Re: PDSA & Enumerative -vs.- analytical approaches

[John <jsdwd@ispwest.com>]

On 1/31/04 4:56 AM, "John Constantine" <thesfg1@cox.net> wrote:

> What better to ask for; it's one more reason for why counting doesn't
> provide much support for improvement, but can often lead to disaster,
> especially when the "counts" are not random elements. N-1 vs. N, is
> another way of looking at things; N gives you control of the entire
> count, the whole population, whereas N-1 assumes that you do not have
> the entire population, but always something other.


I am not sure how or if the difference between n and n-1 is relevant here
and pretty sure any idea of 'population' is not.

Enumerative studies are those done on a particular sampling frame when full
and complete coverage of the frame is possible (future samples, for example,
would be excluded as they hasn't happened yet and are, therefore,
unavailable).  Such a study (enumerative) if properly done (e.g.
Randomization used if sampling is done) can be of some use and provide the
relevant characteristics or estimates of same of the frame being studied.

Since full and complete coverage is needed and results are limited to the
sample under study, prediction is essentially, impossible from a statistical
point of view.  Other methods of prediction are still available, for
example, Tarot cards.

Analytic studies are, simply stated, those intended for prediction.

Deming said repeatedly that "Management is prediction".  Therefore while
enumerative studies are not without some use, one is almost always in the
'analytic' situation.

At least that's my take.  Others may have a clearer or more complete
definition or disagree.

John Dowd
jsdwd@ispwest.com