Texts Or Ratings:A Research On Information Modality Bias In Online User Reviews

As the development of the Internet and E-commerce,online reviews generated by product users becomes a new kind of elctronic word of mouth,which,compared with traditional word of mouth,have a lot of advantages such as sustainability,availability,easy to record,low cost,easy to use and universal appeal,and thus gains consumers’ much attention and becomes the focus of the researchers and practitioners because of its many advantages related to the Internet.Because online user reveiws consist of qualitative information(texts)and quantitative information(ratings,volume,length),researchers have conducted many studies involving diverse qualitative textual characteristics and quantitative numerical characteristics and focusing on antecedents and consequences of this electronic word of mouth since the beginning of this century.However,no light has been shed on the comparison between different informatin modalities of online user reviews,i.e.qualitative information and quantitatie information,especially when these different information modalities point to the same objective,such as valence.Review valence represents the positivity or negativity of review information,which can be embodied by different information modalities,i.e.qualititive texts and quantitative ratings.Of these two information modalities,the former qualitative texts are product and service descriptions and evaluations using positive or negative words,while the latter qantitative ratings are product and service evaluations based on numerical scales.Because review valence is a symbol of the favorability or unfavorability of product alternatives,this review characteristic has an important impact on consumer’s purchase decision.In order to evaluate the favorability or unfavorability of product alternatives,readers can make their purchase decision by reading texts or browsing ratings.In extant literature,many studies focus on the comparisons in the direction of impact(favorable impact vs.umfavorable impact)and the amount of impact(strong vs.weak)between different levels of review valence(positive vs.neutral vs.negative),however,no studies focus on the comparison between different valence information modalities,i.e.do textual valence and numerical valence have the same influence over readers?In many extant studies focusing on review valence,some researchers specialize in textual valence,others in quantitative ratings,and more in both.Of these studies incorporating both valence modalities,one common practice is to view textual valence and numerical valence as one whole and treat them equally by focusing on textual valence in the theoretical model but employing the corresponding ratings in the data processing,implying that they are consistent in valence level as well as in the impact they have on readers.Two problems arise from this common practice.First,some researchers have found that review texts and ratings are not always consistent in valence level and amount of information contained,therefore,to treat them equally and take one(such as ratings)as a proxy of the other(such as texts)leads to biased conclusions.Second,due to the fact that in information communication,texts and numbers are two totally different information modalities with different characteristics and values and hence require different processing tactics and efforts,text and rating should have different values to consumers.In spite of the fact that some theories claim that consumers actively search for and use information to make informed choices,a great deal of consumer behavior does not involve extensive search for information or a comprehensive evaluation of the choice alternatives and their information search process can be highly selective,this research proposes that qualitative texts and quantitative ratings have different impacts on consumers in different situations.Therefore,it is not without fault to consider them equal and totally replaceable.Based on these points,it is necessary to explore consumer’s information modality bias in a review system incorporating both texts and ratings to further reveal the mechanism of online review valence and push forward the related research in this area.The research topic of this study is to examine the different impacts of online user review’s qualitative texts and quantitative ratings on reader’s purchase intention as well as the underlying process of this impact,i.e.consumer’s information modality bias.In order to investigate this topic,this study firstly reviews related literature on online user review’s diverse qualitative and quantitative characteristics,review valence and’ its two information modalities,information search and processing,information modality,involvement and uncertainty.By reviewing the extant literature,this research summarizes the findings and contributions of prior studies and point out their research gap and shortcomings.Secondly,based on the reviewing of related literature,by theoretical analyses and deduction,this study analyzes the different impacts of review textual valence and numerical valence(ratings)on reader’s purchase intention and identifies boundary variables from the persptive of reader’s involvement and perceived uncertainty as well as the underlying processes based on the benefit-cost model of information economics,which composes the theoretical model of this study.Lastly,this study testifies the theoretical model with the interviewing method and experimentation method.In the interviewing part,the interviewees include undergraduates and MBA students,and the purpose of the interview is to conduct an initial exploration on the differential impacts of the two valence information modalities and provide a guideline for the following formal experiments.Experimentation is the main method for the testification of the theoretical model.In the two experiments of this research,the information modality biases under the influence of two factors are examined,with two different product categories,two different data sources,and two different experiment manipulation methods.The findings of this study are as follows.Firstly,the internal factor,i.e.reader’s involvement,impacts reader’s review information modality bias in information searching and processing.Review texts impact high-involved reader’s purchase intention greater than ratings while ratings impact low-involved reader’s purchase intention greater than texts.Secondly,the external factor,i.e.uncertainty,impacts reader’s review information modality bias information searching and processing.Review texts impact reader’s purchase intention greater than ratings when uncertainty is high while ratings impact reader’s purchase intention greater than texts when uncertainty is low.Thirdly,this study reveals the underlying process of this information modality bias.In high involvement and high uncertainty conditions texts impacts readers through reader’s motivation to increase benefits while in low involvement and low uncertainty conditions ratings impact readers through reader’s motivation to reduce costs.Highlights and contributions of this study are as follows.Firstly,this study deepens and perfects research on online user reviews by revealing the differenct impacts of two review information modalities,texts and ratings,and demonstrating the diverse factors influencing this bias such as consumer’s internal factor(involvement)and external factor(uncertainty).These findings demonstrate that consumer’s information bias includes not just the biases for contents,such as positivity bias/negativity bias,subjectivity bias/objectivity bias,but also the bias for information modality.Although this research focuses on online user reviews,it enhances consumer information search and processing literature as well.The findings of this research reveal that information contents are not the only important problem,but the information modality is also very crucial to consumer information decision,which provides new insights for consumer decision making literature and firms’ communication strategy.Secondly,this research contributes to the research of review valence by bringing to light the inequality between textual valence and quantitative valence(ratings).This research finds,due to the differences in information modality characteristics and processing techniques,textual valence and quantitative valence have different values to consumers,and thus have different impacts on them.Therefore,the common practice in many prior studies that treats these two information modalities as one valence whole without differentiating one from the other is not without faults.More importantly,because their disparity in valence level in some situations has been found by many prior researchers,consuemer’s bias for information modaility will make the purchase decision caused by this disparity uncertain,which is actually a selective Bucket Effect:in spite of the higher valence information modality,the one which is lower in valence level may or may not drag down consumer purchase intention,which depends on whether this lower information modality is consumer’s favorite or not.Therefore,treating them two equally and totally replaceably,which is very prevalent in some prior studies,is not reasonable and easily leading to biased conclusions.Therefore,these findings lay stress on the attention on the disparity between review textual valence and numerical valence.Thirdly,by combining economic theory with consumer behavior,this study explains the underlying process of this information modality bias from the perspective of benefit-cost model.Although many prior researchers explored the differential directions of impacts(positive vs.negative)that benefits and costs have on consumer information acquisition behavior,this research focuses on the relationship of these two from a different perspective,i.e.the different amount of impacts.The findings of this reseaarch shows that,being rational man,consumers put different emphases on cost versus benefit in different situations when they search for information.Consumers are more concerned about increasing benefits regardless of the corresponding higher costs sometimes and they are more concerned about reducing costs and ignoring the achievement of extra benefits at other times.Therefore,firms should realize that,in spite of the fact that both are economic incentives to consumer information acquisition behavior,they do not need the same treatment,and firms should make corresponding market strategies by differentiating whether consumers are benefit-oriented or cost-oriented.