Page 307 - Six Sigma Advanced Tools for Black Belts and Master Black Belts

OTE/SPH
OTE/SPH
August 31, 2006
JWBK119-18
Taguchi Methods
292 3:6 Char Count= 0
Basically, Taguchi methods address all the major quality improvement expectations,
shaping a strategy that promises to integrate the management, technology, and infor-
mation capabilities in an organization. They throw up an unambiguous bottom line
in a language that everyone in the profession understands: on-target, least-variation
and cost-effective performance throughout the life of a product or process in the real
world. An offering like this, designed with a mix of oriental idealism and pragmatism,
presented in Japanese consumer-oriented packaging and promoted with American
63
marketing prowess, was nothing short of formidable in a country looking for a way
out of the quality crisis. With parameter design at its core, Taguchi methods draw
heavily from techniques of statistical design and analysis in a manner that facilitates
immediate adoption by nonstatisticians. In so doing, the American quality profession
was offered what may be termed ‘statistical engineering’, of which a practice-oriented
approach has won the approval of those still uncomfortable with the ‘engineering
statistics’ available from statisticians.
As it turned out, Taguchi methods’ demonstration effect grew entirely out of pro-
portion to their technical superiority. Controversies ensued before long over a number
of ways in which statistical techniques are deployed in Taguchi methods. Disagree-
ments have led to new studies and research in academia that could eventually increase
the effectiveness of practical techniques, for example the use of graphical techniques
for experimental design 64−66 and the exploitation of response surface methodology
for variance reduction. 67,68 However, debates have also in the meantime led to bewil-
derment in the quality profession and hinded the application of useful experimental
design techniques.
For the benefit of practitioners, this chapter has highlighted the nature of Taguchi
methods as well as a number of issues frequently raised about them. The subject has
been interpreted from several angles in the hope that understanding is the most log-
ical approach to resolving differences. Managers and engineers in particular have a
need to get out of the woods; they cannot afford to wait for the dust to settle among
the advocates and critics of Taguchi methods. There has been, however, some evi-
dence of abatement of excesses. One example is the recent ‘balanced’ assessment of
Genichi Taguchi (score: 10 out of 10). 19 On the other side of the Atlantic, the UK
Taguchi Club has now been renamed the Quality Methods Association in recogni-
tion of the fact that no single guru should claim monopoly over the faith of qual-
ity professionals. The new robust design software from AT&T Bell Laboratories has
intentionally discarded the flawed analysis of variance routine in Taguchi-style ex-
periments. These are encouraging signs suggesting that reason and objectivity will
indeed win the day. When that happens, nothing but good will be brought to manage-
ment, engineering, and operations personnel in industry and, by extension, society
as a whole -- which of course means that the mission to cut loss to society is finally
accomplished.

18.9 EPILOGUE

Since there can be no argument as to the usefulness of statistics in quality improvement
efforts -- despite differences in the means of its deployment, some of which have