Decision making - the Analytic Hierarchy and Netwo

ISSN1004-3756/04/1301/1JOURNALOFSYSTEMSSCIENCEANDSYSTEMSENGINEERINGCN11-2983/N©JSSSE2004Vol.13,No.1,pp1-35,March,2004DECISIONMAKING–THEANALYTICHIERARCHYANDNETWORKPROCESSES(AHP/ANP)[e-mail]tpartofanintroductiontomulticriteriadecisionmakingusingtheAnalyticHierarchyProcess(AHP)anditsgeneralization,theAnalyticNetworkProcess(ANP).ThediscussioninvolvesindividualandgroupdecisionsbothwiththeindependenceofthecriteriafromthealternativesasintheAHPandalsowithdependenceandfeedbackintheentiredecisionstructureasintheANP.ThispartexplainstheAnalyticHierarchyProcess,withexamples,andpresentsinsomedetailthemathematicalfoundations.AnexpositionoftheAnalyticNetworkProcessanditsapplicationswillappearinlaterissuesofthisjournal.Keywords:Decisionmaking,AnalyticHierarchyProcess(AHP),AnalyticNetworkProcess(ANP)1.Introduction(Saaty1977,1994,2000a,2000band2001)Decisionmakinginvolvescriteriaandalternativestochoosefrom.Thecriteriausuallyhavedifferentimportanceandthealternativesinturndifferinourpreferenceforthemoneachcriterion.Tomakesuchtradeoffsandchoicesweneedawaytomeasure.Measuringneedsagoodunderstandingofmethodsofmeasurementanddifferentscalesofmeasurement.Manypeoplethinkthatmeasurementneedsaphysicalscalewithazeroandaunittoapplytoobjectsorphenomena.Thatisnottrue.Surprisinglyenough,wecanalsoderiveaccurateandreliablerelativescalesthatdonothaveazerooraunitbyusingourunderstandingandjudgmentsthatare,afterall,themostfundamentaldeterminantsofwhywewanttomeasuresomething.Inrealitywedothatallthetimeandwedoitsubconsciouslywithoutthinkingaboutit.Physicalscaleshelpourunderstandinganduseofthethingsthatweknowhowtomeasure.Afterweobtainreadingsfromaphysicalscale,theystillneedtobeinterpretedaccordingtowhattheymeanandhowadequateorinadequatetheyaretosatisfysomeneedwehave.Butthenumberofthingswedon’tknowhowtomeasureisinfinitelylargerthanthethingsweknowhowtomeasure,anditishighlyunlikelythatwewilleverfindwaystomeasureeverythingonaphysicalscalewithaunit.Scalesofmeasurementareinventionsofatechnologicalmind.Ourmindsandwaysofunderstandingwehavehadwithusandwillalwayshave.ThebrainisanelectricaldeviceofneuronswhoseDecisionMaking–TheAnalyticHierarchyandNetworkProcesses(AHP/ANP)2JOURNALOFSYSTEMSSCIENCEANDSYSTEMSENGINEERING/Vol.13,No.1,March,2004firingsandsynthesismustperformmeasurementwithgreataccuracytogiveusallthemeaningandunderstandingthatwehavetoenableustosurviveandreachouttocontrolacomplexworld.Canwerelyonourmindstobeaccurateguideswiththeirjudgments?Theanswerdependsonhowwellweknowthephenomenatowhichweapplymeasurementandhowgoodourjudgmentsaretorepresentourunderstanding.Inourownpersonalaffairswearethebestjudgesofwhatmaybegoodforus.Insituationsinvolvingmanypeople,weneedthejudgmentsfromalltheparticipants.Ingeneralwethinkthattherearepeoplewhoaremoreexpertthanothersinsomeareasandtheirjudgmentsshouldhaveprecedenceoverthejudgmentsofthosewhoknowlessasinfactisoftenthecaseinpractice.Judgmentsexpressedintheformofcomparisonsarefundamentalinourbiologicalmakeup.Theyareintrinsicintheoperationsofourbrainsandthatofanimalsandonemightevensayofplantssince,forexample,theycontrolhowmuchsunlighttoadmit.Weallmakedecisionseverymoment,consciouslyorunconsciously,todayandtomorrow,nowandforever,itseems.Decision-makingisafundamentalprocessthatisintegralineverythingwedo.Howdowedoit?TheHarvardpsychologistArthurBlumenthaltellsusinhisbookTheProcessofCognition,Prentice-Hall,Inc.,EnglewoodCliffs,NewJersey,1977,thattherearetwotypesofjudgment:“Comparativejudgmentwhichistheidentificationofsomerelationbetweentwostimulibothpresenttotheobserver,andabsolutejudgmentwhichinvolvestherelationbetweenasinglestimulusandsomeinformationheldinshorttermmemoryaboutsomeformercomparisonstimulioraboutsomepreviouslyexperiencedmeasurementscaleusingwhichtheobserverratesthesinglestimulus.”Whenwethinkaboutit,boththeseprocessesinvolvemakingcomparisons.Comparisonsimplythatallthingsweknowareunderstoodinrelativetermstootherthings.Itdoesnotseempossibletoknowanabsoluteinitselfindependentlyofsomethingelsethatinfluencesitorthatitinfluences.Thequestionthenishowdowemakecomparisonsinascientificwayandderivefromthesecomparisonsscalesofrelativemeasurement?Whenwehavemanyscaleswithrespecttoadiversityofcriteriaandsubcriteria,howdowesynthesizethesescalestoobtainanoverallrelativescale?Canwevalidatethisprocesssothatwecantrustitsreliability?Whatcanwesayaboutotherwayspeoplehaveproposedtodealwithjudgmentandmeasurement,howdotheyrelatetothisfundamentalideaofcomparisons,andcantheybereliedonforvalidity?Theseareallquestionsweneedtoconsiderinmakingadecision.Itisusefultorememberthattherearemanypeopleintheworldwhoonlyknowtheirfeelingsandmayknownothingaboutnumbersandneverheardofthembutcanstillmakegooddecisions,howdotheydoit?Itisunlikelythatbyguessingatnumbersandassigningthemdirectlytothealternativestoindicateorderunderacriterionwillyieldmeaningfulprioritiesbecausethenumbersarearbitrary.Eveniftheyaretakenfromascaleforaparticularcriterion,howwouldwecombinethemacrossthecriteriaastheywouldlikelybe