Re: Cause source - system v. employee

["John Constantine" <thesfg1@cox.net>]

In the context of this discussion, as other DEN members have pointed out, we
live in at least two worlds. The dream world of experiment and probability,
and the conscious world of empiricism and "reality". We operate as spiritual
beings in a material setting, grasping for links to the "real", and applying
our beliefs by experiment in our daily lives, adjusting as we go.

I think we are always struggling to better understand what we believe we
know and what we need to know, and we don't always succeed.

As Stein mentioned, "...can any cause be OBJECTIVELY identified? Isn't
probability essential?"

It's a bit disconcerting to come to grips with the notion that we are
creating our own reality all the time, and that is why the use of
operational definitions is so very important. If we can agree on what we
mean, then we might be able to say that "this" is probably happening. And in
this context, probability is "probably" essential (sorry, I just felt the
need.)

If we must reproduce that which we have said has been "objectively
identified", we may then do so, for the purpose of the experiment only. When
it comes to drawing inferences from what we have identified, as it might
apply to other, future realities, that probabilistic no longer holds. It no
longer serves the purposes of our daily lives, and is relegated to another
world.

As I read Deming and Wheeler, in the case of apparent trends, what is "seen"
is not what one actually sees, but merely a representation, and the reality
demands the use of different methods to identify that the trend actually
does or does not exist. Hence, the use of the phrase "no true value". From
TNE, p.100, (distinguishing between enumerative and analytic studies), "The
interpretation of results of a test or experiment is something else. It is
prediction that a specific change in a process or procedure will be a wise
choice, or that no change would be better. Either way the choice is
prediction." (continuing)"...Tests of significance, t-test, chi-square, are
useless as inference-i.e.., useless for aid in prediction.

>From STATISTICAL ADJUSTMENT OF DATA, p.12 (quoting J.M. Keynes) "But to
argue, without analysis of the instances, from the mere fact that a given
event has a frequency of 10 percent in the thousand instances under
observation, or even in a million instances, that...it is likely to have
frequency near to 1/10 in a further set of observations, is...hardly an
argument at all."

This appears to me to be just one of many such instances of the reason why
we use, or should use, empirical data. It does not depend on probability
theory, or theories. Paraphrasing Wheeler, if the probability theory being
applied approximates the empirical data, it will come close to the chart of
that data. If not, then it will not, and the empirical data being charted
will be a better representation, more suitable for predictive purposes. To
me this is a powerful explanation, and very helpful to one using control
charts to see what is occurring in a process.

This is what my dim light casts on the subject, and I hope the use of these
discussion does not end.

Stein has been helpful in asking the question. I look forward to others'
comments on whether probability is indeed essential.

One last thing, referring to Wayne's comment as noted by Stein, "verifying
that a special cause
variation occurs on the control chart immediately following the change..." -
what I think Wayne was using was a moving range chart, perhaps. This helps
to give the observer a rapid response to special cause events.

________________
John Constantine
thesfg1@home.com
Phoenix, AZ