Running head: EXPERT VERSUS PEER REVIEWS ********************* DRAFT February 11, 2009 ********************* Preferences for Expert Versus Peer Reviews on the Internet L. Mark Carrier a, Nancy A. Cheever b, and Larry D. Rosen a a Department of Psychology, California State University-Dominguez Hills, 1000 E. Victoria St., Carson CA 90747, United States b Department of Communications, California State University-Dominguez Hills, 1000 E. Victoria St., Carson CA 90747, United States E-mail addresses: L. Mark Carrier (lcarrier@csudh.edu), Nancy A. Cheever (ncheever@csudh.edu), Larry D. Rosen (lrosen@csudh.edu). Corresponding Author: L. Mark Carrier, Telephone: 00-1-310-243-3499; FAX: 00-1-310-516-3642. DRAFT, Feb 11, 2009 Preferences for Expert 1 Preferences for Expert Versus Peer Reviews on the Internet An online questionnaire queried consumer preferences for online expert versus peer reviews at different types of websites and measured consumer reactions to artificial expert and peer reviews. The data showed that negative reviews were more important and more helpful than positive ones and that website type affected preferences. The results suggest a model in which consumers discount products or services with negative reviews, seek peer-provided information as a substitute for direct physical contact with products that normally require experience to evaluate, and seek expert-provided information for evaluating products whose full attributes can be ascertained from technical information about the products. Keywords: Electronic Word-of-Mouth, Online Reviews, Consumer Decisions, Internet, Consumer Choice DRAFT, Feb 11, 2009 Preferences for Expert 2 Preferences for Expert Versus Peer Reviews on the Internet 1. Introduction Online consumers have access to a variety of opinions about products, including those by the sellers themselves, unbiased experts, and peers (Crabtree, 2007; Voight, 2007), and evidence shows that shoppers make use of this online review information (Freedman, 2008; Senecal & Nantel, 2004; Westerman, Tuck, Booth, & Khakzar, 2007) and that customer word-of-mouth affects purchasing behavior (Chevalier & Mayzlin, 2006; Hu, Liu, & Zhang, 2008; Riegner, 2007). Online reviews of products or services could benefit shoppers for several reasons. First, reviews might match customers’ natural inclinations to seek out additional information when using the computer to shop (Schlosser, 2003). Second, online “recommendation agents” might reduce information overload in the user’s cognitive system (Hu et al., 2008; Swaminathan, 2003). Third, user- and expert-generated reviews might foster website “trust,” leading to later purchasing behavior (Sillence & Briggs, 2007). Fourth, online reviews by consumers provide both informational and product popularity data that impact purchaser behavior (Park, Lee, & Han, 2007). 2. Peer versus expert online reviews Peer and expert reviews represent two different sources of online information for consumers (Swaminathan, 2003). Peer reviews are tied to subjective customer word-of-mouth and expert reviews are tied to mostly technical seller-generated information (Chen & Xie, 2008). There is little research as to how these separate information sources are valued by consumers and under what conditions they are valued differently. Cheung, Lee and Rabjohn (2008) hinted that the difference between peers and experts as sources of reviews might have only a minimal effect DRAFT, Feb 11, 2009 Preferences for Expert 3 upon customer reactions to reviews. On the other hand, books recommended online by peers are more preferred than books recommended by experts, all other things being equal (Chen, 2008). 2. 1. Linking product type to preference for peer review All information about “search” products can be acquired prior to purchase (e.g., a portable stereo), whereas “experience” products are those for which the consumer must have experience with the product in order to acquire all relevant information about it (e.g., a song download from the Internet) (Klein, 1998; Nelson, 1970). Websites offering peer and expert reviews vary in the types of products offered. Experience products may be associated with a need for subjective information about the products and search products may be associated with a need for technical information. Thus, a consumer’s preference for peer versus expert reviews may depend upon whether technical information is being sought or subjective information is being sought, leading to the following hypothesis. H1: Variation in the type of website upon which reviews are offered will affect consumers’ preferences for peer versus expert reviews. 2.2. The role of user experience Increased experience with a website probably will be associated with increased experience with the types of products or services offered. Increased experience with a type of product or service could contribute to a consumer’s mental database of subjective knowledge about that type of product or service, resulting in less of a need for seeking out subjective information from others. Therefore, the following hypothesis was generated. H2: Users with increased experience at a website will show reduced preference for peer reviews. 2.3. The role of review valence DRAFT, Feb 11, 2009 Preferences for Expert 4 Prior research suggests that the valence (negative versus positive) of online reviews contributes to the value of reviews for online consumers, although the research results are mixed. Positive reviews have a stronger effect than negative reviews on heeding advice from movie reviews (Gershoff, Mukherjee, & Mukhopadhyay, 2003). In contrast, user evaluations of the value of reviews are higher for negative reviews than for positive reviews (Park & Lee, 2009). Because the review valence might contribute to the evaluation of review usefulness by consumers, the valence of artificial reviews was manipulated in the present study, and the following research question was generated. RQ1: What is the effect of review valence upon consumers’ preferences for peers versus experts? An online survey was used to assess Internet users’ preferences for online reviews that varied in their source (peer versus expert), their valence (negative or positive), and their subject (different types of products and services). Respondents were asked to evaluate hypothetical reviews from experts or from peers. Respondents indicated how valuable the reviews appeared to be. Additionally, respondents rated their preference for peer (over expert) reviews on various types of websites. 3. Method 3.1. Participants A convenience sample of 414 respondents was formed in Spring 2007 by students in an upper-division General Education course at a four-year university in Southern California. Potential respondents were contacted through informal channels via the students. Respondents’ ages ranged from 18-years to 55-years old. The largest age group was 18-24 years old (38% of participants) and the next largest age group was 25-30 years old (25%). The respondents were DRAFT, Feb 11, 2009 Preferences for Expert 5 mostly females (58%). The ethnic backgrounds of the respondents were Asian (9%), Black (25%), Hispanic (33%), White (27%), and other (6%). 3.2. Materials Respondents indicated how often they used each of several different online information sources: movie reviews, technology reviews of products costing less than $100, technology reviews of products costing more than $100, book reviews, health issues, professor ratings, Wikipedia, hotel ratings and evaluations, travel site evaluations, video hosting sites such as YouTube, and blogs about politics. For each source, participants indicated their frequency of use: 1 = “At least a few times per week,” 2 = “About once a week,” 3 = “A few times a month,” 4 = “About once a month,” 5 = “I rarely use it,” and 6 = “I never use it.” (“At least a few times per week” and “About once a week” were later combined into “Weekly” to balance the cell sample sizes.) Participants also indicated preferences for expert commentary versus peer commentary on each online information source (1 = “Only Experts,” 2 = “Mostly Experts,”, 3 = “Equally Experts and Peers,” 4 = “Mostly Peers,” and 5 = “Only Peers”). Artificial review scenarios presented negative and positive hotel and professor reviews differing by type of author (peer versus expert) (Table 1). The review presentation order started with the professor reviews. After reading each review, participants rated its helpfulness and importance in their decision-making process on a 5-point scale, from most important or helpful (1) to least important or helpful (5). After each review pair, subjects chose which author (peer versus expert) was most valuable in making their decision. Last, participants read an artificial online review of a book called ''Professional Chef 30Minute Recipes'' by Carol Laschober, and rated the helpfulness and importance of the review in two different situations, one identified as a peer (jackson9999@yahoo.com) and the second DRAFT, Feb 11, 2009 Preferences for Expert 6 identified as an expert (foodeditor@hawthornepress.com). Helpfulness was rated on a 1 (“Very Helpful”) to 4 (“Very Unhelpful”) scale. Importance was rated on a 1 (“Very Important”) to 4 (“Very Unimportant”) scale. Participants also chose which reviewer they would “trust” more. 3.3 Procedure The General Education students solicited the participants. Respondents had to be at least 18 years of age and regular Internet users (at least five hours per week). A web link was provided that linked to the questionnaire posted on SurveyMonkey.com, an online provider of survey services. The participants remained anonymous and no incentive was provided for participation. 4. Results 4.1. Effects of demographic variables Correlations between preferences and age, sex, and education level, were not statistically significant, except for two results indicating that on Wikipedia.com and on video hosting websites, males were more likely to prefer experts than were females. Based on these nonsignificant relationships, demographics were not considered further. 4.2. Preference for peer reviews by source A between-subjects, 11 X 4 analysis of variance (ANOVA) was performed on the preferences with information source as the first factor and frequency of use as the second factor (Figure 1). There was a significant main effect of information source, F(10, 3275) = 76.53, p < 0.001, indicating that the overall preferences for peers versus experts depended upon the type of information source (Table 2). The information sources formed five groups (post-hoc analyses using Tukey’s B test). Health Issues websites showed the strongest preference for experts, followed by Wikipedia. A next group of information sources, including Tech Reviews > $100, Political Blogs, and Tech Reviews < $100, showed a smaller preference for experts. The next DRAFT, Feb 11, 2009 Preferences for Expert 7 group of sources—Travel Sites, Hotel Ratings, Book Reviews, and Movie Reviews—showed close to no preference for either experts or peers. A last group of websites that included Video Hosting and Professor Ratings showed the strongest preferences for peers. Statistically, the number of visits to an information source also impacted the preference for peers versus experts, F(3, 3275) = 15.13, p < 0.001. Increasing frequency of use was related to increased preference for experts in nearly all information sources (Figure 1). Website and frequency did not interact, F(30, 3275) = 1.19, p = 0.224. 4.3. Artificial reviews Two hundred and ninety-three participants (70.8% of the sample) completed the artificial review items. The helpfulness ratings for each professor and hotel review scenario were subjected to a 2 X 2 X 2 within-subjects ANOVA with valence (negative versus positive), author (peer versus expert), and category (professor versus hotel) as factors (Figure 2). Negative reviews were rated more helpful than positive reviews, F(1, 291) = 10.45, p < 0.005, peers were more helpful than experts, F(1, 291) = 15.06, p < 0.001, and category and valence interacted with each other, F(1, 291) = 7.06, p < 0.01. The effect of the negative reviews was reduced for the professor reviews. The main effect of category was not significant, F(1, 291) = .067, p = 0.796, nor was the interaction between category and author, F(1, 291) = .512, p = .475, nor the interaction between valence and author, F(1, 291) = 3.37, p = 0.072, nor the three-way interaction between the factors, F(1, 291) = .177, p = 0.674. A 2 X 2 X 2 ANOVA with valence, author, and category as factors examined the artificial review importance ratings (Figure 3). Negative reviews were deemed more important than positive reviews, F(1,291) = 62.61, p < 0.001, peers again were preferred over experts, F(1,291) = 21.90, p < 0.001, and there was a significant interaction between valence and author, DRAFT, Feb 11, 2009 Preferences for Expert 8 F(1,291) = 4.77, p < 0.05, showing that there was a reduced effect of peer versus expert within the negative reviews. Further, hotel ratings were viewed as more important than professor ratings, F(1,291) = 6.14, p < 0.05. The interaction between category and valence was not significant, F(1, 291) = .324, p = 0.570, nor was the interaction between category and author, F(1, 291) = .081, p = 0.776, nor the three-way interaction between the factors, F(1, 291) = .022, p = 0.882. For the negative professor review, 29.7% of the participants chose the expert review as most valuable and 70.3% chose the peer review. For the positive review, 22.2% chose the expert review and 77.8% chose the peer review. The preference for peers did not differ depending on valence, 2(1) = 0.04, p = 0.85. Thus, three measures of preference for peer versus expert reviews show a preference for peers, although the effect is slightly reduced with ratings of importance of negative reviews. 4.4. Book review scenario The expert review (M = 1.72; SD = 0.71) was significantly more helpful than the peer review (M = 2.19; SD = 0.85); t(292) = 8.81, p < 0.001. This result contrasts with the professor and hotel reviews, in which the peers were found to be more helpful than the experts. A similar pattern was found for the importance ratings where the peer as reviewer (M = 2.35; SD = 0.85) was rated as significantly lower than the expert (M = 1.95; SD = 0.75); t(292) = 7.76, p < 0.001. Two hundred and forty-two people chose the expert as the reviewer to trust most, while only 51 people chose the peer. Thus, the expert as reviewer was more trusted than the peer as reviewer; 2(1) = 124.51, p < 0.001. 5. General discussion, implications, and limitations 5.1. General discussion DRAFT, Feb 11, 2009 Preferences for Expert 9 This study measured consumer preferences for peer versus expert reviews in online settings. Preferences for peer reviews were hypothesized to vary by the type of information source and experience with an information source was hypothesized to increase the preference for expert reviews. Additionally, a research question asked about the role of review valence in preferences for peers versus experts. The study results showed that preferences for peers varied by the kind of information source (Table 2) and those preferences were somewhat consistent for information sources and artificial reviews, supporting Hypothesis 1. Experienced Internet users more strongly preferred expert reviews than peer reviews, supporting Hypothesis 2. Finally, there was a pattern of preference for negative over positive reviews (Research Question 1). 5.1.1. Theoretical explanations Consumers might prefer experts for their objective knowledge about a product or service such as medical information or encyclopedic information queries, but prefer peers for their subjective knowledge associated with experience products (e.g., taking a college course). Negative reviews were preferred over positive reviews in both helpfulness and importance ratings suggesting that participants are using what Reed (2004) defined as a non-compensatory decision-making model: “a strategy that rejects alternatives that have negative attributes without considering positive attributes” (p. 356). The finding that experienced Internet users tend toward preferring experts over peers shows that repeated use of an information source increases the weight of expert reviews in the decision-making process. The user might gain individualized subjective experience and exposure to the class of products through practice and now seek objective information. For example, a user who has purchased enough books online to learn about her own book preferences could then focus on objective information—hence, expert review—that contains factual information about a book and maximizes the chances that her next DRAFT, Feb 11, 2009 Preferences for Expert 10 book purchase matches her preferences. The development of “preference structures” of product categories that represent the consumer’s evolving preferences within that product category (Gershoff et al., 2003) could underlie this effect. A different search strategy might be employed when subjective information is being sought than when objective information is being sought. The data showed a smaller preference for negative over positive reviews with subjective information seeking (for professor ratings which showed that peers were preferred over experts), hinting at a decision-making strategy balancing positive and negative information (i.e., a compensatory approach). 5.2. Limitations The sample used in this study might not be representative of all Internet users. The majority of the sample was young (18 to 30 years old) and the respondents were sampled from the Southern California region. This region is known for having a relatively large Hispanic population, as well as a relatively large proportion of persons who are immigrants or children of recent immigrants into the United States. 5.3. Implications The results of the present study have implications for consumers and for sellers in the online environment. For consumers, it appears that positive reviews might be undervalued, so paying more attention to positive reviews could be recommended for optimizing decisionmaking outcomes. For online businesses, one implication of the present results is that there should be product alternatives offered for products with negative reviews, since consumers appear to be influenced by such reviews. Another implication is that there should be reviews posted (peer versus expert) that match the type of information that consumers might be seeking. DRAFT, Feb 11, 2009 Preferences for Expert 11 References Chen, Y. (2008). Herd behavior in purchasing books online. Computers in Human Behavior, 24, 1977-1992. Chen, Y., & Xie, J. (2008). Online consumer review: Word-of-mouth as a new element of marketing communication mix. Management Science, 54(3), 477-91. Cheung, C. M. K., Lee, M. K. O., & Rabjohn, N. (2008). The impact of electronic word-ofmouth: The adoption of online opinions in online customer communities. Internet Research, 18(3), 229-247. Chevalier, J. A., & Mayzlin, D. (2006). The effect of word of mouth on sales: Online book reviews. Journal of Marketing Research, XLIII, 345-54. Crabtree, P. (November 25, 2007). Staying power. The San Diego Union Tribune. [retrieved from the World-Wide Web at http://ww.signonsandiego.com/uniontrib/20071125/news_lz1b25stay.html on April 10, 2008] Freedman, L. (February, 2008). Merchant and customer perspectives on customer reviews and user-generated content. Published online by the e-tailing group and Power Reviews. [retrieved from the World-Wide Web at http://www.etailing.com/graphics/2008_WhitePaper_0204_4FINAL.pdf on July 2, 2008] Gershoff, A. D., Mukherjee, A., & Mukhopadhyay, A. (2003). Consumer acceptance of online agent advice: Extremity and positive effects. Journal of Consumer Psychology, 13(1&2), 161-170. DRAFT, Feb 11, 2009 Preferences for Expert 12 Hu, N., Liu, L., & Zhang, J. J. (2008). Do online reviews affect product sales? The role of reviewer characteristics and temporal effects. Information Technology and Management, 9, 201-214. Klein, L. R. (1998). Evaluating the potential of interactive media through a new lens: Search versus experience goods. Journal of Business Research, 41, 196–203. Nelson, P. (1970). Information and consumer behavior. Journal of Political Economy, 78, 311– 329. Park, C., & Lee, T. M. (2009). Information direction, website reputation and eWOM effect: A moderating role of product type. Journal of Business Research, 62, 61-67. Park, D-H., Lee, J., & Han, I. (2007). The effect of on-line consumer reviews on consumer purchasing intention: The moderating role of involvement. International Journal of Electronic Commerce, 11(4), 125-48. Reed, S. K. (2004). Cognition: Theory and Applications (Sixth Edition). Belmont, CA: Wadsworth/Thomson Learning. Riegner, C. (2007). Word of mouth on the Web: The impact of Web 2.0 on consumer purchase decision. Journal of Advertising Research, 47(4), 436-47. Schlosser, A. E. (2003). Computers as situational cues: Implications for consumers product cognitions and attitudes. Journal of Consumer Psychology, 13(1&2), 103-12. Senecal, S., & Nantel, J. (2004). The influence of online product recommendations on consumers’ online choices. Journal of Retailing, 80, 159-169. Sillence, E., & Briggs, P. (2007). Please advise: Using the Internet for health and financial advice. Computers in Human Behavior, 23, 727-48. DRAFT, Feb 11, 2009 Preferences for Expert 13 Swaminathan, V. (2003). The impact of recommendation agents on consumer evaluation and choice: The moderating role of category risk, product complexity, and consumer knowledge. Journal of Consumer Psychology, 13(1&2), 93-101. Voight, J. (November 12, 2007). Shoppers want more customer reviews. Adweek. [retrieved from the World-Wide Web at http://www.adweek.com/aw/national/article_display.jsp?vnu_content_id=1003671155 on April 10, 2008] Westerman, S. J., Tuck, G. C., Booth, S. A., & Khakzar, K. (2007). Consumer decision support systems: Internet versus in-store application. Computers in Human Behavior, 23, 2928-44. DRAFT, Feb 11, 2009 Preferences for Expert 14 Table 1. Artificial professor reviews. Positive review by expert (a professor) I observed Professor Jones several times this semester in his History 101 class. He gives clear lectures that include good handouts and information displayed on a screen in front of the class. His syllabus is clear and sets all dates for tests. His tests are fair and not too easy nor too difficult. He does give weekly homework which requires several hours outside of class per week. Positive review by peer (a student) I took History 101 from Professor Jones. I found him pretty easy to listen to and enjoyed the lectures. It was helpful that he used both paper handouts and slides. His tests were difficult but if you study you will do well. There is a lot of homework but mostly it just requires time. It is not difficult. Overall, I recommend that you take Dr. Jones for this class. Negative review by expert I observed Professor Smith several times this semester in his Political Science 100 class. Overall, I found it difficult to follow his lectures. He would start out with one idea and then mid-way through that, start talking about something else that may or may not be relevant to that topic. His syllabus did list dates for the midterm and final but there was no other information given about these tests. The tests themselves were, in my opinion, pretty difficult and the grades were not very good. Negative review by peer I took Political Science 101 from Professor Smith last semester. This guy can’t seem to stay on topic. He starts talking about something and then jumps to another topic and then back to the first. Very hard to follow. His tests were much harder than I expected and I don’t think people did very well on them. Overall, I would stay away from him. He is too confusing and too difficult. NOTE: Artificial hotel reviews were similarly constructed. DRAFT, Feb 11, 2009 Preferences for Expert 15 Table 2. Mean Preference Ratings for Experts Versus Peers Information Source Mean SE Health Issues 1.93 0.05 Wikipedia 2.35 0.06 Technology Reviews > $100 2.61 0.05 Political Blogs 2.67 0.06 Technology Reviews < $100 2.76 0.05 Travel Site Evaluations 2.98 0.04 Hotel Ratings 3.00 0.05 Book Reviews 3.03 0.05 Movie Reviews 3.15 0.05 Video Hosting 3.57 0.05 Professor Ratings 3.60 0.08 Note. All ratings used the following scale: 1=Only Experts, 2=Mostly Experts, 3=Equally Experts and Peers, 4=Mostly Peers, 5=Only Peers DRAFT, Feb 11, 2009 Preferences for Expert 16 Figure Captions Figure 1. Preference data for the different information sources ranging from 1 (only experts) to 5 (only peers). The lines represent different subsets of users categorized on the basis of their frequency of use of the information sources. Figure 2. Average participant ratings of helpfulness of artificially-constructed reviews of professors and hotels. Figure 3. Average importance ratings of artificial hotel and professor reviews provided to participants. ch Te ch Te R R ok > < R s s s 00 $1 00 $1 ie w ev R H ev ie ea w s lth Is su Pr es of R at in gs W ik ip H ed ot ia el R at in Tr gs av el Vi Si de te s o H os Po tin lit g ic al Bl og s Bo ie w ev ie w ev ie M ov Prefer Experts (1) vs. Peers (5) DRAFT, Feb 11, 2009 Preferences for Expert 17 4 3.5 3 2.5 2 Weekly 1.5 A Few Times a Month About Once a Month I Rarely Use It 1 Website FIGURE 1. DRAFT, Feb 11, 2009 Preferences for Expert 18 1.8 1.7 Helpful (1) vs. Unhelpful (4) 1.6 1.5 1.4 Professor Review - Bad 1.3 Professor Review - Good Hotel Review - Bad Hotel Review - Good 1.2 Expert Peer Writer FIGURE 2. DRAFT, Feb 11, 2009 Preferences for Expert 19 2 Professor Review - Bad Professor Review - Good Hotel Review - Bad Hotel Review - Good Important (1) vs. Unimportant (4) 1.9 1.8 1.7 1.6 1.5 1.4 Expert Peer Writer FIGURE 3.