標題：Consumers’Decision-MakingProcessandTheirOnlineShoppingBehavior原文：IntroductionAmongallpossibleadvantagesofferedbyelectroniccommercetoretailers,thecapacitytoofferconsumersaflexibleandpersonalizedrelationshipisprobablyoneofthemostimportant(Wind&Rangaswamy,2001).Onlinepersonalizationoffersretailerstwomajorbenefits.Itallowsthemtoprovideaccurateandtimelyinformationtocustomerswhich,inturn,oftengeneratesadditionalsales(Postma&Brokke,2002).Personalizationhasalsobeenshowntoincreasethelevelofloyaltyconsumersholdtowardaretailer(CyberDialogue,2001;Srinivasan,Anderson,&Ponnavolu,2002).Whilethereareseveralwaystopersonalizeanonlinerelationship,thecapacityforanonlineretailertomakerecommendationsiscertainlyamongthemostpromising(Thee-tailingGroup,2003).Online,recommendationsourcesrangefromtraditionalsourcessuchasotherconsumers(e.g.,testimoniesofcustomersonretailwebsitessuchasA)topersonalizedrecommendationsprovidedbyrecommendersystems(Westetal.,1999).Todate,nostudyhasspecificallyinvestigatedandcomparedtherelativeinfluenceoftheseonlinerecommendationsourcesonconsumers’productchoices.Therefore,themainobjectiveofthisstudyistoinvestigatetheinfluenceofonlineproductrecommendationsonconsumers’onlineproductchoices.Inaddition,weexplorethemoderatinginfluenceofvariablesrelatedtorecommendationsourcesandthepurchasedecision.LiteraturereviewResearchontheuseandinfluenceofrecommendationsonconsumershastypicallybeensubsumedunderpersonalinfluenceorword-of-mouth(WOM)research.Inaddition,asnotedbyRosenandOlshavsky(1987),researchonopinionleadershipandreferencegroupsalsorelatestothestudyofecommendationsandtoinfluenceingeneral.Recommendationsourcesareconsideredprimarilyasinformationsources.Andreasen(1968)proposesthefollowingtypologyofinformationsources:(1)ImpersonalAdvocate(e.g.,massmedia),(2)ImpersonalIndependent(e.g.,ConsumerReports),(3)PersonalAdvocate(e.g.,salesclerks),and(4)PersonalIndependent(e.g.,friends).AlthoughresearchonpersonalinfluenceandWOMfocusesonthelattertwoinformationsources,itisnoteworthythatimpersonalindependentinformationsourcessuchasConsumerReportscanalsoserveasrecommendationsources.Moreover,theInternetcanprovideconsumerswithanadditionaltypeofimpersonalinformationsource.Forinstance,electronicdecision-makingaidssuchasrecommendersystemsareimpersonalinformationsourcesthatprovidepersonalizedinformationtoconsumers(Ansari,Essegaier,&Kohli,2000).InanefforttoextendAndreasen’s(1968)typologytocomputer-mediatedenvironments,weassertthatinformationsourcescanbesortedintooneoffourgroups:(1)Personalsourceprovidingpersonalizedinformation(e.g.,“Mysistersaysthatthisproductisbestforme.”);(2)Personalsourceprovidingnon-personalizedinformation(e.g.,“Arenownedexpertsaysthatthisproductisthebest.”);(3)Impersonalsourceprovidingpersonalizedinformation(e.g.,“Basedonmyprofile,therecommendersystemsuggeststhisproduct.”);(4)Impersonalsourceprovidingnon-personalizedinformation(e.g.,“AccordingtoConsumerReports,thisisthebestproductonthemarket.”).Inconsumerresearch,studiesonpersonalinfluence,socialinfluence,orWOM,canbecategorizedasstudiesinvestigatingpersonalsourcesprovidingpersonalizedornon-personalizedinformation.Furthermore,studiesdealingwithreferencegroupsencompasssuchsourcesaswellasimpersonalsourcesthatprovidenon-personalizedinformation.Thus,anewareahasemergedinconsumerresearch,arisingmainlyfrominformationtechnologiessuchastheInternet:thatofimpersonalsourcesthatprovidepersonalizedinformation(Albaetal.,1997;Ansarietal.,2000;H?ubl&Trifts,2000;Maes,1999;Urban,Sultan,&Qualls,1999;Westetal.,1999).Researchoninformationsourcessuggeststhatpersonalandimpersonalinformationsourcesinfluenceconsumers’decision-making(Ardnt,1967;Duhanetal.,1997;Gillyetal.,1998;Olshavsky&Granbois,1979;Price&Feick,1984).Forinstance,PriceandFeick(1984)foundthatconsumersplannedtousethefollowinginformationsourcesfortheirnextdurablegoodpurchase:(1)Friends,relatives,andacquaintances,(2)Salespeople,(3)PublicationssuchasConsumerReports.However,ifmuchisknownabouttherelativelikelihoodofconsumerstoconsiderrecommendationsinthecourseoftheirdecisionmakingprocess,littleisknownabouthowrecommendations,especiallyinacomputer-mediatedenvironment,impactconsumers’productchoices.DeterminantsofrecommendationinfluenceThecurrentstudyfocusesonthreedeterminantsthatcouldinfluencetheimpactofcomputer-mediatedrecommendationsonconsumers’onlineproductchoices:thenatureoftheproductrecommended,thenatureofthewebsiteonwhichtherecommendationisproposed,andthetypeofrecommendationsource.Priorresearchhasshownthatthetypeofproductaffectsconsumers’useofpersonalinformationsourcesandtheirinfluenceonconsumers’choices(Bearden&Etzel,1982;Childers&Rao,1992;King&Balasubramanian,1994).Nelson(1970)suggeststhatgoodscanbeclassifiedaspossessing.eithersearchorexperiencequalities.Searchqualitiesarethosethat“theconsumercandeterminebyinspectionpriortopurchase,”andexperiencequalitiesarethosethat“arenotdeterminedpriortopurchase”(Nelson,1974,p.730).Sinceitisdifficultorevenimpossibletoevaluateexperienceproductsbeforepurchase,consumersshouldrelymoreonproductrecommendationsfortheseproductsthanforsearchproducts.Insupportofthisview,KingandBalasubramanian(1994)foundthatconsumersassessingasearchproduct(e.g.,a35-mmcamera)aremorelikelytouseown-baseddecision-makingprocessesthanconsumersassessinganexperienceproduct,andthatconsumersevaluatinganexperienceproduct(e.g.,afilm-processingservice)relymoreonother-basedandhybriddecision-makingprocessesthanconsumersassessingasearchproduct.Thenatureofthewebsitecanalsoinfluencetheimpactofagivenrecommendation.Basedonpreviouswebsiteclassifications(Hoffman,Novak,&Chatterjee,1995;Spiller&Lohse,1998),SenecalandNantel(2002)suggestthatrecommendationsourcescanbeusedandpromotedbythreedifferenttypesofwebsites:sellers(e.g.,retailerormanufacturerwebsitessuchasA),commerciallylinkedthirdparties(e.g.,comparisonshoppingwebsitessuchasMyS),andnon-commerciallylinkedthirdparties(e.g.,productormerchantassessmentwebsitessuchasC).Moreindependentwebsitessuchasnon-commerciallylinkedthirdpartiesthatfacilitateconsumers’externalsearcheffortbydecreasingsearchcostsareassumedtobepreferredbyconsumers(Albaetal.,1997;Bakos,1997;Lynch&Ariely,2000).Byprovidingmorealternativestochoosefromandmoreobjectiveinformation,independentwebsitesshouldbeperceivedasmoreusefulbyconsumers.Inaddition,priorresearchonattributiontheorysuggeststhatconsumersdiscreditrecommendationsfromendorsersiftheysuspectthatthelatterhaveincentivestorecommendaproduct(forreviews,refertoFolkes,1988;Mizerski,Golden,&Kernan,1979).Accordingtothediscountingprincipleoftheattributiontheory(Kelley,1973),whichsuggeststhatacommunicatorwillbeperceivedasbiasediftherecipientcaninferthatthemessagecanbeattributedtopersonalorsituationalcauses,consumerswouldattributemorenon-productrelatedmotivations(e.g.,commissionsonsales)torecommendationsourcesthatarepromotedbycommerciallylinkedthirdpartiesandsellersthanindependentthirdpartywebsites.Consequentlyconsumerswouldfollowproductrecommendationsinagreaterproportionwhenshoppingonmoreindependentthanonlessindependentwebsites.Inlightofresearchonconsumers’useofrelevantothersintheirpre-purchaseexternalsearchefforts(Olshavsky&Granbois,1979;Price&Feick,1984;Rosen&Olshavsky,1987)andinconsiderationoftheemergenceofonlineinformationsourcesprovidingpersonalizedrecommendations(Ansarietal.,2000),SenecalandNantel(2002)assertthatonlinerecommendationsourcescanbesortedintothreebroadcategories:(1)otherconsumers(e.g.,relatives,friendsandacquaintances),(2)humanexperts(e.g.,salespersons,independentexperts),and(3)expertsystemssuchasrecommendersystems.Wepositthattheseonlinerecommendationsourceswillhavedifferentlevelsofinfluenceonconsumers’onlineproductselection.BrownandReingen(1987)suggestthatinformationreceivedfromsourcesthathavesomepersonalknowledgeabouttheconsumerhavemoreinfluenceonthelatterthansourcesthathavenopersonalknowledgeabouttheconsumer.Thus,arecommendationsourceprovidingpersonalizedinformationtoconsumers(e.g.,recommendersystem)shouldbemoreinfluentialthanarecommendationsourceprovidingnon-personalizedinformation(e.g.,otherconsumers).Thefactthatbothfactors,theorigin(source)ofarecommendationaswellasthetypeofwebsiteonwhichitismade,haveanimpactonthelikelihoodithastobefollowedmayfinditsexplanationinKelman’s(1961)workonsourcecredibility.Kelman(1961)suggeststhatcredibilityisaproductofexpertiseandtrustworthiness.Expertisecanbeviewedastheperceivedabilityofaninformationsourcetoknowtherightanswerandtrustworthinessastheperceivedinformationsource’smotivationtocommunicatethisexpertisewithoutbias(McGuire,1969).Althoughmoderatedbycontextualfactors(forareview,refertoSternthal,Phillips,&Dholakia,1978),sourceexpertiseandtrustworthinesshavebeenfoundtobepositivelycorrelatedwithconsumers’attitudetowardthebrand,andconsumers’behavioralintentionsandbehaviors(Gillyetal.,1998;Harmon&Coney,1982;Lascu,Bearden,&Rose,1995;Tybout,1978).HypothesesBasedontheprecedingreviewoftheliteraturewepostulatethatpersonalinformationsourcesaswellasimpersonalinformationsourcesprovidingproductrecommendations(Price&Feick,1984)willinfluenceconsumersincomputer-mediatedenvironmentssuchastheInternetandtheWorldWideWeb.Wethusformulatethefollowinggeneralhypothesis.H1.Consumerswhoconsultanonlineinformationsourcerecommendingagivenbrandwillselectthatbrandinagreaterproportionthanconsumerswhodonotconsultanonlinerecommendationsource.Asfortheimpactthatsucharecommendationwillhaveonconsumers’choice,weformulatethreeadditionalhypotheses.First,wepositthatthenatureoftheproductforwhicharecommendationisprovidedwillinfluencethelikelihoodthatitwillbefollowed.Basedonpriorresearchontherelationshipbetweenproducttypeandpersonalinformationsourceinfluence(Bearden&Etzel,1982;Childers&Rao,1992;King&Balasubramanian,1994),weputforwardthefollowinghypothesis.H2.Onlinerecommendationsforexperienceproductswillbefollowedinagreaterproportionthanonlinerecommendationsforsearchproducts.Second,basedonAlbaetal.(1997),Bakos(1997)andLynchandAriely(2000),weproposethatonlineproductrecommendationsfrommoreindependentwebsitesaremoreinfluentialthanthosefromlessindependentwebsites.Wethereforeputforththefollowinghypothesis.H3.Onlineproductrecommendationsconsultedon“non-commerciallylinkedthirdparty”websiteswillbefollowedinagreaterproportionbyconsumerthanifconsultedon“commerciallylinkedthirdparty”websites,andonlineproductrecommendationconsultedonthelattertypeofwebsiteswillbefollowedinagreaterproportionthanifconsultedon“seller”websites.Finally,webelieve,basedontheliteraturewhichhasdealtwiththeissueofconsumers’useofrelevantothersintheirpre-purchaseexternalsearchefforts,thatpersonalizedrecommendationswillhaveagreaterinfluenceonconsumersthannon-personalizedones(Brown&Reingen,1987).Thusfollowshypothesesfour.H4.Recommendationsfrominformationsourcesofferingpersonalizedrecommendations(e.g.,recommendersystem)willbefollowedinagreaterproportionbyconsumersthanrecommendationsfrominformationsourcesprovidingnon-personalizedrecommendations.Inadditiontothissetofhypotheses,whichpertainstothevariablesthatmoderatetheinfluenceofanonlinerecommendation,weformulateasetofthreehypotheseswhichconsiderpotentialreasonsforwhichvariousonlinerecommendationsourcesmaydifferintheirinfluenceonconsumers’choices.First,weexpectthattherecommendationsource“otherconsumers”willbeperceivedaslessexpertthan“humanexperts”and“recommendersystems”.However,basedonthediscountingprincipleofattributiontheory(Kelley,1967),therecommendationsource“otherconsumers”shouldbeperceivedasmoretrustworthythanhumanexpertsandrecommendersystemssincethelattertworecommendationsourcesaremoresusceptibletonon-productrelatedattributions.Second,sinceconsumersmayalsoattributenon-productrelatedmotivationsmoreeasilytorecommendationsourcespromotedbywebsitesthatarenotclearlyindependent,wepredictthatthetypeofwebsitewillhaveanimpactontheperceptionoftherecommendationsource’strustworthiness.Forinstance,ahumanexpertwhorecommendsaproductonasellerwebsitemaybeperceivedbyconsumersaslesstrustworthythanifthatpersonrecommendedthesameproductonanindependentthirdpartywebsite.Thus,thefollowinghypothesesareposited.H5a.Theonlinerecommendationsources“humanexperts”and“recommendersystem”willbeperceivedaspossessingmoreexpertisethantheonlinerecommendationsource“otherconsumers.”H5b.Theonlinerecommendationsources“humanexperts”and“recommendersystem”willbeperceivedaslesstrustworthythantheonlinerecommendationsource“otherconsumers.”出處：SenealS．，J．Nante1．influenceofonlineproductrecommendationsonconsumers’onlinechoices[J]．JournalofRetail．1ing，2004，80標題：消費者的決策過程和在線購買行為譯文：摘要:由電子商務(wù)向零售商提供的所有可能的優(yōu)勢中，有能力向消費者提供靈活和個性化的關(guān)系可能是最重要的(Wind&Rangaswamy,2001)。在線個性化為零售商提供兩大好處。這使他們能向客戶提供準確和及時地信息，反過來，經(jīng)常產(chǎn)生額外的銷量(Postma&Brokke,2002)。個性化也被證明能提高消費者對零售商的忠誠度水平(CyberDialogue,2001;Srinivasan,Anderson,&Ponnavolu,2002)。雖然存在著多種發(fā)展個性化關(guān)系的方法，然而對于網(wǎng)上零售商提供建議無疑是最有前途的(Thee-tailingGroup,2003)。在線建議范圍從傳統(tǒng)的來源，如推薦系統(tǒng)為消費者（例如，在亞馬遜等零售商網(wǎng)站提供證詞的顧客）所提供給的個性化建議。(Westetal.,1999)迄今為止，還沒有研究專門調(diào)查并比較這些在線推薦來源對消費者產(chǎn)品選擇的相對影響。因此，本研究的主要目的是探討網(wǎng)上推薦產(chǎn)品對消費者在線產(chǎn)品選擇的影響。此外，我們還探討了有關(guān)建議的來源與購買決策變量的干擾影響。文獻：推薦對消費者的作用和影響研究通常歸結(jié)為個人的影響力或口碑的研究。此外，Rosen和Olshavsky在1987年指出，領(lǐng)導(dǎo)和相關(guān)群體觀點的研究大體上也與在線推薦的課題和影響有關(guān)。推薦來源，被認為是首要的信息來源。Andreasen(1968)提出以下幾種信息來源：（1）非個人的提倡（例如，大眾媒介），（2）客觀獨立的（例如，消費者報告），（3）人員建議（例如，銷售員），（4）個人獨立的（例如，朋友）。雖然對個人影響力和口碑研究的重點是后兩種信息源，值得注意的是，如消費者報告的獨立信息源也可以作為建議的來源。此外，互聯(lián)網(wǎng)可以向消費者提供額外類型的的客觀信息來源，比如，為消費者提供個性化信息的推薦人系統(tǒng)這樣的電子決策幫助(Ansari,Essegaier,&Kohli,2000)。在一個延伸Andreasen(1968)類型學理論對以計算機作為媒介的環(huán)境中，我們主張信息資源可分為四大類中的一種：（1）提供個性化信息的個人資料來源（比如，“我妹妹說，本產(chǎn)品對我是最適合的?！保?；（2）提供非個性化信息的個人資料來源(例如，“一位著名的專家說,本產(chǎn)品是最好的”）；（3）提供個性化的客觀源(例如,"根據(jù)我的推測，推薦系統(tǒng)表明這種產(chǎn)品?！保唬?）提供非個性化信息的客觀源（例如，“根據(jù)消費者報告，這是市場上最好的產(chǎn)品?！保?。在消費者調(diào)查中，對個人影響、社會影響以及口碑的研究，可以歸類為提供個性化和非個性化的個人信息來源的研究。此外，研究處理涵蓋這些信息來源的相關(guān)組以及提供非個性化信息的客觀來源。因此，消費者研究的一個新領(lǐng)域出現(xiàn)了，它主要從信息技術(shù)，比如互聯(lián)網(wǎng)——提供個性化信息的客觀來源中產(chǎn)生(Albaetal.,1997;Ansarietal.,2000;H?ubl&Trifts,2000;Maes,1999;Urban,Sultan,&Qualls,1999;Westetal.,1999)。信息來源的研究表明：個人和客觀信息資源影響消費者決策(Ardnt,1967;Duhanetal.,1997;Gillyetal.,1998;Olshavsky&Granbois,1979;Price&Feick,1984)。舉例來說，PriceandFeick(1984)發(fā)現(xiàn)，消費者計劃利用以下信息來源服務(wù)于將來的耐用品購買：（1）朋友、親戚、相識，（2）銷售人員，（3）消費者報告等刊物。然而，如果在消費者決策過程中，消費者考慮相對可以被更多了解的推薦的話，那么推薦對消費者產(chǎn)品選擇的影響，尤其是在電腦媒介的環(huán)境中，將會被更少的了解到。影響推薦的因素目前的研究主要集中在以計算機為媒介的可能會影響消費者在線產(chǎn)品選擇的三個推薦因素:：產(chǎn)品的性質(zhì)的建議，推薦網(wǎng)站的性質(zhì)，推薦來源的類型。之前的研究表明產(chǎn)品的類型會影響消費者對個人信息資源的使用以及其影響對消費者的選擇(Bearden&Etzel,1982;Childers&Rao,1992;King&Balasubramanian,1994).。Nelson(1970)認為商品可以看作是擁有。由于它很難甚至無法評估購買前體驗的產(chǎn)品，消費者應(yīng)該更多地依靠這些產(chǎn)品的產(chǎn)品推薦而不是搜索產(chǎn)品。為了支持這一觀點，KingandBalasubramanian(1994)發(fā)現(xiàn)，消費者評估一個搜索的產(chǎn)品（比如，一個35毫米相機）時，更有可能根據(jù)自己的決策過程而不是其他消費者對體驗的產(chǎn)品的評價，而消費者評價一個體驗產(chǎn)品（如電影處理服務(wù)）則更多地依靠其他和混合型的決策過程而不是消費者對一件搜索的產(chǎn)品的評估。網(wǎng)站的性質(zhì)也會影響到一個既定建議的效果。在(Hoffman,Novak,&Chatterjee,1995;Spiller&Lohse,1998)以往的網(wǎng)站分類的基礎(chǔ)上，SenecalandNantel(2002)提出推薦來源可以通過三種不同的網(wǎng)站類型進行使用和推廣，賣方（例如，零售商或者制造商網(wǎng)站如亞馬遜），商業(yè)聯(lián)系第三方（例如,，MyS一類型的比較網(wǎng)站）和非商業(yè)聯(lián)系第三方（例如,，C這樣的產(chǎn)品或商家的評估網(wǎng)站等）。更多的比如連接第三方團體的獨立網(wǎng)站，通過降低消費者的搜尋成本的外部努力，被假定為消費者的首選(Albaetal.,1997;Bakos,1997;Lynch&Ariely,2000)。通過提供更多的以供選擇的方案和更客觀的信息，獨立網(wǎng)站應(yīng)被消費者認為更加有用。此外，先前的歸因理論表明，如果消費者懷疑代言人有誘因才去推薦產(chǎn)品，則不會相信他們的推薦(forreviews,refertoFolkes,1988;Mizerski,Golden,&Kernan,1979)。根據(jù)貼現(xiàn)(Kelley,1973)歸因理論的原則，如果接收人可推測該消息歸因為個人或情景原因，則溝通者是被