Three essays of online word of mouth.
详细信息
文摘
Online word of mouth WOM) has been playing an increasingly significant role in consumer purchase decisions. Prior studies illustrate that a variety of aspects of online WOM influence sales Chevalier and Mayzlin 2006； Liu 2006； Duan,Gu,and Whinston 2008； Dellarocas,Awad,and Zhang 2004； Godes and Mayzlin 2004). Given its rising importance,firms need to understand and manage online WOM to benefit from the phenomenon. Accordingly,several auto brands,such as Scion xB,Ford Fiesta,and Chevy Tahoe,tried to use online WOM campaigns to generate product awareness as well as sales. One unique feature of online WOM marketing is that firms initiate it but consumers implement it. This means that online WOM can serve as a new type of marketing communication mix and work as free "sales assistants" Chen and Xie,2008). However,it can also turn against firms when the tone and content become negative. The key to the success of such a campaign,therefore,is to understand the mechanism of online WOM contributions. This mechanism could be categorized into two decisions: a posting decision what motivates people to spread online WOM?) and an evaluation decision what influences the tone - positive,negative,or neutral - of online WOM?). Few studies so far have examined the two questions. This is,clearly,a major gap. It is this gap that we attempt to fill in this dissertation. In the first essay,we propose and empirically test a theory to explain the volume and type of online WOM for automobiles. We find that if products are not satisfactory on utilitarian attributes,consumers are likely to express their unhappiness to other consumers. Satisfying consumers on utilitarian attributes,however,does not necessarily lead to positive WOM. Hedonic attributes,on the other hand,are primarily responsible for positive WOM in that,if they are pleased with the performance of products on those attributes,consumers are likely to discuss the positive aspects of the products with other consumers. Not pleasing consumers on hedonic attributes,on the other hand,is unlikely to stimulate negative WOM. In addition,innovativeness can lead to both negative and positive WOM. In the second essay,we propose hierarchical Bayes HB) Poisson Rate and HB Binomial models of the volume of positive online WOM for automobiles. Our models have the ability to account for both observed and unobserved heterogeneity and produce brand-model level parameter estimates. We find that the proposed models have a higher predictive accuracy than five non-hierarchical models,i.e.,Regression,Density Regression,Poisson,Poisson Rate and Binomial models. Substantively,we find that an automobiles design/performance level is positively associated with positive online WOM whereas its quality is not. In other words,improving the design/performance level of a car may generate positive online WOM,whereas improving the quality level of a car may not necessarily do so. In general,innovation i.e.,designing new models or redesigning existing models) also stimulates positive online WOM. However,this strategy may not be as effective for luxury and large cars. In the third essay,we empirically examine the relationship between the online opinion for an automobile and its attributes. As opposed to essay one and two,however,we focus on the dynamics of this relationship,i.e.,we examine whether the attributes that affect the opinions towards automobiles vary over their life. We also control for the role of reviewers characteristics,and social influences on reviewers,in our investigation of these dynamics. Our key substantive finding is that different attributes play different roles in influencing consumer opinions at different stages of an automobiles lifecycle. Specifically,the design and performance play a role when opinions are stated soon after the purchase. Over time,however,this effect weakens while the role of quality increases in opinions stated long after the purchase. In terms of social influences,we find that the opinions expressed previously by other consumers do affect consumer ratings: a higher percentage of positives in the previously expressed opinions lead to more positive consumer ratings while a larger percentage of negatives worsen the same. In addition,we also find that positive expert opinions improve ratings while negative expert opinions hurt them. Further,the negative effect of negative expert opinions is larger than the positive effect of positive expert opinions.