Control Charts: An Introduction to Statistical Quality Control. by Edward S.
Smith
Review by: Joseph M. Juran and Mason E. Wescott
Journal of the American Statistical Association, Vol. 43, No. 242 (Jun., 1948), pp. 343-346
Published by: American Statistical Association
Stable URL: http://www.jstor.org/stable/2280385 .
Accessed: 15/06/2014 20:06
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
American Statistical Association is collaborating with JSTOR to digitize, preserve and extend access to Journal
of the American Statistical Association.
http://www.jstor.org
This content downloaded from 62.122.76.45 on Sun, 15 Jun 2014 20:06:56 PM
All use subject to JSTOR Terms and Conditions
BOOK REVIEWS 343
containedin themanual.At presentthislittlevolume,Time-Series
Charts,
is stillthebestguideforthosewhowantto drawchartsoftheconventional
type.
ControlCharts: An Introduction
to StatisticalQualityControl.EdwardS. Smith
(Professorof Mathematics,Universityof Cincinnati,Cincinnati,Ohio), New
York 18: McGraw-HillBook Co., Inc. (330 West 42nd St.), 1947. Pp. xi, 161.
$3.00. (London, W.C. 2: McGraw-Hill PublishingCo. Ltd. [AldwychHouse,
Aldwych].15s.) Two reviews
follow:
REVIEW BY JOSEPH M. JURAN
Professorof Administrative
Engineeringand Chairman of theDepartment
New York University,New York,N. Y.
HERE is another of the numerous books now emergingto expound the
use of statisticalmethodsto controlquality of product.
There is a good discussionof the use of controlchartsforaverages, ranges,
and number or fractionof defectives.Visual aids in the formof charts or
tables are well prepared. The exercisesare well chosen.
The claim that the book is "a test forthe busy executive" is wishfulthink-
ing. In this work,and forthat matter,in others,thereis little evidence that
the authors have trulygrasped the quality problemsof the industrialexecu-
tive.
Industry'sproblemhas been and still is to make productto meet specifica-
tions. The statisticalmethods,if they are to be grasped by industry,must be
presentedas a means to the solutionof that problem. The emphasis must, in
the firstinstance, be noton (a) how to achieve controlledprocesses; the em-
phasis must ratherbe on (b) the fact that througha controlledprocess it is
easier to make productsto meet specifications.
To a statistician, (b) seems only too obvious; ergo the executive should
take (b) for granted, and thus only (a) remains to be learned. The precise
opposite is true. The practical problem is to convince the executive of the
truthof (b). Even afterhe is convinced of (b) he will ordinarilynot be con-
cernedwith (a); instead he will delegate the learningof (a) to another.
With the exception of the unrealisticfirstchapter ProfessorSmith's book
can well serve as a supplementaltext forundergraduateengineers.
REVIEW BY MASON E. WESCOTT
AssistantProfessorof Mathematics,Northwestern
University
Evanston,Illinois
PROFESSOR SMITH has long been a successful teacher of college mathe-
matics. During and since the war, he has participated as an instructorin
numerous 8- and 10-day intensive training courses in statistical quality
control for industrial people. He has conducted several successful in-plant
trainingcourses and has served as technical advisor forthe installation and
operation of statistical quality control programs. Drawing on this back-
This content downloaded from 62.122.76.45 on Sun, 15 Jun 2014 20:06:56 PM
All use subject to JSTOR Terms and Conditions
344 AMERICAN STATISTICAL ASSOCIATION
groundof theoreticaland practicalexperience, he has contributed Control
Chartsto theMcGraw-HillIndustrialOrganization and ManagementSeries.
It has been the author'sintentionto writea text "formenand women
whowishto acquireknowledge ofthemethodsand chartsofstatisticalqual-
ity control,but who are too busyto have an interestin the mathematical
aspectsof the methods."He believesthat "it is moreimportantthan ever
thateveryoneconcerned withsuchproduction (production of goodsofhigh
quality)shouldbe able to makeand read controlcharts."
Withthe latterpointof view,this revieweris in completeagreement,
but he does notfeelthatsuchunderstanding can be acquiredwithoutmore
exposureto thingsmathematical thanthe 'busy executive"generallyfeels
he can or wantsto endure.As a matteroffact,theinnocentreaderwhobe-
ginsProfessor Smith'sbookexpectantly hopingto be painlessly initiatedinto
the mysteries of statisticalqualitycontrol,soon findshimselfface to face
withformulasand expositionthatare prettylikelyto appearfrighteningly
mathematical to him.That thisis so, shouldnotbe heldagainsttheauthor,
because ProfessorSmithhas verynearlysucceededin keepingto a mini-
mumthe formalaspectsof his subject.Rather,it is convincing testimony
to thefallacyoftryingto undertakean expository discussionofa technical
subject,be it evenso simpleas controlcharts,withoutsooneror laterhaving
to use symbolsand languagethat to a greatmanynonprofessional people
mustseemdecidedlymathematical.
Chapter1 is probablythebestchapterinthebookfromthepointofview
oftheaudienceto whomthebookis addressed.Here theauthorintroduces
the definitions he proposesto followforsuch conceptsas "control,""qual-
ity,"and "satisfactory controlof quality,"distinguishes well betweenin-
spectionforthe purposeof controllingqualityand inspectionwhichmerely
sortsgood productfrombad, outlinespointedlythe fundamental roleof a
controlchartas a criterionfortellingwhento takeactionand whento leave
well enoughalone, and lists ten benefitsto be expectedfroma properly
administered controlchartoperation.One mayverywell take exceptionto
partsofthedefinitions given,but on the whole,theyare acceptableforthe
purposesintended.The statementthat "A production processis said to be
producing goodsof satisfactoryqualityif 99.7 per centoftheitemsare satis-
factoryin quality" is, to this reviewer, the kindof a statementhe would
rathernotsee usedwiththeemphasisit is madeto carryhere.
The remaining chaptersare,forthemostpart,wellorganizedand clearly
presented.One becomesconsciousofa tendencyto emphasizedetailsrather
thanto paintthepicturein broadoutline.The lattertreatment wouldhave
seemedmoreconsistentwiththe objectiveof "selling"the subjectto an
audiencewhichis notsupposedto be interested in details.However,Profes-
sor Smithis understandably morethe teacherthan the salesmanin these
chapters, and as such,he anticipatesand triesto removethementalhazards
thatexperience has taughthimblockthestudent'sprogress in learninghow
to make and use controlcharts.Certainly,if the readerwill exercisethe
This content downloaded from 62.122.76.45 on Sun, 15 Jun 2014 20:06:56 PM
All use subject to JSTOR Terms and Conditions
BOOK REVIEWS 345
patienceand intellectualcuriosity to studythroughthesechapters,working
theproblemsproposedas he goes,he willendup withat leastthemechanical
skillto makea controlchart,and a somewhatbetterthanbowingacquaint-
ance withtheoperationalmeaningofsucha chart.
In thesechaptersa numberof technicalpointsappearedquestionableto
thisreviewer, but mostof themare not seriousenoughto warrantdetailed
discussionhere.Threepoints,however,cannotbe ignored.
a) Thereis an irritating loosenesswithwhichtheword"normal"is used.
Such phrasesas "normal,symmetrical distribution," "normalaverage,"
"normaland approximately symmetrical frequencydistribution," "normal
process"occurrepeatedly withoutthebenefit ofany consistently adheredto
definition; in fact,withoutany "definition" at all! Unlessthereaderhas un-
usualinsightor sophistication, he mayverywellend up withthehappyde-
lusionthateverything willbe just dandyifonlyit can be describedas "nor-
mal." The wordis simplyoverworked in the last fourchapterswithoutre-
servingto it thedistinction it rightfully possessesin statistics.
b) It is theopinionofthisreviewerthaton pages48-52,and elsewhere in
subsequentdiscussions, Professor Smithoverplayshishandin thematterof
testingcontrolagainstproductspecifications. In thefirstplace,thepractice
ofputting"dangerlimits"(three-sigma limitsforindividuals)on a chartfor
averagesis questionable,particularly if the chartis to be keptat the ma-
chine.Exhibitingthe to-be-expected spreadforindividualson a chartfor
averages simplyinvitestrouble,becausetheoperatorfindsit hardto under-
standwhyhe mustkeepthepointswithinthenarrower band set by control
limitsforaverageswhenthe "dangerlimits'are wayout beyondthisband.
In thesecondplace,thediscussionmanagesto conveytheimpression (al-
thoughthecontrary is explicitlystatedon page 52) thattherewillfrequently
be hair-splitting closenessbetweena 'dangerlimit"and a blueprint specifi-
cationwhichhas to be resolvedbeforea controlledprocesscan be adjudged
in satisfactory control.The criteriondescribedis not in error:it simply
overemphasizes a relativelyunimportant point. Time mightbetterhave
beenspentin showinghowto estimatethepercentofproduction likelyto be
outoftolerancewhencontrolis notsatisfactory, and in discussing morefully
theproblemof balancingthe costof sortingand/orreworkagainstthat of
makingprocessor specification changesthatwillyieldcontrolsafelywithin
acceptableblueprint tolerancesoncetheoptimumcapabilitiesoftheprocess
have beendetermined.
c) The discussionin Chapter4 uses theword'lot" wheretheword"sam-
ple" wouldordinarily be preferable. The beginner readingthischaptermust
surelyexperience a senseofconfusion when,afterdigesting 21 pagesinwhich
he has had theword"lot" drilledintohimat everystage,he suddenlydis-
coverson page 106 thata boxfullof beads is a "lot,"but a paddlefullof
beadsis a "sample,"and thatnowit is thesamplethatis to be inspected100
per centratherthan the lotas he has previously been led to expect!Evi-
dentlytheauthorintendedthischapterto deal withattributeinspection for
This content downloaded from 62.122.76.45 on Sun, 15 Jun 2014 20:06:56 PM
All use subject to JSTOR Terms and Conditions
346 AMERICAN STATISTICAL ASSOCIATION
processcontrolratherthan foracceptancesampling,although,of course,
thetwo are not mutuallyexclusive.But ifinspectionforprocesscontrolis
intended,it would seem more consistentand realisticto characterizea
"day'sproduction" (whichtheauthorsaysmustbe inspected100percent!)
as a "sample"fromthe cause systemrepresented by theproduction stream
thanit is to describetheseitemsas a "lot."
Furthermore, it would be entirelypossibleforthe beginnerto get the
impression fromthischapterthat 100 per centinspection(in the ordinary
sense) is requiredto keep a fractiondefectivechart.Many plantsexercise
verysatisfactory processcontrolthrough themaintenanceoffraction defec-
tive chartsbased on samplesthatare a verylongwayfrombeingwhatthe
averagemanhas in mindwhenhe findshimselfreadingabout 100 per cent
inspection ofa lot.It is thesamplethatis inspected100percent,and it would
be wiserto say so, ratherthanto use languagethatcan be so easilymisin-
terpreted.
Table II appearsto be potentiallyusefulin simplifying the calculations
requiredforcontrollimitson p and np charts.Its use is amplyillustrated in
Chapter4.
The typography is good; tables,charts,and diagramsare well doneand
legible.The bookis easy on theeyes.Exceptfora fewmissingpunctuation
marks,onlytwo errorswerefound:on page vi, the name shouldbe Paul
Peach,and on page 142,thebarhas beenomittedoverthe R in thethirdline
fromthe top.
Control Chartscouldproveusefulas a textfora shorttrainingcourseat an
elementary level.As a textbookforuntutored home-study or as a "quickie"
forthecasual readerwhowantsto see theforestbut notthetrees,it is the
opinionofthisreviewer thatthebookis likelyto provedisappointing.
Federales aux Etats-Unis.
Methods StatistiquesModernes des Administrations
Pierre Thionet. Paris: Hermann & Co. (6 rue de la Sarbonne), 1946, Pp.
117.
REvIEW BY W. EDWARDS DEMING
Adviserin Sampling, Bureau of theBudget
Washington
HIS book is a unique and definitecontributiontoward improvedstatistical
1 practice,as wellas a delightto read.The authorexhibitsunusualability
in theoryand extraordinary He recognizesthe need
clarityin presentation.
forproperfieldorganization in conducting samplesurveys,and appreciates
theneedforreliablefigures in government administration. The studyofthis
bookhas beena pleasingexperience forthereviewer.It is interesting
to see
howwe areviewedfromacrosstheAtlantic,and gratifying to findtheessen-
tial pointsofprogress in the UnitedStatesso clearlypointedout. Here,we
are too closeto givea coherentaccountofwhathas beenhappening.More-
This content downloaded from 62.122.76.45 on Sun, 15 Jun 2014 20:06:56 PM
All use subject to JSTOR Terms and Conditions
