Information Services and Products

SPC Trick 5, Know the "68, 95, 99.7 Rule" - A Wayworld Tutorial
Back | Tutorial Content Guide | Next Page

There's something very important we can know about the Normal Distribution. I call it the "68 95 99.7 Rule". (It is also called the "Empirical Rule" and "standard deviation").
I call it the "68, 95, 99.7 Rule" because you should remember these numbers (and because many people simply won't go anywhere near standard deviation... usually some kind of bad experience in juniour high school).
The rule goes like this: In a normal distribution (a process that is running in "control"): · 68% of the results will be near the Average. · 95% of the results will be in the main body of the curve. · 99.7% of the results will be in the curve, edge to edge.
Here's a picture of the rule:
The "68 95 99.7 Rule" tells you how tight or loose a process is. It tells you how many of your units will be near the Average, which is hopefully right near your center Target.
Here's what this means: There's three chances in 1,000 (that's 100% - 99.7% = .3%) that the process will produce one measurement at the edge of the curve (far from average, or target). For the process to do it twice, the odds are 9 in 1,000,000, and three times the odds are 27 in 1,000,000,000!! So we know that if we measure very many results (like two) outside the bell curve , the odds are we have a process that is not in control, not doing what we want it to do. We know it right now, not days or weeks later when we find out the results from our customers.
If you get the "68, 95, 99.7 Rule", now you get "standard deviation" too. On the picture above, the distance from the center line (average) to the first dotted line is "one standard deviation", and 68% of the results will fall that distance from the average.
In this tutorial we'll use a little trick with the ranges between measurements (trick six) that gives us a pretty good estimate of the "68, 95 99.7 Rule" for a process in action. (SPC software will do this for you with a precise calculation of standard deviation).

Back | Tutorial Content Guide | Next Page - Trick 6, Find Ranges Between Measurements

1997-2011 © Wayworld Inc.