消费者选择模型在市场营销中的个案分析

1AbstractThispaperintroduces2000EconomicsNobelPrizewinningdiscretechoicemodelanditsestimationmethod.Itdevelopsaconsumerchoicemodelingframeworkbasedonthediscretechoicetheoryandestimatesfactorsthataffectarticlereadingprobabilityonadatasetof57505readingdecisionsextractedfromthesurveyof465readersofthreemajorShanghainewspapersin2005.Thepaperidentifiesinterestingandsignificantreadingpatterns.Insummary,therearemainlythreegroupsofarticlefactorsthataffectreadingdecisionsofnewspaperreaders.Thefirstgroupincludescontentrelatedcharacteristicsandtheyaffectarticlereadingprobabilitybyabout20%to60%.Thesecondgroupincludesfactorsrelatedtopublicationtime,pageandsectionandtheyaffectarticlereadingprobabilitybyabout5%to20%.Thethirdgroupincludesfactorsrelatedtoarticledisplayformatsuchasitspositiononapageandthesefactorsaffectarticlereadingprobabilitybyabout2%to5%.KeyWordsNewspaper,ChinaMediaStudy,DiscreteChoiceModel,ConsumerChoiceModeling1Dr.Wangisanappliedeconometricianandanexpertindiscretechoicemodels.Dr.WangobtainedhisPh.D.ineconomicsfromBostonUniversity.HisdesignofConsumerChoiceModelingframeworkhasbeenappliedinChinabySinomonitorShanghaiforcompaniessuchasShanghaiVolkswagon,DongfengNissanPassengerVehicleCompany,ShanghaiMobileCommunicationsCo.,andJiefangDailyGroup.2004,234kjkjxuβ=∂∂/∑==JjjjjVVP1)exp()exp(2F(x)=Pr(εjx)=exp{-exp[-((x/ab))]}a0,b=0.5772561111iiiieYXu+++=ηγβ2222iiiieYXu+++=ηγβijijijijeYXu+++=ηγβ1111−−−−+++=ininiinineYXuηγβinineu=},{1.....21−iniiηηη7},...,{21Lγγγγ=γ≥≥))ln(Pr(1},{jdMaxiNi=∑=γβ∑∑==⎪⎪⎩⎪⎪⎨⎧⎪⎪⎭⎪⎪⎬⎫++++==NSnsnjnsijjjnsijjjiYXYXNSjd11)exp()exp(1)Pr(ηγβηγβγjkiXjd∂=∂)Pr(3SimulatedMaximumLikelihoodEstimation4Gourieroux,ChristianandAlainMonfort(1996)Wang(2001)8520051121~25;9111iiieYXu++=γβ00iieu=},...,{21Lγγγγ=γ;;;31/*31/*×3×5465;25~45//3163115465757505153762055321576109122263455740921227633484557437112741356539375239105713162393752391100530383658430910≥)exp(1)exp()1Pr(111γβγβiiiYXYXd+++==)exp(11)0Pr(11γβiiYXd++==)exp(1)exp()1Pr(γβγβijijijYXYXd+++==)exp(11)0Pr(γβijijYXd++==118930101211121313⎭⎬⎫⎩⎨⎧=−+==∑∑∑∑jiijijjiijijddddMaxLLnMax)0Pr(ln)1()1Pr(ln)(,,γβγβ14γγγγ15jkiXd∂=∂)1Pr(141615171819Berry,StevenT.,JamesLevinsohn,andArielPakes,1995“AutomobilePricesinMarketEquilibrium”Econometrica63(July)841-90Berry,StevenT.,JamesLevinsohn,andArielPakes,2004“DifferentiatedProductsDemandSystemfromaCombinationofMicroandMacroData:TheNewCarMarket”JournalofPoliticalEconomyvol112,FebGourieroux,ChristianandAlainMonfort(1996),SimulationBasedEconometricMethods,OxfordUniversityPressMcFadden(1974)ConditionallogitanalysisofqualitativechoicebehaviorI.Zarembaka(ed.)FrontiersinEconometrics,105-142.NewYork:AcademicPressChang,J.(1991)ASimultaneousApproachtotheWhether,What,andHowMuchtoBuyQuestions”,MarketingScience,10,4(Fall),p.297-315Chintagunta,P(1993)InvestigatingPurchaseIncidence,BrandChoiceandPurchaseQuantityDecisionsofHouseholds.MarketingScience,Vol.12,No.2Spring1993Luce,R.D.andP.Suppes.1965Preference,UtilityandSubjectiveProbability.HandbookofMathematicalPsychology,vol.3,ed.NewYork:WileyWang,Fenghua(2001)EstimatingtheEffectsofAdvertisingonBrandandQuantityChoices.BostonUniversityPh.D.DissertationShenYujing,AnnHendricks,FenghuaWang,JohnGardner(2005)Theimpactofprivateinsurancecoverageonveterans’useofcare:insuranceandselectioneffectsunderrevisionatHealthServicesResearchJournal