depressed, or hopeless. Response options are 'Not at all,' 'Several days,' More than half the days,' and 'Nearly everyday.' Approximately 23 million people did not respond to this question.

I included in my analysis only on the people that responded to the question and grouped together the people that responded “Not at all” and “Several days” so that I would only have three categories or slices in my pie. I was most interested in the extreme responses, so I kept those separate.


Here is the pie chart I created:


Pie chart with one large gray slice showing that 75% of respondents selected one of the two options that was fewer than half of the last 7 days. There is a dark purple slice for the 13% of respondents that selected 'Nearly everyday' and a light purple slice for 12% of respondents that selected 'More than half of the days.' The title summarizes that 1 in 4 Americans felt down, depressed, or hopeless 4 or more of the past 7 days, which is the sum of the two purple slices. A link to the data source is also provided.

I applied all the best practices for pie charts that I described above. In addition, I used shades of the same purple to visually group the two most extreme response options together while keeping the responses distinct. These two categories together add up to 25%, which I summarized in the chart title.


What other strategies can improve a pie chart?