Surveys

[Stephen Kobrin <kobrins@wharton.upenn.edu>]

While I am not a statistician, I use statistics frequently.  All of the
comments about sampling, confidence intervals and the like are certainly
relevant, but not the whole story.  There is a big difference between being
able to say precisely that 26.5% of all M6s have some sort of defect and
that something does not smell quite right in the state of Denmark.  I
certainly would like  an M6, but spending the $1900 or so a new one would
cost would be a big step for me.  While perfect quality control exists only
in one's imagination, defects should be very few and far between in Leica
production.  A defect rate of 10% or even 5% is much too high for Leica,
given the cost of the camera.  Furthermore, they have been doing this for a
while; the data are not for the first few models of a new run.  If you pull
a few marbles out of a big jar that is supposed to contain only white
marbles and the three of the first six are red, you certainly cannot say
with any certainty that 50% of the marbles in the jar are red.  You do
know, however, that there are red marbles where none should be.

Moral; don't throw the baby out with the bath water.  There is a reasonable
chance that there are problems with recent production that exceed what we
have a right to expect as "normal" under the circumstances.

Steve