RE: Info

["Prevette, Steven S" <Steven_S_Prevette@rl.gov>]

Linda Ortberg stated:  Has the absence of statistical thinking always been
the case in academic statistics classes?  If so, why did Dr. Deming say that
one should have a statistician on their staff if they could not use what
they knew in applied statistics such as we use in quality?

My perspective in answering this is as: an engineering student (my undergrad
work was in Civil Engineering), an operations research student, a
follower/practitioner of Dr. Deming's theories, and a teacher of statistics
courses.

Knowledge of statistical theory is important to applying Dr. Deming's
methodologies.  Certainly the operations research courses (and one
undergraduate course in civil engineering on probabilistic design)
introduces one to the concept that the world is not deterministic.  Most
engineering courses (and certainly most management courses) teach a very
deterministic view of the world - linear cause and effect.  Do this and you
get that.  Design a bridge with a safety factor of 2 and it will stand.
(but some bridges designed to such a safety factor have fallen).  Dangle
this carrot in front of the workers, set this goal, and it will happen.
(but what about the Red Bead Experiment?)  Knowledge of statistical theory
helps you to avoid falling in the MBO traps, the belief that if you got 11
work items done this week and you got 10 items done last week you can
celebrate a 10% productivity improvement.

Statistical degree work, with its understanding of concepts like standard
deviation, distributions, Tchebychev Inequality, queuing, linear programming
starts you on the path of a world view of probabilities and stochastics.
Yes, there is a leap from the "data" of statistical ideas to the
understanding and wisdom of "statistical thinking", but I would not quite
say that statistical thinking has been absent.  It is true that most
statistics courses don't push you to maintain the time sequence of the data,
but with statistical background, one can come to understand the need for it.

I do think probably the biggest downfall of statistics courses (and having
to teach the dreaded required statistics course to business majors has
opened my eyes to this) is the statistics instructor who has never practiced
the art of statistics.  Or alternately, one who has practices the art of
statistics, but has not practiced the art of teaching.  The first failing is
probably more common than the second, but I should acknowledge the second
can exist.

Without a good grounding in statistical theory it is difficult to apply
statistical thinking, and it is very easy to fall in the trap of becoming a
"hack".  To avoid depending upon a hack, Dr. Deming did recommend in Out of
the Crisis several key attributes for statistical leadership, including at
least a masters level degree work in statistics.  My own addenda would be to
look for people with understanding and wisdom of statistical thinking - and
that understanding and wisdom may not necessarily exactly coincide to what
piece of paper the person holds, but what paper (degrees or certifications)
the candidate holds is at least a piece of data.

Steve Prevette
Site Technical Authority for Statistical Trending
Safety and Health
Fluor Hanford, A Fluor Global Services Company
ASQ Certified Quality Engineer
steven_s_prevette@rl.gov
509-373-9371