Improvement Insights Blog

After a hundred years, people still aren’t using control charts. Here’s why:

Download my free eBook, Agile Process Innovation-Hacking Lean Six Sigma for Results.

“Well hi, I’m Jay Arthur, author of “Lean Six Sigma Demystified” and QI Macros [software].

“Control charts just turned a hundred years old, but if you look at the adoption rate for control charts it’s been pretty flat. You see it mainly in manufacturing, you don’t see it much in healthcare.

“We attend two big healthcare quality conferences: At the Magnet conference, out of hundreds of posters, 89% use line and bar charts (which I consider to be last-century technology). It’s the Dumb and Dumber of charts: line and bar charts. Then we went to the IHI (Institute for Healthcare Improvement) Conference and 69% were line and bar charts. I’ve been measuring this since 2015 and it hasn’t changed, so people are actively ignoring technology that can help them. Now, if you look at other things like TVs, cell phones, PCs, the adoption rate is just {woo sound effect} exponential, right? So what’s going on with control charts?

“I thought maybe we should do a root cause analysis; that might be fun.

“So my first big bone is going to be ‘Jargon’. We use words… you know, Shewhart wanted to control manufacturing quality, so he called it ‘control charts.’ Now, in America, nobody likes control, right? So the actual word ‘control’ causes problems. Now, some people call them Shewhart charts, but unless you know who Shewhart is, so what, right? That’s like saying ‘Beyonce charts’ or something. You know who Beyonce is but you don’t know who Shewhart is, so you might like a ‘Beyonce chart’ better… I don’t know.

“Anyway, Dr. Donald Wheeler calls them ‘process behavior charts’… well, ‘behavior’… right? And then ‘statistical’; somebody put statistical in front of it: ‘statistical process control.’ I’m telling you: people are afraid of math, they are especially afraid of statistics, and so it makes everything sound super complicated. I’m promoting that we change SPC from ‘statistical process control’ to ‘smart performance charts’ because everybody wants to improve performance. They’ll go out and buy Fitbits and Apple watches and everything else so they can figure out what’s going on in their life.

“So there… why don’t we change the jargon? Change the jargon to ‘smart performance charts,’ all right? That’s a countermeasure. I’m suggesting maybe we need to rethink all this nonsense we’ve been saying for 100 years because it’s not working.

“All right, now let’s talk about training. In training, very often people teach run charts before they teach control charts. Now why is that? Because in the old days they were trying to ease people into it because you had to calculate everything manually, so you would calculate a center line and then draw your charts. Then you had to learn a different set of run rules, because run charts have different run rules than the than the control charts. Now you’re teaching people stuff they don’t need to know beforehand.

“The problem as I see it is once people get used to a run chart they never change; you get to 15 points and you should shift to a control chart… they don’t. They just stay with a run chart. I’ve seen examples of this at all these conferences where they have improvement posters and here’s a nice run chart and guess what? Nada.

“So let’s stop teaching run charts; just tell people to ignore the control limits until you get 15 points. Just start doing control charts and then that will make everybody’s life easy. In the QI Macros we have templates where you can just drop data into a control chart template and so once it gets to 15 points then you can start to believe your limits.

“Now one of the other things that’s nasty in this whole process is there’s what’s called ‘a priori’ (know before you do something) and ‘a posteriori’ knowledge (know after you do something). The way we’ve always taught it (because you had to do it that way) was that you had to have a decision tree: You had to decide if it was attribute or variable, and how many samples and subgroups and blah blah blah… Oh my gosh! People don’t think in decision trees; they don’t think in decision trees.

“They used to teach formulas for all of these charts which nobody can calculate accurately. About the only thing you learn is it’s really complicated and you’re not going to get it right the first time. Then you have to know the run chart rules… I mean the control chart rules for control charts. That’s a whole different set of rules, so stop teaching run charts and run chart rules. Start teaching control charts and control chart rules.

“Here’s the deal: the QI Macros are the only software that ask you to select the data first. Because you select the data first, (back in 2006 I built that decision tree into the software) the Control Chart Wizard will pick the right chart for you based on whatever data you select. It doesn’t matter if it’s an X chart or an attribute chart or a c chart, p chart and u chart, it’ll go pick the right chart for you so you don’t need to know decision trees. The formulas are all built into the software. All of us use the same formulas – get over it, all right?

“So that’s built in and the QI Macros will automatically look for all of the control chart rules that are violated and highlight those in red so that you don’t have to know all that either; the software will do that for you, so you can start doing. Then later, I found it’s much easier to come back and teach you how the decision tree worked or what control chart rule was violated. It’s much easier ‘a posteriori’ after you’re doing than it is before; if you do it before the doing, everybody thinks it’s way too complicated… just nasty, all right?

“Then there’s also too many charts, right? Everybody tries to teach everything, so an XmR, I-mR chart, an XbarR chart, an XbarS chart, a C, NP, P and U chart and sometimes a G and a T chart… blah blah blah blah blah! This is where people get lost and they throw up their hands in disgust. You’ve exceeded their ‘seven plus or minus two’ limit of whatever.

“The great thing about attribute data is you can convert those… numerators and denominators into ratios, and then you can draw an XmR chart. Dr. Donald Wheeler calls that the ‘Swiss Army Knife’ of control charts, so if we can get all the data in some sort of format where we can just start using an XmR chart, that’d be great. Wheeler really likes using the actual data because those limits are based on the data but things like the c, mp, p and u chart are all based on theoretical limits, they are not based on the data. You know data doesn’t always behave the way the theory says, right?

“If you can convert it into numbers that you can put into an XmR chart it’s much more useful, at least that’s my observation. Later you can come back and learn another type of chart or another chart and another chart, but master the XmR chart to begin with.

“The other thing that’s a problem out there is software. Line, bar and pie charts are readily available. They’re umpteen years old; they’re from the last century. Maybe we ought to grow up and get some new tools, some automated tools like QI Macros. But you can draw a line, bar or pie chart in Excel, Google Sheets or whatever the heck it is, but they’re still the Dumb and Dumber of charts. They are not going to help you achieve excellence, they are not going to get you below a 1% error rate, and in healthcare you want to get below 1% error rate because you want to stop killing one person out of every hundred unnecessarily. (That’s actually a real statistic… I apologize.) So all you need to do is spend a little money, buy the QI Macros; you can buy [software from] somebody else if you want, you can spend a lot more money if you want; I don’t care, but start using the Tools of Quality.

“How are we going to get this control chart level of adoption to start to move up even a little bit? It’s not going to go exponential – that’d be sweet because then everybody would be using control charts to manage everything and I think we’d have a lot smoother world out there if we did that.

“So that’s my Improvement Insight for this week: If you look at the root cause of things, we can change the jargon, we can change how we teach things. Let the software do all that stuff for you. Once you’ve used the Control Chart Wizard for a little while it starts to embed that decision tree in your brain, and then I can explain it to you and it won’t be a struggle. Also, maybe we ought to just rethink how we’ve been going at this, because they’ve been around for a long time and nobody’s using them except in manufacturing in general. Remember, in healthcare 69 – 89% of them are still using line and bar charts… that’s like using bloodletting or something from the 1800s. You know: “Let the evil blood out…” No! Come on! Grow up, all right?

“Let’s go out and improve something this week. Let’s start using the Tools of Quality, and let’s use jargon and software that makes it super easy.”