Re: M6 problem survey

[ABreull@aol.com]

In einer eMail vom 05.09.1997  02:26:05, schreiben Sie:

> how closely any
>sample is likely to reflect the frequencies of the whole population
>depends on the sample size.  The smaller it is, the less confident one
>can be of its similarity to the whole.  And twenty-some units out of
>several hundred thousand is a very small sample indeed.

Thanks for covering the response bias, and sample sizes. Further, there are
several additional effects, which have influence on the interpretation of
results, e.g.

1) the interpretation of the results depends on the effect size also, e.g.
you don't need a medium to large sample size to find out, that strychnine is
dangerous to persons's health.

2) the revealed event of broken M6s might follow the law of large numbers,
meaning (also): there is a large (absolute) _number_ of broken M6s, if the
number of produced M6s is large, even if the garbage _rate_ (percentage) is
low.

3) the revealed event of broken M6s might agree to the law of rare events,
which usually indicates that a certain event is related to/ takes place with
a certain period or person (or similar), but follows a different theoretical
distribution than the "normal" one. Thus event will happen again with p=1.0
(absolute security),  even if you would exclude the special period or person
from future production or investigation, as long as the sample is "open" and
you don't change the situational conditions.

4) you don't generalize from sample results anymore since middle of the '60s,
specially not from the observation of rare events, e.g. you would not
conclude from the color of three _black_ swans observed from your local lake
onto the color of all swans on earth.

5) the _subjective_ importance, significance or meaningfullness of the
observed result depends on different criteria than the statistical approach,
e.g. some persons would take into account even a propability of p=0.90
against them or of things going wrong, if the event has a strong enough
meaning for them [example: you might marry a certain person, although the
chances might be  0.90 (~90%) that the relation will go wrong]. 

OTOH, it's good to know, that nobody of the respondents mentioned a broken M6
before 1995, no matter whether the garbage rate might be 0.01 or 50 percent.
It means that LUGers, who follow the survey's indication, pay less for a fine
(used) camera. It's similar to select a divorced partner for your life - you
don't care, that she/ he has been living before ...

Besides, there are still those fine and highly reliable M2s and M3s on the
market ... giving you more and longer pleasure than a lot of the younger
bodies  ;-)
Alf