Subscribe Bookmark RSS Feed
XanGregg

Staff

Joined:

Jun 23, 2011

Progress on #OneLessPie

Pi Day (3/14) is next week, and once again, we are encouraging everyone in the data visualization community to use this day as motivation to help clean up ineffective pie charts within their organization or in public spaces like Wikipedia. I think there is still a mindset outside of our community that a pie chart is the embodiment of data visualization, and such thinking will continue if we don't take action when pie charts are used poorly. Replacing such a pie chart is a small but positive step to take, and we've been using #onelesspie as a moniker for the effort.

 

Which pie charts are ineffective or poorly used? Some would say "all of them." John Tukey and Stephen Few have gone so far as saying there is always something better that a pie chart. I detailed many ways pie charts commonly go wrong in my original call to action in 2014. However, as I argued in my keynote presentation at last year's JMP Discovery Summit conference, "all graphs are wrong," and there are always trade-offs to consider when choosing how to represent data.

 

Ann K. Emery offers good advice on the pie charts in her blog post, "When Pie Charts are Okay (Seriously)." The short answer: almost never.

 

I will have a pie chart makeover post to go out on Pi Day next week, but today I want to celebrate positive progress in pie chart usage. While looking around the English-language Wikipedia, I noticed that some of the less effective pie charts have been retired or have evolved into better visualizations.

 

Here are a few examples:

 

This 3D exploding pie chart was once used in the Usage Share of Web Browsers article on Wikipedia. Now it's been removed in favor of an overlaid line chart showing usage share over time.
before_browser_share.pngAfter_browser_share.png 

For a number of years, the Wikipedia pages for the federal budgets of particular years featured a pie chart with way too many slices and disconnected labels (2010 budget page), but the federal budget page now has a saner pie chart with fewer wedges and better labels.

 

Before_US_spending.jpgafter_US_spending.png

 

It's a bad sign when a table of numbers is more readable than your visualization, but I think that's the case with this Vancouver demographics pie chart that was replaced with a table:

Vancouver_pie-1024x1021.pngVancouver_table.png

 

Finally, this pie chart on the usage share of operating systems is now a more readable bar chart.

 

Wikimedia_OS_share_pie_chart-400x300.pngOS_share_bar.png

 

I hope these examples of progress will inspire you to remake an ineffective pie chart yourself. You have a week until Pi Day. Share your efforts using #OneLessPie.

4 Comments
Community Member

Michael Clayton wrote:

Thanks. Always need good and bad examples. Would like to see more good example of JMP Phased SPC charts (for example) and JMP Variability Chart for non-gauge studies and variance components analysis.

Staff

Xan Gregg wrote:

Thanks. I've often wondered about improving those charts versus how much of them are immutable because of de facto industry standards.

Community Member

Peter Bartell wrote:

Xan raises a good point. I think one voice crying out in the wilderness is Don Wheeler in this space. His EMP approach to measurement system analysis always raises the hackles of those wedded to the traditional Gauge R & R approach...made de facto standard in some measure at least by the Automotive Industry Action Group.

Community Member

Excel Roundup 20160314 - Contextures BlogContextures Blog wrote:

[…] honour of Pi Day, the JMP Blog reminds you to cut back on the pie charts. Can you go a whole day without […]