Almost embarrassed to ask... SPC UCL/Data Range

K

krichmond0306

I should know the answer to this off the top of my head but I am lost right now and need an explanation.

I have a data set n=218, Mean=1.6927 anf StDev=3.0210. To me this means that 95% of my data lies within the interval 0 and 7.7347 (mean+2*Stdev), a quick count confirms this. Yet a Minitab control chart gives me an UCL equal to Mean+2.33*(StdDev/2)=5.98.

Why? I know I should understand this but I think I am over thinking it.
 
Steve Prevette

Steve Prevette

Deming Disciple
Leader
Super Moderator
Aside from reaffirming my distrust of Minitab, the odd thing is that 2.33 standard deviations is the 99% point on the normal distribution (one tail). Since the LCL is well below zero, is it trying to do some sort of probability based one sided control limit? Though it makes no sense to divide the standard deviation by 2 as far as I can tell.
 
K

krichmond0306

Some relief, I thought I had lost it. It made no sense to me at all and if I understand you right I should be confused.
 
K

krichmond0306

I'm not sure I can post the data or not but I used an I-mR chart, the data set was a years worth of individual measurements representing a distance from a datum, basically a measurement of how "off" of design you are.

The data was supposed to be both + and - values representing whether you were "left" or "right" of the design but I found out that was not done consistently so I had to use the Abs value of all the readings. 0 is my target meaning the 'critical' point matches what was called for by the model (design).

So the data was supposed to be two tailed but I had to turn it into one tailed and I think that is what got me all mixed up, or Minitab mixed up.

I just want to make sure I explain it correctly to others when asked or adjust my run chart accordingly.
 
You must log in or register to reply here.
Top Bottom