• Votes for this article 99 people voted for this
  • Dashboard Insight Newsletter Sign Up

The Case Against 3D Charts in Dashboards

by David Sprague, Visualization Research, University of VictoriaWednesday, August 10, 2011

Faux 3d

There seems to be a strange disconnect with many dashboard designs.  Most dashboard data is 1 dimensional (using gauges and pie charts) or 2 dimensional (line graphs and scatter plots) and yet many people seem to want widgets that look three dimensional (which I term faux 3D visualizations).  Essentially, some people want an extra dimension that displays no information.  Three dimensional visualizations may be justified when the data itself has an inherent 3D spatial meaning, as is the case for many anatomy and fluid dynamics visualizations.  However, dashboard data is typically not spatial, and addition of a 3rd dimension can actually obscure the data.

So what benefit is there to using 3D charts?  Some people argue there is an improved aesthetic appeal.  Others say the faux 3D looks more professional.  However, using faux 3D charts reduces the readability of the dashboard without most people knowing it.  The presence of faux 3D charting widgets on a dashboard demonstrates a designer’s propensity for valuing marketing glitz over readability.

Why Faux 3d is poor visualization

Arguments against 3D charts are not new.  In fact, I feel like I am simply parroting Stephen Few or Ben Shneiderman as I write this.  Faux 3D charts and graphs have been demonstrated in the research literature to negatively affect the accuracy and speed at which one can interpret data.   The image below demonstrates how the addition of a 3rd dimension can cause distortions in the perceived size, and thus the value, of the data.


3D pie charts tend to skew the size, and thus the perceived value, of the data.

Notice in the upper right image how the slices of the 3D pie representing 17 units are not the same size, whereas the 13 and 17 look equivalent?  Even if we take a 2D pie chart and distort it to match the shape of the 3D pie chart (bottom), you’ll notice that the sections maintain the correct relative sizes.  It is the perspective projection that is causing distortion.  If you must use faux 3D charts, minimizing the 3D effect will reduce these interpretation errors. 

In most cases, dashboards are used for monitoring and should be designed for rapid comprehension, particularly in business contexts.  Dashboards are typically designed for repetitive, short duration use instead of long, in-depth analysis.  Effective dashboards need to be clear, easy to read and accurate for this reason.  So why sacrifice accuracy for aesthetic appeal?  If some of your client base is aware of the faux 3D chart perceptual issues, how does a 3D chart look more professional?  For a visualization researcher like me, faux 3D charts are essentially the black velvet paintings of the visualization world.  You may still enjoy the look of a 3D pie chart, but others may not share your perspective.

For your consideration

I do not expect faux 3D charts to disappear any time soon.  My hope is that some of you will begin to see that not all dashboards (particularly 3D dashboards) are designed for maximum performance.  The flashiest kitchen gadgets are not necessarily the most effective tools for cooking.  Similarly, faux 3D charts are not as effective as their 2D counterparts.  Ultimately, identifying issues associated with different chart designs is a critical first step to creating effective dashboards.

Tweet article    Stumble article    Digg article    Buzz article    Delicious bookmark      Dashboard Insight RSS Feed
 
Other articles by this author

Discussion:

Trent Ballew said:

While extremely skewed 3D charts (as in your example) are definitely a faux-paux, there are types of "asthetic" 3D treatments that are acceptable: a slight drop-shadow behind a chart or a 3D skew of 5% or less can add appeal to a chart without distorting the proportions or altering the message behind the chart.

Regardless of what is theoretically "pure" as far as 3D is concerned, there are also realities of what will sell. I was recently at a user conference showing my company's wares against a competitor who was all flash and no substance. We followed good visualization design techniques (no 3D, no pies, no gratuitous colors, etc) while our competitor broke every rule in the book (3D pies, 3D cones, guages, colors and distracting bouncing graphics everywhere). Customers would come up to us and say they liked our product because it was more useful but it just didn't produce dashboards that were as "pretty" and "exciting" as our competitor's. I quickly did some slight modifications to some of our examples (added slight drop-shadows and minimal 3D treatments) and got rave reviews.

Bland, non-distracting charts are theoretically superior and one should still strive to embrace the "minimalist" principles, but the market sometimes requires us to bend the rules to compete.

Alexander 'Sandy' Chiang said:

@Trent - I was a product manager for a dashboard company and I totally get your point. I had a very candid interview with Stephen Few and he argued: if the data is relevant, who really cares if it is flashy? I opine that packaging and presentation do matter. Apple is a great example but they do it with sleek and minimalistic design principals. However, as you alluded to, dashboards can be stylistic and practical at the same time. It's just more work to find that balance.

Naomi Robbins said:

Nice post! Faux 3D bar charts are at least as bad. In Creating More Effective Graphs, http://amzn.to/qHpaFq, I show that besides distorting the data, the way to read these charts depends on the data used to draw them. I find that to be unacceptable.

Mark Stewart said:

Your example is quite simple. A bar chart sorted in descending order would have provided much more readability.

Alexander 'Sandy' Chiang said:

@Mark - I agree, a bar chart generally makes the data more readable. However, pie charts are better at showing ratios than bar charts (unless you use a bar chart with a scale of 100%).

Site Map | Contribute | Privacy Policy | Contact Us | Dashboard Insight © 2017