-
-
Marketplace
-
Channel Resources
Articles from this Site
-
One brand, one Web site! DM Review is now the home of all the content you're used to at BIReview.com and much more. If you are registered at BIReview.com, you're already registered at DM Review. If not, take just a moment to sign up for all the free services we have for you at the new DMReview.com.
Beware the Gratuitous 3rd Dimension
Speaking at a recent medical conference symposium on risk communication, I decided to use examples from recent issues of the host society's own journal. Perusing the most recent issues, I found a number of images that presented substantial risk to the effective visual communication of the reality inherent in the research data.
Without embarrassing the authors or journal, there is much we can learn-as they graciously did-from reviewing some of the problems found. Figure 1 is a close facsimile of one of those published images.
Figure 1
The caption began: "This bar graph shows the distribution of individual patient residuals ...." Below six categories of patients across the x-axis, a data table listed the values that should be represented by bars rising up the y-axis scale. Your eyes are not playing tricks on you. Amazingly, there are no bars! In the final pre-press review, no one caught the fact that there were no bars in the bar graph because there was so much of what Ed Tufte calls "chart junk," including extra text on the back plane of the original image. There was so much "ink" that a casual observer felt there really was a chart in the visual field, despite the fact that there was absolutely no graphic "data ink."
When contacted for permission to discuss this image at the medical society's symposium, the authors graciously agreed, and also offered to send the chart that should have been published in the article. A close facsimile is shown in Figure 2.
Figure 2
My last column (March 2006 BI Review) looked at "skyscraper charts," or 3-D bars on a 2-dimensional base-plane. Here we explore bars on a 1-dimensional base-line, where a "gratuitous" extra dimension has been added to create 3-D bars and a 3-D base plane. Gratuitous means "free" here -free of information, free of meaning.
The rationale for creating an image like Figure 2 in the medical journal was to give the reader a sense of the comparative values of the heights of the bars, and a sense of the population distribution. Imagine an image such as this presented at a meeting you are attending. The speaker is making a point comparing the second and fourth bars from the left. You look quickly, and try to sense the heights from the scale and judge how they compare. Some of your colleagues perceive the fourth bar to be well over five times larger than the second. Some perceive it to be less than four times larger. Many other perceptions are scattered in between. Of course, you rarely discuss your perceptions; you only discuss the business, or research, implications of what you perceive to be the truth.
To see how this could occur, let's look at the second bar and determine its height. If your eyes focused on the back of the bar, right up against the "1,000" line, you would have a lot of company amongst your colleagues around that conference table. But then, others may notice the front top-left corner and how it prominently stops only 80 percent of the way up from the bottom of the scale to that 1,000 line. They may quickly get the feeling that the value is around 800.
A close-up of this bar is shown in Figure 3. Very few people in your meeting would probably realize the bar is sitting out in front of the back plane, the plane with the scale. To read this bar against the scale you actually have to let your eye run back along the diagonal shown between the arrows at the base of the bar, and then extrapolate that same projection from the top of the bar back to the back plane. Only then would you come up with the value that the author conveniently put in the data table below the chart.
Figure 3
In the blink of an eye, many people gain the misperception of a value of 800, which is only 63 percent of the actual value of 1261. Or, maybe in that blink they gained the misperception of a value of 1,000, which is only 79 percent of the correct value. Yes, they can read the Arabic numerals in the data table, but their misperception has been formed in that blink, and they are on to thinking about other issues and the implications of the wrong perception that they have begun to trust.
The cause of this misperception is due solely to the gratuitous use of a 3-D projection. Thrown in for "free," the third dimension adds nothing. In fact, it is a major detriment to effective communication. Almost every viewer has a misperception, yet the amount varies widely among different individuals.
The consequences of this phenomenon go even further. We saw that many people could easily perceive the second bar to be about 461 or 261 units lower than actual, if they focused on those left corners. These differences are "constants" in this chart, caused by: the angle of the 3-D projection, the distance the bar stands in front of the back plane, and the reality of human perception. For the fourth bar, reducing the true value by these "constants" means the bar is perceived to be around 4357 or 4557-90 or 95 percent of its actual value. Thus, even for a given individual, the perceived distortion in the height of 3-D bars varies across the chart. The distortion is greater for short bars than for tall bars.
Those who perceived the second bar to be 800 may perceive the fourth bar to be 5.4 times larger [4357/800]. Those who perceived 1,000 may perceive the fourth bar to be 4.5 times larger [1000/4557]. And those who did the arithmetic would know that 4818 is actually only 3.8 times larger than 1261.
We observed earlier that a rationale for creating this image in the first place was to give a sense of the comparative values of the heights of the bars. Now we can see that the use of the gratuitous 3rd dimension has completely discombobulated our ability to make intelligent comparisons of these values. It is hard to believe there is any redeeming virtue for 3-D bar charts.
At the next presentation where you see images with a gratuitous 3rd dimension, realize that you, and everyone else in that room, are subject to building misperceptions of the data. Furthermore, your colleagues probably have different misperceptions than you, thus creating a terrible environment for effective, efficient, decision-making, and a great waste of organizational resources.
For the MDs and PhDs at that medical conference, their responsiveness to issues in graphicacy, such as this, was impressive to behold. Now, business leaders and Business Intelligence practitioners need to focus on graphicacy issues as well, educating all members of their organization, to reduce the "risk" in their communication.
Howard A. Spielman, M.B.A., Ph.D., President of Management Semiotics International Inc., can be reached at HASpielman@ManagementSemiotics.com.
For more information on related topics, visit the following channels:
