My apologies to all, and my thanks to Rob, for the catch of my mistake on
stdev vs stdeva vs. stdevp.
The reason we were using std dev. was that we were running x bar/ s
charts instead of x bar/ r charts.
I selected x bar/s based on recommendations that s was a more efficient
(statistical jargon) indicator of variation. This is confirmed in the AIAG
spc manual. In an SPC class long ago, I was told that the best motivation
for the r chart vs. s chart was simplicity in calculation; assuming that the
charting was automated, that s was a better measure. (Shewhart didn't
have Excel!)
I confess that I have never really tested the assumption in real life. It is
usually hard enough just to get accurate, timely data, and none of my
clients ever challenged me on statistical issues.
Has anyone else used the x bar/s instead of x bar r? In particular, did
anyone find that it makes a real world difference?
Regards,
Brad
stdev vs stdeva vs. stdevp.
The reason we were using std dev. was that we were running x bar/ s
charts instead of x bar/ r charts.
I selected x bar/s based on recommendations that s was a more efficient
(statistical jargon) indicator of variation. This is confirmed in the AIAG
spc manual. In an SPC class long ago, I was told that the best motivation
for the r chart vs. s chart was simplicity in calculation; assuming that the
charting was automated, that s was a better measure. (Shewhart didn't
have Excel!)
I confess that I have never really tested the assumption in real life. It is
usually hard enough just to get accurate, timely data, and none of my
clients ever challenged me on statistical issues.
Has anyone else used the x bar/s instead of x bar r? In particular, did
anyone find that it makes a real world difference?
Regards,
Brad