🚀 go-pugleaf

L. H. Garlinghouse posted (in part):

>...there are
>some special cases of unitlateral specs which are naturally bounded
>by zero.  For example, flatness where you are tracking flatness by
>TIR (Total Indicator Reading aka FIM, etc).
While this is certainly true, many such characteristics are quite well
modeled by the Folded-normal or Rayleigh distributions. These
distributions are well-characterized and a Cp, if necessary can be
readily derived without recourse to the specious assumption that the data
are normally distributed.

>... you can invoke
>Wheeler's "Empirical Rule" part 3 (see "Understanding SPC" by Wheeler
>and Chambers, p 49 ), which says that  99 to 100%  of the data will
>be located within 3 sd of either side of the average.
Such an invocation is quite unnecessary. If the data are unimodal and
apparently non-normal, Chebycheff's inequality will serve quite well to
make a reasoned estimate of the upper (or lower) limit of the
distribution. If the data are bimodal (or worse) then some serious,
additional analysis must take place before reporting any capability index.

Kwenda mzuri,

Frank Isackson
re:SOLUTIONS
205 West Walnut Avenue
Unit A
Monrovia, CA 91016
(626) 358-1340 (voice)
(626) 358-1322 (fax)
FIsackson@aol.com