Improvement Insights Blog
Control Chart Rules
Statistical textbooks disagree on what constitutes Control Chart Rule 2 – Zone A: 2 of 3 points above/below 2 sigma. Which one is right?
Download my free ebook, Agile Process Innovation – Hacking Lean Six Sigma for Results.
“Hi, I’m Jay Arthur, author of “Lean Six Sigma Demystified” and QI Macros [software].
“Got a call from a customer this week. He was all upset because of control chart rule two, which is about having points above two sigma or below two sigma (two out of three).
“Well, guess what? Some people in their books say those two points have to be on the same side of the center line. And other books say that it can be on either side of the center line. And if you think about it, if you got one that’s wacky way up high and one that’s wacky way down low, those are both outside of the center [zone], which is… two thirds of them should be in the in the center, right? And so this is… (I guess it’s 95% [in the center two zones.] 66% in the center [zone], and then between minus two and two sigma, that’s 95 point something percent.) So very few should be outside [two sigma.]
“I always think about it like this: If there’s one point high and next point’s way low, that might be tampering. We said “Oh, that’s way off,” and then we tampered with the thing and threw it way off. That might be a type of error [that] Deming calls tampering. So who are you going to believe? Right?
“In the Healthcare Data Guide, they say “one on either side.” There’s another textbook that says the same thing. Then other textbooks say “It has to be on the same side of the center line.” This is, again, a “quarrel” of statisticians. The plural noun of statistician is a “quarrel.”
“And so who’s right? I don’t know. But at least the one where it’s on both sides makes a little more sense to me and predicts certain kinds of behaviors that would involve tampering.
“So that’s my Improvement Insight for this week. Don’t get overly concerned about all this. Look at what it’s telling you. Investigate. Maybe it is a problem, maybe it isn’t. But without that investigation, you can’t tell.
“Let’s go out and improve something this week.”
