Skip the Pie

 

I don´t like pie graphs.

There are many reasons why I don´t like, and hence don´t use them, but I have come to realize that they seem to be popular. Often times when I have said that I don´t like the pie charts, I have been met with a surprised reaction followed by a “why?”.

The first reason why I avoid pies is because our eyes are designed to comprehend and understand square objects better than round objects, but the main reason why I actively choose not to use them is that the risk for incorrect information increases with pie graphs, and you can easily mislead your audience.

Skipping the pie, you have a number of other choices at your disposal. A stacked or unstacked, horizontal or vertical bar graph, a line graph or slopegraph, a table or even a block of text – all depending on the purpose and the audience – would serve better than a pie graph.

As often times, it is better to describe the issue with a visual and practical example as below. Assume that we are looking into the best-selling car brands in the UK Q1-3 2014, and we are looking at the top ten brands.

 

Number of Cars sold in the UK by brand, Q1-3 2014

Cars, Table

 

 

 

 

 

 

 

 

Source: Best Selling Cars, Link  

The table above ranks the top ten most selling car brands in the UK Q1-3 2014, plus the additional information of how many cars were sold in total and the share of top ten biggest brands.

Let’s show the above table as a pie graph.

 

Number of Cars sold in the UK by brand, Q1-3 2014

 

Cars, pie

 

 

 

 

 

 

 

 

 

 

 

As you can see, it is not easy to determine if and by how much Ford is bigger than Vauxhall. The same goes for Peugeot and Toyota. While it is possible to some degree to interpret the ranking, there is some important information lost in the pie graph, for example the number of cars sold by each brand (volume) or percentage of the brand vs the top ten or the total market. There is no easy way to compare the individual values.

Other issues are that pie graphs often by default need to use many colors to differentiate one category from another , which means that our eyes more quickly get tired looking at a pie graph, or the way that our eyes are designed we can grasp and comprehend square objects better than round ones. Round figures also take up more space, and therefore are not very effective in dashboards.

In worst cases, pie graphs even might convey wrong information. Let’s look at the pie graph below, where instead of the number of cars sold we show the % Market Share column of the above table.

 

Sold Cars Market Share % in the UK by brand, Q1-3 2014

Cars, pie 2

 

 

If you combine the individual values, i.e. 13.4% for Ford, 10.7% for Vauxhall, etc., you will notice that the total will land at 67.2%, which obviously is not 100%. The values shown in the pie graph are the market size by brand vs the total car market, which includes all brands and not only the top ten.

 

What to use instead?

Best practice has shown that alternatives to pie graphs work better.

A table as above would go a great job if your parameters are few and individual values have been ranked among themselves.

If you are interested in changes over two periods, a slopegraph would be a great choice. You can read about slopegraphs here.

A very powerful alternative is a horizontal or vertical bar graph as below.

 

Number of Cars sold in the UK by brand, Q1-3 2014, Thousand Units

Cars, Bar

The advantage with the bar graph is the versatility that comes with it. Suppose that you want to look into the the German brands that made the top ten.

 

German car brands in the UK top ten, Q1-3 2014, Thousand UnitsCars, Bar 2

 

And if you still insist to use a pie graph, make sure you get your measures right. As you can see below, 193% of the Republicans support Palin, Romney and Huckabee…

 

Fox's+Pie+Chart