By Harry Rever, MBA, PMP, CSSMBB, CQM, CQC | Director of Six Sigma, IIL
I want you to stand up.
Yep, get out of your chair and stand up straight, put your right arm out to the side with your hand up at a 90 degree angle at the elbow and say aloud:
“I solemnly swear that I will no longer use pie charts. I will no longer use shaded area charts, and I will no longer use stacked bar charts. Furthermore, when making charts I will not be a ‘chartoonist,’ I will not make cutesy charts, nor will I make my charts difficult for my audience to understand. I promise to use the correct charts going forward, I promise to keep my charts clean, and I promise to make it easy for the reader to interpret any chart I use for now on. I do solemnly swear to uphold all of these resolutions.”
Now, you may be seated. Thank you.
No More Pie Charts!
Pie charts are the enemy. Actually, any chart which has the potential to add confusion or misinterpretation is the enemy but pie charts are right at the top of the list. Why, you ask? Well, let’s take a look at a sample pie chart. Oh the colors, the slices, the associated key. It’s like an amalgamation of pastels arranged in random order to drive the reader to a state of dizziness and despair. The horror! Well, maybe it isn’t that bad, but it’s close.
Pie charts are simply brutal.
First of all, the golden rule for any chart or graphic is it should be helpful. If the chart is confusing or hard to interpret, the purpose of the chart has been defeated. By their very nature, pie charts are risky graphical investments.
Is this chart “easy” for the reader? Does a person immediately know where to go first or what the point is? Obviously not. How about that key? Which color goes with which slice? Hmmm, let me spend my time figuring out your chart. I don’t think so. What happens if someone prints the chart in black and white? Could that happen, printing and making copies? Let’s take a look.
The black and white version is even worse.
Pie charts simply spell disaster. My advice, just do not use them. Maybe, maybe if there are just 2-4 categories would I use a pie chart, but only rarely.
Instead, use the illustrious, the infamous, the easy to read and easy to understand Pareto chart. Same data, easier to interpret.
The Pareto chart is nothing more than a bar chart arranged from biggest to smallest. The Pareto chart separates the vital few from the trivial many. It makes no difference if it is in color, black and white, or if the reader happens to be color blind.
You simply focus on the big bars. How easy can it get? For best results, make sure the bars are horizontal like in the chart above. It’s easier for the reader. Vertical bars often require the reader to turn their head sideways to read the category. We don’t want anyone getting hurt now, do we? Remember, make it easy on your audience. Charts should help you, not cause you to have to explain what they mean.
To make a horizontal Pareto chart in Excel, simply do the following:
- Arrange your data in descending order
- Highlight the category and count
- Open Chart Wizard
- Select “Bar” chart and make the chart
Here is a nice comparison of the two charts. Which one do you like better? Be careful, don’t look at the pie chart too long. It may cause dizziness and disorientation.
No More Shaded Area Charts!
What in the world is a shaded area chart? Well, let’s take a look at one.
It looks more like a panoramic view of the sun setting on a mountain range than a chart. How do you interpret it? Could it cause confusion? Are yellow issues smaller than red issues because the width of the yellow band is smaller, or, since it is on top of the red and orange, does that mean it’s the biggest category? I have no idea and neither does your audience. But hey, the chart looks cool doesn’t it? It’s colorful, it’s wavy, and it’s soothing. Oh sorry, I was becoming a “chartoonist’ for a moment.
The basic rule of thumb is pretty simple:
If you are looking at categories of information like reasons for complaints or types of errors, use a Pareto chart.
If you are looking at something over time, use a line chart or a control chart.
“Bars are for buckets; lines are for trends.” You can quote me on that. Let’s take the same data and make a control chart for the red system issues and the orange system issues.
The control charts below are much more informative than our panoramic mountain chart shown above. We now have a better understanding of trends, normal variation, special cause variation, and the stability of our process.
Control charts are almost always the preferred chart when looking at data over time. Control charts not only chart the measure, but also include the average line as well as upper and lower control limits which are three standard deviations from the average line. Anything between the control limits is viewed as normal variation within a process and anything outside the control limits is unusual and is worth investigation.
Normal, or common cause, variation does not imply the results are good enough; it only means its normal output for the current process. If you don’t like the results, change the process which gave you those results. But, of course, that is another topic for another day.***
No More Stacked Column Charts !
Now these may truly be the worst charts ever. Really, they are that bad. Talk about confusing. These are the worst. At least the shaded area chart looked like a mountain range.
These stacked columns are really bad.
It’s like a geological survey of the layers of a canyon wall. I don’t even know what to say – the worst. Yet, people use these charts all of the time. Again, is it EASY to read, or is it potentially hard for the recipient of the chart? Could mistakes be made when interpreting the results? Why risk it? I think I see a dinosaur bone over there in week 9 in the orange layer; do you see it?
Charts are supposed to be helpful; not require your audience to be members of Mensa. My advice is to simply never, ever make a staked column chart. Well, I guess if you wanted to make your boss’s job harder, perhaps confuse him or her, you could use these charts as part of a passive aggressive “get even” strategy. But, I don’t necessarily recommend that approach.
Charts are your friends. Keep them simple. Make them understandable. Avoid the tendency to become a “chartoonist.” Charts should help you communicate effectively. If you have to explain them; they’ve failed you. So remember our motto: “No More Pie Charts.” Come on, stand up and shout it out loud, my friends. Let your neighbors hear you. “NO MORE PIE CHARTS.” If you want, send me some examples of some just “awful” charts you have to deal with. I’ll post the winner in a future article.
If you have further questions or comments, Harry can be reached at firstname.lastname@example.org.
Harry Rever is Director of Six Sigma for International Institute for Learning. He is a dynamic presenter and practitioner of Six Sigma and Project Management with an innate ability to teach the concepts of quality improvement in an understandable and more importantly, applicable manner. With over twenty years as a project manager, process improvement consultant and trainer, Harry has numerous examples of what works (and what doesn’t) when managing projects and applying statistical process improvement concepts.