RE:salmon oil's

Andrew Strak (abstrak@auracom.com)
Fri, 6 Nov 1998 09:26:10 -0400

Thank you, Peter, for the explanation. It was very helpful in understanding
the issue and I am quite surprise to see so much variability of fat content
from one salmon to another within the lot of someting which ones would think
be a reasonably standarized though natural product. I do not question the
importance of fat content for smoking at all. I was trying to say that
knowing only mean values of inherently variable critical attributes
characterized by a large spread of distribution can sometimes lead to poor
production practices, where on average we are right while in 100%
individual cases being invariably wrong. Therefore it is always nice and
practical to know the measures of variability in addition to the measures of
central tendency and thank you for bringing this issue to the fore.

We have talked about sampling methods and large numbers of samples leading
to better precision of the estimates. Actually, I was a bit surprised by
the 'n' numbers in many cited by you sources, in my opinion a little bit
high if the purpose would have been just to estimate the mean and variance
of a single lot. Therefore I assumed they would be composite numbers coming
from several independed evaluations as it may be in the case of Bell's data
for example. But I understand that even within any particular lot or
sub-lot the individual variances were stronger than the variances between
the means of those lots.

Best regards,

Andrew Strak