Re: Response to SPC theory problem

[Jean-Marie Gogue <jmegogue@wanadoo.fr>]

On Wed, 29 Aug 2001 Jim Clauson forwarded comments from a student who asked
the binomial -vs- Poisson differentiation question.

>It is a worry that Don Wheeler says "in my experience, very few
>practitioners are sophisticated enough to correctly judge when the
>binomial or Poisson conditions are satisfied".
>
>But I did like him saying that you could simplify by using Xmr for
>attribute data  (I think I did read that somewhere previously).

I like these comments. Your student makes me remind of this statement made
by JDK in his post Dated Mon, 27 Aug 2001 (Re: SPC Theory):

>For many people there are psychological and cultural barriers
>amounting to a deep distrust of anything remotely mathematical.
>This distrust has to be overcome somehow [any ideas ?] However,
>it goes further than this.  As well as the need to accept some new
>habits (like not being scared of maths), there is a need to discard some
>harmful old habits.  Some examples are :
>
>- the fascination with tables of numbers
>- the love-affair with the bar-chart (second only to the pie-chart for
>concealing and distorting the truth ),
>- the apparent belief that only data since January is of any importance.
>
>You have to start with small steps like these if you're going to persuade
>people to take the whole journey.

Is our aim to get people adopt the Deming philosophy, including
acquaintance with control charts? In this case, we have to find the best
way for breaking down their psychological and cultural barriers. The method
which Deming provided us with is the PDSA cycle. Any approach (such as any
electronic gimmick about control charts) may be helpful. The usefulness
must be tested with the PDSA cycle.

About XMR chart, the viewpoint of the Western Electric SQCH, 2nd edition
1958, page 21, is very interesting. This can help to clarify the situation:

===== begin quoted text  =====

The principal kinds of data for which this chart should be used are the
following:

a. Accounting figures of all kinds, including shipments, efficiencies,
absences, losses, inspection ratios, maintenance costs, accidents reports,
records of medical tests, etc.

b. Production data such as temperatures, pressures, voltages, humidity,
conductivity, furnace heats, gas composition, the results of chemical
analysis, etc.

In all of these cases only one number is available to represent a given
condition.

=====  end quoted text  =====

Best wishes,

Jean-Marie Gogue

Versailles  France

jmegogue@wanadoo.fr