Do You Buy What They Write?
How Source Similarity and Reviewer Expertise Affect the
Relationship between Online Consumer Reviews and
Consumer Purchase Intentions
2
Do You Buy What They Write?
How Source Similarity and Reviewer Expertise Affect the
Relationship between Online Consumer Reviews and
Consumer Purchase Intentions
Author Wouter Neef
Eendrachtskade Zuidzijde 14a 9726 CW Groningen
The Netherlands +31 624 63 70 41 w.neef@student.rug.nl
University of Groningen
Faculty of Economics and Business Research Master Marketing Master’s Thesis
Supervisors
3 PREFACE
This master’s thesis is a product of the research project I have conducted for my graduation from January 2011 until September 2011. The thesis is the final part of my research master in marketing at the Faculty of Economics and Business at the University of Groningen. The research was carried out at the Caroll School of Management at Boston College.
4 ABSTRACT
Online consumer reviews, as the most accessible and prevalent type of word-of-mouth, have gained significant importance in the field of marketing. However, the credibility of online consumer reviews is becoming a major concern for consumers. As a result, consumer review websites are rethinking ways to gain the trust of consumers and are transitioning from an anonymous to a more personal environment. Prior research has found a relationship between online consumer reviews and consumer purchase intentions. The current research extends work on the effects of online consumer reviews by including source similarity and reviewer expertise as moderating factors in this relation. In addition, we are the first to use source credibility theory to explain the effects of source similarity and reviewer expertise.
We confirm in an online experiment that online consumer reviews significantly influence consumer purchase behavior. In addition, we find that the perceived source credibility of a review strengthens the relationship between review valence and consumer purchase intentions. Furthermore, we show that both source similarity and reviewer expertise affect perceived source credibility.
Moreover, we find that, for positive reviews, source credibility mediates the relationship between reviewer expertise and consumer purchase intentions. For negative reviews, we find a mediating effect of source credibility in the relationship between source similarity and consumer purchase intentions. We explain these findings with the negativity bias phenomenon and social identity theory.
5 INDEX
1. INTRODUCTION ... 7
1.1STUDY BACKGROUND... 7
1.2RESEARCH PURPOSE AND RESEARCH QUESTION ... 9
1.3RELEVANCE ... 9
1.4THESIS ORGANIZATION ... 10
2. LITERATURE REVIEW ... 11
2.1WORD-OF-MOUTH ... 11
2.2ONLINE CONSUMER REVIEWS... 12
2.3CONTEXTUAL FACTORS ... 12
3. HYPOTHESES AND CONCEPTUAL MODEL ... 14
3.1REVIEW VALENCE ... 14 3.2SOURCE CREDIBILITY ... 14 3.3SOURCE SIMILARITY ... 15 3.4REVIEWER EXPERTISE ... 18 3.5CONCEPTUAL MODEL ... 19 4. METHODOLOGY ... 20 4.1EXPERIMENTAL DESIGN ... 20 4.1.1 Subjects ... 20 4.1.2 Experimental product ... 20 4.1.3 Experimental procedure ... 21 4.1.4 Manipulations ... 21 4.1.5 Manipulation checks ... 22 4.1.6 Dependent variables ... 22 4.1.7 Control variables ... 22 4.1.8 Consumer demographics ... 23 4.1.9 Validity issues ... 23 4.2PLAN OF ANALYSIS ... 23 5. RESULTS ... 25 5.1SUBJECTS ... 25 5.1.1 Invalid responses ... 25 5.1.2 Manipulation checks ... 25 5.1.3 Randomization ... 26 5.2SCALE RELIABILITY ... 26
5.2.1 Source credibility dimensions ... 26
5.2.2 Consumer purchase intentions ... 26
5.2.3 Product involvement ... 26
5.2.4 Consumer expertise ... 27
5.3EFFECTS OF REVIEW VALENCE AND SOURCE CREDIBILITY ... 27
5.3.1 Main effect of review valence ... 27
6
5.4EFFECTS OF SOURCE SIMILARITY AND REVIEWER EXPERTISE ON SOURCE CREDIBILITY ... 29
5.4.1 ANCOVA analyses trustworthiness dimension ... 29
5.4.2 ANCOVA analyses expertise dimension ... 30
5.5MEDIATING EFFECTS OF SOURCE CREDIBILITY ... 31
5.5.1 Mediating effects of the trustworthiness dimension ... 31
5.5.2 Mediating effects of the expertise dimension ... 32
6. CONCLUSION ... 34
6.1DISCUSSION ... 34
6.2THEORETICAL CONTRIBUTIONS ... 36
6.3MANAGERIAL IMPLICATIONS ... 36
6.4LIMITATIONS ... 37
6.5DIRECTIONS FOR FUTURE RESEARCH ... 38
REFERENCES ... 40
APPENDIX I: SCREENSHOTS ... 47
7 1. INTRODUCTION
1.1 Study background
In the last two decades, online consumer reviews have grown from a new phenomenon in marketing into a critical resource for consumers to acquire information about product quality. Since the start of the first consumer review websites in the 1990s, this new type of electronic word-of-mouth has become more and more popular (Hu and Lui 2004). Research shows that review websites grow almost four times faster than the overall Internet (comScore 2007). One out of every four consumers reads reviews before purchasing a product online. More than three-quarters of those consumers believe that online consumer reviews are influential in their purchase decision (comScore 2007). Different from traditional word-of-mouth, of which the influence is limited to a local social network (Brown and Reingen 1987; Biyalogorsky et al. 2001; Shi 2003), the impact of online consumer reviews can reach far beyond the local community, because consumers from all over the world can access the reviews via Internet.
Online consumer reviews are common for services (e.g. hotels and restaurants) and products, such as books, electronics and games (Chen and Xie 2008). In general, an online consumer review consists of a rating a consumer assigns to a product and written comments about the product. Consumers can post reviews on the websites of online stores (e.g. Amazon.com), third-party websites (e.g. Expedia.com) or independent review websites (e.g. Epinions.com). On most websites, the average consumer rating can be found next to the name of the product. More detailed information about the reviews, such as the written statements or the rating distribution is displayed less prominently. Most of the online consumer reviews are anonymous, i.e. consumers are allowed to use a user name instead of their real name to post reviews.
8
consumer reviews on product sales using data from the video game industry. In contrast, Chen, Wu, and Yoon (2004) and Duan, Gu, and Whinston (2008) conclude that online consumer reviews do not influence sales, but act as predictors.
Although these studies have generated mixed results, there is a growing belief in both marketing science and practice that it is important to manage this type of word-of-mouth. Research shows that firms motivate their consumers to spread the word about their products online (Godes and Mayzlin 2004). According to comScore (2007), 30% of the consumer reviewers write reviews because they are asked to do so. For example, television networks in the United States actively induce viewers to post online reviews about their television shows (Godes and Mayzlin 2004). Some firms are even manipulating consumers by posting anonymous online consumer reviews that praise their products or bad-mouthing those of their competitors (Dellarocas 2006). For example, when Amazon.com mistakenly revealed the true identities of some book reviewers, it turned out that a sizeable proportion of the reviews were written by the books’ own publishers, authors and competitors (Harmon 2004). Another example is the music industry, which is known to hire professional marketers who visit websites to post positive opinions on behalf of new albums (White 1999; Mayzlin 2006).
9
know that highly expert sources are more influential than sources that have less expertise (Hovland and Weiss 1951). Therefore, we want to examine if reviewer expertise affects consumer purchase intentions. We want to use source credibility theory to explain the relations between source similarity and reviewer expertise and the consumer purchase decision.
1.2 Research purpose and research question
According to Cheung et al. (2009), the believability of electronic word-of-mouth is becoming a major concern for consumers. Hence, consumer review websites are rethinking ways to gain the trust of consumers and are changing from an anonymous to a more personal environment. The consequences of these changes on consumer behavior are unknown so far. Therefore, the purpose of this research is to examine how source similarity and reviewer expertise influence consumer decision making. We formulate the following problem statement:
How do source similarity and reviewer expertise affect the relationship between online consumer reviews and consumer purchase intentions?
In our research we focus on services, because we expect larger effects on consumer purchase intentions for services than for products. According to Bharadwaj et al. (1993), it is more difficult to evaluate the value of a service and therefore consumers consider word-of-mouth as more important in a service context (Zeithaml 1981).
1.3 Relevance
10
we are the first to use source credibility theory to explain the effects of source similarity and reviewer expertise on consumer purchase intentions
The present research gives managers more insights in the important effects of online consumer reviews on consumer purchase intentions. This study enables managers to distinguish between less and more influential online consumer reviews. As a result, managers can allocate resources more effectively to maximize the effects of positive reviews and minimize the effects of negative reviews.
1.4 Thesis organization
11 2. LITERATURE REVIEW
In this chapter, we discuss the relevant literature on online consumer reviews. First, we introduce important research on traditional and electronic word-of-mouth. After that, we discuss literature on the relationship between online consumer reviews and consumer behavior. Next, we give an overview of relevant contextual factors in research on online consumer reviews. We end this chapter with a brief discussion of the application of source credibility theory in research on word-of-mouth.
2.1 Word-of-mouth
Since the 1950s, word-of-mouth has been considered as one of the most important resources in marketing communication (Godes and Mayzlin 2004). Bone (1995) defines word-of-mouth as person-to-person verbal communication between two or more consumers, none of whom represents a marketing source. Word-of-mouth is a result from previous experiences of consumers with products or services. These experiences can either be direct (based on own experiences) or indirect (based on opinions of others). Previous research has shown the important effects of word-of-mouth in marketing. For example, Brown and Reingen (1987) suggest that word-of-mouth communications are more effective than conventional advertising. In addition, Villanueva et al. (2008) show that customers acquired through word-of-mouth contribute almost twice as much long-term value to the firm, compared to customers acquired through marketing.
12
word-of-mouth (Dellarocas 2003). Although there are important differences between traditional and electronic word-of-mouth, authors found that participants in both types of word-of-mouth have the same motivations (Hennig-Thurau et al. 2004) and that the effects on consumers might be similar (Gruen et al. 2006).
2.2 Online consumer reviews
Online consumer reviews can be considered as the most accessible and prevalent form of electronic word-of-mouth (Chatterjee 2001). Several researchers have examined the effects of online consumer reviews on the purchase behavior of consumers. In their award-winning paper, Chevalier and Mayzlin (2006) find that online consumer reviews have an important impact on online consumer behavior. In addition, the authors conclude that an improvement in a book’s reviews at a website leads to an increase in relative sales at that website. Other authors confirm the relationship between online consumer reviews and product sales. Lui (2006) find that online consumer reviews explain a significant part of movie revenue, especially in the early weeks after the movie opens. However, the author argues that most of this effect comes from the number of reviews and not from the valence of the reviews. Zhang and Dellarocas (2006) present a diffusion model to forecast product sales and find a positive relationship between online consumer review valence and volume and sales for movies. Zhu and Zhang (2010) find that product and consumer characteristics moderate the influence of online consumer review (ratings and volume) on product sales using data from the video game industry. In contrast, Chen, Wu, and Yoon (2004) and Duan, Gu, and Whinston (2008) conclude that online consumer reviews do not influence sales, but act as predictors.
2.3 Contextual factors
13
effect found is consumer involvement. Park and Lee (2008) find that low-involvement consumers are mainly affected by review quantity, while high-involvement consumers are affected by review quantity mainly when the review quality is high. The same authors find that the national culture of a consumer (South-Korean vs. American) influences the relationship between online consumer reviews and the consumer purchase decision (Park and Lee 2009).
As far as we know there are no studies on the role of source similarity in the relationship between online consumer reviews and consumer purchase intentions. However, we do find evidence for the role of reviewer expertise in this relationship. Vermeulen and Seegers (2009) include reviewer expertise in their model on the effect of online consumer reviews on the consumer purchase decision. The authors find that reviewer expertise has a minor positive influence on the impact of the online reviews. Hu, Liu, and Zhang (2008) examine how reviewer quality (a proxy for the expertise of a reviewer) and reviewer exposure (the number of reviews a reviewer has written) affect the relation between online consumer reviews and product sales. The authors find that consumers who read online reviews also pay attention to this contextual information. Consumers respond more favorably to reviews written by reviewers with a better reputation and higher exposure.
14 3. HYPOTHESES AND CONCEPTUAL MODEL
In this chapter, we propose our hypotheses and present our conceptual model. First, we discuss the effects of review valence on consumer purchase intentions. After that, we propose the hypothesis on the moderating effects of source credibility on the relationship between review valence and consumer purchase intentions. Next, we propose the hypotheses on the effect of source similarity on consumer purchase intentions and the mediating effect of source credibility in this relationship. Then, we propose the hypothesis on the mediating effect of source credibility in the relationship between reviewer expertise and consumer purchase intentions. We present our conceptual model at the end of the chapter.
3.1 Review valence
In the current research, we make a distinction between online consumer reviews with a positive valence towards services and reviews with a negative valence towards services. The reason for this is straightforward, positive reviews enhance consumer purchase intentions, whereas negative reviews reduce consumer purchase intentions (Chevalier and Mayzlin 2006).
3.2 Source credibility
We define the perceived source credibility as the amount of credibility (believability) the recipient attributes to a source of information. According to the source credibility theory, the credibility of a source consists of two dimensions (Hovland, Janes, and Kelly 1953). Expertise refers to the extent to which a source is perceived to be capable of making correct assertions. Trustworthiness refers to the degree to which the audience perceives the assertions made by the source to be ones that the source considers as valid.
15
persuasive than a low-credibility source (Hovland and Weiss 1951; Johnson, Torcivia, and Poprick 1968; Kelman and Hovland 1953; Miller and Baseheart 1969).
Research in the field of marketing has shown that sources with a higher credibility are more persuasive than sources with low credibility on attitudes and behavior towards firms, brands, products and services (Fireworker and Friedman 1977). Woodside and Davenport (1974) find in an experiment that the source credibility of salesmen positively affects the purchasing behavior of consumers. In a comparable study on price and credibility, the same authors confirm these results (Woodside and Davenport 1976). Friedman and Friedman (1979) find an effect of source credibility on consumer purchase intentions in their study on product endorsers in advertisements.
Although numerous studies have found the superiority of highly credible sources, some studies have found that sources with low credibility have more influence on attitudes than sources with high credibility (Dholakia and Sternthal 1977). In the literature, we find two factors which can cause the superiority of low credibility sources. Johnson, Torcivia, and Poprick (1968) find that a low credibility source is more influential when the audience is highly authoritarian. Johnson and Izzett (1972), Johnson and Scileppi (1969) and Rhine and Severance (1970) find a similar effect for recipients who are highly involved in the communication issue.
Research in source credibility theory, although not unanimous, shows that higher source credibility results in more persuasion than low source credibility. Both the expertise and the trustworthiness dimension affect the attitudes and behavior of consumers. We find these effects in marketing and consumer behavior as well. Therefore, we propose that,
H1: The perceived source credibility of an online consumer reviewer has a positive effect on the relationship between review valence and the consumer purchase intention
3.3 Source similarity
16
profiles on consumer review websites, nowadays it is often possible for consumers to assess source similarity.
From previous literature, we know that there exists a positive relation between source similarity and source credibility (Swartz 1984; Brock 1965). People see other people who share the same attributes or attitudes as more credible (Byrne and Griffitt 1966; Newcomb 1956). However, it is important that the similarities are relevant with regard to the issue (Simons, Berkowitz, and Moyer 1970). According to Gilbert et al. (1998) the effect of source similarity on source credibility is caused by acting as a simple persuasion cue when elaboration is low or by biasing processing when elaboration is high.
We find evidence that source similarity influences both dimensions of the source credibility construct. Marsh (1967) argues that source similarity affects the trustworthiness of the source as perceived by the recipient. Consumers who perceive reviewers as similar to themselves could expect such reviewers to have similar attitudes. As a result, trustworthiness is higher, because consumers expect benevolent intentions from reviewers who have similar attitudes (Doney and Cannon 1997). Furthermore, the similarity between the source and the recipient contains information about the source’s tastes and preferences in a product category. Therefore, consumers can infer that their tastes and preferences are similar to a source based upon similar attributes (Eagly, Wood, and Chaiken 1978; Kelley 1967; Simons, Berkowitz, and Moyer 1970). As a consequence, the source is more capable of making correct assertions about a product, and therefore source similarity affects the expertise dimension of source credibility. We propose that,
H2a: The similarity between the reviewer and the consumer has a positive effect on the perceived source credibility of the consumer
17
H2b: In case of a high-similar source, disclosure of personal information has a positive effect on the perceived source credibility of the consumer
For low-similar sources, we can follow the same line of reasoning. The disclosure of personal information of the reviewer has a positive effect on the source credibility of the review. However, we expect a larger negative effect of the dissimilarity between the reviewer and the consumer. Therefore, we propose that,
H2c: In case of a low-similar source, disclosure of personal information has a negative effect on the perceived source credibility of the consumer
A number of studies have shown that source similarity is one determinant of the effectiveness of the communicator (e.g. Brock 1965). Research shows that an increase in the similarity between the source and the recipient enhances the persuasiveness of the message of the source (Simons, Berkowitz, and Moyer 1970). In other words, similarity between the source and the recipient increases the degree to which the source is able to influence the recipient (Berscheid 1966). For example, in an experiment high school students were given nutrition messages attributed to a high, medium, or low similarity source (Feldman 1984). The author found that the greater the perceived similarity, the greater the influence on the nutrition behavior and attitudes of the participants.
In marketing, Woodside and Davenport (1974) show the positive effect of source similarity between a salesman and the consumer on consumer purchase intentions. A salesman who expressed prior purchase versus non-purchase by himself of musical tapes being bought by a consumer leaded to higher purchase likelihood by the consumer. Also, Murray (1991) finds that service consumers prefer the opinions and experiences of other comparable individuals in making service purchase decisions.
In literature we find evidence that source similarity positively affects source credibility, which in turn results in higher persuasion (Mugny 1982). Clark and Maass (1988) show an example of the mediating role of source credibility in their study on social categorization and perceived source credibility in minority influence. The authors find that in-group minorities (attending the same university) were perceived as more credible than were out-group minorities (attending a different university), and that this perceived greater credibility leaded to greater persuasiveness. We propose that,
18
3.4 Reviewer expertise
Expertise can be obtained in different ways, for example, through formal training in a subject matter (Feick and Higie 1992). Another way to obtain expertise is through product-related experiences, including product purchase and usage (Alba and Hutchinson 1987). Reviewer expertise had been included as a contextual factor in prior research on the effects of electronic word-of-mouth (Senecal and Nantel 2004; Vermeulen and Seegers 2009). These studies distinguish between the opinions from professionals and the opinions from consumers, where professional reviewers are seen as a high expertise source, while consumers are seen as a low expertise source. For example, Vermeulen and Seegers (2009) use ‘a former hotel manager and six-year veteran hotel reviewer’ as an expert and ‘a secretary’ as a non-expert in their research. In the present research, we use another definition of reviewer expertise. We differentiate between consumers with and consumers without an in-depth mastery of a field of knowledge. We focus on the different effects of consumer reviewers with high expertise in a product category and consumers with low expertise in a product category.
Research on source expertise in persuasive communication indicates that the perceived expertise of a source has a positive effect on attitude change (Mills and Harvey 1972; Ross 1973). It is known that, in general, experts are more persuasive than non-experts (Petty, Cacioppo, and Goldman 1981). In the field of marketing, both advertising and interpersonal influence studies have demonstrated the importance of experienced sources in changing consumer attitudes. For example, Ohanian (1991) found that the expertise of celebrity endorsers with respect to the product was significantly related to consumer purchase intentions. In addition, Woodside and Davenport (1974) found a similar effect in their research on the influence of the expertise of salesmen on consumer purchase intentions. In electronic word-of-mouth Vermeulen and Seegers (2009) found that the expertise of hotel reviewers has a positive influence on the impact of the reviews.
The effects of reviewer expertise on consumer purchase intentions can be explained with source credibility theory. Previous literature indicates that the expertise of a source positively affects perceived credibility, which in turn results in higher attitude change (McGuire 1969). Individuals who are seen as experts are seen as more capable of making correct and valid assertions about products in a category and significantly influence consumer purchase intentions (Ohanian 1991). Therefore, we propose that,
19
3.5 Conceptual model
The conceptual model (figure 1) shows the hypothesis for the moderating effect of perceived source credibility on the relationship between review valence and consumer purchase intentions (H1). In addition, the model shows the hypotheses for the effects of source similarity on the perceived source credibility (H2a, H2b and H2c). We test the mediating effects of perceived source credibility with hypotheses H3 (reviewer expertise) and H4 (source similarity) for both positive and negative valence reviews. Figure 1 Conceptual model Perceived source credibility Reviewer expertise Source similarity H1
Review valence Consumer purchase
intentions H2a,
H2b, H2c
20 4. METHODOLOGY
In this chapter, we present the methodology for this study. We first present the experimental design and after that we discuss the plan of analysis.
4.1 Experimental design
For the current research, we perform a 3x2x2 between-subject experiment with 12 conditions (table 1) and one control condition. We manipulate the three variables source similarity, reviewer expertise and review valence. For source similarity, we have a non-disclosure condition (no assessment of similarity), a low-similarity condition and a high-similarity condition. For reviewer expertise, the two conditions are expert and non-expert. For review valence, we have a negative review condition and a positive review condition.
Table 1
Overview of experimental conditions
Expert Non-expert High- similarity Low-similarity Non-disclosure High- similarity Low- similarity Non-disclosure
Positive valence Group 1 Group 2 Group 3 Group 4 Group 5 Group 6
Negative valence Group 7 Group 8 Group 9 Group 10 Group 11 Group 12
4.1.1 Subjects
For the online experiment, we recruit 405 participants on the online labor marketplace Amazon Mechanical Turk. We use Mechanical Turk, because it enables us to manipulate source similarity. Due to this manipulation, participation is restricted to United States citizens only. In addition, the use of Mechanical Turk allows us to gather a sufficient amount of participants in a relatively short time period. Participants receive $0.70 for the completion of the experiment.
4.1.2 Experimental product
21
4.1.3 Experimental procedure
The subjects begin the experiment by reading an introduction which explains that they have to simulate the booking of a hotel room for a vacation to Edinburgh. After that, the subjects are exposed to a webpage of a fictional independent review website, which shows information about and pictures of a hotel in Edinburgh (see Appendix I). In addition, the webpage shows an online consumer review about the hotel and information about the reviewer who wrote the review. We do not show the consumer review to the subjects in the control condition. However, the presentation of the other information is similar across conditions. The subjects have a few minutes to read the information on the webpage. After that, the subjects have to answer questions about the dependent variables, consumer characteristics and demographics. Before the subjects finish the experiment, they have the opportunity to give comments or ask questions about the experiment. After that, we thank and reward them for their participation.
4.1.4 Manipulations
For the valence of the review, we have two conditions. We manipulate both the review rating and the written comments. For the positive review, we assign 4.5 stars to the hotel and describe five positive aspects and one negative aspect of the hotel and the service. For the negative review, we assign 2 stars to the hotel and describe five negative aspects and one positive aspect of the hotel and the service. We avoid extremes to increase the helpfulness of the online consumer review (Mudambi and Schuff 2010). In the review, we use informational statements, which are seen as more objective, useful, understandable and persuasive than transformational statements (Park, Lee, and Han 2007).
From previous research we know that demographic similarity affects the persuasive effects of word-of-mouth (Gilly et al. 1998). Therefore, we use the reviewer’s nationality to manipulate source similarity. In the non-disclosure condition, we only show the user name of the reviewer and provide no information on the nationality of the reviewer. As a result, it is not possible for subjects to assess source similarity. In the low-similarity and high-similarity conditions, we show the name and the nationality of the reviewer (both in words and with an image of the national flag) and show the text “I am a typical [nationality] traveler”. For the low-similarity condition, we use a Turkish reviewer named Ahmet. For the high-similarity condition, we use an American reviewer named Michael.
22
badge under to the name of the reviewer and show the text “83 (frequent poster)” under the header “Number of Reviews posted”. In the non-expert condition, we show no badge and show the text “3 (infrequent poster)” under the header “Number of Reviews posted”.
4.1.5 Manipulation checks
To assess if the manipulations in the experiment are effective, we ask the subjects questions about the valence of the review, the similarity with the reviewer and reviewer expertise. For the valence of the review, we ask subjects to rate on a seven-point scale (“strongly disagree” to “strongly agree”) to what extend the review is negative or positive. For source similarity, we ask a question using a seven-point scale from “strongly disagree” to “strongly agree”: “To what extend do you consider the reviewer as similar to yourself?” For reviewer expertise, we ask a question with the same scale: “To what extend do you consider the reviewer as an expert?”. We check with independent sample t-tests if our manipulations are successful.
4.1.6 Dependent variables
We measure the perceived source credibility and consumer purchase intentions. To measure source credibility, we use a multiple measure design of Harmon and Coney (1982). We use six seven-point semantic differential scales with the adjectives: trustworthy/not trustworthy, good/bad, open-minded/close-minded, trained/untrained, experienced/not experienced and expert/not expert. These items load highly on the trustworthy and expertise dimensions first identified by Hovland, Janes, and Kelly (1953). To measure the consumer purchase intention, we use a multiple measure design from previous research (Juster 1966; Singh and Cole 1991). We ask subjects three questions using a seven-point scale from “strongly disagree to “strongly agree”: “I’ll consider booking the hotel room,” “I’m glad to book this hotel room,” and “I’m glad to recommend others to book the hotel room.”
4.1.7 Control variables
We ask subjects questions about the consumer characteristics which might influence the effect of source credibility on the relationship between review valence and the purchase decision. We measure product involvement and consumer expertise.
23
second question is: Making one’s selection of hotels is...” with a scale from “A very important decision” to “A very unimportant decision.” The third question is: “The decision about what hotel to book requires...” with a scale from “A little thought” to “A lot of thought.”
Consumer expertise refers to a consumer’s experience, product knowledge, and familiarity in a product category (Biswas and Sherrell 1993). To measure consumer expertise, we use two questions with a seven-point scale from “strongly disagree” to “strongly agree”: “Compared to other people, I often book a hotel room” and “Compared to my friends and acquaintances, I often book a hotel room.”
4.1.8 Consumer demographics
We measure consumer demographics to assess if the randomization is successful. We ask subjects questions on their age and gender.
4.1.9 Validity issues
We try to improve the external validity of our online experiment by designing the fictional webpage of an independent review website to appear as realistic as possible. To improve internal validity, we take into account possible confounding variables. For example, we measure the subjects’ experience with online consumer reviews. In addition, we ensure that both the reviewer in the low-similarity condition and the reviewer in the high-similarity conditions have a male name.
We include several questions and checks to assess the validity of the responses. First, we measure the time subjects spend on the webpage of the fictional independent review website. In addition, we ask a question to ascertain whether or not the subjects read the instructions. Furthermore, we ask the subjects about their nationality. Also, we ask the subjects a question to check whether they are able to memorize the nationality of the reviewer.
4.2 Plan of analysis
24
25 5. RESULTS
In this chapter, we discuss the results from the experiment carried out. In the first paragraph, we discuss the validity of the responses. In the second paragraph, we discuss the scale reliability of the measures in the experiment. We cover the moderating effects of source credibility on consumer purchase intentions in the third paragraph. In the fourth paragraph, we discuss the effects of source similarity and reviewer expertise on source credibility. We discuss the mediating effects of source credibility in the fifth paragraph.
5.1 Subjects
Four hundred and five subjects participated in the online experiment. In this paragraph, we discuss the validity of the responses. In addition, we discuss whether our manipulations and randomizations were successful.
5.1.1 Invalid responses
In the experiment, we included several questions and checks to assess the validity of the responses. Out of 405 subjects, 18 spent less than 10 seconds on the webpage of the fictional review website. Twenty four subjects were not able to correctly answer the question which checked whether they read the instructions or not. Thirty seven subjects filled out a different nationality than American and three subjects participated more than once in the experiment. Twenty nine subjects were not able to correctly memorize the nationality of the reviewer. After we removed these invalid subjects from the dataset, there were 294 subjects left. There are at least 14 subjects and at most of 22 subjects in each experimental condition. There are 72 subjects in the control condition (table 2).
Table 2
Number of responses per condition
Group 1 2 3 4 5 6 7 8 9 10 11 12 Control
Responses 17 14 18 20 19 20 19 17 21 22 16 19 72
5.1.2 Manipulation checks
26
negative valence and the positive valence condition (F = 7.492, p < 0.05). We can draw the conclusion that our manipulations were successful.
5.1.3 Randomization
There are no significant differences in age (F = 1.452, p < 0.142) and experience with online consumer reviews (F = 0.823, p < 0.627) between the conditions. This indicates that the random assignment was successful.
5.2 Scale reliability
In this paragraph we discuss the scale reliability of the measures in the experiment. We discuss the measures of source credibility, consumer purchase intentions, product involvement and consumer experience.
5.2.1 Source credibility dimensions
In the experiment, we asked the subjects six questions on perceived source credibility. A correlation analysis (Appendix II, table 1) showed that there are substantial and significant correlations between the variables. We executed a principal component analysis (PCA), which showed that a two-factor solution is most appropriate. We applied the orthogonal rotation method VARIMAX to create a simpler and clearer factor solution. As a result, a large part of the variance of each variable is explained by the factors. We labeled the first source credibility factor as ‘expertise’ and the second factor as ‘trustworthiness’. In addition to PCA, we computed the Cronbach’s α in order to ensure “consistency of the entire scale” (Hair et al. 1998, p. 118), a value of 0.79 was found.
5.2.2 Consumer purchase intentions
We executed a correlations analysis to examine if we were allowed to combine the three questions on consumer purchase intentions into one variable. The correlation analysis (Appendix II, table 2) showed significant correlations between the three variables. In addition, the Cronbach’s α of the three questions is 0.94. Therefore, we combined the three questions into one consumer purchase intentions variable. We averaged the variables to produce a consumer purchase intentions scale that ranges from 1 to 7.
5.2.3 Product involvement
27
three questions is 0.84. Therefore, we combined the three questions into one product involvement. We averaged the variables to produce a product involvement scale that ranges from 1 to 7.
5.2.4 Consumer expertise
We measured consumer expertise with two questions. We executed a correlation analysis to assess if we were allowed to combine these two questions into one variable. The correlation analysis showed a significant correlation between the two variables (r = 0.841, p < 0.01). Therefore, we combined the two questions into one consumer expertise variable. We averaged the variables to produce a consumer expertise scale that ranges from 1 to 7.
5.3 Effects of review valence and source credibility
In this paragraph, we discuss the effects of source credibility on the relationship between review valence and consumer purchase intentions. First, we discuss the main effect of review valence on consumer purchase intentions. After that, we discuss the moderating effect of source credibility on the relationship between review valence and consumer purchase intentions.
5.3.1 Main effect of review valence
The mean of consumer purchase intentions is highest in the positive valence group (5.30), and lowest in the negative valence group (2.69). The mean consumer purchase intention of the control group is 4.65. An ANOVA analysis (Appendix II, table 4) showed that the differences between the groups are significant (F = 217.81, p < 0.01). Comparisons between groups (Appendix II, table 5) confirmed that there are significant differences between the positive valence group and the control group (p < 0.05) and the negative valence and the control group (p < 0.05). We find evidence that a review with a negative valence towards a service has a negative effect on consumer purchase intentions and a review with a positive valence towards a service has a positive effect on consumer purchase intentions.
5.3.2 Moderating effect of source credibility
28
Table 3 shows that the significance levels for review valence (t = -1.631, p < 0.104), the trustworthiness dimension (t = -0.876, p < 0.382) and the expertise dimension (t = -0.242, p < 0.809) are well above 0.05. However, the interactions between the trustworthiness dimension and valence (t = 5.524, p < 0.01) and the expertise dimension and valence (t = 4.601, p < 0.01) are significant and positive. Therefore, we can draw the conclusion that the source credibility dimensions trustworthiness and expertise strengthen the relationship between review valence and consumer purchase intentions.
Table 3
Regression analysis on consumer purchase intentions
Variable Beta Standard Error T Significance
Constant 4.615 .370 12.468 .000 Review valence* -.455 .279 -1.631 .104 Trustworthiness dimension -.059 .067 -.876 .382 Expertise dimension -.015 .063 -.242 .809 Trustworthiness. x valence .367 .066 5.525 .000 Expertise x valence .287 .062 4.601 .000 Involvement (control) -.097 .051 -1.906 .058 Expertise (control) -.026 .037 -.685 .494
*Negative valence is coded as -1, positive valence is coded as 1
Figure 2 Figure 3
29
To examine these results in greater depth, we conducted two spotlight analyses. In a spotlight analysis, the researcher can focus the ‘spotlight’ on a particular level of a continuous independent variable by shifting the mean of this variable up or down (Fitzsimons 2008). A spotlight analysis at one standard deviation above and one standard deviation below the mean of trustworthiness (figure 2) shows the interaction effect between review valence and the trustworthiness dimension of source credibility. Similarly, a spotlight analysis at one standard deviation above and one standard deviation below the mean of expertise (figure 3) shows the interaction effect between review valence and the expertise dimension of source credibility.
We draw the conclusion that the perceived source credibility has a positive effect on the relationship between review valence and consumer purchase intentions. Therefore, we find support for hypothesis 1.
5.4 Effects of source similarity and reviewer expertise on source credibility
In this paragraph, we discuss the effects of source similarity and reviewer expertise on the expertise and trustworthiness dimensions of source credibility. First, we present the ANCOVA analyses on the effects of source similarity and reviewer expertise on the trustworthiness dimension of source credibility. After that, we discuss the ANCOVA analyses on the effects of source similarity and reviewer expertise on the expertise dimension of source credibility.
5.4.1 ANCOVA analyses trustworthiness dimension
We estimated two ANCOVA models to assess the effects of source similarity and reviewer expertise on the trustworthiness dimension of source credibility. In both models, we included review valence as a covariate to account for the difference in mean trustworthiness between the negative and positive valence groups (Appendix II, table 6). In the first ANCOVA model, we included dummy variables for the non-disclosure and the low similarity group. In the second ANCOVA model, we included dummy variables for the low similarity and the high similarity group.
30
of high source similarity compared to non-disclosure (F = 3.621, p < 0.058). The interactions in the model are both insignificant. The second model is significant (p < 0.01) and the R2 of the model is 0.172. Table 4 shows that non-disclosure compared to low similarity has a significant effect (F = 4.377, p < 0.05) on consumer purchase intentions. Therefore, we find support for hypothesis h2b, in case of a low-similar source, the disclosure of personal information has a negative effect on the perceived source credibility of the consumer. The interactions in the model are both insignificant.
Table 4
ANCOVA models for the trustworthiness dimension of source credibility
Source Model 1 Model 2
F Significance F Significance Corrected model 7.835 .000 7.432 .000 Intercept 1261.95 .000 977.471 .000 Review valence 35.234 .000 31.839 .000 Reviewer expertise .537 .465 .809 .369 Non-disclosure dummy* 10.426 .001 - - Low similarity dummy* 3.621 .058 - - Expertise x non-disclosure dummy .168 .682 - - Expertise x low similarity dummy .290 .591 - - Low similarity dummy** - - 4.377 .038 High similarity dummy** - - 1.948 .164 Expertise x low similarity dummy - - .030 .862 Expertise x high similarity dummy - - .028 .868
R2 .179 .172
* Compared to high similarity group ** Compared to non-disclosure group
5.4.2 ANCOVA analyses expertise dimension
We estimated two ANCOVA models to assess the effects of source similarity and reviewer expertise on the expertise dimension of source credibility. In both models, we included review valence as a covariate to account for the difference in mean expertise between the negative and positive valence groups (Appendix II, table 7). In the first ANCOVA model, we included dummy variables for the non-disclosure and the low similarity group. In the second ANCOVA model, we included dummy variables for the low similarity and the high similarity group.
31
addition, we found no effect of high source similarity compared to non-disclosure (F = 1.458, p < 0.229). The interactions in the model are also insignificant. The second model is significant (p < 0.01) and the R2 of the model is 0.141. Table 5 shows that non-disclosure compared to low similarity has no significant effect (F = 1.329, p < 0.25). The interactions in the model are also insignificant.
Table 5
ANCOVA models for the expertise dimension of source credibility
Source Model 1 Model 2
F Significance F Significance Corrected model 5.946 .000 5.905 .000 Intercept 2348.95 .000 1704.44 .000 Review valence 7.974 .005 7.713 .006 Reviewer expertise 23.827 .000 21.521 .000 Non-disclosure dummy* .277 .599 - - Low similarity dummy* 1.458 .229 - - Expertise x non-disclosure dummy .404 .526 - - Low similarity dummy** - - 1.329 .250 High similarity dummy** - - .090 .764 Expertise x low similarity dummy - - .563 .454 Expertise x high similarity dummy - - .400 .528
R2 .142 .141
* Compared to high similarity group ** Compared to non-disclosure group
5.5 Mediating effects of source credibility
In this paragraph, we examine whether source credibility mediates the relationship between source similarity and reviewer expertise and consumer purchase intentions. We follow the mediation procedure recommended by Baron and Kenny (1986). First, we examine for both negative and positive valence reviews if the source credibility dimension trustworthiness mediates the relationship between source similarity and consumer purchase intentions. Second, we examine for negative and positive valence reviews if the source credibility dimension expertise mediates the relationship between reviewer expertise and consumer purchase intentions.
5.5.1 Mediating effects of the trustworthiness dimension
32
we confirm that the high similarity group has a positive effect on trustworthiness compared to the low similarity group (t = -2.766, p < 0.01). In the third regression, we find that trustworthiness predicts consumer purchase intentions (t = -4.110, p < 0.01), whereas the effect of high similarity compared to low similarity is no longer significant (t = 1.371, p < 0.173). A Sobel test confirms these mediating effects (z = 2.29, p < 0.05). For positive reviews, we do not find mediating effects of trustworthiness in the relationship between source similarity and consumer purchase intentions. In addition, we do not find mediating effects of trustworthiness in the relationship between the disclosure of personal information and consumer purchase intentions for both positive and negative valence reviews.
Table 6
Mediation analysis trustworthiness dimension for negative reviews
Dependent variable Independent variables Beta Standard Error T Significance
CPI Constant 2.366 .173 13.697 .000 Non-disclosure dummy* .434 .246 1.766 .080 Low similarity dummy* .604 .259 2.335 .021 Trustworthiness
dimension
Constant -.016 .156 -.101 .920 Non-disclosure dummy* -.432 .222 -1.947 .054 Low similarity dummy* -.646 .234 -2.766 .007 CPI Constant 2.359 .162 14.605 .000 Non-disclosure dummy* .260 .234 1.110 .269 Low similarity dummy* .343 .250 1.371 .173 Trustworthiness dim. -.404 .098 -4.110 .000 * Compared to high similarity group
From the results, we can conclude that there is partial support for hypothesis 3. We find that only for negative reviews, the trustworthiness dimension of source credibility mediates the relationship between source similarity and consumer purchase intentions.
5.5.2 Mediating effects of the expertise dimension
33
indirect effects (Preacher and Hayes 2004; Zhao, Lynch, and Chen 2010) does confirm the mediation effect. The estimated 95% confidence interval around the indirect effect of reviewer expertise on consumer purchase intentions does not contain zero (-0.2161 to -0.0105), supporting mediation. For negative reviews, we do not find mediating effects of expertise in the relationship between reviewer expertise and consumer purchase intentions.
Table 7
Mediation analysis expertise dimension for positive reviews
Dependent variable Independent variables Beta Standard Error T Significance
CPI Constant 5.153 .107 48.216 .000 Reviewer expertise .330 .159 2.083 .040 Expertise dimension Constant -.020 .112 -.178 .859 Reviewer expertise .416 .166 2.506 .014 CPI Constant 5.156 .105 48.925 .000 Reviewer expertise .254 .161 1.578 .118 Expertise dimension .183 .091 2.005 .048
34 6. CONCLUSION
In this chapter, we first discuss the results of our empirical research. Next, we discuss the contributions we make to practice with this study. In addition, we explain how marketing managers can use our results in practice. We end the chapter with a discussion of the limitations of our research and directions for future research.
6.1 Discussion
In the current research, we find evidence for the relationship between review valence and consumer purchase intentions. We find that a positive valence review has a positive effect and a negative valence review has a negative effect on consumer purchase intentions. With these results, we confirm the findings of Chevalier and Mayzlin (2006) and others that online consumer reviews influence consumer purchase behavior.
We also find evidence that perceived source credibility acts a moderator in the relationship between review valence and consumer purchase intentions. The higher the perceived source credibility the stronger the relationship between review valence and consumer purchase intentions. The results in the current research show that the conclusion of Hovland and Weiss (1951) that sources with a high credibility are more persuasive than sources with a low credibility is also valid in an online environment. Also, in line with the findings of Woodside and Davenport (1974) and Friedman and Friedman (1979), we show the importance of perceived source credibility in marketing situations.
In addition, we find that the similarity between the reviewer and the consumer positively influences the trustworthiness dimension of source credibility. We show that the relation between source similarity and source credibility (Swartz 1984; Brock, 1965) also exists in an online environment. Also, we find that the expertise of a reviewer has a positive impact on the expertise dimension of perceived source credibility. Although this seems straightforward, it shows that consumers actually use badges and other cues to assess the credibility of reviews. With these cues consumers can distinguish between consumer reviewers with and consumer reviewers without in-depth knowledge in product categories.
35
In addition, we find that for positive valence reviews, source credibility mediates the relationship between reviewer expertise and consumer purchase intentions. In other words, when a review is written by an expert, the positive effect of a positive review on consumer purchase intentions is stronger than when a review is written by a non-expert. As far as we know, we are the first to explain the effects of reviewer expertise on consumer purchase intentions with source credibility theory. For negative valence reviews, the effect of reviewer expertise on consumer purchase intentions is close to significant.
Furthermore, results made clear that for negative reviews, source credibility mediates the relationship between source similarity and consumer purchase intentions. In other words, when a review is written by a reviewer highly similar to the consumer, the negative effect of a negative review on consumer purchase intentions is stronger than when a review is written by a low-similar reviewer. This effect can be explained by a higher perceived source credibility of the review. For positive reviews, we find no effect of source similarity on consumer purchase intentions.
The question then is why we only find an effect of source similarity with negative valence reviews. One possible explanation can be derived from the negativity bias phenomenon (the negativity effect). Cannon (1929) states that negative stimuli are more likely to prime or activate some degree of fight-or-flight response than positive stimuli. Also, Baumeister et al. (2001) state that as a result of evolutionary adaption, organisms better attuned to negative things have a higher probability of survival throughout evolutionary history. Survival requires that organisms become more alert to negative information than to positive information. As a result, the psychological system responds more strongly to negativity than to positivity. An additional effect is that the need for trust in the source of the message becomes more important (Goleman, McKee, and Boyatzis 2002). According to the social identity theory, individuals want to communicate with others who are similar to themselves to reduce uncertainty and create trust (Brown and Reingen 1987). Hence, in situations with negative information, there is a higher need for similarity between the source and the recipient. Therefore, a higher similarity between the reviewer and the consumer results in higher persuasiveness and thus higher consumer purchase intentions.
36
6.2 Theoretical contributions
By means of conceptual and empirical research, we contribute to marketing theory on online consumer reviews, electronic word-of-mouth and consumer decision making. First, where other authors mainly focus on consumer characteristics and product characteristics (e.g. Zhu and Zhang 2010), we are the first to include source similarity and reviewer expertise in a model on the influence of online consumer reviews on consumer purchase intentions. In addition, instead of conceptualizing reviewer expertise as the difference between consumer and professional reviewers, we differentiate between consumers with and consumers without in-depth mastery of a field of knowledge.
Second, we contribute to research on electronic word-of-mouth with the use of source credibility as a mediator in our conceptual model. Whereas other research in electronic word-of-mouth often focuses on the basic relationship between electronic word-of-mouth and consumer behavior, we use an established theory in research on persuasiveness to explain the relationships in our model. Our results show a strong relationship between the two source credibility dimensions trustworthiness and expertise and consumer purchase intentions. In addition, we show that source credibility acts as a mediator in the relationship between source similarity and purchase intentions for negative reviews and in the relationship between reviewer expertise and purchase intentions for positive reviews.
Third, we extend theory on consumer decision making by showing different effects of source similarity for positive and negative information. For positive valence reviews we find no effect of source similarity on consumer purchase intentions, while we find an effect of source similarity on consumer purchase intentions for negative reviews. We explain these findings with use of the negativity bias phenomenon and social identity theory. As a result, the current study gives more insights in the consumer’s decision process and how negative and positive arguments affect consumer purchase decisions.
6.3 Managerial implications
The present research confirms the important effects of online consumer reviews on consumer purchase intentions. Managers cannot ignore the effects of both negative and positive reviews on the purchase behavior of consumers. They should be actively involved in and allocate resources to managing online consumer reviews to influence product sales.
37
intentions. Two important influencers of the credibility of a review are the similarity between the reviewer and the consumer and the expertise of the reviewer.
In case of negative reviews, the similarity between the reviewer and the consumer influences consumer purchase intentions. In other words, a negative review has a stronger negative effect on the purchase behavior of another consumer when it is written by a similar consumer. As a result, a negative review written by a consumer in the target audience of a business can have stronger negative effects of business performance than a review from a reviewer who is not a member of a segment of interest. Therefore, managers should give consumers in their target audience a higher priority with regard to service recovery in case of service failure to prevent negative reviews from these consumers. In addition, managers can use ‘owner response’ features on online consumer websites to respond on negative reviews written by these consumers.
For positive reviews, the expertise of a reviewer influences consumer purchase intentions. A positive review from an experienced consumer has a stronger positive effect on purchase intentions than a review from a consumer with less experience. Therefore, when managers want to encourage consumers to write reviews on their products, they should especially focus their efforts on reviewers who are experienced in writing reviewers and are experts in their product category. For example, consumers with a high number of reviews written or consumer with certain badges (such as an ‘Elite’ badge on Yelp.com or a ‘Top 500 Reviewer’ badge on Epinions.com) are more influential.
6.4 Limitations
The results of the present research are subject to a number of limitations. First, in the experiment we exposed only one review to the subjects. However, previous research finds that the review quantity influences consumer purchase intentions (e.g. Chatterjee 2001). The higher the reviewer quantity, the more important the product is and the more persuasive the reviews are. Also, Lui (2006) states that review quantity is a determinant of product sales. In addition, several subjects commented that they usually use more than one review in their purchase decision.
38
expertise manipulation influenced the source similarity manipulation. For example, a consumer who is not an expert might perceive a reviewer who is an expert as less similar. Feick and Higie (1992) found these effects in their study on ad processing and judgments about endorsers.
Third, in the current study we used demographic similarity and dissimilarity to test the effects of similarity on source credibility and consumer purchase intentions. We did not take into account other types of similarities that consumers can assess on consumer review websites, such as similarities in age or gender. For example, Gilly et al. (1998) find that next to demographic similarity, perceptual similarity (similarities in values, preferences and lifestyles) affects word-of-mouth processes.
Fourth, we did not take into account the initial opinions of consumers towards the product. As a result, we were not able to measure the change in purchase intentions after review exposure. Other studies on online consumer reviews find significant changes in pre- and post-attitudes (e.g. Vermeulen, Das, and Swager 2008).
6.5 Directions for future research
Additional research on online consumer reviews might include more than one review in an experiment to see if our conclusions hold in such a situation. For example, researchers can expose one group of subjects to multiple reviews with negative reviews from dissimilar sources and positive reviews from similar sources and one group of subjects to multiple reviews with positive reviews from dissimilar sources and negative reviews from similar sources and see if there are differences between groups in consumer purchase intentions.
Also, future research could study different types of electronic word-of-mouth, such as weblogs and social networks and consumer forums. Follow-up studies on the effects of source similarity, expertise and source credibility in electronic word-of-mouth will contribute to a more systematized body of knowledge in the field. As a result, instead of scattered facts, practitioners can benefit from a coherent body of knowledge.
39
factors between negative and positive arguments are related to the negativity effect and social identity theory.
40 REFERENCES
Alba, Joseph W. and J. Wesley Hutchinson (1987), “Dimensions of Consumer Expertise,” Journal of
Consumer Research, 13 (4), 411-454.
Baron, Reuben M. and David A. Kenny (1986), “The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations,” Journal of Personality and
Social Psychology, 51 (6), 1173-1182.
Baumeister, Roy F., Ellen Bratslavsky, Catrin Finkenauer and Kathleen D. Vohs (2001), “Bad is stronger than good,” Review of General Psychology, 5 (4), 323-370.
Bem, Daryl J. (1972), “Self-perception theory,” in Advances in Experimental Social Psychology, ed. L. Berkowitz, San Diego, CA: Academic Press, 6, 1-62.
Berscheid, E. (1966), “Opinion change and communicator-communicatee similarity and dissimilarity,”
Journal of Personality and Social Psychology, 4, 670-680.
Bharadwaj, Sundar G., P. Rajan Varadarajan and John Fahy (1993), “Sustainable Competitive Advantage in Service Industries: A Conceptual Model and Research Propositions,” Journal of Marketing, 57 (4), 83-99.
Birnbaum, Michael H. and Steven E. Stegner (1979), “Source credibility in social judgment: Bias, expertise, and the judge's point of view,” Journal of Personality and Social Psychology, 37 (1), 48-74. Biswas, Abhijit and Daniel L. Sherrell (1993), “The influence of product knowledge and brand name on internal price standards and confidence” Psychology and Marketing, 10 (1), 31-46.
Biyalogorsky, Eyal, Eitan Gerstner, and Barak Libai (2001), “Customer Referral Management: Optimal Reward Programs,” Marketing Science, 20 (Winter), 82-95.
Bone, Paula F. (1995), “Word-of-mouth effects on short-term and long-term product judgments,”
Journal of Business Research, 32 (3), 213-223.
Brock, Timothy C. (1965), “Communicator-recipient similarity and decision change,” Journal of
Personality and Social Psychology, 1 (6), 650-654.
Brown, Jacqueline Johnson and Peter H. Reingen (1987), “Social Ties and Word-of-Mouth Referral Behavior,” Journal of Consumer Research, 14 (3), 350-362.
Brown, Jo, Amanda J. Broderick and Nick Lee (2007), “Word of mouth communication within online communities: Conceptualizing the online social network,” Journal of Interactive Marketing, 21 (3), 2-20. Byrne, Donn and William Griffitt (1966), “A Developmental Investigation of the Law of Attraction,”
Journal of Personality and Social Psychology, 4 (6), 699-702.
41
Chatterjee, Patrali (2001), “Online Reviews: Do Consumers Use Them?” in ACR 2001 Proceedings, eds. M. C. Gilly and J. Myers-Levy, Association for Consumer Research, 129-134.
Chen, Yubo, and Jinhong Xie (2008), “Online Consumer Review: Word-of-Mouth as a New Element of Marketing Communication Mix,” Management Science, 54 (3), 477-491.
Chen, Pei-Yu, Shin-Yi Wu, and Jungsun Yoon (2004), “The Impact of Online Recommendations and Consumer Feedback on Sales,” in Proceedings of the International Conference on Information Systems,
ICIS 2004. Seattle: Association for Information Systems, 711-724.
Cheung, Man Y., Chuan Luo, Choon L. Sa, and Huaping Chen (2009), “Credibility of Electronic Word-of-Mouth: Informational and Normative Determinants of on-line consumer Recommendations,”
International Journal of Electronic Commerce, 13 (4), 9-38.
Chevalier, Judith A. and Dina Mayzlin (2006), “The Effect of Word of Mouth on Sales: Online Book Reviews,” Journal of Marketing Research, 43 (August), 345-54.
Clark, Russel D. and Anne Maass (1988), “The role of social categorization and perceived source credibility in minority influence,” European Journal of Social Psychology, 18 (5), 381-394.
comScore (2007), “Online Consumer-Generated Reviews Have Significant Impact on Offline Purchase Behavior,” (accessed May 2011), [available at
http://www.comscore.com/press/release.asp?press=1928].
Crano, William D. (1970), “Effects of Sex, Response Order, and Expertise in Conformity: A Dispositional Approach,” Sociometry, 33 (3), 239-252.
Dellarocas, Chrysanthos (2003), ”The Digitization of Word of Mouth: Promise and Challenges of Online Feedback Mechanisms,” Management Science, 49 (10), 1407-1424.
Dellarocas, Chrysanthos (2006), “Strategic Manipulation of Internet Opinion Forums: Implications for Consumers and Firms,” Management Science, 52 (10), 1577–1593.
Doney, Patricia M. and Joseph P. Cannon (1997), “An Examination of the Nature of Trust in Buyer-Seller Relationships,” Journal of Marketing, 61, 35-51.
Duan, Wenjing, Bin Gu, and Andrew B. Whinston (2008), “Do Online Reviews Matter? An Empirical Investigation of Panel Data,” Decision Support Systems, 45 (4), 1007–1016.
Duhan, Dale F., Scott D. Johnson, James B. Wilcox, and Gilbert D. Harrell (1997), “Influences on Consumer use of Word-of-Mouth Recommendation Sources,” Journal of the Academy of Marketing
Science, 25 (4), 283-295.
