Donald Marquardt - The Importance of Statisticians
Transcription
Donald Marquardt - The Importance of Statisticians
The Importance of Statisticians Author(s): Donald W. Marquardt Source: Journal of the American Statistical Association, Vol. 82, No. 397 (Mar., 1987), pp. 1-7 Published by: American Statistical Association Stable URL: http://www.jstor.org/stable/2289118 Accessed: 24/08/2010 18:58 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=astata. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. 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 support@jstor.org. American Statistical Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of the American Statistical Association. http://www.jstor.org The Imporlanceof Statisticians DONALD W. MARQUARDT* 1. INTRODUCTION we can take greatpridein the accomplishanniversary, mentsofourdiscipline and ourassociation. Mankindhasbeendoingappliedstatistics forthousands Nevertheless, thereare some basicproblemswiththe ofyears.Whenfacedwitha problem-almostanyprobof andwithstatisticians. I willexplore lem-people can be observeddecidingwhat data are discipline statistics some of thoseproblemsin thisaddress.I have selected needed and are practicalto collect;thenobtainingthe thetitle"The Importance ofStatisticians." data,analyzing thedata,developing a conceptualframeworkthatis consistent withthedata (a predictive model, 2. PERCEIVED LACKOF VISIBILITY ifyouwill);and thenchoosinga courseofactionbasedAND INFLUENCE at leastin part-upon thedata. Statistics has beena recofstatistics ognizabledisciplineforseveralcenturies. That is, there One oftheproblemsis thelackofvisibility a as discipline. During my tenure as President-Elect and hasbeensomegeneralized theoretical basisandsomeform I President, have recommended widespread study of Paul of documentation. In fact,the disciplineof statistics is olderthanmanyof theothermajordisciplines we know Minton's(1983) article,whichdescribesthescopeof the visibility problem.Justone statistic fromthatarticle:Of today. the approximately 2,000 four-year in theUnited colleges is nowwidelyemployedin all areasofsociety Statistics States and Canada, only about one-tenth offer degreeproincluding industry and government. For example,in the in grams statistics, and only half of these include thename 1986 President'sInvitedAddressgivenby RichardE. in "statistics" the department In own title. my observation, Heckert,ChiefExecutiveofthedu PontCompany, at the is accompanied byanothersympASA AnnualMeetings,Dr. Heckertmade it clearthat theproblemofvisibility a lack tom, perceived of influence. Just one example:Most themoderncorporation is dependent uponthediscipline disciplines today have commercial thatare cirmagazines ofstatistics in a widevarietyofways.The statistical data culated free of to in charge professionals the field whoare suppliedby government agenciesis butone exampleof to and judged influence decisions for serpurchase goods thedependence ofcorporations, andthewholeofsociety, vices related to has the field. Statistics no such magazine upon statistical services.Statisticsand statisticians also havevitalinternal rolesin production, research, market- to myknowledge.Doesn't thissuggestthattheexternal is thatstatisticians havelittleinfluence? ing,and supportfunctions ofthemoderncorporation. perception I know that will be to respondthatthese many quick The AmericanStatistical Associationwas foundedin not statisticians problems are to butare affecting unique 1839.OnlytheAmericanPhilosophical Society,theforeand scientific technical many disciplines, especiallythe runner umbrellaorganization forall scientific disciplines, mathematical You forexample, sciences. be may familiar, is olderamongAmericanprofessional associations (Wallis with in this the recent study reports area, notablythe 1966).Anyonewhohas followedthegrowth oftheASA National of David Sciences the Academy Comreport by during thelastseveraldecadesmustbe impressed withthe mittee related include Other (David 1985). reports Fey, breadthof itsactivities, whichmirror thegrowth of staand and Neal Albers, Fleming (1981) (1986). tisticsas a discipline.As the ASA approachesits 150th We do have to acknowledgethatwe are caughtup * DonaldW. Marquardt is withtheEngineering E. I. du Department, Pontde Nemoursand Company,Wilmington, DE 19898.Thisarticle waspresented as thePresidential Addressat the1986AnnualMeetings oftheAmerican Statistical Association in Chicago. I ? 1987 American StatisticalAssociation Joumal of the American StatisticalAssociation March 1987, Vol. 82, No. 397, PresidentialAddress 2 Journalofthe American StatistioalAssociation,March 1987 all of the sciences,but thatare affecting in difficulties individual are Fromthisperspective, statistical methods as an excuseforstat- viewedas toolsor component we cannotacceptthosedifficulties elementsin an overallsysisticians.In fact,we have specialproblemsof our own temsapproach.Some statisticians and statistical service making. organizations todayreallydo operateinthismodeandare theroleof seen thisway by theirclients.I willcome back to this thatwe needto reexamine I am convinced I thinkwe willfindthatwe statisticiansperspective thestatistician. and expandupon it, but firstlet's examine how howwe see ourselves. havesold ourselvesshort.Let's beginbyexamining otherssee us. 4. HOW WE SEE OURSELVES 3. HOW OTHERSSEE US 4.1 The NarrowView 3.1 The NarrowView First,thereis thenarrowview.These people see stais whatmaybe calledthecock- tisticsas a branchof theoretical or appliedmathematics. insight One penetrating member Statisticians thatanother Uponlearning tailpartyperspective: withthisviewoftenoperateas if is thecomment groupis a statistician, of a conversation * onlyabstract problemsare permitted withinthedisHow veryessential.One is, afterall, a "A statistician? cipline all ofone's life,isn'tone?" statistic * onlyteachingand researchare honorablecareers. is widelyheld.We are Thisnarrowviewofstatisticians must Thisis a vitalpartofstatistics, of data. Everysportsannouncer butitis certainly notthe seen as compilers forsupport.Social and governmentalwhole.We mustnurture have a statistician and supporta minority of statofdataonallmanner isticiansengagedin theseactivities. Indeed,we mustbe agenciesareseenas simplecompilers of subjects.People mayviewthisas a highlyimportantevercertainthatmanyof our brightest and bestare foor challengingcusingtheirefforts as a difficult butnotnecessarily on teachingand research.It is crucial function, roleinsociety.Weknowthisviewisverynarrow.Weknow to thelong-term healthandsurvival ofourdiscipline. But viewtherearedeepandcom- thisnarrowviewcannotbe theviewof the majority thisnarrow thatevenwithin of If thateverhappened,thenstatistics plexissueson howto designandcollectthesamples,how statisticians. would of data,howto makere- lose all hope of visibility and influence, and soon thefito processthelargequantities Weknowthatthestatisticalnancialsupportfromeducational andso forth. liablepredictions, institutions andfunding policychoicesbygovern- agencieswoulddryup as well.Thus,theverylivelihood resulting choicescan influence butthepublicat large of teachersand researchers mentand businessorganizations, is ultimately tiedto thevisichoices. bilityand influence ofstatistical of appliedstatisticians. seemstobe unawareofthisinfluence Some statisticianscanoperatewithin thenarrow view,butall statisti3.2 The Middle View ciansshouldphilosophically a widerview.History support seems to that the most show influential statisticians Takingthenextstep,thereare manywhosee us from in tend to roots and have both strong theory application. as whatI willcall themiddleview.Theysee statisticians Theyknow fordefined problems. orconsultants specialists 4.2 The Consensus View that Let's turnourattention to theconsensusview,heldby * we can providedesignsforsurveys and experiments moststatisticians see statistics as today. They models * we do data analysesand developpredictive * we can make quantitative, from * a uniquediscipline reliableinferences andprobability usingmathematics data. methodas a phitheoryas a tool and thescientific losophy todaywouldbe verycontentifonly Manystatisticians * providing a varietyof satisfying careersin abstract oftheirclientswouldoperatefromthismidthemajority or practicalapplications in theory industry, governitwouldbe! AndyetI havechardleview.Howwonderful the or educational institutions service ment, sector, acterizedit as the "middleview." You can inferthatI * provingusefulin anyfieldof application. whichotherscanorshould mustenvision a widerviewfrom see us. 5. MARKETSEGMENTSAND COMPETITORS 3.3 The WiderView Fromthisconsensusview,statisticians (and statistical can see theworldaroundthemin guides serviceorganizations) are systems The widerviewis thatstatisticians thecollectionand inter- termsof variousmarketsegments.One wayof dividing fortough,fuzzyissueswherever inwhichstatistical services ofdatashouldbe involved.The widerviewex- theworldintomarket segments pretation fieldssuchas are neededis on thebasisofsubjectmatter to providethefollowing: pectsstatisticians inproblemdi* fullserviceguidanceandparticipation agnosisandsolution in anyarea ofapplication forproblems * services * servicesat anylevelofdetailor abstraction. * * * * physicaland engineering sciences lifeand healthsciences businessand economics socialsciences. Marquardt:The Importanceof Statisticians 3 is thatthesefieldshavecreateda largebachelor'ssegments ferent theworldintomarket wayofdividing Another eachsub- levelbase ofpeople.Whatdoes thataccomplish? thatwithin servicesis to recognize forstatistical suchas jectmatterfieldthereare genericfunctions andvisibleas a career * The fieldbecomeslegitimized ofpeopleavailableto be hiredinto andas a category standardjobs (Minton1983). becomeawareofthefield. * Employers * The brightest and bestfromthebachelor'sprogram can provideneededservicesin all of those Statisticians forgraduatetraining. becomeavailableas candidates marketsegments. * The fieldcan growin statureand importance in sosee Withthisas theconsensusviewofhowstatisticians to cietybecause therecan be enoughstatisticians we areina positiontoexaminemorecritically themselves, as a discipline. and influence achievevisibility wherewe are. Let's takea look at whoourcompetitors to become standtodayitis almostimpossible As things are in thejob market.Thereare many.Theyinclude without havingmadea detrainedstatistician a formally * engineers(industrial,electrical,mechanical,elec- liberaterevisionofcareerpathafterhavingfirstattained tronic,chemical,etc.) a bachelor'sdegreeinsomeotherfield.I suspectthatmost * physicists fromsomeotherfield,perhapsone ofus cameto statistics such as thosewithM.B.A. of the fieldsmentionedin Section5. For manyof us, * businessprofessionals, degrees has nowbecomeourprimary field.For others, statistics * operations researchers field. statistics maybe a secondary * mathematicians This currentmarketing approachis rootedin the or* socialscientists Itisinstructive ofthefieldofstatistics. culture ganizational * computer specialists. culture"is. Schein(1985) what"organizational toconsider to stat- definedcultureas a "patternof basic assumptions-intheseas "competitors" Whydo I characterize or developedby a givengroupas it isticiansin the job market?Thinkaboutwhatmanyof vented,discovered, and ofexternal adaptation as deciding learnstocopewithitsproblems thesepeopledo onthejob. Theydo suchthings has workedwellenoughto be integration-that whatdata are neededfora specificproblemand howto internal tonewmemtobe taught validand,therefore, thedatain tabularand considered collectthedata. Theysummarize form.Theyanalyzethedata.Theydeveloppre- bers as the correctway to perceive,think,and feelin graphical tothoseproblems"(p. 9). He notedthat"because and relation dictive modelsfromthedata.Theyderiveconclusions theyarelikely haveworkedrepeatedly, onthedata. suchassumptions coursesofaction,allbasedlargely recommend andtohavedroppedoutofawareWhatare theydoingon thejob? Theyare doingstatistics to be takenforgranted fora largeproportion oftheirtime.Manyoftheseprofes- ness" (p. 9). thenarrowview thatsurround perceptions area primary The cultural functions sionalsholdjobsinwhichstatistical teachingand research(Sec. 4), theconcept workcomponent. In fact,don'twe findin anyindustrial, of statistical mustalwaysbe trainedor expethatpeopleoftenare thatappliedstatisticians or serviceorganization government, andthe statistics, fieldsare differentriencedin anotherfieldbeforestudying assignedto jobs whosesubjectmatter a graduate-level isinherently subject fromthefieldstheystudiedin school?The bottomlineis conceptthatstatistics theera whenwidespread thatfitted applisociety are inventions thattheactualcontentof manyjobs throughout viwas onlya futuristic methodology thanwithanyothersingle cationof statistical has moreto do withstatistics atthedawnofthecomputer ageina period ofthesejobs that sion,occurring field.Yet theproportion subjectmatter have Allofthesecharacteristics funding. is negli- ofheavyresearch trainedin statistics are heldbypersonsformally changed.It is timeto adoptsomenewcultural gradually gible.Why? andinternal normsthataremoreintunewiththeexternal now faces. statistics that problems APPROACH 6. OUR CURRENTMARKETING areat theheartofmanyofthemostsevere Thesethings It all comesbackto ourviewofourroleas statisticians.problemsin thefieldof statistics today.In myopinion, approachin thefieldofstatistics ourcurrent Ourstandard marketing is marketing approachinthefieldofstatistics has beento producea smallnumberofpeopletrainedat doomingus to perpetualstatusas "hangers-on" to the fromtheuni- mainstream thegraduatelevel.Thosewhohavestrayed ofsociety. available as themselves specialists havepositioned versities to thevariousmarketsaroundthem.But I ask,Whydo 7. A WIDERVIEW we see ourroleas limitedto providing helpfromoutside Instead,I holdoutto youa widerviewofourselves:a are thepritoolsand functions If statistical thesystem? vision ofpeopletrainedas statisticians occupying jobs at trained can't of people marycomponents manyjobs,why ofsociety, forcein themainstream thecenter,a unifying holdsomeofthosejobs? in statistics withpeopletrainedinothermainstream together aboutthemarketingworking We shouldask,Whatis different use the scientific method.Consultant that technologies comor business as engineering such fields in approach ifthey also willhave theirgreatestinfluence The biggestsinglethingthatis dif- statisticians paredwithstatistics? * marketing * production * researchand development. 4 usingtheprinoperateat thecenterof themainstream, thatI have espousedelseciplesof "totalinvolvement" where(Marquardt1979). So we cometo theessentialquestions:Whatshouldthe the be? Whatis the confluence, role of the statistician ofthedisoftheuniquecharacteristics together, flowing and themarketopportunities? ciplineofstatistics Journal of the American Statistical Association, March 1987 sequenceofstepsinthe aroundtheoperational organized method. scientific The firstmajorthemewouldbe approachto as a strategic experimentation * deliberate includes theme This first examinea targetenvironment. (b) of the experiment, goals the as subtopics(a) defining of goals to the in relation environment target the defining identify to purpose the is forexample, the experiment, 8. STATISTICSIS THEGENERIC FIELD variablesor to optimizea response?(Strategy important OF TECHNOLOGY facthecontrolled 1975),(c) defining ofExperimentation, of goals to the in relation response is thegeneric torsand theobserved is thatstatistics The uniquecharacteristic in involved the issues and (d) clarifying thatuse the theexperiment, amongall thedisciplines fieldoftechnology, presbe cannot data the of all otherfields cases in whichthestructure method.Fromthisperspective, scientific pecified. andtheunifysubjectmatter, arespecialcasesforspecific method. ingconceptis thescientific Othermajorthemesincludeissuesin partof all experimethodis an inherent The scientific * collecting thedata for responsible is thediscipline mentalsciences.Statistics thedata * and manipulating recording intensity methodwiththegreatest thescientific studying data the and displaying * analyzing to otherdisciplines. expertise in-depth and forproviding thedata(thisincludes models from * creating predictive If we are willingto acceptthisrole,we mustalso accept and model criticism both the model development thatgo withit.Nobodyis goingto lay theresponsibilities phases) thisrolein ourlap. We mustgo outand workforit. But fromthedata * makinginferences whatwe must understand wecanonlydo so ifweproperly customers * to results presenting in expectof ourselves.As Ed Deminghas so aptlysaid * adequacyof servicesand marketplace monitoring is that theproblem ofqualityandproductivity, thecontext results do notunso manyworkersat all levelsoforganizations newconceptsleadingto newexperimen* formulating whattheirjobs are. derstand in a tationin relationto a new targetenvironment ofthisroleinthree theimplications I willdiscussbriefly iterative sequence. continuing areas: specific coursecan teachmanyof the eleThis introductory * theeducationofstatisticians ina first course, statistical methods usuallytaught mentary * theroleof appliedstatisticians From the theme. main must their support but presence Association. * theroleoftheAmericanStatistical methodsare the individualstatistical thisperspective, elementsin an overallsysviewedas toolsor component 9. EDUCATIONOF STATISTICIANS approachis theembodiment temsapproach.The systems and subject method.It drawsstructure wouldbe required of the scientific education Somechangesinstatistical philosophy To beginwith,the matterfromsuch fieldsas epistemology, undertherolemodelI havedescribed. managementscience, in of science, strategicmarketing, experience theireducational generalpublicthroughout orconcepts psychologyand dynamicsof people interaction, gradesK-12 wouldlearna varietyof statistical and, of course, statistical andtoolsand wouldemergefromtheirschoolyearsfully ganizationaleffectiveness, tools,es- methodology. manysimplestatistical able to use effectively studyto designan experilife.Theywouldcarry It is obvioushowa statistical tools,ineveryday graphics pecially It is almostequally framework. and use them mentfitsintothissystems theseskillsintotheirworkenvironments obviousthata surveysuchas a populationcensus,or an there. studyall fitinto Thosewhogo on to collegewouldcontinueto develop opinionpoll,or a medicalepidemiology of In everycase, theviewpoint framework. theseskills.Especiallyat thecollegelevel,and mostes- thissystems us to defineclearlythe methoddisciplines as theirbaccalau- the scientific peciallyforthosewho selectstatistics and thecontrolled factors, would goals,thetargetenvironment, reatemajor,theirviewof theroleof statisticians theobservedresponse. frombefore. be different methodhelpsus viewof thescientific But thesystems educationare alSomeof thesechangesin statistical as a Shewhart simpleapplications to takeplace,andI ampleasedto report evenin suchseemingly readybeginning process.Too often,appliis takinga lead- controlchartforan industrial Association Statistical thattheAmerican cationsofsuchsimpletoolshavefallenshortoftheirpoershiprole.But moreneedsto be done. changethatI wouldsuggest tential.For example,everycontrolchartapplication Thesinglemostimportant of the targetenvironeducationwouldbe in thefirstintroductoryshouldincludeexplicitdefinition in statistical theobservedresponse,and factors, 101. Its focuswouldbe to ment,thecontrolled course,Statistics college-level in thesamplingplan. The graphicalcontrolchartitselfcan conceptof the disciplineof statistics teacha system How manypracin itsapparentsimplicity. method.The coursewouldbe be misleading relationto thescientific Marquardt: The Importance of Statisticians 5 titioners understand thestrengths and thelimitations of business, themarket to be served,developa maridentify themathematical modelthatis embeddedin the chart, ketingplan,andsharpenandshapethestatistical toolsto theoperating characteristics thatareimpliedbytheaction fitthemarket. theentrepreneur statistician will Moreover, lines,or the diagnostic strengths and weaknessesof the be sureto knowtheclient'sproducts, business,andtechembeddedsequenceplotofthedata?I believea systems nicalproblems;the client'speople at all levels;and the viewpoint of statistics in relationto thescientific method technology of the client'sproductsand services(Marcan elucidatetheissuesin all applications, including such quardt1979,1981). seemingly simpleones as controlcharts. Let me citetwoillustrative examplesof ourfailureto Students willcometothisintroductory college-level sta- actas entrepreneurs center instatistics. Does thecomputer tisticscoursealreadyfamiliar withsimplestatistical con- inyourorganization users havea "helpdesk"forcomputer ceptsand tools fromtheirexperiencein gradesK-12. to giveadviceon use ofstatistical packagesandthelike? whentheyemergefrom Students, thisintroductory course, Have you stoppedto thinkthatthathelpdeskis one of willhavetheperspective that staintotheimportant themostusefulpointsofentrance is assigned to organization? Who tistical markets of your * thediscipline ofstatistics dealswiththecentralissues the desk? Is it a clients eneven statistician? If the help ofthemainstream oflife countera shoeclerkapproach,we aretellingthemthatis * thestatistician's roleis atthecenter, fully comparable what statistics is all about.We haveno one to blamebut in visibility and influence to theroleof othermainourselves. streamtechnologies And now,the"helpdesk"situation has explodedinto * individual statistical methods arerelatedtoeachother withtheproa more complex problem bigger and much and to thediscipline, in termsofwhereand howthe and the computing hardware liferation of decentralized methodsfitintotheoverallsystems concept. reality.We as a functioning emergence ofexpertsystems Thisunifying conceptwillthenenablefaculty to shape havea majorchallenge aheadtobuildintothesesoftware/ subsequent courseofferings and students to digestcourse hardware thesystems approachto thescientific creations understandings so thattheyelucidatetheirroleinthesys- method.For example,letme remind youthatthepeople temsframework. Subsequentcoursescan thenget into writing manyof thesesoftwaresystemsare not highly finepointsof detailin a givenarea withoutlosingper- trainedand experienced statisticians. Here, again,is the spectiveon the wholeprocess.Subsequentcoursescan helpdeskproblemin a newguise.Whata fielddaythis delve more deeplyinto the intricacies of the scientific couldbe forteamsoftheoretical and appliedstatisticians methodin variouscontexts.For example,discussions of and software experts!And yetwe haveso fewteamson theassumptions requiredforvalidapplication ofa statis- thefieldthatwe are in dangeroflosingbydefault. ticalmethodshouldbe seen as one facetof applying the ofaccepting an entrepreneurial One oftheimplications scientific method. is a responsibility role at the centerof the mainstream to considerthe consequencesof our guidanceto clients 10. ROLE OF APPLIEDSTATISTICIANS and colleagues.I have emphasizedon otheroccasions ofthequalityof statisticians are (Marquardt1984a)thatthemanagement in the life of Changes manyapplied to threeinattention this a or service In requires equal product role model I have described. under the required as terrelated facets: cannot statisticians rolemodel,applied operatesimply mustbecomeenshoe clerks(Bross 1974). Statisticians thatunderlies 1. the qualitymanagement philosophy cannotsimply trepreneurs. BythisI meanthatstatisticians ina forideallyexpressed theoperationoftheenterprise, sitandwaitfora clientto comein andrequesta multiple malpolicy ora timeseriesanalysis. regression analysis, Instead,those thatare used to achieve, 2. the management systems who do and statistical serviceorganizations statisticians monitor, improve,and controlthe financialand human are responsible for consulting oftheenterprise relations performance thestatissystems and,inparticular, * identifying 3. thetechnology servicesneededbyclients thestatistical imthatareusedto achieve,monitor, ticalmethodologies * convincing clientsto tackleproblems or services and re- prove,and controlthequalityoftheproducts * developing proposedsolutions(on statistical producedbytheenterprise. latedaspects) * convincing theresults clientsto implement Withinthe thirdfacet,the technology facet,mostreal * long-term ofperformance ofsolutions monitoring problemscan be solvedadequatelywithanyof several * totalcostcontrolofservices. In theirroleas sysvariations ofstatistical methodology. tems to issues,statisticians trained as statisticians resolving tough, fuzzy guides those Likewise, personsoriginally have discovered who take an is within line function the approach entrepreneurial whosecurrent job organization enstatistical that some in out approaches to bear acceptable technically willbringthesesystems concepts carrying improvements in thephilosophy couragecomplementary theirwork. systems facetsoftheclient'senterprise. In everycase,theappliedstatistician, especially thecon- andmanagement sultantstatistician, mustdeveloptheself-image ofan en- Other,equallygood technicalapproachestendto cause responsesin the otherfacets(Martrepreneur. The entrepreneur statistician willdefinethe counterproductive 6 Journalofthe American StatisticalAssociation,March 1987 quardt1984b).These"reportcard"consequences should be carefully analyzedand anticipated by statisticians in developingtheirrecommendations and guidance to clients. Boardsof Directorshave been theprimary participants, withcontributions fromSections,Committees, Chapters, and Council.In theJune1986issueofAMSTATNews,I co-authored withPastPresident JohnNeterandPresidentElectBarbaraBailara description ofthestrategic planning 11. THE FOUR SUMMARIZINGTHEMES pathand wherewe are today.Twenty-one common-inin the strategic planning. I have examinedseveralwaysin whichthisexpanded terestgroupswere identified groupsare marketsegments that perspective shouldhave an impacton the educationof Thesecommon-interest statisticians andtheroleofappliedstatisticians. Thereare theASA mustserve.Theyencompassa widerangefrom departments to federalgovernment manyotherissuesto be consideredin an overallgame academicsinstatistics industrial statisticians, teachersin gradesKplan.I wantto leave withyoufoursummarizing themes: statisticians, 12,usersofstatistical software, thegeneralpublic,and 15 1. The importance ofstatisticians is thattheyshouldbe othergroups.In ourstrategic planning we reaffirmed that systems guidesto tough, fuzzyissueswherever thecollec- the ASA has responsibilities to all of thesegroupsand tionand interpretation ofdatashouldbe involved. mustprovideservicessuitedto theneedsof all of these 2. The fundamental role of statisticians is to be pur- groups.Thisrequiresa marketing approachwithproducts veyorsofthescientific method.Thisis thecommonintel- and servicesthatare different forthe different groups. lectualcontent thatunitesall statisticians. Although prac- Manyof the high-priority willbe publications activities ticing statisticians shouldhaveanunderstanding ofgeneral and otherservicestargeted forresearch-oriented statististatisticalmethodology, includingdistribution theory, cianstrainedat thegraduatelevel,butwe shouldprovide probability, and inference, and be skilledin suchwidely also forpriority needsof otherpractitioners. The ASA usedmethodologies as regression analysis,varianceanal- has a verylargenumberof activities alreadyunderway ysis,experimental design,and samplingprocedures, no formostofthecommon-interest groups.We setup three one statistical methodology is decisiveas commonintel- taskforceson organizational on programs, and flexibility, lectualcontentforall statisticians. on interactions withotherorganizations. Thesetaskforces 3. The businessofstatistics is themarketing ofthesci- have developedpreliminary proposalsforchangesto reentificmethod.Specificstatistical methodsare products moveobstaclesand to createnew opportunities forthe withinthe businessof statistics. As withothertechnical ASA to servestatisticians andtheirpublic.In thecoming disciplines,we find some statisticians are generalists months thethreepresidents andthetaskforcechairswill whereasothersbecomespecialists in certainproducts. preparean integrated forstrategic document planning pur4. The challengeto appliedstatisticians is to become poses.I hopetheASA inthecomingyearscanweavethe entrepreneurial practitioners and consultants in the sci- themesof thisarticleintoall of theseactivities so that entific method. statisticians can achieve,andbe perceivedas deserving, a oursociety. throughout Radicalas thesethemes maysoundinterms oftheevery- statureofhighimportance daybehaviorof manystatisticians, theyreallyare totally [Received September 1986.] consistent withthetheoretical ofourfield. underpinnings REFERENCES Forexample,theyepitomizetheapproachtakenbyearly statistical leaderssuchas Karl Gauss and RonaldFisher Box, GeorgeE. P. (1979),"SomeProblemsofStatistics andEveryday Life"(ASA Presidential Address),Journal oftheAmerican Statistical (1925) and by recentpresidents of the ASA who have Association, 74, 1-4. operatedin a diversity of applicationareas suchas W. Bross, IrwinD. J.(1974),"TheRoleoftheStatistician: orShoe Scientist Allen Wallis (Wallis and Roberts1956), Leslie Kish Clerk,"TheAmerican Statistician, 28, 126-127. David,EdwardE. (1985),"The FederalSupportofMathematics," Sci(1978),GeorgeBox (1979),and others. entific American, 252,45-51[basedon"Reviewing U.S. Mathematics: On a theoretical and philosophic basisthesubjectsof CriticalResourceforthe Future,"by the Ad Hoc Committee on thephilosophy ofscienceandthefoundations ofstatistics ResourcesfortheMathematical Sciences(NationalResearchCouncil), are, in principle,indistinguishable except,possibly,for publishedbytheNationalAcademyPressin 1984]. Encyclopaedia Britannica (1975),"Science,Philosophy of,"16,375-393. areas of emphasisin describing or developingthebases Fey, JamesT., Albers,Donald J., and Fleming,WendellH. (1981), forthe scientific method(see EncyclopaediaBritannica "Undergraduate Mathematical Sciencesin Universities, Four-Year Colleges,andTwo-YearColleges,1980-81,"report, Washington, DC: 1975). On an appliedbasis it does not matterthatthe Conference BoardoftheMathematical Sciences. scientific methodis nota fixed,immutable setofconcepts Fisher, RonaldA. (1925),"Statistical MethodsforResearchWorkers," and procedures.It embodiesa unifying and philosophic Edinburgh: Oliver& Boyd. and Statisticians" (ASA Presiproceduralbasis.Appliedstatisticians, theoretical statis- Kish,Leslie(1978),"Chance,Statistics Address),Journal oftheAmerican Statistical Association, 73, ticians,and philosophers can makethejourneytogether dential 1-6. withfarbetterresultsthaniftheyattempt Marquardt, cooperatively, inIndustry," DonaldW. (1979),"Statistical Consulting The 33, 102-107. theirjourneysseparately, towardseparatedestinations. AmericanStatistician, 12. THEROLE OF THEASA I mustsaya fewwordsabouthowallthisrelates lFinally, to theAmericanStatistical Association. In 1985we began a broad-based strategic planning effort. The 1985and1986 of Statistical (1981), "CriteriaforEvaluatingthePerformance Consultants in Industry," TheAmerican Statistician, 35,216-219. forMan(1984a),"New Technicaland EducationalDirections agingProductQuality,"TheAmerican Statistician, 38, 8-14. -~(1984b),"ReportCardIssuesin QualityManagement," unpublishedShewellAwardpaperpresented at theASA/ASQCFall TechnicalConference, London,Ontario. Marquardt: The Importance of Statisticians 7 Minton,Paul D. (1983),"The Visibility of Statistics as a Discipline," Strategy of Experimentation (1975),Wilmington, DE: E. I. du Pontde TheAmerican Nemours. Statistician, 37, 284-289. Neal,HomerA. (1986),"ReportoftheNSB TaskCommittee on Un- Wallis,W. Allen (1966), "EconomicStatistics and EconomicPolicy" (ASA Presidential dergraduate Scienceand Engineering Address),Journalof theAmerican Education,"ReportNSB-86Statistical As100,Washington, DC: NationalScienceBoard. 61, 1-10. sociation, Schein,EdgarH. (1985),Organizational Culture and Leadership, San Wallis,W. Allen,and Roberts,HarryV. (1956), "Statistics, A New Approach,"Glencoe,IL: FreePress. Francisco:Jossey-Bass.