11

What is the UX rationale behind picking a doughnut chart vs a pie chart or viceversa?

I'm asking because I'm intrigued as to why a designer would prefer to represent data with a pie instead of a doughnut. Are there any "hidden" advantages of one over the other? Did the designer of the doughnut chart just felt like adding white-space to the pie chart would make it more aesthetic and truly has no added value?

Looking around the web, most if not all of the articles I've found just treat both charts as if they were the same.

|improve this question
  • 2
    "What has triggered the question?" In making a choice, as simple it may be, I'm asking myself "Why?". Digging through this site, I've understood that is important to ask "Why" when designing an interface having in mind the user expectations. As for research, I did a Google Search, found a lot of people just claiming them to be "the same". I cam e here to see if there was indeed differences between the two, so I could make a choice in one of my designs. – RainierMallol Mar 13 '17 at 17:39
  • 4
    You may find this an interesting read on why both types are bad and should rarely (contextually) ever be used: perceptualedge.com/articles/08-21-07.pdf – mrchaarlie Mar 13 '17 at 19:41
  • 2
    @mrcharlie Totally agree. Pie charts if you have 2 values to chart, something else if you have >2. They're really not that usable for 3 or more values. – JonW♦ Mar 14 '17 at 9:40
  • 1
    @AndrewMartin: With all due respect, and not trying to pick a fight here as you are clearly much more experienced and professional than myself in these matters, I disagree. At least for this particular question. When presented with two choices that seem equal, what would be the rationale to choose one over the other? Why did someone create new type of chart equal to another? Was it purely aesthetic and in fact it doesn't matter? Or does it? Some have posted benefits of one over the other, or not to use them at all. If its the way the question is asked that is troubling you, I have edited it. – RainierMallol Mar 14 '17 at 13:17
  • 1
    Neither - pie charts and doughnut charts are poor for comparing things if you have more than 2 segments - mainly because it is hard to work out relative size if the segments are a similar size. A bar graph is a much better choice for comparing multiple things. Read any book by Stephen Few. – SteveD Mar 14 '17 at 14:52
15

A doughnut (or donut) chart serves a similar purpose to a pie chart, except that it is able to show more than one set of data. Think of it as a pie chart with an additional dimension.

See the example below, where a donut chart shows sales for different regions for two separate years. This allows to compare data more easily with respect to time (see how easy is to see that the US sales decreased in 2006), and takes much less space than two pie charts put one aside the other.

It works well between 2 and 4 sets of data (after that limit, it's going to appear crammed) and for a limited number of slices for each set (since only two slices are guaranteed to be aligned, such as US and Europe in the picture below, comparing other slices is going to be difficult).

enter image description here

|improve this answer
  • 5
    Yeah, and see how much more difficult it is to see whether the sales for Asia increased or decreased. To be honest, I wouldn't consider this to be a figure that is working particularly well. – Schmuddi Mar 14 '17 at 7:46
  • 1
    This is poor data representation. 2006 data appears to be "larger" than 2005, as it takes up more area. For example, look at the United States data between the two years--it appears, visually, that the value in 2006 is greater than 2005 because it takes up more area. I would strongly discourage the use of this type of graph for comparisons. There are many other much clearer ways to depict this data. – maxathousand Mar 14 '17 at 13:55
  • 1
    If you have to put numbers on the pie, you are already failing to communicate with the data visualisation. A bar graph would be so much better in your pie example. – SteveD Mar 14 '17 at 14:56
  • 1
    A bar chart needs numbers, too; moreover, it fails to communicate that the values sum up to 100%. In a pie/donut chart, this information is immediately conveyed. – dr01 Mar 14 '17 at 15:13
  • 1
    @dr01 I agree that it's useful for showing parts of a single, whole data set, but my point is that I don't think it's an effective way to compare two different data sets. Again, when "nesting" these data sets, the 2006 data set occupies more visual space, so it incorrectly implies that the values are greater than the 2005 data. – maxathousand Mar 14 '17 at 18:27

Not the answer you're looking for? Browse other questions tagged or ask your own question.