The Problem with Animated Charts
Something I once heard which has always stuck with me was that charts and data visualisations can be divided into two broad categories - charts which are intended to reveal information vs. charts which are intended to obscure it.
A good example of the latter can be found in Glen’s recent article about pie charts. I took issue with it for a couple of reasons: partly because I’m part of the "other" segment, but also because I don’t feel like pie charts were the ideal way to show this dataset. Sure the pie chart was pretty but it did obscure the detail of the data set.
Pies are one example but animated charts are the ones that I want to focus on. Before I begin let me say I don’t think all animations are bad for information clarity. Used sparingly to provide a sense of time they can be great.
When we were testing the chart animations in Yellowfin we used a dataset which contained the population of each Australian state at the start of each decade. The upper bound of the Y-axis was hard set to the highest value in the entire dataset, and the year was used as the controlling field for the animation. This was fantastic to watch, you could see when states went through periods of growth, when states stagnated and so forth. Although the data could have been presented using a line chart with a series for each state it wouldn’t have been nearly as absorbing.
The same effect can be seen on the Google Public Data Explorer, although I think the descending order makes the data harder to follow it’s good to watch as well as informative.
Now for animations which I don’t think present data very well.
Firstly animation for animations sake – those tedious charts that render and bounce till they reach equilibrium. Imagine having to use it every day – time wasted. These should never be used.
Secondly, animated charts that are trying to be beautiful. During each World Cup game The Guardian plotted the density of tweets of key words relating to each match. I think these visualisations are very pretty, and in terms of aesthetics they really are works of art. During the World Cup I found myself quite addicted to watching the animations after games. But in terms of presenting the information in a useful way I don’t think they’re very good. It may just be me, but I actually find it quite stressful keeping an eye on particular topics and trying to do size comparisons between bubbles over the course of an entire animation.
The data could have been presented in a more accessible way using a line chart of key topics (say the 10 most popular) with annotations along the X-axis for important moments.
Being pretty and being informative are not mutually exclusive for data visualisation. Unfortunately there will be times, particularly when information is very dense, when you may have to choose between them. For my part in the contexts Yellowfin is usually used in I would always choose an informative chart over a pretty one if the tradeoff had to be made. It’s very easy to get sucked into making a pretty chart for your own gratification, but always consider what your target audience will make of it.
A good example of the latter can be found in Glen’s recent article about pie charts. I took issue with it for a couple of reasons: partly because I’m part of the "other" segment, but also because I don’t feel like pie charts were the ideal way to show this dataset. Sure the pie chart was pretty but it did obscure the detail of the data set.
Pies are one example but animated charts are the ones that I want to focus on. Before I begin let me say I don’t think all animations are bad for information clarity. Used sparingly to provide a sense of time they can be great.
When we were testing the chart animations in Yellowfin we used a dataset which contained the population of each Australian state at the start of each decade. The upper bound of the Y-axis was hard set to the highest value in the entire dataset, and the year was used as the controlling field for the animation. This was fantastic to watch, you could see when states went through periods of growth, when states stagnated and so forth. Although the data could have been presented using a line chart with a series for each state it wouldn’t have been nearly as absorbing.
The same effect can be seen on the Google Public Data Explorer, although I think the descending order makes the data harder to follow it’s good to watch as well as informative.
Now for animations which I don’t think present data very well.
Firstly animation for animations sake – those tedious charts that render and bounce till they reach equilibrium. Imagine having to use it every day – time wasted. These should never be used.
Secondly, animated charts that are trying to be beautiful. During each World Cup game The Guardian plotted the density of tweets of key words relating to each match. I think these visualisations are very pretty, and in terms of aesthetics they really are works of art. During the World Cup I found myself quite addicted to watching the animations after games. But in terms of presenting the information in a useful way I don’t think they’re very good. It may just be me, but I actually find it quite stressful keeping an eye on particular topics and trying to do size comparisons between bubbles over the course of an entire animation.
The data could have been presented in a more accessible way using a line chart of key topics (say the 10 most popular) with annotations along the X-axis for important moments.
Being pretty and being informative are not mutually exclusive for data visualisation. Unfortunately there will be times, particularly when information is very dense, when you may have to choose between them. For my part in the contexts Yellowfin is usually used in I would always choose an informative chart over a pretty one if the tradeoff had to be made. It’s very easy to get sucked into making a pretty chart for your own gratification, but always consider what your target audience will make of it.