Quality Progress - May 2014 - 6
for trying to explain the likelihood of defective
parts to a manager.
In full disclosure, I cannot take credit for
William Hooper's article, "Probing Prob-
the simple analysis above; that credit belongs
abilities" (March 2014, pp.18-22), provides
to my wife, a nurse. As I was working through
an excellent review of Bayes' theorem and
Bayes' formula and searching for my calcula-
its implications for conditional probabilities.
tor, I asked her the question and she immedi-
But if his results "defy instinct," as the is-
ately identiﬁed the correct answer using the
sue's cover proclaims, it might be because
logic above. Sometimes simpler really is better.
our instinct has been undermined by the
sophistication of our knowledge.
The article opens with a medical question
asking what the likelihood is that a person
who tests positive for a disease that occurs
In response to "Will to Live" (March 2014, pp.
in only 1% of the population actually has the
32-37): This is an impressive story. My quality
disease, assuming the test is 95% accurate.
manager experienced something similar. He
The author explains Bayes' theorem and
was diagnosed with pulmonary ﬁbrosis and
works through the math to determine the
was lucky to get a lung transplant. By that time,
probability is only 16%-not 80% or 95%, as
he could not walk and was lacking oxygen.
most Harvard Medical School doctors appar-
When I went to visit him one week after the
ently responded. But the answer could have
operation, I was amazed with his breathing,
been arrived at much more simply.
which was easy and normal. He reacted to the
Consider a representative group of 100
situation like a quality professional-he told me
people, one of whom has the disease and 99
that the quality is not an accident, it's a result
who do not. The test, being 95% accurate,
of intelligent work. That famous quality quote
will almost certainly yield a positive result for
could not be used in a better circumstance.
the person with the disease. In addition, it
can be expected to generate about ﬁve false
sample. So, in six positive results, only one
is accurate. The likelihood of a person with
The "Learning From Experience" (February
a positive test actually having the disease
2014, pp. 14-19) interviews with lean experts
would be one in six, or 16.7%, a result that
are inspiring. The common theme from all in-
differs only slightly from the exact answer of
terviewees is that lean should be viewed as a
16.2% obtained from Bayes' theorem.
"common sense revolution" in recognizing that
There is certainly a time for precision.
people learn by doing. Only people who are
But if our intent is to help doctors with no
capable, motivated and aligned in the gemba
knowledge of statistics understand why the
of an organization can make a difference-not
answer is not 80% or 95%, or to reassure a
opportunistic lean consultants only motivated
patient who has just obtained a positive test
by exploiting this movement to ensure they
result, the simple example above may be
have a steady stream of customers.
far more effective than trying to teach them
Bayes' theorem. The same might be true
QP * www.qualityprogress.com
The latest ASQ TV episode focuses on
lean and Six Sigma. In the episode:
* Learn the difference between
lean and Six Sigma methods.
* Hear how a global fi nancial fi rm
improved response times and
* Examine waste analysis.
Watch for a new episode available May 13 discussing the global
impact of quality.
asq.org to access
the full video library.
Find the latest news, quips and targeted
content from QP staff.
Executive Editor & Associate
Publisher Seiche Sanders:
positives from the 99 healthy people in the
Associate Editor Mark Edmund:
Assistant Editor Amanda Hankel:
Contributing Editor Megan Schmidt:
Table of Contents for the Digital Edition of Quality Progress - May 2014
Mr. Pareto Head
Making the Rounds
Plan of Attack
Measure for Measure
Quality in the First Person
One Good Idea
Quality Progress - May 2014