The Effects of Expert and Consumer Endorsements on Audience Response

ABSTRACT This study examines the process by which audiences integrate expert and consumer endorsements into their product evaluations and how endorsement consensus affects this process. The results suggest that positive expert and consumer endorsements both enhance audiences' attitudes toward the endorsed product. However, positive consumer endorsements and higher perceived credibility of consumer endorsements, rather than expert endorsements, enhance audiences' behavioral intents when audiences are already interested in the endorsed product.

disagreement on audiences' attitudes and behavioral intents in the hedonic, pleasureseeking, consumption context. Consistent with previous studies, this study focuses on overall ratings rather than on the information content contained in reviews, because overall ratings have been shown to be more influential than information content in affecting audiences' interests (Wyatt and Badger, 1990). In addition, the visual impact study of third-party endorsement suggests that information conveyed in movie reviews can be sometimes ambiguous while average ratings could be more visually effective (Dean, 1999). That is to say that audiences are familiar with ratings for movies, so the form appears to be a succinct visual conveyer of information while quoted statement form is the least visually interesting, consisting only of text (Dean and Biswas, 2001).
The importance of understanding how to use expert and consumer endorsements strategically motivates the investigations proposed in this study regarding the effects of third-party endorsement. By understanding the effects of expert and consumer endorsements, this study can provide practical implications for advertisers to strategically use third-party endorsements to their benefits when advertising their products.

Effects of third-party endorsement
Two types of third-party endorsers, relevant to this study, have been identified in the endorsement literature: experts and regular consumers. They generally parallel, respectively, the two different source characteristics: credibility and perceived similarity to the audience (Frieden, 1984;Wilson and Sherrell, 1993). Friedman and Friedman (1979) proposed that expert endorsers influence through the process of internalization, suggesting that expert endorsers persuade through the credibility dimension. That is, the endorser is perceived to have credible information that may be used to solve the consumer's problem. The process of internalization occurs when the audience adopts an attitude because it is useful for the solution of a problem.
Third-party endorsements that employ regular consumers to recommend products and provide testimonials appear to be related most closely to persuasion through the similarity to the audience dimension (Dean and Biswas, 2001). In addition, audiences are particularly attracted to chunks of information about products that efficiently convey meaning (Jacoby, Szybillo, and Busato-Schach, 1977). By rating products on experience and credence characteristics, a third-party endorsement by either institutions or regular consumers may function as a chunk of information about a product and be perceived as a cost-efficient guide to product quality.
In this case, it is hypothesized that: H1: Regular consumers' positive endorsements will generate better attitude toward the endorsed product than regular consumers' negative endorsements. Institutions such as the Boston Globe and the New York Times are probably perceived to have access to testing facilities, equipment, and information to a greater degree than do regular consumers. Because institutions are probably perceived to have more than one expert on staff, this characteristic may imply that a consensus was reached prior to endorsement. Such a check-and-balance system may not be attributed to regular consumers (Dean and Biswas, 2001). Therefore: H3: Audiences will perceive institutions as having more expertise than regular consumers.
. . . third-party endorsements [are certainly a] factor in brand reputation-especially in light of the fact that advertisers are increasingly using their websites to obtain feedback on their products, services, and overall reputation among their customers, and also as a tool to improve their public relations and image.
H4: Audiences will perceive institutions' average rating as more credible than regular consumers' average rating.
This study argues that even though audiences might perceive institutions' average rating as more credible than regular consumers' average rating, this does not mean audiences will rely on institutions more to make their decisions. Dean and Biswas (2001) proposed that third-party endorsements should be tested for their ability to enhance perceptions of products that have hedonic, pleasure-seeking, consumption characteristics. They also argued that third-party endorsements might perform very differently for such prod-  (Hogarth, 1989). Research also has shown that audiences respond negatively to such uncertainty (Jaccard and Wood, 1988). Audiences may completely reject an alternative with conflicting opinions or ignore the endorsement disagreement and use a discounted average value for the category as a default valuation for the alternative (Ross and Creyer, 1992). Meyer (1981) conducted a study in which participants were asked to evaluate restaurants given critic ratings. He found that for restaurants whose average critic rating exceeded the mean value across all restaurants, participants did not exhibit utility when the critics disagreed about the restaurant quality. This study argues that endorsement consensus should influence audiences' movie evaluations in a positive way, yielding a strengthening effect on audiences' attitudes toward the movie as well as their behavioral intents. Therefore: H5a: Positive consensus between endorsements will generate better attitude than low consensus between endorsements.
H5b: Positive consensus between endorsements will generate better attitude than negative consensus between endorsements.
H6a: Consequently, positive consensus between endorsements will generate higher behavioral intent than low consensus between endorsements.
H6b: Consequently, positive consensus between endorsements will generate higher behavioral intent than negative consensus between endorsements.

Experimental design
A 2 ϫ 2 factorial design, measuring two levels of average rating by regular consumers (positive versus negative) and two levels of average rating by expert institutions (positive versus negative), was used in this study. A positive average rating given by either regular consumers or expert institutions was defined as an average rating that was perceived as positive by participants, whereas a negative average rating given by either regular consumers or expert institutions was defined as an average rating that was perceived as negative by participants. In this case, positive endorsement consensus between regular consumers' average rating and expert institutions' average rating materialized when both average ratings were perceived as positive by participants. Negative endorsement consensus between regular consumers' average rating and expert institutions' average rating materialized when both average ratings were perceived as negative by participants. Consequently, low endorsement consensus materialized in two conditions: (1) consumers' average rating was perceived as positive and expert institutions' average rating was perceived as negative by participants; (2) consumers' average rating was perceived as negative and expert institutions' average rating was perceived as positive by participants.  Participants logged onto the Yahoo Movie's website that featured Yahoo consumers' average rating and institutions' average rating of the movie (please see Figure 1 for a low consensus example). The desktops used in this study were typical computers found in computer stores and participants' homes. Because 16 computers at the lab were all the same type with exactly the same monitors, this nullified the possibility that participants would have performed the experimental task differently due to differences in the computers used.
Participants were free to decide the order that they reviewed the average rat-  Six dependent variables were measured in this study. Perceived credibility of institutions' average rating and regular consumers' average rating was measured by asking participants "how credible is institutions' (regular consumers') average rating" and using a 1-item scale where 7 indicated credible and 1 indicated not credible. Perceived expertise toward either institutions or regular consumers was measured by asking participants to rate institutions' (regular consumers') expertise in rating a movie and using a 1-item scale where 7 indicated high expertise and 1 indicated low expertise. Bipolar, 7-point semantic differential scales were used to measure participants' attitudes and behavioral intents. Attitude toward the movie was measured by asking participants to complete the sentence, "I would describe this movie as . . . ," using a 6-item scale composed of good/bad, pleasant/unpleasant, high quality/low quality, like it/don't like it, desirable/not desirable, and favorable/unfavorable (Hallahan, 1999). Cronbach's a value for participants' attitudes toward the movie is .92, which indicated that the measures were reliable. For this construct, a mean index was computed and used as the basis for analysis. Behavioral intent was measured by asking participants how likely they would go to see the movie and using a 1-item scale (Hallahan, 1999).

ENDORSEMENT EFFECTS ON AUDIENCE RESPONSE
regular consumers' positive average rating and higher credibility of regular consumers as perceived by the audiences enhanced behavioral intent. Second, even though the average rating provided by institutions, perceived as having higher expertise than regular consumers, was considered as more credible than regular consumers' average rating, this study observed audiences shifting their reliance on regular consumers' positive endorsements and credibility to decide the likelihood of going to see the movie.
These findings substantiated scholars' arguments that third-party endorsements may perform very differently for products that have hedonic consumption characteristics when audiences may have idiosyncratic tastes and follow thirdparty endorsements differently (Dean and Biswas, 2001;Goldsmith, Lafferty, and Newell, 2000).  White and Ramana (1999) argued that competition and the desire to establish an internet presence were the driving forces for advertisers to develop a website because many advertisers view the internet as a desirable medium for advertising their products.
The first practical implication is that advertisers can use the web for not only advertising and marketing activities, but also to obtain feedback and improve public relations. Thus, strategic thinking of integrating endorsements via websites will be an important task and channel for educating, informing, and persuading diverse audiences (Wang, 2005a) the internet allows audiences to exchange information with many third parties. Advertisers can include a link to a regular consumer's endorsement that lets audiences contact the source with permission.
In this way, advertisers could encourage audiences to seek feedback from sources after endorsements have appeared and leave this channel entirely in audiences' hands without intruding upon their privacy. By encouraging the audiences to verify the endorsements, the credibility dimension of regular consumers' endorsements could be greatly enhanced because trust, one important dimension of building endorsement credibility, can be built (Grazioli and Jarvenpaa, 2000;Wang, 2003).

Next steps
Audiences are more likely to face uncertainty for the experiential products stud-