The Next Decimal Place

Improvement Insights Blog

The Next Decimal Place

SPC software companies keep trying to gain some edge by tweaking the formulas for various statistics, but is the quest for the next decimal place of precision useful? If you’re trying to send a rocket to Mars? Absolutely. If you’re trying to tune up a business process, not really. Here’s why.

“Hi, I’m Jay Arthur, author of “Lean Six Sigma Demystified” and QI Macros [software]. I’m going to talk to you about what I call The Next Decimal Place.

“Because I’ve been in the software business, [I’ve seen] people and try and change formulas. Let’s say it’s Cp and Cpk. Well, we used to do the upper spec minus the lower spec over six standard deviations for CP, and guess what? That would give you one reading, but then you started using estimated standard deviation using the variance between samples, and so something like that, and that’ll give you CPK. The other one gave you PPK using standard deviations.

“Then about four or five years ago, Minitab started using pooled standard deviation to calculate these things. Well, the funny thing about it is when you change from an estimated standard deviation using Rbar or something like that (or Sbar over C4) whatever happens, then you change to a pooled standard deviation instead of like 1.33 for a CP you get maybe 1.333 or 1.335 or 1.326, right? It’s the thousandth decimal place that gets changed, though 1.33 does not change radically.

“So what are we doing? We’re making things more complex. But are we getting more value for the last decimal place, the next decimal place. Although that’s more statistically correct, you know, if you’re at 1.33 you’re kind of at the floor or manufacturing quality, right? You need 1.66 or 2.0 for a Six Sigma level process to minimize defects and returns and all the other stuff and associated costs with it, so that last decimal place does not matter, right? You’re trying to use an electron microscope or micrometer instead of a just a plain old micrometer or a yardstick (most people need a yardstick).

“I want you to be cautious when you hear people saying, “Well, blah blah blah blah blah… we use this special thingamabob that you can’t calculate yourself because it’s too damn complex.” Hey, well, why are we doing that? Well we’re trying to stop you from being able to do these things. Yes, I understand; however, it’s not going to change your end result if you’re a 1.33 and 1.335 is not going to make you look better. You’re still too close to the floor for manufacturing quality.

“So just be cautious, okay? The answer, is it good enough? Now, if I’m sending rockets into space and trying to hit Mars, sure; I need like twelve thousand decimal places. But for the average business, you do not.

“I’m Jay Arthur. That’s my Improvement Insight for this week. Watch out for the next decimal place.”