【作者】 郝媛媛；

【导师】 李一军；

【作者基本信息】 哈尔滨工业大学 ， 管理科学与工程， 2010， 博士

【摘要】 过去十几年中,随着互联网技术发展与应用的深化,以互联网为平台的在线口碑的传播凸显其巨大的影响力和重要性。用户发表的对特定商品的在线评论(简称在线评论),作为在线口碑的一种,正深刻改变着消费者的经济行为模式、企业的赢利性及其市场策略。尽管在线评论的实践已经开始,但学术界对这个新兴的信息媒体如何影响消费者决策的回答并不充分。而弄清此问题对消费者参考在线评论提高决策的效率和效果,企业有效地开展在线评论的管理与营销,最终推动企业实现更大赢利具有重要的现实意义和价值。本论文以行为导向为研究范式,结合定性和定量分析,从消费者的感知和购买行为两个层面出发探讨在线评论对消费者的影响,对在线评论如何影响消费者对评论有用性的感知以及消费者最终购买行为进行了实证研究。其中,围绕消费者对在线评论的有用性如何感知开展了两个方面的研究:(1)针对带有有用性投票统计信息的在线评论,结合文本挖掘技术和统计回归分析,基于评论的文本特征及发表时间等静态线索探索了在线评论感知有用性(以评论的有用性投票统计信息作为测度指标)的影响因素,并在此基础上研究了自动快速识别评论有用性的方法。(2)针对缺少有用性投票统计信息的一类特殊的在线评论,基于在线评论信息处理阶段,定性分析了这类特殊评论缺少有用性投票的原因及影响因素,并基于二分类Logistic回归比较这类评论与带有有用性投票信息的评论在影响因素上的差异,据此定量化地揭示消费者对这类评论是如何感知的以及这类特殊评论的特征是什么。另外,围绕这些有价值的在线评论如何进一步影响消费者的最终购买行为,从消费者个体与总体两个层面进行了研究:(1)以消费者个体为研究单元,根据归因理论、前景理论、感知风险理论、信息可获得性及诊断性理论,研究了在商品类型、接收者专业知识两个因素的调节作用下,在线评论的情感极性(正向或负向)对消费者个体购买行为的影响,基于实验设计和协方差分析对研究假设进行了实证检验。(2)以消费者总体为研究单元,采用面板数据建模,以商品销售额作为消费者总体购买行为的测度指标,探索了在线评论的数量和情感极性两个维度对消费者总体购买行为的绝对影响以及这种影响随商品生命阶段变化的动态性,另外通过与其他信息媒体专家评论的影响相比较,进一步揭示在线评论的相对影响和价值。本论文试图在以下方面做出有理论和实践价值的创新性成果。第一,基于面板数据建模,实证检验了在线评论的数量和情感极性两个维度对消费者总体购买行为的绝对影响以及(与专家评论相比)的相对影响,并探索和发现了影响随时间变化的动态性特点。研究模型可以更准确地揭示在线评论与消费者总体购买行为之间的现实关系及其随时间变化的一般规律。研究结果可为企业在多种信息推广方式中是否应该重点关注在线评论、在商品不同生命阶段如何有效地管理在线评论提供指导。第二,首次围绕在线评论的情感极性,探索了商品类别、接收者专业知识对在线评论情感极性影响消费者个体购买行为的调节作用及其具体形式,发现了商品类别调节作用的不对称性。研究结果可为企业在资源有限的情况下基于特定的现实情境更有效地管理不同情感极性的在线评论提供具体指导。第三,将搜索型商品的相关研究成果扩展到体验型商品领域,在已有研究基础上考察评论文本语义特征等静态线索的影响,分析并建立了在线评论感知有用性的影响因素模型,所建模型对评论有用性具有更强的分类识别能力。此研究一方面可实现从海量评论中自动识别有用评论,帮助消费者快速识别高质量商品,促进企业销售增长;另一方面可指导企业发表有影响力的评论,开展更有效的口碑营销。第四,首次针对没有有用性投票的一类特殊评论,定量化地研究了影响消费者未对在线评论有用性进行投票的原因及因素,揭示消费者对这类特殊评论有用性的感知。另外,尝试基于在线评论文本特征等一手数据揭示消费者复杂的心理行为,为消费者在线行为领域相关研究提供了新的视角、方法和手段。更多还原

【Abstract】 Over the past ten years, as the growing depth of Internet technologydevelopment and application, the online word-of-mouth with Internet ascommunication platform has begun to show its tremendous influence andsignificance. User-generated online product reviews, as one form of onlineword-of-mouth, is bringing a profound change in consumer behavior patterns, theprofit margin and market strategy of the enterprise. Related theoretical study onhow this emerging media affect consumers’ perception and purchasing behavior isnot sufficient. Clearing this question can help consumers improve decision makingefficiency and effectiveness, help enterprise manage online reviews effectively topromoteprofits, andmake a good use of the advantage of online reviews.Taking behavior orientation as the research paradigm, from two views ofconsumers’perceptions and buying behaviors, applying both qualitative andquantitative analysis, this paper empirically studys how online reviews impact theconsumers’perception on reviews helpfulness and consumers’purchase behaviors.The four questions are studied. Two parts refer to reviews helpfulness: (1)Combining text mining technologywith regression analysis, we explored the impactfactors for consumers’perception on reviews helpfulness (including the textcharacteristic and published time of reviews, etc.), and further studied the methodof automatically identify review helpfulness. (2) We qualitiviely discussed thecause and impact factors of some online reviews having no helpfulness votes, andquantitatively analyzed the differences in these impact factors between the reviewswith no helpfulness votes and ones with helpfulness votes based on binaryclassification Logistic regression, the results of which are studied to further revealwhy no votes. Contents of the other two parts include: (1) Taking the individualconsumer as the research unit, we studied the moderating effects of product type,consumer expertise on the impact of the online reviews valence, according toattribute theory, prospect theory, perceived risk theory and informationavailability/diagnosticity theory.Applying the experiment design andANCOVA, wetested the moderating effects related hypothese. (2) Takingthe consumer populationas the research unit, we explored the effect of online reviews numbers and valence on product sales (used to measure the buying behaviors of the overall consumergroup) and the variation ofthe impacts with time using panel data. We also revealedthe relative effect of online reviewscompared with expertise reviews.This paper attempts to achieve theoretical and practical innovation in thefollowing four aspects.First, based on the panel data model, we empirically tested the absolute andrelative (compared with expertise reviews) effects of online reviews numbers andvalences on the overall consumer purchase behaviors. We also explored andvalidated the dynamic variation nature of the impacts with time. Our study modelcan reveal the realistic relation between online reviews and consumer populationbehaviors, offering an expalanation for present disconsistent research conclusions.These researches can also richen the relative theories and researches on the impactof online reviews and overall consumer behaviorsSecond, we first explored how product type, consumer expertise moderate theimpact of online reviews valence on individual consumer purchase behaviors. Wealso analyzed the detailed forms of moderating effects and found the asymmetriccharacteristics of moderating role of product type. Our study offers detailedsuggestions for enterprise to effectively manage online reviews under limitedresources.Third, extending researches for search goods to experience goods, weestablished one model for the impact factors of reviews perceived helpfulness. Thismodel significantly improves the forcast effect for the helpfulness of online reviews.This study can help consumers identify high quality products quikly which furtherenhance product sales.Besides, our findings can offer suggestions for enterprise onhow to post high-impact reviews and improve online word-of-mouthmarketing.Fourth, we first fill the research gap in quatitatively analyzing the causes andimpact factors of reviews not being voted for helpfulness. Besides, we attempted toreveal consumers’complex psychology and behaviors through first-hand textfeatures data of reviews, which provides a new perspective, method and means forstudying online consumer behaviors.更多还原