The Differential Effects of Online Peer Review and Expert Review on Service Evaluations

Abstract

The present research investigates the differential effects of online peer review and expert review on consumers’ evaluations of experience and credence services. We propose that these effects are mediated by consumers’ confidence in their service evaluation and moderated by information convergence. We conduct three studies to test our hypotheses. Study 1 shows that consumers evaluate experience (vs. credence) services more favorably when exposed to peer review (vs. expert review). Across the three studies, we show that the interaction effects between information source and service type on service evaluation are mediated by consumer confidence. Importantly, we identify the moderating role of information convergence on these effects (Studies 2 and 3). Convergent positive reviews substantiate the interaction effects between information source and service type on service evaluation. Interestingly, when consumers see mixed information from either similar or different sources, negative expert review has greater influence than negative peer review in lowering consumer confidence and their evaluations of both experience and credence services. These findings contribute to the literature on information processing in the services domain and also have significant practical implications on managing consumer expectations of third-party information.

Keywords

information source experience service credence service confidence online marketing

Consumers often seek and rely on information from a third party to reduce the uncertainty that may accompany goods or services consumption and to make more informed choices (Chakravarty, Liu, and Mazumdar 2010). This is because, compared to direct communication by the marketer (e.g., advertising), information from a third party is often perceived to be more credible due to a lack of vested interest by the source (Bansal and Voyer 2000; Smith, Menon, and Sivakumar 2005). While other researchers have investigated the effects of third-party information sources in the context of goods (e.g., Gilly et al. 1998; Huang and Chen 2006; Senecal and Nantel 2004), the present research focuses on services. This is because services are associated with greater uncertainty and variability compared to goods (Murray and Schlacter 1990; Zeithaml, Parasuraman, and Berry 1985), leading consumers to approach goods purchases differently from services purchases (Carter and Gilovich 2010).

Prior research shows that consumers tend to pay greater attention to what others have to say when making service evaluations (Bansal and Voyer 2000; Holbrook and Addis 2007). In today’s vibrant Internet environment, social networks, blogs, recommendation sites, and online communities greatly facilitate information search. Online reviews that are available prior to purchase and consumption have a profound impact on consumer decision-making (Chakravarty, Liu, and Mazumdar 2010; Sotiriadis and van Zyl 2013). In general, online third-party reviews can come from two sources: (1) ordinary consumers who purchased, used, or experienced the service and wanted to share their comments with their peers, which we term “peer review” and (2) experts aligned with institutions or organizations (e.g., Consumer Reports in the United States and Which? in the United Kingdom) who reviewed or tested the service and provided their objective assessments, which we term “expert review.”

Yet, with few exceptions, prior research on third-party information tends to focus on the effects of either peer review (e.g., Chevalier and Mayzlin 2006; Duhan et al. 1997; Lee and Bradlow 2011; Liu 2006; Sotiriadis and van Zyl 2013; Zhu and Zhang 2010) or expert review (e.g., Basuroy, Chatterjee, and Ravid 2003; Eliashberg and Shugan 1997) on consumer responses or market performance. Thus, it is not readily apparent what factors determine consumers’ adherence to peer versus expert reviews. This gap in the literature does not reflect market reality, whereby consumers are regularly exposed to both online peer and expert reviews for the same service (e.g., http://rottentomatoes.com, http://zagat.com, and http://tripadvisor.com).

Furthermore, there are different categories of services. A widely used categorization is based on the ease by which services could be evaluated; that is, services have varying levels of experience and credence attributes (Ostrom and Iacobucci 1995). Experience attributes are those that require actual experience of the service to be evaluated, and credence attributes are those that are difficult to evaluate even after consumption—they have to be taken on faith. Thus, credence services tend to carry greater uncertainty and risks compared to experience services (Keh and Pang 2010; Mitra, Reiss, and Capella 1999; Murray and Schlacter 1990).

Accordingly, the present research seeks to examine the differential effects of online peer review and expert review on consumer evaluations for experience and credence services. We propose that these effects can be explained by the confidence mechanism. That is, consumers’ reliance on peer review versus expert review would vary depending on their confidence level in evaluating the service. Specifically, consumers have lower confidence in evaluating credence (vs. experience) services, leading them to rely more on expert review (vs. peer review). As it is common for consumers to come across multiple reviews of the same service, which could be convergent or mixed, we further propose that these effects are moderated by information convergence. If supported, findings from this research contribute new insights to the services marketing literature and have significant managerial implications.

The remainder of this article is organized as follows. We first review the relevant literature to develop hypotheses on the differential effects of peer versus expert reviews, experience and credence services, and information convergence. We then conduct three studies to test the hypotheses. Study 1 examines the differential effects of peer review and expert review on consumer evaluations of experience and credence services and the mediating role of confidence underlying these effects. Study 2 further tests the mediation effect of confidence and unveils the boundary condition of information convergence on the interaction effects between information source and service type. Drilling down, Study 3 investigates the moderating effect of mixed information from similar (i.e., mixed peer reviews or mixed expert reviews) as well as different sources (i.e., mixed peer and expert reviews). We conclude by discussing the theoretical and managerial implications of the findings.

Research Background

Peer review and expert review

Existing research on advice giving and taking mainly examines two message sources: ordinary consumers and experts (Chakravarty, Liu, and Mazumdar 2010; Libai et al. 2010; Palmeira, Spassova, and Keh 2015). While recognizing that some ordinary consumers may be experts and are able to provide knowledgeable and insightful reviews (Bettman and Sujan 1987), in this article, peer review refers to reviews by nonexpert consumers (Holbrook and Addis 2007). A common form of peer review is word of mouth (WOM), defined as the “informal communications directed at other consumers about the ownership, usage, or characteristics of particular goods and services and/or their seller” (Westbrook 1987, p. 261). While the scope of WOM was traditionally confined to oral, face-to-face, and other means of direct communication, the rapid growth of the Internet and associated technologies has amplified the impact of peer reviews online. We operationalize peer reviews as coming from other novice consumers who have no commercial intentions or vested interests in the focal brands (Bansal and Voyer 2000; Smith, Menon, and Sivakumar 2005).

In contrast, expert review here refers to professional review by a third party (Reinstein and Snyder 2005). The key difference between expert review and peer review lies in domain expertise (Spence and Brucks 1997). Compared to peer review, expert review is more professional, authoritative, and formal. We operationalize expert review as impersonal, professional reviews of services by critics and experts from institutional entities (e.g., Consumer Reports; Chen and Xie 2008).

On some review websites (e.g., http://rottentomatoes.com), consumers can see both peer review and expert review simultaneously. Surprisingly, there is scant research examining the differential effects of peer review and expert review on consumers’ service evaluations. For instance, Chakravarty, Liu, and Mazumdar (2010) find that expert reviews have stronger influence on prerelease movie evaluations than peer reviews for frequent moviegoers. Seiders et al. (2015) show that customer efficacy and service provider efficacy moderate consumers’ adherence to expert advice for medical services. Notably, movies and medical service are exemplars of experience service and credence service, respectively, which have significantly different characteristics and effects.

Using secondary data from yelp.com, in which expertise was operationalized using the number of reviews made by a person, regression analyses by Racherla and Friske (2012) yielded the negative effect of reviewer expertise. They speculated that reader disbelief in reviewer expertise could have led to this anomalous finding. In addition, the effects of review valence are not clearly established, with some studies indicating that negative reviews influence consumer behavior more strongly than positive reviews, and other studies finding no asymmetrical effects (Kimmel and Kitchen 2014). Furthermore, the effect and weight of review valence are moderated by consumer expertise (Ketelaar et al. 2015). Taken together, prior research on peer and expert reviews does not unequivocally explain how these reviews affect service evaluations.

Experience service and credence service

The experience-credence typology is based on the extent to which customers are able to evaluate services based on the service attributes (Darby and Karni 1973; Mitra, Reiss, and Capella 1999).

Experience services can be more easily evaluated following their consumption, whereas credence services are often associated with a higher degree of variability that makes them difficult to evaluate even after consumption (Keh and Pang 2010; Ostrom and Iacobucci 1995). Thus, credence services are perceived to be more risky and uncertain than experience services (Mitra, Reiss, and Capella 1999). Consumers’ lack of confidence in evaluating credence services originates from the greater uncertainty associated with them (Mattila and Wirtz 2002; Sun, Keh, and Lee 2012).

Past research shows that consumers process experience and credence services differently (Ostrom and Iacobucci 1995). Mitra, Reiss, and Capella (1999) find that consumers rely heavily on impersonal information (e.g., third-party reports, newspaper, TV, and pamphlet) and to some extent on weighted personal information (e.g., WOM and personal experience) when evaluating credence services. However, their findings are equivocal on the information source consumers rely on when evaluating experience services. While prior research has studied various factors influencing consumers’ evaluations of credence services (Mattila and Wirtz 2002; Seiders et al. 2015), to our knowledge, the present research is the first to examine the differential effects of peer review versus expert review on consumers’ service evaluations.

Hypotheses Development

The differential effects of peer review and expert review on consumer evaluations

Research based on social comparison theory reveals the positive effect of source similarity on persuasion (Faraji-Rad, Samuelsen, and Warlop 2015; Yaniv, Choshen-Hillel, and Milyavsky 2011). Specifically, homophily, or similarity between individuals, facilitates information transfer, attitude formation, and interpersonal interaction, leading to better communication and greater influence on the information seeker’s decision (Gilly et al. 1998; McPherson, Smith-Lovin, and Cook 2001; Yaniv, Choshen-Hillel, and Milyavsky 2011). The reasoning is that homophilous ties make it easier for an individual to relate to similar others, thus increasing the likelihood of activating a peer review flow of information (Brown and Reingen 1987). Faraji-Rad, Samuelsen, and Warlop (2015) argue that similarity influences the advice-taking process by shaping the feeling of certainty. Huang and Chen (2006) show that online peer recommendations have greater influence than expert recommendations on consumer choices for books, which represent an experience product. As consumers can more confidently evaluate experience services, we propose that they would tend to relate to and rely more on peer reviews (vs. expert reviews).

While consumers can more easily determine the value of experience attributes prior to purchase, information search for credence attributes can be quite lengthy, difficult, and frustrating (Darby and Karni 1973). By nature, credence attributes make the service (e.g., medical, education, financial, and legal services) difficult to evaluate even after consumption (Keh and Pang 2010; Ostrom and Iacobucci 1995). As consumers often do not have enough knowledge to confidently evaluate credence services, they would perceive greater uncertainty and risk for such services (Sun, Keh, and Lee 2012). Decisions that involve high uncertainty warrant more knowledgeable and credible information to mitigate the uncertainty (Racherla and Friske 2012). Thus, peer review cannot mitigate consumers’ lack of confidence in evaluating credence services. Unlike novice consumers, experts are deemed to have authoritative knowledge about credence services (Spence and Brucks 1997). We propose that as credence services are complex and their outcomes tend to be more uncertain, consumers would rely more on expert review as a cue for decision-making (De Bruyn and Lilien 2008; Seiders et al. 2015). More formally:
Hypothesis 1: For experience services, consumers’ service evaluations will be higher when consumers see a peer review than when they see an expert review.

Hypothesis 2: For credence services, consumers’ service evaluations will be higher when consumers see an expert review than when they see a peer review.

The Mediating Role of Confidence

Past research on dual-process theories (e.g., the heuristic-systematic model of persuasion and the elaboration likelihood model) indicates that when consumers feel uncertain and less confident, they will continue to search for and process more information until they attain the sufficiency threshold (Chaiken and Maheswaran 1994; Trope and Chaiken 1999). Consumers considering a highly risky purchase would seek additional information from others who are perceived to have a high level of expertise and knowledge (Bansal and Voyer 2000), and the influence of expert comments is greater in a risky and complex purchase context (De Bruyn and Lilien 2008). Indirect evidence from Price and Stone (2004) shows that decision makers rely more on knowledgeable and confident financial advisors than on less knowledgeable ones when choosing a credence product. Thus, the extent to which consumers seek advice or information from experts versus novices depends on confidence in their own judgment.

Specifically, confidence in judgment is a cognition-based evaluation, which refers to the degree of certainty people hold about the appropriateness or optimality of their decisions (Thomas and Menon 2007). It reflects the perceived quality and correctness of consumers’ beliefs in their judgments, which are highly malleable and subject to the context in which the judgment is made (Tsai and McGill 2011). That is, confidence in judgment is a state rather than a trait. Confidence in judgment is highly consequential and is a reliable predictor of consumers’ attitudes or actual behavior (Chaxel 2016; Sun, Keh, and Lee 2012), such as the extent of advice taking (Palmeira, Spassova, and Keh 2015). Faraji-Rad, Samuelsen, and Warlop (2015) show that the feeling of certainty serves as a mediator in the advice-taking process.

Accordingly, we propose that consumers’ varying responses to peer review and expert review for experience and credence services can be explained by their confidence in service evaluation. As consumers have higher confidence in evaluating experience services, they would rely more on peer review. Conversely, consumers have lower confidence in evaluating credence services (Mattila and Wirtz 2002), leading to greater reliance on expert review. More formally:
Hypothesis 3: Consumers’ confidence mediates the interaction effects between information source and service type on their service evaluations.

The Moderating Role of Information Convergence

The effects predicted in Hypotheses 1–3 are based on the premise that consumers view a positive review (i.e., either peer review or expert review). In reality, it is more common for consumers to come across multiple reviews of the same service, which could be convergent or mixed (i.e., some reviews are positive while others are negative). Thus, we propose that information convergence in the reviews serves as a boundary condition for these effects. Specifically, convergent positive information enhances consumers’ confidence in their service evaluation, resulting in the effect hypothesized in Hypothesis 3. More formally:
Hypothesis 4a: When the multiple reviews are positively convergent, consumers’ evaluations of the experience (vs. credence) service will be higher when they see peer reviews (vs. expert reviews).

In contrast, mixed reviews can increase consumers’ cognitive load and lower their confidence (Erber and Fiske 1984). This is because information incongruity violates expectancies, which leads to surprise, captures attention, and promotes cognitive elaboration (Karmarkar and Tormala 2010). In particular, mixed or conflicting judgments between experts have a stronger negative effect on confidence than mixed judgments between ordinary consumers. The reasoning is, while disagreements between novice consumers tend to be based more on opinion and less on authoritative knowledge, individuals often look up to experts for guidance and would be more confused and less confident in their own evaluations when the experts disagree. Thus, we propose that mixed expert reviews would have stronger negative impacts than mixed peer reviews on consumers’ evaluations for both experience and credence services. More formally:
Hypothesis 4b: When the multiple reviews are mixed, mixed expert reviews have stronger negative impacts than mixed peer reviews on consumers’ evaluations for both experience and credence services.

Study 1: Underlying Effects of Confidence on Information Source

Study 1 investigated the differential effects of online peer review and expert review on consumers’ service evaluations (Hypotheses 1 and 2). We also tested the mediating role of confidence underlying these effects (Hypothesis 3).

Following the procedure by Sun, Keh, and Lee (2012), we first conducted a pretest with 72 undergraduates to rate 36 services and classify them as either experience or credence services based on the definitions provided. Specifically, if the service could be evaluated confidently postconsumption, it would be categorized as an experience service. In contrast, if the service could not be evaluated confidently even after consumption, it would be classified as a credence service (Ostrom and Iacobucci 1995). Based on this pretest, Study 1 used hotel and dental care to represent experience and credence services, respectively.

Method

Design and Sample

Study 1 used a 2 (information source: peer review vs. expert review) × 2 (service type: experience vs. credence) between-participants design. We recruited 120 undergraduate students (74.2% female, mean age = 20.5 years) at a major university in China who were compensated for their participation.

Procedure

Participants were randomly assigned to one of the four conditions. They first read a service consumption scenario (i.e., hotel or dental care) for which they needed to search for additional information. They then read a review purportedly from either an ordinary consumer (i.e., peer review) or an expert (i.e., expert review) as shown in the Appendix.

Following that, participants evaluated the service on a 3-item scale (Sun, Keh, and Lee 2012), rated their confidence in evaluating the service on a 2-item scale (Chaxel 2016; Sun, Keh, and Lee 2012), and rated source expertise on a 2-item scale (Karmarkar and Tormala 2010). Participants also responded to several control variable questions, including their involvement in processing the information (Wang and Lee 2006), purchase importance (Ostrom and Iacobucci 1995), and perceived homophily (Faraji-Rad, Samuelsen, and Warlop 2015). All items were measured on 7-point scales and are summarized in Table 1.

Table 1.
Variables and Measures Used in Studies 1–3.

Reliability Coefficients

Variables Measures Study 1 Study 2 Study 3

Service evaluations Please evaluate the service on the following measures: satisfactory-satisfactory, unfavorable-favorable, dislike-like (7-point scales) α _hotel = .93, α _dental = .91 α _{hair salon} = .96, α _insurance = .96

Expected service quality Based on the scenario, how would you expect the service quality of the service to be? (1 = very low, 7 = very high)Overall, what are your expectations of the service quality of this service? (1 = very poor, 7 = outstanding)” r_restaurant = .82, r_language = .77

Purchase likelihood How likely are you to patronize the service firm? (1 = very unlikely, 7 = very likely)How likely are you to purchase the service from this firm? (1 = very unlikely, 7 = very likely) R _restaurant = .87, r _language = .86

Confidence in service evaluation Please indicate how confident you feel about your decision given the review you read. (1 = not at all confident, 7 = extremely confident)Please indicate how certain you feel about your decision given the review you read. (1 = not at all certain, 7 = extremely certain) r_hotel = .77, r_dental = .79 r_{hair salon} = .86, r_insurance = .63 r_restaurant = .87, r_language = .89

Source expertise (Study 1) How knowledgeable do you think the expert/consumer is about the service firm? (1 = not at all knowledgeable, 7 = very knowledgeable)How expert do you think the expert/consumer is about the service firm? (1 = not at all expert, 7 = highly expert) r_hotel = .77, r_dental = .84

Source expertise (Studies 2 and 3) How knowledgeable do you think peer/expert reviewer A is about the service firm? (1 = not at all knowledgeable, 7 = very knowledgeable)How expert do you think peer/expert reviewer A is about the service firm? (1 = not at all expert, 7 = highly expert)How knowledgeable do you think peer/expert reviewer B is about the service firm? (1 = not at all knowledgeable, 7 = very knowledgeable)How expert do you think peer/expert reviewer B is about the service firm? (1 = not at all expert, 7 = highly expert) α _{hair salon} = .87, α _insurance = .70 α _restaurant = .60, α _language = .60

Involvement How involved were you in processing the information? not at all involved-very involved, skimmed it quickly-read it carefully (7-point scales) r_hotel = .79, r_dental = .84 r_{hair salon} = .70, r_insurance = .67 r_restaurant = .60, r_language = .76

Purchase importance How critical do you think is the purchase decision? (1 = very low purchase criticality, 7 = very high purchase criticality)

Perceived homophily (Study 1) In general, how similar is the ordinary consumer/expert to yourself? (1 = not at all similar, 7 =very similar)In general, how similarly does the ordinary consumer/expert behave like yourself? (1 = not at all similar, 7 = very similar) r_hotel = .82, r_dental = .78

Perceived homophily (Studies 2 and 3) In general, how similar is Peer/Expert Reviewer A to yourself? (1 = not at all similar, 7 = very similar)In general, how similarly does Peer/Expert Reviewer A behave like yourself? (1 = not at all similar, 7 = very similar)In general, how similar is Peer/Expert Reviewer B to yourself? (1 = not at all similar, 7 = very similar)In general, how similarly does Peer/Expert Reviewer B behave like yourself? (1 = not at all similar, 7 = very similar) α _{hair salon} = .72, α _insurance = .86 α _restaurant = .60, α _language = 60

Emotional difficulty I would feel the decision described in the scenario is an emotionally difficult one to make (1 = not at all, 7 = extremely so)I would feel the decision described in the scenario is stressful. (1 = not at all, 7 = extremely so) r_hotel = .60, r_dental = .61 r_{hair salon} = .60, r_insurance = .77 r_restaurant = .78, r_language = .81

Results

Manipulation Checks

A 2 (information source) × 2 (service type) analysis of variance (ANOVA) on source expertise revealed only the main effect of information source, F(1, 116) = 17.03, p < .001, η² = .12, such that participants perceived the expert review to be significantly more knowledgeable than the peer review—hotel: M_PR = 4.42 vs. M_ER = 4.98, F(1, 58) = 4.38, p < .05, η² = .07; dental care: M_PR = 3.85 vs. M_ER = 4.90, F(1, 58) = 13.75, p < .001, η² = .19. A 2 × 2 ANOVA on involvement and purchase importance revealed no significant effects (all Fs < 2.13, all p values >.10). Similarly, a 2 × 2 ANOVA on perceived homophily (r _hotel = .82, r _dental = .78) yielded only a significant main effect of information source, F(1, 116) = 8.01, p < .01, η² = .06, such that participants’ perceived similarity was significantly higher for peer review than for expert review—hotel: M_PR = 4.67 vs. M_ER = 3.98, F(1, 58) = 4.11, p < .05, η² = .07; dental care: M_PR = 4.62 vs. M_ER = 4.03, F(1, 58) = 3.94, p = .05, η² = .06.

Service Evaluation

As the 3 items measuring service evaluation had high reliability scores (α_hotel = .93, αdental = .91), we averaged them to form a service evaluation index. A 2 × 2 ANOVA revealed a significant interaction between information source and service type, F(1, 116) = 12.17, p < .001, η² = .10 (Figure 1a). Planned contrasts showed that participants had higher evaluation of the hotel service when exposed to peer review than to expert review (M_PR = 5.74 vs. M_ER = 5.14), F(1, 58) = 6.31, p < .05, η² = .10, whereas they had higher evaluation of the dental care service when exposed to expert review than to peer review (M_PR = 4.67 vs. M_ER = 5.09); F(1, 58) = 6.18, p < .05, η² = .10. Thus, Hypotheses 1 and 2 were supported.

Figure 1.
Interaction effects of information source and service type (Study 1). (a) Service evaluation, (b) confidence rating.

Confidence as a mediator

As the measurement items for confidence were highly correlated (r _hotel = .77, r _dental = .79), they were averaged to form a confidence index. Results of a 2 × 2 ANOVA showed that the main effect of service type, F(1, 116) = 8.42, p < .01, η² = .07, was qualified by the predicted information Source × Service Type interaction, F(1, 116) = 8.42, p < .05, η² = .07 (Figure 1b). Specifically, for the hotel service, participants had higher confidence ratings when exposed to peer review than to expert review (M_PR = 5.43 vs. M_ER = 4.87); F(1, 58) = 4.15, p < .05, η² = .07. In contrast, for the dental service, participants had higher confidence ratings when exposed to expert review than to peer review (M_PR = 4.33 vs. M_ER = 4.87); F(1, 58) = 4.29, p < .05, η² = .07.¹

We tested the mediating effect of confidence using bootstrapping analysis with 5,000 samples (PROCESS Model 8, Hayes 2013). A 95% confidence interval (CI) for the indirect effect was significant and excluded zero (95% CI: [.14, .84]), and the partial information Source × Service Type interaction effect was also significant, β = .58, t(115) = 2.22, p < .05 (Figure 2). That is, consumers’ feelings of confidence partially mediated the interaction effects of information source and service type on their service evaluations, supporting Hypothesis 3.

Figure 2.
Mediated moderation analysis (Study 1). p < .001. p < .01.

Discussion

Study 1 shows that consumers’ service evaluations of an experience (vs. credence) service are higher when they see a peer (vs. expert) review, consistent with Hypotheses 1 and 2, respectively. Study 1 also offers evidence that consumers’ confidence in their evaluations mediates the interaction effects between information source and service type, supporting Hypothesis 3.

A possible alternative explanation for our findings is the level of perceived emotional difficulty associated with the service situation. In particular, White (2005) finds that participants tend to solicit advice from experts when their decisions are low in perceived emotional difficulty but favor the advice of ordinary consumers when their decisions have high perceived emotional difficulty. In our scenarios, requiring participants to imagine booking a hotel for a family holiday may potentially induce high emotional difficulty, while booking a dental service for themselves may induce low emotional difficulty. To rule out this possible confound, we conducted a follow-up study to assess participants’ emotional difficulty for the hotel and dental care services. We measured perceived emotional difficulty using 2 items from White (2005) as shown in Table 1. Results indicated that participants’ perceived emotional difficulty did not significantly differ between the hotel and the dental care conditions (M _hotel = 3.61 vs. M _dental = 4.18); t(24) = −1.68, p > .10, η² = .07. Thus, our results were not confounded by perceived emotional difficulty.

Nonetheless, a limitation in Study 1 was that participants were exposed to one positive review from either an ordinary consumer or an expert. As it is more common for consumers to come across multiple reviews of the same service, information convergence could serve as a boundary condition for the results found in Study 1.

Study 2: Moderating Role of Information Convergence

Study 2 tested the moderating role of convergent (vs. mixed) information on the interaction effects of information source and service type. Specifically, we predicted that convergent positive information would enhance consumers’ confidence in their service evaluations (Hypothesis 4a) as found in Study 1. In contrast, mixed information, particularly between expert reviews, would attenuate consumers’ confidence in their service evaluations (Hypothesis 4b).

Method

Design and Sample

Study 2 used a 2 (information source: peer review vs. expert review) × 2 (service type: hair salon vs. insurance) × 2 (information convergence: convergent vs. mixed) between-participants design. Participants were recruited online at Amazon’s Mechanical Turk (M-Turk; N = 126, 42.9% females, 85.7% between ages 20 and 39).

Materials and Procedure

Participants were randomly assigned to one of the eight conditions. They first read a service consumption scenario (i.e., hair salon or insurance agency representing experience and credence services, respectively) for which they needed to search for additional information as shown in the Appendix. In each service situation, they read either two convergent reviews (i.e., two positive peer reviews or two positive expert reviews) or two mixed reviews (i.e., two conflicting peer reviews or two conflicting expert reviews).² For the mixed reviews, the presentation order of positive and negative information was counterbalanced. Following that, participants responded to questions measuring their service evaluation, confidence in service evaluation, source expertise, involvement, purchase importance, and perceived homophily as in Study 1.

Results

Manipulation Check

A 2 (information source) × 2 (service type) × 2 (information convergence) ANOVA on source expertise revealed only the main effect of information source, F(1, 118) = 9.48, p < .01, η² = .07. Specifically, participants perceived the expert reviews to be more knowledgeable than the peer reviews—hair salon: M_PR = 4.51 vs. M_ER = 4.95, F(1, 60) = 3.44, p = .06, η² = .05; insurance: M_PR = 4.64 vs. M_ER = 5.05, F(1, 62) = 3.82, p = .06, η² = .06, indicating successful manipulations of information source.

A 2 × 2 × 2 ANOVA on participants’ involvement and purchase importance revealed no significant effects (all Fs < 2.60, all p values >.10). In addition, a 2 × 2 × 2 ANOVA on perceived homophily (αhair salon = .72, αinsurance = .86) showed only the significant main effect of information source, F(1, 118) = 20.94, p < .001, η² = .15, such that participants reported higher perceived similarity for the peer reviews than for the expert reviews—hair salon: M_PR = 4.75 vs. M_ER = 4.11, F(1, 60) = 8.16, p < .01, η² = .12; insurance: M_PR = 4.70 vs. M_ER = 4.14; F(1, 62) = 9.24, p < .01, η² = .13. Furthermore, results showed that participants’ perceived emotional difficulty (r _{hair salon} = .60, r _insurance = .77) did not significantly differ between the two services (Mhair salon = 3.80 vs. Minsurance = 3.95), t(32) = −.44, p > .10.

Service Evaluation

As the measurement items for service evaluations were highly correlated (αhair salon = .96, αinsurance = .96), they were averaged to form a service evaluation index. Importantly, a 2 × 2 × 2 ANOVA on service evaluation revealed the predicted three-way interaction effect, F(1, 118) = 8.72, p < .01, η² = .07, indicating that information convergence moderated the interaction effects between information source and service type.

We then separately examined the convergent and mixed information conditions. In the convergent positive information condition, there was a significant two-way interaction between information source and service type, F(1, 62) = 11.62, p < .001, η² = .16 (Figure 3a), supporting Hypothesis 4a. Planned contrasts showed that for the hair salon, participants had higher service evaluations when exposed to peer reviews than to expert reviews (M_PR = 6.15 vs. M_ER = 5.69), F(1, 31) = 4.72, p < .05, η² = .13, while for the insurance agency, participants had higher service evaluations when exposed to expert reviews than to peer reviews (M_PR = 5.56 vs. M_ER = 6.20), F(1, 31) = 6.91, p < .05, η² = .18. These results further supported Hypotheses 1 and 2.

Figure 3.
Moderating effect of information convergence (Study 2). (a) Convergent information, (b) mixed information.

Relative to the convergent positive information condition, we expected lower service evaluations for the mixed information condition. ANOVA results revealed only the significant main effect of information source, F(1, 52) = 11.32, p < .001, η² = .18, with no interaction between information source and service type (Figure 3b). For both services, participants had lower service evaluations when exposed to mixed expert reviews than to mixed peer reviews—hair salon: M_PR = 4.07 vs. M_ER = 3.71, F(1, 25) = 7.23, p < .01, η² = .22; insurance: M_PR = 4.31 vs. M_ER = 3.81, F(1, 27) = 5.36, p < .05, η² = .17. These findings implied that mixed expert reviews had stronger negative effects than mixed peer reviews on service evaluations for both experience and credence services, supporting Hypothesis 4b.

Confidence

Similarly, as the measurement items for confidence were highly correlated (r _{hair salon} = .86, r _insurance = .63), they were averaged to form a confidence index. A 2 × 2 × 2 ANOVA on confidence indicated a significant three-way interaction effect, F(1, 118) = 4.05, p < .05, η² = .03. As information convergence moderated the interaction effects of information source and service type, we decomposed the results to gain further insights.

In the convergent positive information condition, a 2 × 2 ANOVA showed a significant interaction between information source and service type, F(1, 62) = 11.56, p < .001.

Specifically, for the hair salon, participants had higher confidence ratings when exposed to peer reviews than to expert reviews (M_PR = 6.03 vs. M_ER = 5.26), F(1, 31) = 4.88, p < .05, η² = .14. In contrast, for the insurance agency, participants had higher confidence ratings when exposed to expert reviews than to peer reviews (M_PR = 5.09 vs. M_ER = 5.91), F(1, 31) = 6.92, p < .05, η² = .18.

In the mixed information condition, a 2 × 2 ANOVA showed only the significant main effect of information source, F(1, 52) = 12.18, p < .001, η² = .19, with no interaction between information source and service type. For both services, participants were less confident in their evaluations when exposed to mixed expert reviews than to mixed peer reviews—hair salon: M_PR = 4.36 vs. M_ER = 3.60, F(1, 25) = 6.55, p < .05, η² = .21; insurance: M_PR = 4.21 vs. M_ER = 3.79; F(1, 27) = 5.76, p < .05, η² = .18. These findings implied that mixed expert reviews had stronger negative effects than mixed peer reviews on participants’ confidence for both experience and credence services.

Confidence as a Mediator

We conducted separate mediation analyses for the convergent and mixed information conditions to test the mediating effect of confidence. In the convergent information condition, using information source, service type, and Information Source × Service Type interaction terms as independent variables and confidence rating as a mediator, results showed that the Information Source × Service Type interaction was diminished after controlling for the mediator (Figure 4a). Specifically, a 95% bootstrap CI for the indirect effect (PROCESS Model 8, Hayes 2013) did not include zero (95% CI: [0.21, 1.11]), confirming that confidence rating partially mediated the interaction effects between information source and service type on consumers’ service evaluations, supporting Hypothesis 3.

Figure 4.
Mediated moderation analysis (Study 2). (a) Convergent information p < .001. p < .01. p < .05. (b) Mixed information. p < .001. p < .01. p < .05.

In the mixed information condition, as there was only the main effect of information source, the mediation analysis showed that the indirect effect was significantly different from zero (95% CI: [−.28, −.01]), supporting the mediating role of confidence rating (Figure 4b).

Discussion

Using multiple reviews, Study 2 provided further support for the mediating role of confidence in the interaction effects of Information Source × Service Type on service evaluations, consistent with Study 1. Importantly, Study 2 revealed the moderating effect of information convergence. In the convergent positive information condition, the findings confirmed results in Study 1. However, in the mixed information condition, mixed expert reviews had a more negative impact than mixed peer reviews on consumers’ service evaluations and confidence for both experience and credence services.

Nonetheless, in Study 2, the convergent or mixed reviews were from similar information sources (i.e., either peer reviews or expert reviews). A more intriguing situation would be when the mixed reviews were from different information sources (i.e., mixed peer and expert reviews) as commonly seen on review websites (e.g., http://rottentomatoes.com).

Study 3: Mixed Information From Similar Versus Different Sources

The goal of Study 3 was to further examine the moderating effects of mixed information from both similar (i.e., mixed peer reviews or mixed expert reviews) and different information sources (i.e., mixed peer and expert reviews).

Method

Design and Sample

Study 3 used a 4 (information source: mixed peer reviews vs. mixed expert reviews vs. positive peer review and negative expert review [PPR&NER] vs. positive expert review and negative peer review [PER&NPR]) × 2 (service type: experience vs. credence services) between-participants design. Participants were recruited on Amazon’s M-Turk (N = 225, 46.7% females, 56.9% between ages 20 and 39).

Materials and Procedure

Participants were randomly assigned to one of the eight conditions. They first read a service consumption scenario (i.e., restaurant or language institute, representing experience and credence services, respectively) for which they needed to search for additional information as detailed in the Appendix. In each service situation, the two mixed reviews were manipulated as coming from either similar sources (i.e., two mixed peer reviews or two mixed expert reviews) or different sources (i.e., PPR&NER or PER&NPR).

After reading the scenario, participants responded to questions measuring their expected service quality, purchase likelihood, confidence in service evaluation, emotional difficulty, source expertise, involvement, purchase importance, and perceived homophily (see Table 1).

Results

Manipulation Check

As mixed information could come from both similar and different information sources, we conducted separate data analyses to gain clear insights. For mixed information from similar sources, a 2 (information source: mixed peer reviews vs. mixed expert reviews) × 2 (service type) ANOVA on source expertise revealed only the main effect of information source, F(1, 107) = 8.75, p < .01, η² = .08. Specifically, participants perceived the expert reviews to be more knowledgeable than the peer reviews—restaurant: M_PR = 4.26 vs. M_ER = 4.80, F(1, 55) = 4.60, p < .05, η² = .08; language institute: M_PR = 4.61 vs. M_ER = 5.03, F(1, 52) = 4.36, p < .05, η² = .08. For mixed information from different sources, a 2 (information source: PPR&NER vs. PER&NPR) × 2 (service type) repeated-measures ANOVA on source expertise revealed only the main effect of information source, F(1, 112) = 27.17, p < .01, η² = .20, such that participants perceived the expert review to be more knowledgeable than the peer review (restaurant: M_PR = 4.26 vs. M_ER = 5.17; F(1, 56) = 10.50, p < .01, η² = .16; language institute: M_PR = 4.32 vs. M_ER = 5.25; F(1, 56) = 18.95, p < .001, η² = .25).

We conducted the same analyses for perceived homophily. For mixed information from similar sources, a 2 × 2 ANOVA showed only the significant main effect of information source, F(1, 107) = 8.06, p < .01, η² = .07, such that participants perceived higher similarity for the peer reviews than for the expert reviews—restaurant: M_PR = 4.61 vs. M_ER = 3.98, F(1, 55) = 4.57, p < .05, η² = .08; language institute: M_PR = 4.17 vs. M_ER = 3.84; F(1, 52) = 4.64, p < .05, η² = .08. For mixed information from different sources, a 2 × 2 repeated-measures ANOVA revealed only the main effect of information source, F(1, 112) = 8.67, p < .01, η² = .07; participants perceived higher similarity for the peer review than for the expert review—restaurant: M_PR = 4.27 vs. M_ER = 3.46, F(1, 56) = 4.10, p < .01, η² = .07; language institute: M_PR = 4.31 vs. M_ER = 3.51, F(1, 56) = 4.61, p < .05, η² = .08. These results showed that the manipulations of information source were successful.

In addition, a 4 (information source) × 2 (service type) ANOVA on involvement and purchase importance revealed no significant effects (all Fs < 2.00, all p values > .10). Similarly, a 4 × 2 ANOVA indicated that participants’ perceived emotional difficulty (r _restaurant = .78, r _language = .81) did not significantly differ between the restaurant and language institute scenarios (M _restaurant = 3.23 vs. M _language = 3.61), F(1, 217) = 2.68, p > .10, η² = .01.

Expected Service Quality

A 4 × 2 ANOVA on expected service quality (r _restaurant = .82, r _language = .77) revealed the predicted main effect of information source, F(3, 217) = 10.01, p < .001, η² = .11. We then decomposed the data into mixed information from similar and different sources to gain additional insights. For mixed information from similar sources, a 2 (information source: mixed peer reviews vs. mixed expert reviews) × 2 (service type) ANOVA showed the significant main effect of information source, F(1, 107) = 19.67, p < .001, η² = .16 (Figure 5a), replicating the findings in Study 2. For both types of services, participants’ expected service quality was lower when exposed to mixed expert reviews than to mixed peer reviews—restaurant: M_PR = 4.97 vs. M_ER = 4.24, F(1, 55) = 9.55, p < .01, η² = .15; language institute: M_PR = 4.74 vs. M_ER = 3.98; F(1, 52) = 10.14, p < .01, η² = .16.

Figure 5.
Moderating effects of information divergence (Study 3). (a) Mixed information from similar sources, (b) mixed information from different sources.

For mixed information from different sources, a 2 (information source: PPR&NER vs. PER&NPR) × 2 (service type) ANOVA also showed the significant main effect of information source, F(1, 110) = 12.07, p < .001, η² = .10 (Figure 5b). Consumers’ expected service quality was lower when the negative information came from the expert review (i.e., PPR&NER) than when it came from the peer review (i.e., PER&NPR)—restaurant: MPER&NPR = 4.97 versus MPPR&NER = 4.15, F(1, 55) = 5.83, p < .02, η² = .10; language institute: MPER&NPR = 5.11 versus MPPR&NER = 4.29, F(1, 55) = 6.26, p < .05, η² = .10. That is, for mixed information from different sources, participants’ expected service quality was lowered more by the negative expert review than the negative peer review for both experience and credence services, which supported Hypothesis 4b.

Purchase Likelihood

Similarly, a 4 × 2 ANOVA on purchase likelihood (r _restaurant = .87, r _language = .86) revealed the main effect of information source, F(3, 217) = 6.98, p < .001, η² = .09. For mixed information from similar sources, a 2 (information source: mixed peer reviews vs. mixed expert reviews) × 2 (service type) ANOVA showed the main effect of information source, F(1, 107) = 11.42, p < .001, η² = .10. For both types of services, participants showed lower purchase likelihood for the service when exposed to mixed expert reviews than to mixed peer reviews—restaurant: M_PR = 4.87 versus M_ER = 3.81, F(1, 55) = 7.68, p < .01, η² = .12; language institute: M_PR = 4.17 versus M_ER = 3.48, F (1, 52) = 3.96, p = .05, η² = .07.

For mixed information from different sources, a 2 (information source: PPR&NER vs. PER&NPR) × 2 (service type) ANOVA also showed the significant main effect of information source, F (1, 110) = 7.62, p < .01, η² = .07. Participants’ purchase likelihood was lower when the negative information came from the expert review (i.e., PPR&NER) than when it came from the peer review (i.e., PER&NPR)—restaurant: MPER&NPR = 4.77 versus MPPR&NER = 3.94, F (1, 55) = 3.96, p = .05, η² = .07; language institute: MPER&NPR = 4.86 versus MPPR&NER = 4.03, F (1, 55) = 5.18, p < .05, η² = .09. That is, participants’ purchase likelihood was lowered more by the negative expert review than the negative peer review for both experience and credence services, further supporting Hypothesis 4b.

Confidence

A 4 × 2 ANOVA on confidence (r _restaurant = .87, r _language = .89) revealed the main effect of information source, F(3, 217) = 10.71, p < .001, η² = .13. We then decomposed the data into mixed information from similar and different sources to gain additional insights. For mixed information from similar sources, a 2 (information source: mixed peer reviews vs. mixed expert reviews) × 2 (service type) ANOVA showed the significant main effect of information source, F(1, 107) = 9.82, p < .01, η² = .08. For both types of services, participants had lower confidence ratings when exposed to mixed expert reviews than to mixed peer reviews—restaurant: M_PR = 4.53 vs. M_ER = 3.46, F(1, 55) = 5.60, p < .05, η² = .09; language institute: M_PR = 3.70 vs. M_ER = 2.83, F (1, 52) = 4.28, p < .05, η² = .08.

For mixed information from different sources, a 2 × 2 ANOVA showed only the significant main effect of information source, F(1, 110) = 12.61, p < .001, η² = .10. Specifically, participants had lower confidence ratings when the negative information came from the expert review (i.e., PPR&NER) than when it came from the peer review (i.e., PER&NPR)—restaurant: MPER&NPR = 4.90 vs. MPPR&NER = 3.81, F (1, 55) = 7.01, p < .05, η² = .11; language institute: MPER&NPR = 4.68 vs. MPPR&NER = 3.78, F(1, 55) = 5.61, p < .05, η² = .09. Thus, participants’ confidence was lowered more by the negative expert review than by the negative peer review for both experience and credence services.

Confidence as a Mediator

We then conducted separate mediation analyses for the two mixed information conditions using the bootstrapping approach (PROCESS Model 4; Hayes 2013). With expected service quality as the dependent variable, for mixed information from similar sources, results showed while the direct effect of information source remained significant (−.52, p < .01), the indirect effect of information source through confidence was negative and significant (−.22, 95% CI: [−.43, −.09]; Figure 6a). For mixed information from different sources, results showed that the direct effect of information source remained marginally significant (.40, p = .07), while the indirect effect of information source through confidence was positive and significant (.41, 95% CI: [.19, .70]; Figure 6b). Thus, the effects of information source on expected service quality were partially mediated by confidence, supporting Hypothesis 3.

Figure 6.
Mediated moderation analysis (Study 3). (a) Mixed information from similar sources p < .001. p < .01. (b) Mixed information from different sources ***p < .001. ^⁁ p < .10.

With purchase likelihood as the dependent variable, for mixed information from similar sources, results revealed a significant direct effect of information source (−.50, p < .05) as well as a significant indirect effect of information source through confidence (−.39, 95% CI: [−.72, −.15]). Similarly, for mixed information from different sources, results revealed a nonsignificant direct effect of information source (.36, p > .10) as well as a significant indirect effect of information source through confidence (.46, 95% CI: [.19, .88]). These results supported the mediating role of confidence in the effects of information source on purchase likelihood.

General Discussion

The present research examines the differential effects of peer review and expert review on consumers’ service evaluations. We show that consumers’ evaluations of experience services are higher when they see a peer review (vs. an expert review), while their evaluations of credence services are higher when they see an expert review (vs. a peer review; Study 1). These differential effects can be explained by consumers’ confidence in evaluating services (Studies 1–3). Importantly, we show that the mediating effect of confidence is moderated by information convergence. That is, when there are multiple reviews, convergent positive reviews substantiate the effects found in Study 1. However, mixed reviews lower consumers’ confidence, which reduce their service evaluations (Study 2). In particular, for mixed reviews from different sources, the negative expert review exerts stronger unfavorable effects than the negative peer review on consumers’ confidence and outcome variables (i.e., expected service quality and purchase likelihood) for both types of services (Study 3). These findings have significant theoretical and practical implications.

Theoretical Contributions

The present research builds on and extends the literature on information processing in the services domain. Our review of the literature indicates that prior research tends to separately investigate the effects of peer review (e.g., Chevalier and Mayzlin 2006; Duhan et al. 1997; Huang and Chen 2006; Lee and Bradlow 2011; Liu 2006; Sotiriadis and van Zyl 2013; Zhu and Zhang 2010) and expert review (e.g., Basuroy, Chatterjee, and Ravid 2003; Eliashberg and Shugan 1997) on consumer responses or market performance. Thus, existing research does not unequivocally explain the factors determining consumers’ adherence to peer versus expert reviews.

To this end, we examine the differential effects of both information sources on consumers’ evaluations of experience and credence services. In so doing, we also answer the call by Libai et al. (2010) for research on WOM in customer-to-customer interactions by contrasting peer-to-peer against expert-to-consumer referrals. In particular, for complex and professional services embedded with credence attributes (e.g., financial planning and health care), consumer adherence to the expert service provider’s advice is an important research area (Mitra, Reiss, and Capella 1999; Racherla and Friske 2012; Seiders et al. 2015).

We show that consumers’ situational feeling of confidence underlies the differential effects of peer review and expert review on their evaluations of experience and credence services. Specifically, credence services with high uncertainty lead consumers to rely more on expert reviews (vs. peer reviews), while consumers are more confident evaluating experience services based on peer reviews. This finding extends past research linking low confidence with increased search for information (Chaiken and Maheswaran 1994; Trope and Chaiken 1999).

Importantly, we reveal that these effects are moderated by information convergence, which is new to the literature. That is, when the reviews are positively convergent, consumers’ evaluations of experience (credence) services are higher when they see peer (expert) reviews. However, when the reviews are mixed, consumers’ confidence and evaluations of both experience and credence services are lowered more by the negative expert review than by the negative peer review.

Taken together, findings from the three studies are consistent for a range of experience and credence services, different outcome variables (e.g., service evaluations, expected service quality, and purchase likelihood) as well as across both U.S. and Chinese samples.

Managerial Implications

Beyond the theoretical contributions, the present research also offers considerable managerial implications. It is common for consumers to seek opinions and comments from other consumers as well as experts on various issues, including product and service recommendations online (Karmarkar and Tormala 2010). In this vein, some service firms incorporate reviews from ordinary consumers and experts in their advertising and marketing communications. At the same time, some marketers attempt to influence consumer decision-making using a personalized recommendation system (Yuan et al. 2013). Thus, it is important to understand the psychological mechanisms guiding consumers’ use of these recommendations in deciding whose opinions to trust and use (Yaniv, Choshen-Hillel, and Milyavsky 2011).

For experience services such as movies, we show that peer review exerts greater influence than expert review on consumer evaluations. This finding explains prior empirical research, whereby peer review has been shown to exert strong effects on movie takings (Liu 2006; Moul 2007). Specifically, Moul (2007) finds that approximately 10% of the variation in consumer expectations of movies can be attributed to peer review, which affects consumer behavior quickly. In comparison, expert reviews on movies have only a weak influence on box office revenue (Reinstein and Snyder 2005). Thus, marketers of experience services should pay greater attention to reviews by ordinary consumer comments.

In contrast, we find that consumers’ evaluations of credence services are influenced more by expert review. By nature, credence services are difficult to evaluate even after consumption, leading consumers to rely more on expert opinion. This explains why credence service firms often seek professional third-party reviews for endorsement or accreditation (e.g., financial institutions are rated by Moody’s and Standard & Poor’s,³ and websites are certified by TRUSTe). Another common practice is to “soft sell” by holding information sessions conducted by experts to explain credence services (e.g., complex financial services) in lay language to potential customers. Nonetheless, we extend a caveat to this suggestion; the emphasis on expert review for credence services should not be at the total exclusion of peer review. For example, there is increasing recognition in health care that patients do not only rely on their physicians’ advice but also frequently turn to peer review for recommendations (Kenagy, Berwick, and Shore 1999).

Our findings also have implications for online marketers. In particular, not all social networks are equally effective at harnessing potential peer-to-peer referrals. For instance, the network of friends is more suited to the effective diffusion of experience services, while the network of professionals and colleagues is more suited to credence services. To encourage positive WOM, experience service firms such as Lyft (2017) initiate programs that reward ordinary consumers who recommend the service to others. Similarly, credence service companies (e.g., real estate websites) can establish an expert recommendation system to consumers (Yuan et al. 2013). Importantly, marketers should take steps to boost consumers’ confidence in their service evaluations, such as providing greater transparency and assurance on the objectivity of the peer review and expert review.

Future Research

Our review of the literature indicates a lack of conclusive evidence on the effects of review valence on consumer behavior (Kimmel and Kitchen 2014). While some studies suggest that online reviews affect purchase intention in the same direction as their valence (e.g., Ketelaar et al. 2015), others find no effects for review valence (e.g., Cheung et al. 2009; Liu 2006). Thus, there is scope for further research on the differential effects and underlying mechanisms for positive and negative information (Wangenheim 2005).

In addition, while the present research operationalizes peer review as coming from novice consumers, we recognize that ordinary consumers can be quite knowledgeable (Bettman and Sujan 1987). Thus, it would be interesting to include “expert peer review” as another information source and contrast its effects against those of novice peer review and professional expert review on consumers’ evaluations.

Furthermore, our studies controlled for the amount of information presented to the participants. In reality, the quantity of information by source may differ significantly. For many services, there are far fewer expert reviews than peer reviews. For example, the movie “Hail Caesar” was rated rather favorably, 7.1/10 (83%), by 239 professional critics, while it was rated much lower, 2.9/5 (45%), by 30,453 regular moviegoers (Rottentomatoes 2016). Can the weight of far fewer positive expert reviews be offset by the sheer volume of negative peer reviews? While recent research has explored interactions between volume and valence of online ratings on consumer perceptions of risk and purchase intention for movies (Keh et al. 2015), more research is needed to investigate how consumers react to changes in information quantity, quality, and sources for experience and credence services.

Reliability Coefficients
Service evaluations	Please evaluate the service on the following measures: satisfactory-satisfactory, unfavorable-favorable, dislike-like (7-point scales)	α _hotel = .93, α _dental = .91	α _{hair salon} = .96, α _insurance = .96
Expected service quality	Based on the scenario, how would you expect the service quality of the service to be? (1 = very low, 7 = very high)Overall, what are your expectations of the service quality of this service? (1 = very poor, 7 = outstanding)”			r_restaurant = .82, r_language = .77
Purchase likelihood	How likely are you to patronize the service firm? (1 = very unlikely, 7 = very likely)How likely are you to purchase the service from this firm? (1 = very unlikely, 7 = very likely)			R _restaurant = .87, r _language = .86
Confidence in service evaluation	Please indicate how confident you feel about your decision given the review you read. (1 = not at all confident, 7 = extremely confident)Please indicate how certain you feel about your decision given the review you read. (1 = not at all certain, 7 = extremely certain)	r_hotel = .77, r_dental = .79	r_{hair salon} = .86, r_insurance = .63	r_restaurant = .87, r_language = .89
Source expertise (Study 1)	How knowledgeable do you think the expert/consumer is about the service firm? (1 = not at all knowledgeable, 7 = very knowledgeable)How expert do you think the expert/consumer is about the service firm? (1 = not at all expert, 7 = highly expert)	r_hotel = .77, r_dental = .84
Source expertise (Studies 2 and 3)	How knowledgeable do you think peer/expert reviewer A is about the service firm? (1 = not at all knowledgeable, 7 = very knowledgeable)How expert do you think peer/expert reviewer A is about the service firm? (1 = not at all expert, 7 = highly expert)How knowledgeable do you think peer/expert reviewer B is about the service firm? (1 = not at all knowledgeable, 7 = very knowledgeable)How expert do you think peer/expert reviewer B is about the service firm? (1 = not at all expert, 7 = highly expert)		α _{hair salon} = .87, α _insurance = .70	α _restaurant = .60, α _language = .60
Involvement	How involved were you in processing the information? not at all involved-very involved, skimmed it quickly-read it carefully (7-point scales)	r_hotel = .79, r_dental = .84	r_{hair salon} = .70, r_insurance = .67	r_restaurant = .60, r_language = .76
Purchase importance	How critical do you think is the purchase decision? (1 = very low purchase criticality, 7 = very high purchase criticality)
Perceived homophily (Study 1)	In general, how similar is the ordinary consumer/expert to yourself? (1 = not at all similar, 7 =very similar)In general, how similarly does the ordinary consumer/expert behave like yourself? (1 = not at all similar, 7 = very similar)	r_hotel = .82, r_dental = .78
Perceived homophily (Studies 2 and 3)	In general, how similar is Peer/Expert Reviewer A to yourself? (1 = not at all similar, 7 = very similar)In general, how similarly does Peer/Expert Reviewer A behave like yourself? (1 = not at all similar, 7 = very similar)In general, how similar is Peer/Expert Reviewer B to yourself? (1 = not at all similar, 7 = very similar)In general, how similarly does Peer/Expert Reviewer B behave like yourself? (1 = not at all similar, 7 = very similar)		α _{hair salon} = .72, α _insurance = .86	α _restaurant = .60, α _language = 60
Emotional difficulty	I would feel the decision described in the scenario is an emotionally difficult one to make (1 = not at all, 7 = extremely so)I would feel the decision described in the scenario is stressful. (1 = not at all, 7 = extremely so)	r_hotel = .60, r_dental = .61	r_{hair salon} = .60, r_insurance = .77	r_restaurant = .78, r_language = .81

Footnotes

Appendix

Authors’ Note

The authors are listed alphabetically. We are grateful to Wenbo Ji for his assistance on an early version of this article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by grants from the National Natural Science Foundation of China awarded to Jin Sun (Grant Nos. 71372004, 71772040) and also supported by the Program for Excellent Talents in UIBE (Grant No. 17JQ03).

Notes

References

Bansal

Harvir S.

Voyer

Peter A.

(2000), “Word-of-Mouth Processes within a Services Purchase Decision Context,” Journal of Service Research, 3 (2), 166–177.

Basuroy

Suman

Chatterjee

Subimal

Abraham

Ravid S.

(2003), “How Critical Are Critical Reviews? The Box Office Effects of Film Critics, Star Power, and Budgets,” Journal of Marketing, 67 (4), 103–117.

Bettman

James R.

Sujan

Mita

(1987), “Effects of Framing on Evaluation of Comparable and Noncomparable Alternatives by Expert and Novice Consumers,” Journal of Consumer Research, 14 (2), 141–154.

Brown

Jacqueline Johnson

Reingen

Peter H.

(1987), “Social Ties and Word-of-Mouth Referral Behavior,” Journal of Consumer Research, 14 (3), 350–362.

Carter

Travis J.

Gilovich

Thomas

(2010), “The Relative Relativity of Material and Experiential Purchases,” Journal of Personality and Social Psychology, 98 (1), 146–159.

Chaiken

Shelly

Maheswaran

Durairaj

(1994), “Heuristic Processing Can Bias Systematic Processing: Effects of Source Credibility, Argument Ambiguity, and Task Importance on Attitude Judgment,” Journal of Personality and Social Psychology, 66 (3), 460–473.

Chakravarty

Anindita

Liu

Yong

Mazumdar

Tridib

(2010), “The Differential Effects of Online Word-of-Mouth and Critics’ Reviews on Pre-release Movie Evaluation,” Journal of Interactive Marketing, 24 (3), 185–197.

Chaxel

Anne-Sophie

(2016), “Why, When, and How Personal Control Impacts Information Processing: A Framework,” Journal of Consumer Research, 43 (1), 179–197.

Chen

Yubo

Xie

Jinhong

(2008), “Online Consumer Review: Word-of-Mouth as a New Element of Marketing Communication Mix,” Management Science, 54 (3), 477–491.

10.

Cheung

Man Yee

Luo

Chuan

Sia

Choon Ling

Chen

Huaping

(2009), “Credibility of Electronic Word-of-Mouth: Informational and Normative Determinants of Online Consumer Recommendations,” International Journal of Electronic Commerce, 13 (4), 9–38.

11.

Chevalier

Judith A.

Mayzlin

Dina

(2006), “The Effect of Word of Mouth on Sales: Online Book Review,” Journal of Marketing Research, 43 (3), 345–354.

12.

Darby

Michael R.

Karni

Edi

(1973), “Free Competition and the Optimal Amount of Fraud,” Journal of Law and Economics, 16 (1), 67–88.

13.

De Bruyn

Arnaud

Lilien

Gary L.

(2008), “A Multi-Stage Model of Word-of-Mouth Influence through Viral Marketing,” International Journal of Research in Marketing, 25 (3), 151–163.

14.

Duhan

Dale F.

Johnson

Scott D.

Wilcox

James B.

Harrell

Gilbert D.

(1997), “Influences on Consumer Use of Word-of-Mouth Recommendation Sources,” Journal of the Academy of Marketing Science, 25 (4), 283–295.

15.

Eliashberg

Jehoshua

Shugan

Steven M.

(1997), “Film Critics: Influencers or Predictors?” Journal of Marketing, 61 (2), 68–79.

16.

Erber

Ralph

Fiske

Susan T.

(1984), “Outcome Dependency and Attention to Inconsistent Information,” Journal of Personality and Social Psychology, 47 (4), 709–726.

17.

Faraji-Rad

Ali

Samuelsen

Bendik M.

Warlop

Luk

(2015), “On the Persuasiveness of Similar Others: The Role of Mentalizing and the Feeling of Certainty,” Journal of Consumer Research, 42 (3), 458–471.

18.

Gilly

Mary C.

Graham

John L.

Wolfinbarger

Mary Finley

Yale

Laura J.

(1998), “A Dyadic Study of Interpersonal Information Search,” Journal of the Academy of Marketing Science, 26 (2), 83–100.

19.

Hayes

Andrew F.

(2013), Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach. New York, NY: Guilford.

20.

Holbrook

Morris B.

Addis

Michela

(2007), “Taste versus the Market: An Extension of Research on the Consumption of Popular Culture,” Journal of Consumer Research, 34 (3), 415–424.

21.

Huang

Jen-Hung

Chen

Yi-Fen

(2006), “Herding in Online Product Choice,” Psychology & Marketing, 23 (5), 413–428.

22.

Karmarkar

Uma R.

Tormala

Zakary L.

(2010), “Believe Me, I Have No Idea What I’m Talking About: The Effects of Source Certainty on Consumer Involvement and Persuasion,” Journal of Consumer Research, 36 (6), 1033–1049.

23.

Keh

Hean Tat

Pang

Jun

(2010), “Customer Reactions to Service Separation,” Journal of Marketing, 74 (2), 55–70.

24.

Keh

Hean Tat

Wenbo

Wang

Xia

Sy-Changco

Joseph

Singh

Ramendra

(2015), “Online Movie Ratings: A Cross-Cultural, Emerging Asian Markets Perspective,” International Marketing Review, 32 (3/4), 366–388.

25.

Kenagy

John W.

Berwick

Donald M.

Shore

Miles F.

(1999), “Service Quality in Health Care,” Journal of the American Medical Association, 281 (7), 661–665.

26.

Ketelaar

Paul E.

Willemsen

Lotte M.

Sleven

Laura

Kerkhof

Peter

(2015), “The Good, the Bad, and the Expert: How Consumer Expertise Affects Review Valence Effects on Purchase Intentions in Online Product Reviews,” Journal of Computer-Mediated Communication, 20 (6), 649–666.

27.

Kimmel

Allan J.

Kitchen

Philip J.

(2014), “WOM and Social Media: Presaging Future Directions for Research and Practice,” Journal of Marketing Communications, 20 (1), 5–20.

28.

Lee

Thomas Y.

Bradlow

Eric T.

(2011), “Automated Marketing Research Using Online Customer Reviews,” Journal of Marketing Research, 18 (5), 881–894.

29.

Libai

Barak

Bolton

Ruth

Bügel

Marnix S.

De Ruyter

Götz

Oliver

Risselada

Hans

Stephen

Andrew T.

(2010), “Customer-to-Customer Interactions: Broadening the Scope of Word of Mouth Research,” Journal of Service Research, 13 (3), 267–282.

30.

Liu

Yong

(2006), “Word of Mouth for Movies: Its Dynamics and Impact on Box Office Revenue,” Journal of Marketing, 70 (3), 74–89.

31.

Lyft (2017), “Refer New Passengers and Get Rewarded.” https://help.lyft.com/hc/en-us/articles/214216977-Refer-New-Passengers-and-Get-Rewarded (accessed June 3, 2017).

32.

Mattila

Anna S.

Wirtz

Jochen

(2002), “The Impact of Knowledge Types on the Consumer Search Process: An Investigation in the Context of Credence Services,” International Journal of Service Industry Management, 13 (3), 214–230.

33.

McPherson

Miller

Smith-Lovin

Lynn

Cook

James M.

(2001), “Birds of a Feather: Homophily in Social Networks,” Annual Review of Sociology, 27 (1), 415–444.

34.

Mitra

Kaushik

Reiss

Michelle C.

Capella

Louis M.

(1999), “An Examination of Perceived Risk, Information Search and Behavioral Intentions in Search, Experience and Credence Services,” Journal of Services Marketing, 13 (3), 208–232.

35.

Moul

Charles C.

(2007), “Measuring Word of Mouth’s Impact on Theatrical Movie Admissions,” Journal of Economics & Management Strategy, 16 (4), 859–892.

36.

Murray

Keith B.

Schlacter

John L.

(1990), “The Impact of Services versus Goods on Consumers’ Assessment of Perceived Risk and Variability,” Journal of the Academy of Marketing Science, 18 (1), 51–65.

37.

Ostrom

Amy

Iacobucci

Dawn

(1995), “Consumer Trade-Offs and the Evaluation of Services,” Journal of Marketing, 59 (1), 17–28.

38.

Palmeira

Mauricio

Spassova

Gerri

Keh

Hean Tat

(2015), “Other-Serving Bias in Advice- Taking: When Advisors Receive More Credit than Blame,” Organizational Behavior and Human Decision Processes, 130 (C), 13–25.

39.

Price

Paul C.

Stone

Eric R.

(2004), “Intuitive Evaluation of Likelihood Judgment Producers: Evidence for a Confidence Heuristic,” Journal of Behavioral Decision Making, 17 (1), 39–57.

40.

Racherla

Pradeep

Friske

Wesley

(2012), “Perceived ‘Usefulness’ of Online Consumer Reviews: An Exploratory Investigation across Three Services Categories,” Electronic Commerce Research & Applications, 11 (6), 548–559.

41.

Reinstein

David A.

Snyder

Christopher M.

(2005), “The Influence of Expert Reviews on Consumer Demand for Experience Goods: A Case Study of Movie Critics,” Journal of Industrial Economics, 53 (1), 27–51.

42.

Rottentomatoes (2016), “Hail, Caesar!” http://www.rottentomatoes.com/m/hail_caesar_2016/ (accessed March 8, 2016).

43.

Seiders

Kathleen

Flynn

Andrea Godfrey

Berry

Leonard L.

Haws

Kelly L.

(2015), “Motivating Customers to Adhere to Expert Advice in Professional Services: A Medical Service Context,” Journal of Service Research, 18 (1), 39–58.

44.

Senecal

Sylvain

Nantel

Jacques

(2004), “The Influence of Online Product Recommendations on Consumers’ Online Choices,” Journal of Retailing, 80 (2), 159–169.

45.

Smith

Donnavieve

Menon

Satya

Sivakumar

(2005), “Online Peer and Editorial Recommendations, Trust, and Choice in Virtual Markets,” Journal of Interactive Marketing, 19 (3), 15–37.

46.

Sotiriadis

Marios D.

van Zyl

Cinà

(2013), “Electronic Word-of-Mouth and Online Reviews in Tourism Services: The Use of Twitter by Tourists,” Electronic Commerce Research, 13 (1), 103–124.

47.

Spence

Mark T.

Brucks

Merrie

(1997), “The Moderating Effects of Problem Characteristics on Experts’ and Novices’ Judgments,” Journal of Marketing Research, 34 (2), 233–247.

48.

Sun

Jin

Keh

Hean Tat

Lee

Angela Y.

(2012), “The Effect of Attribute Alignability on Service Evaluation: The Moderating Role of Uncertainty,” Journal of Consumer Research, 39 (4), 831–847.

49.

Thomas

Manoj

Menon

Geeta

(2007), “When Internal Reference Prices and Price Expectations Diverge: The Role of Confidence,” Journal of Marketing Research, 44 (3), 401–409.

50.

Trope

Yaacov

Chaiken

Shelly

(1999), Dual-Process Theories in Social Psychology. New York, NY: Guilford.

51.

Tsai

Claire I.

McGill

Ann L.

(2011), “No Pain, No Gain? How Fluency and Construal Level Affect Consumer Confidence,” Journal of Consumer Research, 37 (5), 807–821.

52.

Wang

Jing

Lee

Angela Y.

(2006), “The Role of Regulatory Focus in Preference Construction,” Journal of Marketing Research, 43 (1), 28–38.

53.

Wangenheim

Florian V.

(2005), “Postswitching Negative Word of Mouth,” Journal of Service Research, 8 (1), 67–78.

54.

Westbrook

Robert A.

(1987), “Product/Consumption-Based Affective Responses and Postpurchase Processes,” Journal of Marketing Research, 24 (3), 258–270.

55.

White

Tiffany Barnett

(2005), “Consumer Trust and Advice Acceptance: The Moderating Roles of Benevolence, Expertise, and Negative Emotions,” Journal of Consumer Psychology, 15 (2), 141–148.

56.

Yaniv

Ilan

Choshen-Hillel

Shoham

Milyavsky

Maxim

(2011), “Receiving Advice on Matters of Taste: Similarity, Majority Influence, and Taste Discrimination,” Organizational Behavior and Human Decision Processing, 115 (1), 111–120.

57.

Yuan

Xiaofang

Lee

Ji-Hyun

Kim

Sun-Joong

Kim

Yoon-Hyun

(2013), “Toward a User-Oriented Recommendation System for Real Estate Websites,” Information Systems, 38 (2), 231–243.

58.

Zeithaml

Valarie Parasuraman

Parsu

Berry

Leonard

. (1985). Problems and Strategies in Service Marketing. Journal of Marketing, 49 (2), 33–46.

59.

Zhu

Feng

Zhang

Xiaoquan

(2010), “Impact of Online Consumer Reviews on Sales: The Moderating Role of Product and Consumer Characteristics,” Journal of Marketing, 74 (2), 133–148.

Reliability Coefficients
Variables	Measures	Study 1	Study 2	Study 3
Service evaluations	Please evaluate the service on the following measures: satisfactory-satisfactory, unfavorable-favorable, dislike-like (7-point scales)	α _hotel = .93, α _dental = .91	α _{hair salon} = .96, α _insurance = .96
Expected service quality	Based on the scenario, how would you expect the service quality of the service to be? (1 = very low, 7 = very high)Overall, what are your expectations of the service quality of this service? (1 = very poor, 7 = outstanding)”			r_restaurant = .82, r_language = .77
Purchase likelihood	How likely are you to patronize the service firm? (1 = very unlikely, 7 = very likely)How likely are you to purchase the service from this firm? (1 = very unlikely, 7 = very likely)			R _restaurant = .87, r _language = .86
Confidence in service evaluation	Please indicate how confident you feel about your decision given the review you read. (1 = not at all confident, 7 = extremely confident)Please indicate how certain you feel about your decision given the review you read. (1 = not at all certain, 7 = extremely certain)	r_hotel = .77, r_dental = .79	r_{hair salon} = .86, r_insurance = .63	r_restaurant = .87, r_language = .89
Source expertise (Study 1)	How knowledgeable do you think the expert/consumer is about the service firm? (1 = not at all knowledgeable, 7 = very knowledgeable)How expert do you think the expert/consumer is about the service firm? (1 = not at all expert, 7 = highly expert)	r_hotel = .77, r_dental = .84
Source expertise (Studies 2 and 3)	How knowledgeable do you think peer/expert reviewer A is about the service firm? (1 = not at all knowledgeable, 7 = very knowledgeable)How expert do you think peer/expert reviewer A is about the service firm? (1 = not at all expert, 7 = highly expert)How knowledgeable do you think peer/expert reviewer B is about the service firm? (1 = not at all knowledgeable, 7 = very knowledgeable)How expert do you think peer/expert reviewer B is about the service firm? (1 = not at all expert, 7 = highly expert)		α _{hair salon} = .87, α _insurance = .70	α _restaurant = .60, α _language = .60
Involvement	How involved were you in processing the information? not at all involved-very involved, skimmed it quickly-read it carefully (7-point scales)	r_hotel = .79, r_dental = .84	r_{hair salon} = .70, r_insurance = .67	r_restaurant = .60, r_language = .76
Purchase importance	How critical do you think is the purchase decision? (1 = very low purchase criticality, 7 = very high purchase criticality)
Perceived homophily (Study 1)	In general, how similar is the ordinary consumer/expert to yourself? (1 = not at all similar, 7 =very similar)In general, how similarly does the ordinary consumer/expert behave like yourself? (1 = not at all similar, 7 = very similar)	r_hotel = .82, r_dental = .78
Perceived homophily (Studies 2 and 3)	In general, how similar is Peer/Expert Reviewer A to yourself? (1 = not at all similar, 7 = very similar)In general, how similarly does Peer/Expert Reviewer A behave like yourself? (1 = not at all similar, 7 = very similar)In general, how similar is Peer/Expert Reviewer B to yourself? (1 = not at all similar, 7 = very similar)In general, how similarly does Peer/Expert Reviewer B behave like yourself? (1 = not at all similar, 7 = very similar)		α _{hair salon} = .72, α _insurance = .86	α _restaurant = .60, α _language = 60
Emotional difficulty	I would feel the decision described in the scenario is an emotionally difficult one to make (1 = not at all, 7 = extremely so)I would feel the decision described in the scenario is stressful. (1 = not at all, 7 = extremely so)	r_hotel = .60, r_dental = .61	r_{hair salon} = .60, r_insurance = .77	r_restaurant = .78, r_language = .81