Consulting. Real Problems, Real Interactions, Real Outcomes

download Consulting. Real Problems, Real Interactions, Real Outcomes

of 4

Transcript of Consulting. Real Problems, Real Interactions, Real Outcomes

  • 8/2/2019 Consulting. Real Problems, Real Interactions, Real Outcomes

    1/4

    Real Problems, Real Interactions, Real Outcomes

    Author(s): Richard Tweedie and Sue TaylorSource: Statistical Science, Vol. 13, No. 1 (Feb., 1998), pp. 1-3Published by: Institute of Mathematical StatisticsStable URL: http://www.jstor.org/stable/2676708

    Accessed: 08/11/2010 21:38

    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 athttp://www.jstor.org/action/showPublisher?publisherCode=ims.

    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 [email protected].

    Institute of Mathematical Statistics is collaborating with JSTOR to digitize, preserve and extend access to

    Statistical Science.

    http://www.jstor.org/action/showPublisher?publisherCode=imshttp://www.jstor.org/stable/2676708?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=imshttp://www.jstor.org/action/showPublisher?publisherCode=imshttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2676708?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=ims
  • 8/2/2019 Consulting. Real Problems, Real Interactions, Real Outcomes

    2/4

    Statistical Science1998, Vol. 13, No. 1, 1-29Consulting:Real Problems,Real Interactions,Real OutcomesRichardTweedie (Resources Appendix by Sue Taylor)

    Abstract.The Pullmanmeetingf MS-WNAR had, as one of tsthemes,"Statistical onsulting."n this overview fthecase studiespresentedthere, n attempt s made to drawtogetherome of he essons of hesepapers, howing hediverse ole of he statisticiann collecting,nalyz-ing and presenting he nformationontained nthe data.Key words nd phrases: Statistical onsulting, lient nteractions,on-sulting ibliography,onsulting esources

    1. THE PULLMAN ESSIONSThe papers n the Pullmanpanel discussionwerepresented y nvitation t the MS-WNAR meetingheld at Washington tate Universityn June 1996.I organized his session. have, over the years,organized nd participatedn a numberof suchsessions at conferences oth n the United Statesand inAustralia, nd it is clearly perennial opicforstatisticalmeetings, nly perhaps rivalled bysessionsonthegap between academic" nd "prac-tical" statisticians, r sessions on how to teachundergraduate ervice coursesin our notoriouslyuncharismaticubject.This time was asked to organize he session n

    order o atonefor, r perhaps amplify, ome com-ments had made on theroleofconsultants omeyearspreviously.n Tweedie 1992)I wrote hat hecomments n consulting f the Committee n NewResearchers CNR; New Researchers ommittee fthe IMS, 1991) were "glib": n particular, dis-agreed rather trongly iththe Committee tate-ment that "... unless you need the data analysisexperience our ole as a consultant]s todispense

    RichardTweedie s Chair,Department fStatistics,Colorado State University,ort Collins, Colorado80523-1877 (e-mail: [email protected]),and Director f theCenter orAppliedStatisticalExpertise here; ewas a consultantn CSIRO andtheprivatesector n Australia (1974-1987). SueTaylorwas Consultant, lindersUniversityf outhAustralia 1990-1995); she s now Ph.D. student,UniversityfColoradoHealthSciencesCenter, en-ver,Colorado, nd an active onsultantnepidemi-ology nd medical tatistics.

    advice. The clientshould be responsiblefor theactual analysis."This seemed to me to be veryoptimistic, r perhaps very pessimistic ependingon one'sviewpoint.n almost ny nteraction haveever been nvolvedwith, statistician,speciallynew researcher,s not usually seenby clients s aguruon a mountaintop.tatisticianswillnotgo farif hey dopt hatrole, t least not n a real consult-ing situationwhere t is critical o understandvariety f client-consultantnteractions hat willinfluence he requirements or effective onsulta-tion:see the Appendix ornumerous iewson thereal complexityfthe consulting ole.Moreover, the CNR article advises that "... ifyouhaveput n substantial ffort,n terms ftimeordeveloping ew techniques, ou should sk tobea co-author"nthepaperthat s assumedtobe theend-productfconsulting. felt hat this also indi-cateda serious misapprehensionboutwhatmostconsulting as about, venfor hosewhoserolewasto be "new researchers" ather han full-timeon-sulting tatisticians.Of course, nemight e usti-fiablycynicalabout the reasons forthese state-ments:giventhe prevailing riteria orpromotionin academia nparticular, heCNR mayhave beenmore ealistic hanonemightwish ntheir dvice.)Even so,in the belief hat consultings valuabletoacademic ndperhapsmoreparticularlyonaca-demic statisticians,with Deb Nolan and LuAnnJohnsonthe conferencerganizers t Pullman)adouble-barrelledession was organized:the firsthalfwould describe ome real consulting roblemsthat might llustrate he range of activities hatconsultants ace,of nterestn theirownright swellas illustrative f he many acets fourprofes-sionbeyondmereadvising nd coauthorship;ndthe secondpartwouldbe a panel discussion, iving

    1

  • 8/2/2019 Consulting. Real Problems, Real Interactions, Real Outcomes

    3/4

    2 TWEEDIE ET AL.anecdotes nd advice and insights,with wehoped)strongudienceparticipation.This workedout surprisingly ell, and the pa-pers thatfollow re from he first art of the ses-sion:regrettably,he nsight, xperiencend wit ofthe secondpartare lost to all butthosewhowerethere,although the bibliographynd WWW re-sources ppendedwill enable nterested eaderstodelve furthernto the available literature n thistopic and share at least one ofthe more tangiblebits of dvicefrom hepanel.

    2. THE PARTICIPANTSIn what followsyou will see four rticles fromfourrather differenterspectives.When invitingspeakers,we lookedfor a spectrum f statisticalbackgrounds:heparticipantsepresentheexperi-ences of graduate students, f university acultyand of tatisticians orkingull ime s consultantsin academia, n-housewith a governmentnstru-mentalitynd in a private ector apacity.Karl Broman, Ph.D. student romU.C. Berke-ley, llustrates he bestof the classic "on-campus"interactions: good scientificollaborationeven apotential co-authorship!); ome new techniquesneeded nd developed; nd an outcome f cademicvalueto bothparties. his s verymuch nthemoldoftheprojects nvisionedn the CNR article,butevenherethe consultants doing he analysis,notjust assuming he clientwillcarryt out.Jennifer oeting, new researcher owat Col-orado State, describesthe role of the academicstatisticians funded dvisor-consultantn a proj-ect. She illustrates he way nwhich arefulnves-tigation f he data sources s vitalfor nyanalysistobemeaningfuland lack ofknowledgef hedatacollection,r a limited lientunderstandingfthedata under tudy, an cause a well-planned naly-sis to be inappropriate).Sue Taylor escribeshe ort f areprojectwherethe statisticians actuallynvolved arly nough obe ableto nfluence ata collection ndenhance he

    ability oanalyze.As a consultant o theAustralianLongitudinal tudyofAgeing, he was able to en-sure withmuchwork) hat nalysiswould ctuallybe carried out on reasonablyclean and reliabledata, thussavinga largeamountof ater work nanalysis.Missing s thepaperfrom ob O'Brienfrom at-telle.From the-perspectivef an in-house onsul-tant,he described majorenvironmentalroblem,and onewheremuchoftheconsultant's ole ay intrying o determinewhat the real goals and con-straintswere,since therewere very manystake-

    holders n the problem: tatistical nalysiswouldprovide the answers if only the questionswereclear. This omission evealsagain the priorities fmany onsultants: is ongoing utiespreclude venthe writingf sole-authoredaper.Finally,wearing n oldishhat as a private ectorconsultant, describe situationwheremodellingdoes work, nd an appropriate nalysisyieldsacounterintuitivend effective olution-but yetagain, only afterthe data are revisited nd thewholecontext fthe problem s understood.

    3. CONCLUSIONSMoststatisticians ill find omethingamiliarnthese case studies.We know thatunderstandingur data and thequestionsof the clientare of paramount mpor-tance: t is reassuring osee an examplewhere hestatistician an control hat process Taylor).Weknow that close examination an revealfar morethan the client originally old us (Hoeting).Weknowthatproblems re rarely tandard, nd thatat its best statistical hinking an come up withnew waysto cope withnew problems Broman), rperhapsmoretypically e can see the roleof ourassumptions nd decidehow well theoldwaysfit(Tweedie).However, n the end, these case studies,andmany therwar stories hatmany fus tell or haveheard, ll illustrate wothings verwhelmingly.First, no matterwhat subject areas we enter,statistics an contribute omethinghat was nottherepreviously,nd we have much to offer oalmosteveryone. hese examplescover benchre-searchproblems,ocialsurveys,managementrac-ticesand environmentalroblems n a local and aglobal scale. Without tatistical hinking, one ofthemwould be solvable.Moreover, hey howthatit is often he mere fact of such thinking,atherthan the specific echnical nput,that proves n-valuable. t is hard to overestimateowpowerfullyourdiscipline rainsus tothink boutcomplicated

    issues in ways that allow us to quicklydiagnosedifficultiesn esoteric isciplines owhichwe havehad onlyseveralminutesof introduction-afactreinforcedytheseexamplesand even further ytherefereesfthis collection.Butsecond, or hatcontributiono be truly t itsbest, hestatisticianmustenter nto hecontext ftheproblem, ot ust as an "advisor," ut as some-one prepared o understand hedata, analyzethedata, interactwiththosewhoreallyowntheques-tionsbeing skedand consider he mpact f tatis-tics withinthe real context f the problem. he

  • 8/2/2019 Consulting. Real Problems, Real Interactions, Real Outcomes

    4/4

    CONSULTING 3Pullman case studies show many of these at-tributes, ut also illustrate ividlyheproblemswehave in achieving uchan idealistic tate.

    RESOURCES APPENDIXBibliography for Consultants

    One of the mostuseful ools for onsultantss abibliographyf nformationut together y otherconsultants. he papers belowdo notpretend o beexhaustive: hey epresent hose that have foundofmostuse inmyprofessional ractice.They also contain ointers o othermaterial ov-ering a wider range if needed.Those referencesmarkedwithan asterisk ontain n extensive istof statistical onsulting eferenceswhich are notduplicated ere.ASAAD Hoc COMMITTEE N PROFESSIONAL THICS 1983). Ethi-cal guidelines or tatistical ractice: eport fthe Ad HocCommitteen Professional thics.Amer. tatist. 7 5-20.*BASKERVILLE,.C. (1981).A systematictudy f he consultingliteratures an integral artofappliedtrainingn statis-tics. Amer. tatist. 5 121-123.BOEN, J. R. (1972). The teaching f nterpersonalelationshipsin statistical onsulting. mer. tatist. 6 30-31.BOEN, J. R. and FRYD, . (1978). Six-state ransactionalnalysisin statistical onsulting. mer. tatist. 2 58-60.BOEN, J. R. andZAHN,D. A. (1982).TheHuman Side of tatisti-cal Consulting. ifetime earning ublications,A.CHATFIELD, C. (1988). Problem olving:A Statistician'sGuide.Chapman ndHall,London.CHATFIELD, . (1991). Avoidingtatistical itfalls. tatist. ci. 6240-268.ELLENBERG, .H. (1983). Ethicalguidelines or tatistical rac-tice: a historicalprospective.Amer. tatist. 7 1-4.HAND, . J. and EVERITT, . S. (1987). TheStatistical onsultantin Action. ambridgeUniv. Press.HUNTER,W. G. 1981).Thepractice f tatistics: hereal world san idea whose time has come. Amer. tatist. 5 72-76.HYAMS,. (1971).Thepractical sychologyfbiostatisticalon-sultation.Biometrics7 201-211.*JOINER, . L. (1961). Consulting, statistical. In EncyclopaediaofStatistical ciences 47-155. Wiley,New York.JOWELL, . (Chairman) (1986). InternationalStatistical Institutedeclaration on professionalethics. InternationaltatisticalReview 4 227-242.KIRK, R. E. (1991). Statistical consulting n a university:dealingwith people and otherchallenges. Amer. tatist. 5 28-34.RUSTAGI, J. S. and WOLFE,D. A. (1982). Teaching fStatisticsand Statistical onsulting. cademic Press, New York.SLOAN, J. A. (1992). How to consultwitha statistician. heStatistical onsultant 3-4.*WOODWARD,W.A. andSCHUCANY,W.R. 1977).Bibliographyorstatistical consulting.Biometrics3 564-565.

    ZAHN, . A. and SENBERG, D. J. 1980). Non-statisticalspects fstatistical onsulting.n Proceedingsfthe tatistical du-cation Section 67-72. Amer. Statist. Assoc., Washington,DC.Electronic Resources

    A searchof the WorldWide Web as of October1996 revealed a large number f sites relevant oconsultants. he followingre just a fewof these;we have found them to be useful entrypoints,although hey re by no means intended o coverthegrowingnformationesource vailable on theWorldWide Web:. The World-WideWeb VirtualLibrary: tatistics,

    http://www.stat.ufl.edu/vlib/statistics.htm. Statistics n the Web,http://www.execpc.com/elberg/

    statistics.html. StatisticsResources n the Web,http://wwwstats.gla.ac.uk/cti/linkstats html. A Guide to StatisticalComputing esourcesonthe nternet,http://asa.ugl.lib.umich.edu/chdocs/statisticsstat_guide_home.html. An "Essential BookList," nd useful fa year ortwoolder handesirable,

    http://www.stat.wisc.edu/statistics/consultstatbook.html

    A veryuseful ndup-to-dateocument,he "ListofStatistics ists" s also availablebysending heone-linemessagesend minitab ist-of-liststo [email protected] pointing ourWeb browser thttp://www.mailbase.ac.uk/lists-k-o/minitabfiles/list-of-listsThis documentontains etailsof ll currenttatis-tics-related -mail ists, ncluding ubscriptionn-formation.heseenableconsultantso share nfor-mation r conduct iscussions n a timely asis.