Anyone who has heard me speak about charts knows that I’m not a fan of three dimensional (3D) charts. Here are the reasons why.
All charts can present problems in conveying information if used improperly. What makes 3D charts unique is that their major problem is inherent in the chart design itself — namely, the confusion induced by the depth of field effect.
Conveying a third dimension on a two dimensional surface creates difficulties for the eye and the brain. Just look at some of the fantastic optical illusions that prey upon the brain’s bewilderment when confronted with a 3D simulated image on a 2D plane. When you try to get information from a three dimensional chart, you have to use mental gymnastics to make allowance for the depth of field effect. My first rule of chart design is that if you have to use any mental gymnastics on a chart to get the information you want, then it’s not a good chart.
The first problem, evident even in a simple 3D clustered column chart (one or more data series all in the foreground), is that the brain automatically estimates the values of the columns from the grid in the background. Unfortunately, this gives a false reading since the actual height of the columns differs, sometimes appreciably, from the value read on the grid in the background.
In a true 3D column chart (with series data presented from foreground to background, see chart below), the confusion is even worse. First, observe the problem noted above. The tallest blue column visually aligns with the gridline for 50 in the background. Yet its actual value is 65.
A second problem is that it is difficult to compare the values of the different columns. The tallest blue column and the tallest green column appear at exactly the same height on the chart. Obviously the blue column is a higher value since it starts at a lower point on the chart but it is difficult to determine with any precision how much higher it is. You could, of course, use data labels to put the value at the top of each column but there are two problems with this, especially with multiple data series charts. First, if you label all the columns in a three dimensional chart of more than one data series, the chart is overly busy. Second, and more importantly, although the labels clearly show that one column is numerically greater than the other, visually there is poor confirmation of this. If you feel compelled to use data labels to overcome the visual confusion inherent in a 3D chart, you would be better off using a non-3D chart or even a simple table of values.
A third problem is data dependent. In a multiple series 3D column chart, a higher value column in the foreground may totally obscure a lower value column in the background, resulting in missing data. Note in the chart above how the tallest blue column totally obscures a green column in the background (the value for the Night shift for Ward 106).
Finally, I find it much more difficult to identify patterns and trends in a 3D chart, especially one with more than one data series. I have to work at it — something that well designed charts don’t require. One of the main purposes of displaying data in a chart is to facilitate the identification of patterns and trends and a non-3D chart does a much better job — at least for me.
By the way, most experts who write books on chart design agree that 3D charts should not be used.
Dr. James M. Smith gives lectures at facilities/colleges and conferences across the country showing healthcare staff how to analyze and present data more effectively. His belief is that data presented as data are meaningless, but data presented as information are priceless. Information on his “largely bullet free” presentations may be found on his website.
Prior to becoming a consultant, James served the Quality Management Officer for Veterans Health Administration (VHA) hospitals in the New York/New Jersey metropolitan area. He has a doctorate in Experimental Psychology from Fordham University and has over 35 publications in professional journals.