Inspired by the Academy Awards, I have decided to create my own new prize for the graphic display of data: The “Spectacularly Dreadful Graphic Display Award” (hey, I just call them like I see them).
Envelope please….and the nominee is the following dashboard that was sent to me by a subscriber to this Newsletter.
I have no clue about what decisions could ever be supported by this dashboard — well, that is not true, I do know — NONE.
But, the “Spectacularly Dreadful Graphic Display Award” is awarded to the: “wind tunnel, we are not in Kansas anymore Toto, ice cream cone graphic” or “cone” for short.
The cone is an impressive example of the most common design mistakes made in all forms of quantitative data presentation — the inappropriate choice of media.
Labeled “Monthly Length of Stay” each band on this graphic represents a month with the accompanying data listed as: Jan 09:3.15/8.4% (for example).
There are twelve months of data listed — January through December (but note that three months are missing a label). I am assuming (based on experience and guessing) that the number after the month is the Average Length of Stay (ALOS) expressed in days. Following the ALOS is a percentage, but of what I have no idea (experience and guessing won’t solve this one). Is this an Occupancy Rate? It may be, but I simply can’t tell.
Assuming I have guessed correctly about what is being reported and I read the fine print — I discover that the ALOS for all twelve months ranges from 3.05 to 3.19 days.
But look at the graph. The visual message is that January’s Length of Stay is “whoppingly” (a technical term) bigger than December’s. And that it has been steadily decreasing since January, resulting in this perfect and flawless cone shaped graphic. (Do you believe in magic?)
The viewer of this dashboard is led to believe that someone had better investigate what the heck happened between January and December to cause the rapidly decreasing and vanishing ALOS (where have the patients disappeared to?)
And they would be wrong.
Consider my quick and simple redesign of the information (now labeled 2010 ALOS) with the addition of the previous year’s data (2009 ALOS — which I made up for this example).
The use of a simple line graph shows how the ALOS has changed over time — notice how the correct message is contained in the shape of the values. When you view this graph it is clear that the 2010 ALOS (i.e., the cone data) is essentially flat at about three days per month. Compared to the previous year’s ALOS it is about one day longer each month.
Additionally, the ALOS was steadily declining in 2009, but at the start of 2010 it increased and has stayed at about three days per month. This is an entirely different (and correct) story than the one imparted by the cone graph and will lead to an entirely different (and correct) investigation by the viewer.
Data visualization and dashboard design is not about using all 64 crayons in the box or creating fanciful designs.
Rather, data visualization is about using our largest sense — sight — to see and understand the most important trends, patterns and variances needed to think and reason and make informed decisions.
Thanks to the reader who submitted this dashboard — I encourage others to do the same — and while you are at it maybe you want to create your own category of award and work on your acceptance speech.
Look out Ms. Streep and Mr. Clooney, we’re taking over the show.
P.S. If you want to see this dashboard in action (and others that will most assuredly be receiving nominations) click here.