The influences of electronicword-of-mouth message appealand message source credibility

on brand attitudePaul C.S. Wu

Department of Business Administration, Aletheia University,Tamsui, Taiwan, and

Yun-Chen WangGraduate School of Management Sciences, Aletheia University,

Tamsui, Taiwan

Abstract

Purpose – Based on message source theory, the purpose of this paper is to use positive electronicword-of-mouth (eWOM) settings to examine the influence of message source credibility on brandattitude (brand trust, brand affection, and purchase intention). The current study also uses theelaboration likelihood model to examine the moderating effect of product involvement on therelationships between message appeals (rational vs emotional) and brand attitude, as well as betweenmessage source credibility and brand attitude.

Design/methodology/approach – Data were collected via internet questionnaires. Based on theinternet investigation and pretest, this study chose the stimuli for research products, virtual brands,eWOM, and message source credibility. Via a 2 (notebooks vs shampoo) *2 (appeal: rational vs emotional)experimental design, 211 effective samples were collected to verify the hypotheses of this study.

Findings – The results, for both notebooks and shampoo, are: the positive eWOM message withhigher message source credibility indicates a better brand attitude than the eWOM message withlower message source credibility, and this effect is not moderated by the degree of productinvolvement, indicating its robustness. Second, with a high degree of product involvement, the rationalappeal indicates a better brand attitude than the emotional appeal; no significant difference is foundwhen product involvement drops to a low level.

Originality/value – Previous research investigating the moderating effect of involvement on theeffects of message-appeal types and the message source credibility of eWOM is rare. The maincontribution of this study is to fill this gap.

Keywords Electronic word-of-mouth (eWOM), Message appeal, Message source credibility,Product involvement, Brand attitude, Brand trust, Brands, Internet

Paper type Research paper

1. IntroductionElectronic word-of-mouth (eWOM) refers to the knowledge exchange consumers carry outonline. This exchange has a direct relationship on customer loyalty, which in turn affectsoverall value of the company. Therefore, the effect of eWOM on consumers is graduallyreceiving increased attention (Gruen et al., 2006; Hennig-Thurau et al., 2004). eWOMdiffers from commercial advertisements because it is a non-commercial message createdby consumers. This message provides direct experiences of consumers and affects thepurchasing decisions of other consumers (Godes and Mayzlin, 2004; Park et al., 2007).

The current issue and full text archive of this journal is available at

www.emeraldinsight.com/1355-5855.htm

APJML23,4

448

Received 16 April 2010Revised 19 November 2010Accepted 27 April 2011

Asia Pacific Journal of Marketing andLogisticsVol. 23 No. 4, 2011pp. 448-472q Emerald Group Publishing Limited1355-5855DOI 10.1108/13555851111165020

Research related to WOM mainly focuses on the effect of WOM on future sales(Godes and Mayzlin, 2004), the motivations of consumers to disseminate WOM(Hennig-Thurau et al., 2004), factors that influence consumers to use WOMcommunication (Duhan et al., 1997), and comparing WOM influences with othermarket information (Bickart and Schindler, 2001). The other research stream of WOMinvestigates influential factors related to its persuasion effects (product attitude andpurchase intention), including message direction (favorable or unfavorable) (Arndt,1967; Burzynski and Bayer, 1997; Mizerski, 1982; Park and Lee, 2009), message sourceaccessibility (Herr et al., 1991), and message source similarity (Feick and Higie, 1992).Consumers generally believe that WOM credibility is higher than commercialadvertisement. Hence, the effect of WOM is typically higher than advertisements(Herr et al., 1991). However, eWOM is unlike WOM. The recommendation messagetypically comes from anonymous individuals in a text-based format. Therefore, it isdifficult for online consumers to use source similarity, i.e. whether or not therecommender is on the same side, to determine the credibility of the message (Chatterjee,2001). Yet, message source credibility has a huge effect on whether or not the consumerbelieves the message. Few researches have investigated whether or not eWOM messagesource credibility influences persuasion effect. Various types of message appeals(rational or emotional) may affect the persuasive effects to the message receivers (Kotlerand Keller, 2008). To establish an effective marketing strategy, marketers need tounderstand that message source credibility and the type of message appeal influence theeffect of eWOM (Park and Lee, 2009).

Laurent and Kapferer (1985) pointed out that involvement consists of fourdimensions: hedonic value, symbol value, interest, and perceived risk. Rossiter et al.(1991) defined involvement purely in terms of purchase risk perceived by the consumer;a low-risk purchase was categorized as low involvement and a risky purchase as highinvolvement. When consumers evaluate the product attribute, they are involved in theproduct and feel disturbed about the uncertainty of purchasing the product, whichmeans they now have perceived risk (Chaudhuri, 2000). Zaichkowsky (1985), Roberson(1976) and Brooker (1981) separately pointed out that consumers with a high degree ofproduct involvement actively look for information related to the product and evaluate allthe alternatives, whereas consumers with a low degree of product involvement do not.Therefore, when degrees of product involvement differ, perceived risk for purchasing aproduct differs (Zaichkowsky, 1985). In sum, product involvement refers to theimportance of the product to an individual, which affects purchasing behavior.Collecting product information reduces perceived risk. Consumers with a high degree ofinvolvement have high perceived risk and consumers with a low degree of involvementhave low perceived risk.

Wangenheim and Bayon (2004) indicated that when consumers perceive high socialor psychological risk of a purchase, they would search for credible information that ismore diagnostic or referential, such as WOM information, from a source they believe tobe reliable. This is particularly true in Taiwan, where independent consumerorganizations are rarely useful sources of reference. Consumers search for credibleeWOM information from popular web sites and WOM from other reference groups.Consumer perceived risk of products has significant moderating effects on the outcomeof WOM (Wangenheim and Bayon, 2004). Under a high degree of perceived risk,consumers invest significant efforts to make a purchasing decision and are easily

eWOM messageappeal

449

affected by WOM; consequently, the effect of WOM on purchasing decision is largerthan the situation with a low degree of risk (Still et al., 1984). Hence, the interactionbetween perceived risk and message source credibility affects the influence ofeWOM/WOM. As the level of perceived risk relates to the degree of involvement, theinteraction between product involvement and message source credibility also affectsthe influence of eWOM/WOM. However, there are few related studies in the literature.

According to the social judgment theory (SJT), involvement moderates the effects onconsumer attitude (Sherif and Hovland, 1961; Wang and Lee, 2005). The elaborationlikelihood model (ELM) points out that the information-processing route of consumersis related to the degree of involvement. Consumers with a high degree of involvementtend to process information through a central route, whereas consumers with a lowdegree of involvement tend to process information through a peripheral route.Consumers of central route information processing are likely to think rationally andconsumers of peripheral route information processing tend to judge by emotions(Petty and Cacioppo, 1981). The cognitive fit theory asserts that when the message typeand consumer information-processing strategy are consistent, the consumer has acognitive fit and a better attitude towards the tagged object (Vessey and Galletts, 1991).Therefore, the level of consumer involvement moderates the influence ofmessage-appeal types (rational or emotional) on the effects of WOM. Although themoderating role of product involvement is very important to understand the in-deptheffect of eWOM message-appeal type on brand attitude, previous studies have notadequately addressed this interaction effect.

Based on the preceding discussion, this work assumes that individual degree ofinvolvement affects the decision-making process and that the persuasive effects of theeWOM message-appeal type and message source credibility vary with individualdegree of involvement. Previous studies investigating the moderating effect ofinvolvement on the persuasive effects of the eWOM message-appeal type and messagesource credibility are rare. The main contribution of this study is to fill in this gap. Thisresearch employed product involvement as the moderator to test whether degree ofinvolvement influences the effect of message-appeal types and message sourcecredibility of eWOM on consumer brand attitude.

The theory of country-of-origin effect (Chao, 1993, 2001; Lampert and Jaffe, 1996;Veale and Quester, 2009) predicts that Asian-branded products in emerging countriesmight be unattractive to Western consumers. eWOM could be a great marketing tool forthose product manufacturers to extend their business to developed countries in otherAsian or Western countries, due to a well-established web infrastructure in thesecountries, which plays an important role in communicating consumption experiencesbetween consumers. The findings of this study contribute to the practitioners ofAsian-based manufacturers to increase their sales to other Asian countries or developedcountries by adequately utilizing eWOM and the degree of consumer productinvolvement.

2. Literature review and hypotheses2.1 Electronic word-of-mouthWOM is oral person-to-person communication between a receiver and a sender, whichinvolves a product, service, or brand. To the receiver, this messenger does not havestrong commercial intent (Anderson, 1998; Harrison-Walker, 2001), which is why

APJML23,4

450

consumers believe the credibility of WOM is higher than commercial advertisements(Herr et al., 1991). WOM is a widely accepted non-commercial information source thathas a huge effect on consumer suggestion formation and purchasing decisions(Richins, 1983). This type of interpersonal communication affects marketer decisions(Hennig-Thurau et al., 2004). Any positive or negative statement effectively made to amultitude of people and institutions via the internet becomes the so-called eWOM oronline WOM (Hennig-Thurau et al., 2004).

Besides, from the viewpoint of message senders, consumers who gather informationfrom online discussions also show greater interest in the product topic than do thoseconsumers who acquire information from the marketer-generated sources (Bickart andSchindler, 2001). It seems that eWOM may have much more influence on a consumer’sbrand’s attitude and judgment than any other source of influence (Godes and Mayzlin,2004). Thus, throughout the eWOM activity, consumers can obtain high levels ofmarket transparency. In addition, they can take on a more active role in the value chainand influence products and prices according to individual preferences (Park and Kim,2008) and the motivation for the consumers to use eWOM is increased.

2.2 Message appealMany studies have claimed that advertising message strategies can be classified aseither informational or transformational (Golan and Zaidner, 2008; Laskey et al., 1989;Wells, 1980). The former refers to ads that appeal to the cognition or logic ofconsumers, whereas the latter refers to ads that appeal to the emotions or senses (Putoand Wells, 1984). Foote, Cone & Belding Communications Inc. explored and developedthe so-called FCB grid. This grid classifies products in terms of high/low involvementand think/feel motivations – to create four types of goods/services, each having adifferent hierarchy of effects and a different advertising approach (Mortimer, 2002;Ratchford, 1987; Vaughn, 1986). Rossiter et al. (1991) argued that emotions could bepositive or negative, and replaced labels of think and feel in the FCB grid withinformational and transformational motivations. Taylor (1999) proposed thesix-segment message strategy wheel, which divides advertising strategies into twocommunication views of Carey (1975): transmission and ritual. This division is similarto the informational and transformational dichotomy (Hwang et al., 2003).The transmission view is composed of three segments: ration, acute need, androutine, whereas the ritual view is composed of the ego, social, and sensory segments.

Kotler and Keller (2008) pointed out that message senders must think over the typeof message they want to send to target receivers to obtain the expected response anddefined that advertisement appeal should show some type of benefit, encouragement,agreement, or reason and why consumers should pay attention to or purchase theproduct. They classified message appeal into two categories:

(1) rational appeal – using consumer benefit as the appeal to present productattribute; and

(2) emotional appeal – appeal that can elicit consumer positive or negativeemotions to arouse purchase willingness.

As for the eWOM message appeal, Voss et al. (2003) suggested eWOM as a way forconsumers to share and discuss product and brand application experience.The consumption experience consumers share on the internet consists of hedonic

eWOM messageappeal

451

and utilitarian elements. Hedonic eWOM refers to the pleasant, gratified productexperience consumers receive; utilitarian eWOM refers to product functionality andeffectiveness. Park and Lee (2008) classified internet reviews into attribute-valuereviews and simple-recommendation reviews. Attribute-value reviews containproduct-relevant attribute/benefit information. These reviews are rational, objective,and concrete, based on facts about a product. Simple-recommendation reviews consistof emotional and subjective recommendations. These reviews present subjectivefeelings, interjection, and non-relative information, therefore they are emotional andabstract based on consumer feelings about a product.

Many studies have applied rational and emotional appeal to categorizeadvertisement appeal (Hotchkiss, 1949; Kotler and Keller, 2008; Simon, 1971), andempirically verify the validity of such categorization (Bang et al., 2005; Chen et al.,2005; Hetsroni, 2000). This type of classification is also similar to theinformational/think/transmission and transformational/feel/ritual dichotomy.Although the categorization by Voss et al. (2003) and Park and Lee (2008) targeteWOM appeals, they lack the opinion and application of follow-up researchers.The definitions of these two appeals for categorizing are similar to the definitionestablished by Kotler and Keller (2008). Hence, according to basic message appealcategorization by Kotler and Keller, this research used rational and emotional appealcategorization for the eWOM message. The purpose was twofold. One was to testwhether basic message appeal categorization can be applied in an eWOM setting, andthe other was to strengthen measurement validity, because this type ofappeal categorization has been used by many studies.

2.3 Message source credibilityMessage source credibility refers to how much the message receiver believes in thesender. It is an attitude towards the message source (Gunther, 1992) that affectsthe receiver level of belief about what the source claims (West, 1994). This credibility isalso an important factor affecting persuasion effectiveness (Hovland and Weiss, 1951). Ifthe source has credibility, consumers will believe the message. Thus, credibility is howmuch the message reflects reality after the consumer’s evaluation (Ling and Liu, 2008).

Message source credibility is a high-order construct consisting of threesub-dimensions, trustworthiness, expertness, and attractiveness. Trustworthinessrefers to the degree of confidence and acceptance receivers have towards the messagesender. Expertness refers to professional knowledge the sender has about the product.Attractiveness refers to when the sender attracts receivers to consume products orservices (Ohanian, 1990).

2.4 InvolvementInvolvement is the level of relevance an individual has towards an event or objectbased on the individual, value, and interest, which is the level of attention a consumerhas for a product or the level of motive and interest a product triggers (Batra and Ray,1985; Goldsmith and Emmert, 1991; Zaichkowsky, 1985).

Involvement can be categorized into three types based on targets:

(1) product involvement – consumer concern or reactions towards products orbrands;

APJML23,4

452

(2) advertisement involvement – consumer concern towards the advertisementmessage or consumer psychological state when encountering an advertisement;and

(3) purchase decision involvement – consumer concern towards a certainpurchasing activity or purchasing decision (Zaichkowsky, 1985).

In this research, involvement refers to product involvement.

2.5 Brand attitudeAttitude is an overall persistent evaluation towards people, things, or goods (Engel et al.,2006); it is a consistent positive or negative reaction towards certain things throughlearning (Hawkins et al., 2007; Schiffman and Kanuk, 2006), which includes favorable orunfavorable evaluation, emotional feeling, and behavioral tendency (Kotler and Keller,2008). Brand attitude is the continuous preference or loathing tendency of the consumertowards a certain brand (Fishbein and Ajzen, 1980) and the overall evaluation theconsumer has towards the brand. The evaluation base is the prominent benefit or imageof the brand. The consumer judges whether the prominent benefit or image is beneficialand then forms an overall brand attitude (Wilkie, 1986), which is the comprehensiveevaluation result of the consumer attitude towards the brand attribute (MacKenzie andSpreng, 1992). A positive individual attitude towards a brand may increase the chance ofusing the brand (Kotler and Keller, 2008).

Several previous studies have suggested three components of attitudes:

(1) cognitive component – the knowledge or cognition of the product formed afterintegrating direct experience or information obtained;

(2) affective component – the emotion or affective reaction of the consumertowards a product or brand, representing the direct or overall evaluation of theconsumer towards the product; and

(3) conation component – the possibility or tendency of an individual act or behavewith regard to the product (Assael, 2004; Guiltinan et al., 1997; Lutz et al., 1986;Rosenberg and Hanland, 1960).

Some studies have treated only cognitive/utilitarian and affective/hedonic componentsas perceptual attitude factors (Syvia and Chan-Olmsted, 2005; MacKenzie and Spreng,1992; Muray and Vogel, 1997; Voss et al., 2003), while few studies have treatedproduct/brand attitude as one dimension, i.e. affective component (Osgood et al., 1957).Because belief, affection, and intentions are components in most accepted hierarchicalconceptualizations of the communication process (Engel et al., 2006; McGuire, 1976),this study adopted the theory of the three components of attitude. This researchinterpreted cognition as brand trust (Delgado-Ballester, 2004), affection as brandaffection (Vakratsas and Ambler, 1999; Howard and Gengler, 2001), and conation aspurchase intention (Bennett and Rundle-Thiele, 2000).

2.6 Message source credibility and brand attitudeThe message source theory asserts that when source credibility is high, the receiver willbe highly persuaded by the message, and when source credibility is low, the receiver willhave doubts about the message (Eagly and Chaiken, 1993; Zhang and Buda, 1999).Ling and Liu (2008) suggested that consumers not only rely on the celebrity credibility

eWOM messageappeal

453

to interpret the message, but also form their own attitude towards the advertisement andbrand. When a celebrity possesses high source credibility, consumer attitude towards,and comprehension of, the advertisement and brand will be positive. Because eWOMsender plays the role of celebrity, the source credibility of eWOM will have similar effecton attitude towards the brand.

Positive WOM attracts new customers (Reichheld and Sasser, 1990). Consumersmay base their judgments of trustworthiness and relevance of the recommendationupon the perceived similarity of peer attitudes and preference, as well as the feelings ofrapport they share with the peer recommender (Bickart and Schindler, 2001; Kieckerand Cowles, 2001). This is similar to the way eWOM works. When eWOM about aproduct is positive, consumers are likely to consider the product desirable inconformity to the eWOM (Park and Lee, 2008). However, source credibility of themessage influences this positive effect. Kiecker and Cowles (2001) recommended that ifconsumers perceived high attraction or expertness towards the eWOM source,the message would have higher persuasiveness and credibility to them. Based on thepreceding discussion, the following hypotheses are advanced:

H1. For positive eWOM, consumer brand attitude of high message sourcecredibility would be better than that of low message source credibility.

H1-1.For positive eWOM, consumer brand trust of high message source credibilitywould be better than that of low message source credibility.

H1-2.For positive eWOM, consumer brand affection of high message sourcecredibility would be better than that of low message source credibility.

H1-3.For positive eWOM, consumer purchase intention of high message sourcecredibility would be better than that of low message source credibility.

2.7 Moderating effects of involvementBased on the concept of SJT, individual current attitude serves as the referenceframework to categorize received information based on this framework (Sherif andHovland, 1961). Individual attitude can be classified according to rejection andacceptance. Moreover, individual degree of involvement is a significant influentialfactor in the large difference among individual attitudes (Wang and Lee, 2005).Thus, involvement has an interference effect on consumer attitude. The ELM pointsout two routes that affect consumer attitude formation and alteration:

(1) the central route – consumers decide whether to accept the message afterrationally thinking through the message that is trying to persuade them; and

(2) the peripheral route – altered consumer attitude is speculated based onextrinsic product cues (Petty and Cacioppo, 1981).

Message appeal can be classified as rational and emotional (Kotler and Keller, 2008).Here, eWOM plays two roles:

(1) informants – online consumers deliver additional user-oriented information;and

(2) recommenders – online consumers provide either a positive or a negative signalof product reputation (Park et al., 2007).

APJML23,4

454

The informant role is similar to rational appeal, whereas the recommender role is similarto emotional appeal. Consumers with high involvement are most likely to apply centralroute information processing and are more willing to elaborate on eWOM contentcarefully to obtain additional product information. They are also more likely to assumethe informant role in eWOM, which means they are more likely to accept the rationalappeal of eWOM. Consumers with low involvement are most likely to apply peripheralroute information processing and do not like the process of elaborating information.They are more likely to assume the recommender role in eWOM, which means they aremore likely to accept the emotional appeal of eWOM. According to the cognitive fittheory, when the information type and consumer information processing are consistent,the consumer obtains a cognitive fit and better attitude towards the object (Vessey andGalletts, 1991). Based on this discussion, the following hypotheses are advanced:

H2. Product involvement has moderating effects on the relationship betweenpositive eWOM message appeal and brand attitude.

H2-1. For positive eWOM, under high degree of product involvement, consumerbrand attitude of rational appeal would be better than that of emotionalappeal.

H2-1-1. For positive eWOM, under high degree of product involvement, consumerbrand trust of rational appeal would be better than that of emotionalappeal.

H2-1-2. For positive eWOM, under high degree of product involvement, consumerbrand affection of rational appeal would be better than that of emotionalappeal.

H2-1-3. For positive eWOM, under high degree of product involvement, consumerpurchase intention of rational appeal would be better than that ofemotional appeal.

H2-2. For positive eWOM, under low degree of product involvement, consumerbrand attitude of emotional appeal would be better than that of rationalappeal.

H2-2-1. For positive eWOM, under low degree of product involvement, consumerbrand trust of emotional appeal would be better than that of rational appeal.

H2-2-2. For positive eWOM, under low degree of product involvement, consumerbrand affection of emotional appeal would be better than that of rationalappeal.

H2-2-3. For positive eWOM, under low degree of product involvement, consumerpurchase intention of emotional appeal would be better than that of rationalappeal.

Likewise, when consumers have low involvement in the product of online consumerreviews, they would apply peripheral route information processing and pay moreattention to non-product-related cues (Park and Lee, 2008). They would not elaborateon the functional attribute of the product and evaluate credibility of the source.On the contrary, when consumers possess a high degree of involvement, they would

eWOM messageappeal

455

apply central route information processing and elaborate the reviews ofproduct-related information and source credibility. High-involvement subjects wouldhave more motivation and higher ability to process the product information thanlow-involvement subjects adopting the peripheral route (Chien et al., 2007; Petty andCacioppo, 1981). Thus, the positive effect of message source credibility on brandattitude would be higher for consumers in a high degree of product involvement thanfor those with a low degree of product involvement. Based on the preceding discussion,the following hypotheses are advanced:

H3. The positive effect of eWOM message source credibility on brand attitudewould enhance as the degree of product involvement increases.

H3-1.The positive effect of eWOM message source credibility on brand trust wouldenhance as the degree of product involvement increases.

H3-2.The positive effect of eWOM message source credibility on brand affectionwould enhance as the degree of product involvement increases.

H3-3.The positive effect of eWOM message source credibility on purchase intentionwould enhance as the degree of product involvement increases.

3. Method3.1 Measurement3.1.1 Message appeal. This study classified message appeals into rational appeal andemotional appeal. The text and images in a rational appeal message may emphasizeproduct value, function, and benefit, whereas the text and images in an emotional appealmessage may create a type of atmosphere and meaning to try to impress the consumer(Kotler and Keller, 2008). This study used two questions to measure rational andemotional cognitions of respondents in terms of eWOM message appeal. The firstquestion (the message mainly describes the product function, benefit, and value) askedhow much the subject agrees that the message is a rational appeal. The second question(the message mainly spreads a certain atmosphere, emotion, and feeling) asked howmuch the subject agrees that the message is an emotional appeal. Reverse scoring wasapplied to emotional appeal questions. For example, a score of 7 was reversely processedinto 1; a score of 6 was reversely processed into 2, and so on. Finally, the scores of thesetwo questions were added up. A high score means the subject thinks the advertisementmessage appeal was rational, whereas a low score means the subject thinks the appealwas emotional. These two questions were measured on a Likert seven-point scaleanchored by “1 ¼ strongly disagree” and “7 ¼ strongly agree.”

3.1.2 Message source credibility. Message source credibility refers to how much themessage receiver believes in the message spread by the source (Gunther, 1992).The measurement was based on the three sub-dimensional source credibility scaledeveloped by Ohanian (1990). This research focused on measuring the sourcecredibility of eWOM. Because the message sender who sent out eWOM cannotbe contacted directly, the attraction sub-dimension could not be measured. Therefore,this research only used the other two sub-dimensions, expertness and trustworthiness,to measure message source credibility. Each sub-dimension includes five seven-pointsemantic differential scales (the Appendix). The average was used as the standard tomark high and low credibility.

APJML23,4

456

3.1.3 Product involvement. Involvement is defined as “the level of relevance anindividual has towards an object based on individual demand, value, and interest,which is the level of attention a consumer has for a product (Zaichkowsky, 1985).”The personal involvement inventory scale developed by Zaichkowsky (1994) wasadopted for measurement, which includes ten seven-point semantic differential scalesfor measurement. The average was used as the standard to mark high and lowinvolvement.

3.1.4 Brand attitude. Many researches claim three components for attitude:cognition, affection, and conation (Guiltinan et al., 1997; Rosenberg and Hanland,1960; Schiffman and Kanuk, 2006). This research used brand trust, brand affect,and purchase intention to represent cognition, affection, and conation, respectively.Brand trust refers to the confidence of consumers that when they take a risk ofbrand choice, they are certain that the reliability and performance of the brand willmeet their expectations (Delgado-Ballester, 2004). Brand affect refers to when thebrand is able to trigger consumer positive affection, which in turn causes theconsumer to use the brand (Chaudhuri and Holbrook, 2002). Purchase intention isthe possibility and probability that the consumer is willing to purchase a certainproduct (Dodds et al., 1991). The measurement scales of these three dimensions wereadopted from these researches, respectively, and all items were measured on aLikert seven-point scale anchored by “1 ¼ strongly disagree” and “7 ¼ stronglyagree.”

3.2 Research designThis research selected notebooks and shampoo as the research products to representelectronics goods and fast-moving consumer goods (FMCG), respectively. The reasonfor this choice was that these two products represent shopping and convenience goods(Kotler and Keller, 2008), by which this study could demonstrate external validity, andthat the eWOM message of these two product categories often includes both rationaland emotional appeals. Virtual brands for these two products were designed as thestimuli to avoid the effect of brand image.

A pretest was conducted to find out the eWOM of rational and emotional appeals.The pretest participants were students from a university in Northern Taiwan. The totaleffective sample was 60, which included four versions of questionnaires for the mainstudy based on two product categories (notebooks and shampoo) and two eWOMmessage appeals (rational and emotional). Online questionnaires were used as the datacollection tool and convenience sampling was employed through forwarding e-mails.The process of data collection was set up such that after respondents received the e-mail,they randomly selected one of the four online questionnaire links. After linking to theselected questionnaire, participants answered the questionnaire. The answeringsequence was:

. the degree of involvement regarding the research products;

. reading the product information and the eWOM; and

. answering other questions.

After completing the questionnaire, the participant was asked to forward this e-mail toother people with the expectation to obtain more samples.

eWOM messageappeal

457

4. Results4.1 Basic data analysis4.1.1 Pretest results. A total of 60 effective pretest samples were collected, 30 notebooksand 30 shampoos. To ensure that the eWOM messages of the questionnaire werecorrectly identified as rational and emotional appeals by respondents, six types ofeWOM messages based on the result of an online field survey were designed for thepretest of each of these two products. Table I shows the results. For “DN-Portable”notebooks, the mean of eWOM (6) was the highest and was chosen as a rational appeal;the mean of eWOM (2) was the lowest and was chosen as the emotional appeal.For “Hi-Young” shampoo, the mean of eWOM (4) was the highest and was chosen asthe rational appeal; the mean of eWOM (3) was the lowest and was chosen as theemotional appeal. Table II shows the chosen messages of the eWOM.

In terms of message source credibility, the current eWOM on the web was used as areference. Finally, the maintenance staff of “DN-Portable” notebooks and the productionstaff of “Hi-Young” shampoo were chosen as the eWOM provider, for respondents toevaluate message source credibility. The means were 4.92 and 4.64, respectively.

4.1.2 Sample. A total of 211 effective samples were collected from the main study,including 102 notebooks and 109 shampoos. The sample contained a higher proportionof males (75.4 percent) than females (24.6 percent). A large percentage of youngerconsumers, aged 21-30 years, were also included in the samples (73.5 percent).Although the sample was skewed towards men and younger consumers, this iscommon in internet survey samples.

4.1.3 Manipulation checks. Manipulation checks were examined for testing whethersubject recognition of rational and emotional level for the eWOM message on the mainexperiment conforms to the pretest or not. The results revealed that for notebooks,rational appeal received a higher score (M ¼ 7.79) than emotional appeal (M ¼ 5.94);and for shampoo, rational appeal obtained a higher score (M ¼ 8.26) than emotionalappeal (M ¼ 6.29). T-tests revealed significances in mean message appeals(t (100) ¼ 4.462, p ¼ 0.000 for notebooks; and t (107) ¼ 5.977, p ¼ 0.000 for shampoo).Therefore, the present experiment adequately manipulated the message appeal ofeWOM. The positive level of eWOM was also examined. The finding revealed it wassignificantly greater than 4 (M ¼ 4.9515) in the seven-point scale at the p , 0.001 level(t (210) ¼ 11.262, p ¼ 0.000).

Products Message of eWOM Means SD

“DN-Portable” notebooks 1 8.43 1.572 6.43 1.913 6.77 1.854 9.00 1.465 6.80 1.906 9.47 1.43

“Hi-Young” shampoo 1 6.80 1.302 9.17 1.863 5.93 1.804 9.67 2.295 7.10 2.146 8.97 2.03

Table I.Evaluation results ofeWOM message appeal

APJML23,4

458

4.1.4 Validation of key constructs. The reliability of key constructs was examined usingconventional methods since established scales were used in the present research. TheCronbach’s a of each construct exceeded the suggested cut-off value of 0.70 (Nunnally,1978) (the Appendix). The results of the exploratory factor analysis revealed thatthe factor loadings of each construct were between 0.501 and 0.960, exceeding thesuggested cut-off value of 0.5.

4.1.5 Sample checks. Convenience sampling via the internet was used for datacollection. Manipulation checks confirmed that respondents did not demonstratesignificant differences on the main variables of interest (including product involvement,rational and emotional message appeals, positive WOM level, message sourcecredibility, and brand attitude) based on their demographic and behavioral data(including gender, age, education, occupation, monthly income, daily number of hoursspent using the internet, and the degree of searching eWOM of products).Among the 84 ANOVA tests (two products*seven demographic variables*six mainvariables), the results revealed that only six tests showed significant differences on themain variables at the p , 0.05 level.

Gender might have moderating effects on brand attitude or affect the degree ofinvolvement, which could interfere with the effect of the eWOM message. Therefore,ANOVA including interaction effects (gender*message appeal and gender*sourcecredibility) was run for each product. The results showed no significant moderatingeffects on brand attitude by gender. The t-test also revealed no effects of gender on thedegree of involvement for each product.

ProductsMessageappeal Message of eWOM

“DN-Portable”notebooks

Rational DN-Portable has the best specifications. It has a 13 in. screen,newest processor, and independent video card. Specifications of anotebook like this are similar to a PC. After using it, I think thesound quality of the speakers is great. It is good for listening tomusic or watching movies. The cooling design is also great. Thenotebook does not overheat easily. It is a brand worth purchasing

Emotional When I first saw a DN-Portable notebook, I was attracted to itshigh-quality appearance. Not only is the external appearance ofDN-Portable beautiful, but the quality is Japanese style, with achoice of five colors. Shiny pink and solid black can satisfy manygirls and boys. Whenever I bring my notebook to school, nowpeople look at me as if they are jealous and ask me what brand thenotebook is. I am so happy!

“Hi-Young”shampoo

Rational My hair is oily and my scalp is sensitive, which means there are notmany shampoos I can use. However, after using Hi-Youngshampoo, my hair is not as oily and is soft and untangled. Theprickly and itchy feeling on my scalp is gone. This shampoo alsohas a moisturizing effect. If your hair is like mine, I recommendthat you use this shampoo

Emotional Can you imagine? After using Hi-Young shampoo, it is like you areembraced by nature or running through the grass with a slightbreeze. The once withered flowers and grass are instantlysupplemented with energy. Your hair shines like the sun to allowyou to show self-confidence everyday. Come and feel it yourself!

Table II.Chosen eWOM for pretest

eWOM messageappeal

459

4.2 Hypotheses test4.2.1 The effect of message source credibility. The means of the message sourcecredibility were used to divide the samples of notebooks and shampoo into high and lowmessage source credibility. Brand attitude and its three sub-dimensions were used as thedependent variables to carry out one-factor MANOVA. The results (Table III) revealedthat for notebooks, H 2 ¼ 0.309, p ¼ 0.000 (high credibility n ¼ 57, low credibilityn ¼ 45); and for shampoo, H 2 ¼ 0.483, p ¼ 0.000 (high credibility n ¼ 70, lowcredibility n ¼ 39). The values were all significant. Table IV shows that the differenceson dependent variables between high and low source credibility for notebooks andshampoo were all positive and significant at the p , 0.001 level. This indicates thatconsumers who perceive high-eWOM message source credibility would have higherbrand attitude along with its three sub-dimensions (brand trust, brand affection, andpurchase intention) than consumers who perceive low message source credibility, thussupporting H1, H1-1, H1-2, and H1-3.

4.2.2 The moderating effect of involvement. The notebooks and shampoo sampleswere divided into high- and low-involvement groups based on the means (notebooks,M ¼ 4.88, n ¼ 54 and 48 for high/low involvement; shampoo, M ¼ 4.42, n ¼ 53 and56 for high/low involvement). Next, two-factor MANOVA was conducted to examine themoderating effect of involvement on the relationship between message appeal and brandattitude, and the result was shown in Table V. For notebooks, the interaction termshowed that H 2 ¼ 0.102, F (4, 95) ¼ 2.433 and was significant at the p , 0.05 level; forshampoo, the interaction term showed that H 2 ¼ 0.125, F (4, 102) ¼ 3.289 and wassignificant at the p , 0.05 level. The contrast comparison results (Tables VI and VII)revealed that for the high-involvement sub-sample of notebooks and shampoo, brandattitude, brand trust, brand affection, and purchase intention of rational appeal were allbetter than those of emotional appeal and were significant at the p , 0.05 level, thussupporting H2-1, H2-1-1, H2-1-2, and H2-1-3. However, for the low-involvementsub-sample of these two products, brand attitude, brand trust, brand affection, andpurchase intention of rational appeal were still better than those of emotional appeal,which were opposite the hypotheses, although the differences were very small andinsignificant. Therefore, H2-2, H2-2-1, H2-2-2, and H2-2-3 were not supported. Thismay be because eWOM is often presented as text, with few images or videos, whichmakes emotional appeal difficult to present and reduce the effect of emotional appeal(Park and Lee, 2009). However, when compared to high involvement, the differences inbrand attitude and its sub-dimensions for emotional and rational appeals decrease. Thisimplies that when product involvement is low, the eWOM of rational appeal may not bebetter than emotional appeal to enhance consumer brand attitude. Based on thisdiscussion of both notebooks and shampoo, hypothesis H2 was supported.

Another two-factor MANOVA test was conducted to examine the moderating effectof involvement on the relationship between message source credibility and brand

Products Hotelling’s trace value F (df1, df2) P

Notebooks 0.309 7.579 (4, 097) 0.000 *

Shampoo 0.483 12.914 (4, 104) 0.000 *

Note: Significant at: *p , 0.001

Table III.MANOVA results ofmessage sourcecredibility towards brandattitude

APJML23,4

460

Dep

end

ent

var

iab

les

Bra

nd

atti

tud

eB

ran

dtr

ust

Bra

nd

affe

ctio

nP

urc

has

ein

ten

tion

Notebooks

Hig

hv

slo

wcr

edib

ilit

yD

iffe

ren

ces

(hig

h2

low

)0.

711

(4.1

552

3.44

4)0.

735

(3.9

952

3.26

0)0.

654

(4.0

882

3.43

4)0.

705

(4.4

532

3.74

8)S

E0.

142

0.13

50.

173

0.20

4F

24.9

7229

.520

14.3

4111

.941

(df1

,d

f2)

(1,

100)

(1,

100)

(1,

100)

(1,

100)

P0.

000

**

0.00

0*

*0.

000

**

0.00

1*

*

Shampoo

Hig

hv

slo

wcr

edib

ilit

yD

iffe

ren

ces

(hig

h2

low

)1.

046

(4.1

662

3.12

0)0.

969

(3.9

952

3.02

6)1.

072

(4.1

192

3.04

7)1.

154

(4.4

692

3.31

4)S

E0.

154

0.16

00.

154

0.22

0F

46.1

2936

.726

48.4

9827

.654

(df1

,d

f2)

(1,

107)

(1,

107)

(1,

107)

(1,

107)

P0.

000

**

0.00

0*

*0.

000

**

0.00

0*

*

Note:

Sig

nifi

can

tat

:* p

,0.

01,

** p

,0.

001

Table IV.Contrast results of highvs low message source

credibility on brandattitude

eWOM messageappeal

461

attitude and its three sub-dimensions. The results revealed that for both notebooks andshampoo, the interaction terms were not significant at the p , 0.05 level (notebooksp ¼ 0.053, shampoo p ¼ 0.279). The ANOVA results also revealed the same pattern foreach dependent variable. Therefore, H3, H3-1, H3-2, and H3-3 were all not supported.This implies that the positive effect of eWOM message source credibility on brandattitude and its sub-dimensions (brand trust, brand affection, and purchase intention) isnaturally robust and is not affected by the degree of product involvement.

5. Conclusion and discussions5.1 Theoretical implicationsThe results indicate a direct and positive relationship between message sourcecredibility and brand trust, brand affection, purchase intention, and brand attitude, andsupport H1-1, H1-2, H1-3, H1, and past research. In particular, high credibilitytowards the eWOM message source increases perceived quality, decreases perceivedrisk (Erdem and Swait, 1998), induces high persuasiveness (Eagly and Chaiken, 1993),and improves brand attitude. H3 predicted enhanced positive effect of eWOM messagesource credibility on brand attitude with an increased degree of involvement. However,the results of this study revealed that the positive effect of eWOM message sourcecredibility on brand trust, brand affection, purchase intention, and overall brandattitude are not affected by the degree of product involvement, which means theserelationships are naturally robust. This also indicates the importance of messagesource credibility on the effect of eWOM on brand attitude.

H2 predicted that product involvement would moderate the relationship betweeneWOM message appeal and brand attitude. This effect was supported by the result ofMANOVA (Table V) and has not been addressed in previous research; hence, thisfinding is a valuable contribution to eWOM and branding research streams. Based on theSJT (Sherif and Hovland, 1961) and the ELM (Petty and Cacioppo, 1981), we argue thatconsumers with high product involvement prefer to apply the central route rather thanthe peripheral route to elaborate eWOM content. They are more likely to assume theinformant role rather than the recommender role in the context of eWOM (Park et al.,2007) and to accept eWOM of rational appeal rather than emotional appeal. According tothe cognitive fit theory (Vessey and Galletts, 1991), this would result in enhancedconsumer attitude towards the object. The results support these theories and indicatethat consumers with a high degree of product involvement have better brand trust,brand affection, purchase intention, and overall brand attitude towards eWOM ofrational appeal than towards emotional appeal. However, under low product

Products Factors Hotelling’s trace value F (df1, df2) P

Notebooks Product involvement 0.036 0.846 (4, 95) 0.500Message appeal 0.122 2.901 (4, 95) 0.026 *

Product involvement*message appeal 0.102 2.433 (4, 95) 0.053Shampoo Product involvement 0.030 0.786 (4, 102) 0.537

Message appeal 0.085 2.236 (4, 102) 0.070Product involvement*message appeal 0.125 3.289 (4, 102) 0.014 *

Note: Significant at: *p , 0.05

Table V.MANOVA results ofproduct involvement andmessage appeal towardsbrand attitude

APJML23,4

462

Bas

eof

con

tras

tD

epen

den

tv

aria

ble

sA

pp

eals

Inv

olv

emen

tC

ontr

ast

Bra

nd

atti

tud

eB

ran

dtr

ust

Bra

nd

affe

ctio

nP

urc

has

ein

ten

tion

Rat

ion

alv

sem

otio

nal

app

eal

Hig

hD

iffe

ren

ces

0.51

50.

441

0.56

90.

597

(Rat

ion

al2

emot

ion

al)

(4.0

782

3.56

3)(3

.967

23.

526)

(4.1

212

3.55

2)(4

.227

23.

631)

SE

0.21

00.

209

0.26

40.

242

F5.

985

4.43

14.

657

6.06

5(d

f1,

df2

)(1

,52

)(1

,52

)(1

,52

)(1

,52

)P

0.01

8*

0.04

1*

0.03

6*

0.01

8*

Low

Dif

fere

nce

s0.

161

0.44

30.

264

0.03

7(R

atio

nal

2em

otio

nal

)(3

.931

23.

770)

(3.9

762

3.53

3)(3

.987

23.

722)

(4.0

172

3.98

0)S

E0.

255

0.25

70.

303

0.30

6F

0.40

02.

979

0.76

30.

014

(df1

,d

f2)

(1,

46)

(1,

46)

(1,

46)

(1,

46)

P0.

530

0.09

00.

386

0.90

5

Note:

Sig

nifi

can

tat

:* p

,0.

05

Table VI.Contrast results of brand

attitude based oninvolvement and message

appeal (notebooks)

eWOM messageappeal

463

Bas

eof

con

tras

tD

epen

den

tv

aria

ble

sA

pp

eals

Inv

olv

emen

tC

ontr

ast

Bra

nd

atti

tud

eB

ran

dtr

ust

Bra

nd

affe

ctio

nP

urc

has

ein

ten

tion

Rat

ion

alv

sem

otio

nal

app

eal

Hig

hD

iffe

ren

ces

0.75

80.

541

0.53

81.

239

(Rat

ion

al2

emot

ion

al)

(4.2

072

3.44

9)(3

.923

23.

382)

(4.0

102

3.47

2)(4

.781

23.

542)

SE

0.23

50.

238

0.26

30.

319

F10

.430

5.15

64.

188

15.0

87(d

f1,

df2

)(1

,51

)(1

,51

)(1

,51

)(1

,51

)P

0.00

2*

*0.

027

*0.

046

*0.

000

**

*

Low

Dif

fere

nce

s0.

254

0.35

50.

340

0.03

9(R

atio

nal

2em

otio

nal

)(3

.809

23.

555)

(3.7

662

3.41

1)(3

.839

23.

500)

(3.8

592

3.82

0)S

E0.

248

0.25

70.

235

0.31

2F

1.05

51.

911

2.08

80.

016

(df1

,d

f2)

(1,

54)

(1,

54)

(1,

54)

(1,

54)

P0.

309

0.17

20.

154

0.90

0

Notes:

Sig

nifi

can

tat

:* p

,0.

05,

** p

,0.

01,

**

* p,

0.00

1

Table VII.Contrast results of brandattitude based oninvolvement and messageappeal (shampoo)

APJML23,4

464

involvement, the differences are insignificant. Whether the appeal is emotional orrational, the effect on brand trust, brand affection, purchase intention, and overall brandattitude does not significantly differ. Even so, the verification process verified that undera low degree of product involvement, although rational appeal still outperformsemotional appeal on brand attitude and its sub-dimensions, the differences diminish andare insignificant. This shows that when product involvement is low, although emotionalappeal has persuasive effect, it cannot exceed the effect made by rational appeal.

5.2 Marketing implicationsThe present research indicates that irrespective of the degree of involvement,message source credibility is an important factor influencing formation of brandattitude, brand trust, brand affection, and purchase intention of eWOM, whether ornot the products are electronics goods such as notebooks or FMCG such as shampoo.Obtaining a positive eWOM effect includes increasing professionalism and reliabilityof the message source. Professionalism emphasizes the experience and knowledge ofthe message provider; reliability perceives the message provider as reliable, honest,trustworthy, and sincere to enhance credibility in the minds of consumers.

The present research provides guideline for selecting an effective message appealfor eWOM. Based on present study findings, rational appeal, rather than emotionalappeal, should be adopted to enhance overall brand attitude and its sub-dimensionsof positive eWOM under a high degree of product involvement, whether or not theproducts are electronics goods such as notebooks or convenience goods such asshampoo. The present research also indicates that under a low degree of involvement,brand attitudes of the eWOM using emotional and rational appeal do notsignificantly differ, whether or not the products are notebooks or shampoo. Thus,either appeal can be utilized to promote the brand disseminated in the eWOM.

Based on the findings of this study, Asian-branded manufacturers can promote theirproducts to global markets by utilizing eWOM, to influence or arrange the sourcecredibility and appeal type of the message. For example, marketing practitioners ofFMCG such as shampoo, or electronics goods such as notebooks, can build a web site inhost countries of target markets for consumers to share consumption experiences andfor potential customers to increase knowledge about the company via eWOM and to tryproducts.

5.3 LimitationsOne limitation of this research relates to generalisability of the findings. This researchonly examined positive eWOM effect. However, previous research argues that the effectof negative eWOM cannot be ignored (Hennig-Thurau and Walsh, 2003; Smith and Vogt,1995). Therefore, examining only positive eWOM effect may not be sufficient. In terms ofmessage source credibility, this research only supplied respondents with provideroccupation as a judgment reference. Other factors may affect eWOM message sourcecredibility. The occupation of a message sender as a proxy also limits measurement ofthe expertness dimension of message source credibility, although this seemsunavoidable in an eWOM setting. This research selected notebooks as electronicsgoods and shampoo as FMCG for the research products. Although consideringdifferences for various types of products, the results can only be confidently applied tothese products. This research used an online questionnaire as the data collection tool.

eWOM messageappeal

465

More frequent use of the internet by males caused a higher proportion of male samples.There were also many younger samples. Although this is common in internet surveys,the findings may not project to other samples such as mail interception. Finally, thisexperiment only used one eWOM. Although it reduced interaction and complexity, it didnot meet realistic eWOM settings and might therefore affect external validity of theresearch.

5.4 Future research directionsThis study offers several directions for future research. Researches can apply thisexperimental model to different products or objects to increase the generalisability of theresults. Because eWOM can be positive or negative (Andrea, 2006; Hennig-Thurau andWalsh, 2003), future research can look into the persuasion effects of negative or evenneutral eWOM and compare the result with positive eWOM. Future research can alsomanipulate factors that may affect eWOM message source credibility, such as detailedinformation of the eWOM provider, message content, webpage design, and web sitereputation to discuss the persuasion effects of eWOM message source credibility onconsumers. Variables related to consumer characteristics such as cognitive risk andproduct knowledge, should be included to examine whether they have any effects oneWOM persuasion effects. This study selected fictive brands as the stimuli to avoid theeffect of brand image. Future research could select real brands, as these reflect dailyconsumer behavior and could influence the effect of eWOM. Finally, for simplification,this research did not manipulate eWOM message source credibility. Future research canmanipulate this variable and examine whether its interaction with message appealaffects brand attitude.

References

Anderson, E.W. (1998), “Customer satisfaction and word of mouth”, Journal of Service Research,Vol. 1 No. 3, pp. 5-17.

Andrea, C.W. (2006), “Word-of-mouth and word-of-web: talking about products, talking aboutme”, Advances in Consumer Research, Vol. 33 No. 1, pp. 573-5.

Arndt, J.A. (1967), “Role of product-related conversations in the diffusion of a new product”,Journal of Marketing Research, Vol. 4, August, pp. 291-5.

Assael, H. (2004), Consumer Behavior: A Strategic Approach, Hunghton Mifflin Company,Boston, MA.

Bang, H.-K., Raymond, M.A., Taylor, C.R. and Moon, Y.S. (2005), “A comparison of servicequality dimensions conveyed in advertisements for service providers in the USA andKorea: a content analysis”, International Marketing Review, Vol. 22 No. 3, pp. 309-26.

Batra, R. and Ray, M. (1985), “Affective response mediating acceptance of advertising”, Journal ofConsumer Research, Vol. 3, pp. 234-49.

Bennett, R. and Rundle-Thiele, S.R. (2000), “Attitudinal brand loyalty – personality trait or brandspecific?”, paper presented at Australia New Zealand Marketing Educators Conference,Griffith University, Gold Coast.

Bickart, B. and Schindler, R.M. (2001), “Internet forums as influential sources of consumerinformation”, Journal of Interactive Marketing, Vol. 15 No. 3, pp. 31-40.

Brooker, G. (1981), “A comparison of the persuasive effects of mild humor and mild fear appeals”,Journal of Advertising, Vol. 10 No. 4, pp. 29-41.

APJML23,4

466

Burzynski, M.H. and Bayer, D.J. (1997), “The effect of positive and negative prior information onmotion picture appreciation”, Journal of Social Psychology, Vol. 101, pp. 215-18.

Carey, J.W. (1975), “A cultural approach to communication”, Communication, Vol. 2, December,pp. 1-22.

Chao, P. (1993), “Partioning country of origin effects: consumer evaluations of a hybrid product”,Journal of International Business Studies, Vol. 24 No. 2, pp. 291-306.

Chao, P. (2001), “The moderating effects of country of assembly, country of parts, and country ofdesign on hybrid product evaluations”, Journal of Advertising, Vol. 30 No. 4, pp. 67-81.

Chatterjee, P. (2001), “Online reviews do consumers use them?”, Advances in Consumer Research,Vol. 28 No. 4, pp. 129-33.

Chaudhuri, A. (2000), “A macro analysis of the relationship of product involvement andinformation search: the role of risk”, Journal of Marketing, Vol. 64, pp. 1-12.

Chaudhuri, A. and Holbrook, M.B. (2002), “Product-class effects on brand commitment and brandoutcomes: the role of brand trust and brand affect”, Journal of Brand Management, Vol. 10No. 1, pp. 33-58.

Chen, C.-N., Ting, S.-C., Lin, S.-Y. and Liu, J.-L. (2005), “The effects of advertising messageappeals, message framing, message source credibility, and need for cognition onadvertising performance”, Sun Yat-Sen Management Review, Vol. 13 No. 1, pp. 107-42.

Chien, Y.-W., Hsiao, C.-C., Chiu, Y.-Y. and Lu, D.-W. (2007), “The influence of messagesource-product category relevance, source credibility, and involvement on productattitudes”, Management Review, Vol. 26 No. 2, pp. 23-38.

Delgado-Ballester, E. (2004), “Applicability of a brand trust scale across product categories:a multigroup invariance analysis”, European Journal of Marketing, Vol. 38 Nos 5/6,pp. 573-92.

Dodds, W.B., Monroe, K.B. and Grewal, D. (1991), “Effect of price, brand, and store informationon buyers’ product evaluation”, Journal of Marketing Research, Vol. 28, pp. 307-19.

Duhan, D.F., Johnson, S.D., Wilcox, J.B. and Harrell, G.D. (1997), “Influences on consumer use ofword-of-mouth recommendation sources”, Journal of Academy of Marketing Science,Vol. 25 No. 4, pp. 283-95.

Eagly, A. and Chaiken, S. (1993),Psychology ofAttitudes, Harcourt Brace Jovanovich, New York, NY.

Engel, J.E., Blackwell, R.D. and Miniard, P.W. (2006), Consumer Behavior, 10th ed., HarcourtBrace Jovanovich, New York, NY.

Erdem, T. and Swait, J. (1998), “Brand equity as a signaling phenomenon”, Journal of ConsumerPsychology, Vol. 7 No. 2, pp. 131-57.

Feick, L. and Higie, R.A. (1992), “The effect of preference heterogeneity and source characteristicson ad processing and judgments about endorsers”, Journal of Advertising, Vol. 73,pp. 412-35.

Fishbein, M. and Ajzen, I. (1980), Understanding Attitudes and Predicting Social Behavior,Prentice-Hall, Englewood Cliffs, NJ.

Godes, D. and Mayzlin, D. (2004), “Using online conversations to study word-of-mouthcommunication”, Marketing Science, Vol. 23 No. 4, pp. 545-60.

Golan, G.J. and Zaidner, L. (2008), “Creative strategies in viral advertising: an application ofTaylor’s six-segment message strategy wheel”, Journal of Computer-MediatedCommunication, Vol. 13, pp. 959-72.

Goldsmith, R.E. and Emmert, J. (1991), “Measuring product category involvement:A multitrait-multimethod study”, Journal of Business Research, Vol. 23, pp. 363-71.

eWOM messageappeal

467

Gruen, T.W., Osmonbekov, T. and Czaplewski, A.J. (2006), “eWOM: the impact ofcustomer-to-customer know-how exchange on customer value and loyalty”, Journal ofBusiness Research, Vol. 59 No. 4, pp. 449-56.

Guiltinan, J.P., Gordon, W.P. and Madden, T.J. (1997), Marketing Management Strategies andPrograms, 6th ed., McGraw-Hill, New York, NY.

Gunther, A.C. (1992), “Biased press or biased public: attitudes toward media coverage of socialgroups”, Public Opinion Quarterly, Vol. 56 No. 2, pp. 147-67.

Harrison-Walker, L.J. (2001), “The measurement of word-of-mouth communication and aninvestigation of service quality and customer commitment as potential antecedents”,Journal of Service Research, Vol. 4 No. 1, pp. 60-75.

Hawkins, D.I., Best, R.J. and Koch, E.C. (2007), Consumer Behavior: Building Marketing Strategy,10th ed., McGraw-Hill, New York, NY.

Hennig-Thurau, T. and Walsh, G. (2003), “Electronic word-of-month: motives for andconsequence of reading customer articulation on the internet”, International Journal ofElectric Commerce, Vol. 8 No. 2, pp. 51-74.

Hennig-Thurau, T., Gwinner, K.P., Walsh, G. and Gremler, D.D. (2004), “Electronicword-of-mouth via consumer-opinion platform: what motivates consumers to articulatethemselves on the internet?”, Journal of Interactive Marketing, Vol. 18, pp. 38-52.

Herr, P.M., Kardes, F.R. and Kim, J. (1991), “Effects of word-of-mouth and product-attributeinformation on persuasion: an accessibility-diagnosticity perspective”, Journal ofConsumer Research, Vol. 17 No. 4, pp. 454-62.

Hetsroni, A. (2000), “The relationship between values and appeals in Israeli advertising:a smallest space analysis”, Journal of Advertising, Vol. 29 No. 3, pp. 55-68.

Hotchkiss, G.B. (1949), Advertising Copy, 3rd ed., Harper, New York, NY.

Hovland, C.I. and Weiss, W. (1951), “The influence of source credibility on communicationeffectiveness”, Public Opinion Quarterly, Vol. 15 No. 4, pp. 635-50.

Howard, D.J. and Gengler, C.E. (2001), “Emotional contagion effects on product attitudes”,Journal of Consumer Research, Vol. 28, pp. 189-201.

Hwang, J.-S., McMillan, S.J. and Lee, G. (2003), “Corporate web sites as advertising: an analysis offunction, audience and message strategy”, Journal of Interactive Advertising, Vol. 3 No. 2,pp. 10-23.

Kiecker, P. and Cowles, D. (2001), “Interpersonal communication and personal influence on theinternet: a framework for examining online word-of-mouth”, Journal of Euro-Marketing,Vol. 11 No. 2, pp. 71-88.

Kotler, P. and Keller, K.L. (2008), Marketing Management, 13th ed., Prentice-Hall, Upper SaddleRiver, NJ.

Lampert, S.I. and Jaffe, E.D. (1996), “Country of origin effects on international market entry”,Journal of Global Marketing, Vol. 10 No. 2, pp. 27-52.

Laskey, H.A., Day, E. and Crask, M.R. (1989), “Typology of main message strategies”, Journal ofAdvertising, Vol. 18 No. 1, pp. 36-41.

Laurent, G. and Kapferer, J.N. (1985), “Measuring consumer involvement profiles”, Journal ofMarketing Research, Vol. 22, February, pp. 41-53.

Ling, I.-L. and Liu, Y.-F. (2008), “Comprehension and persuasion on advertising message:heuristic-systematic model approach”, Journal of Management, Vol. 25 No. 5, pp. 487-503.

APJML23,4

468

Lutz, R.J., MacKenzie, S.B. and Belch, G.E. (1986), “The role of attitude toward the ad asa mediator of advertising effectiveness: a test of competing explanations”, Journal ofMarketing Research, Vol. 8 No. 1, pp. 130-43.

McGuire, W.J. (1976), “Some internal psychological factors influencing consumer choice”, Journalof Consumer Research, Vol. 2, March, pp. 302-19.

MacKenzie, S.B. and Spreng, R.A. (1992), “How does motivation moderate the impact of centraland peripheral processing on brand attitudes and intentions?”, Journal of ConsumerResearch, Vol. 18, pp. 519-28.

Mizerski, R.W. (1982), “An attribution explanation of the disproportionate influence ofunfavorable information”, Journal of Consumer Research, Vol. 9 No. 3, pp. 301-10.

Mortimer, K. (2002), “Integrating advertising theories with conceptual models of servicesadvertising”, Journal of Services Marketing, Vol. 16 No. 5, pp. 460-8.

Muray, K.B. and Vogel, C.M. (1997), “Using a hierarchy-of-effects approach to gauge theeffectiveness of corporate social responsibility to generate goodwill toward the firm:financial versus nonfinancial impacts”, Journal of Business Research, Vol. 38, pp. 141-59.

Nunnally, J. (1978), Psychometric Theory, 2nd ed., McGraw-Hill, New York, NY.

Ohanian, R. (1990), “Construction and validation of a scale to measure celebrity endorsers’perceived expertise, trustworthiness, and attractiveness”, Journal of Advertising, Vol. 19,pp. 39-52.

Osgood, C.E., Suci, G.J. and Tannenbaum, P.H. (1957), The Measurement of Meaning, Universityof Illinois Press, Urbana, IL.

Park, C. and Lee, T.M. (2009), “Information direction, website reputation and eWOM effect:a moderating role of product type”, Journal of Business Research, Vol. 62, pp. 61-7.

Park, D.-H. and Kim, S. (2008), “The effects of consumer knowledge on message processing ofelectronic word-of-mouth via online consumer reviews”, Electronic Commerce Researchand Applications, Vol. 7, pp. 399-410.

Park, D.-H. and Lee, J. (2008), “eWOM overload and its effect on consumer behavioral intentiondepending on consumer involvement”, Electronic Commerce Research and Applications,Vol. 7, pp. 386-98.

Park, D.-H., Lee, J. and Han, I. (2007), “The effects of on-line consumer reviews on consumerpurchasing intention: the moderating role of involvement”, International Journal ofElectronic Commerce, Vol. 11 No. 4, pp. 125-48.

Petty, R.E. and Cacioppo, J.T. (1981), Attitude and Persuasion: Classic and ContemporaryApproaches, William C. Brown, Dubuque, IA.

Puto, C.P. and Wells, W.D. (1984), “Informational and transformational advertising:the differential effects of time”, Advances in Consumer Research, Vol. 11, pp. 638-43.

Ratchford, B.T. (1987), “New insights about the FCB grid”, Journal of Advertising Research,August/September, pp. 24-37.

Reichheld, F.F. and Sasser, W.E. Jr (1990), “Zero defections: quality comes to services”, HarvardBusiness Review, Vol. 68, September/October, pp. 105-11.

Richins, M. (1983), “Negative word-of-mouth by dissatisfied consumers: a pilot study”, Journal ofMarketing, Vol. 47 No. 1, pp. 68-78.

Roberson, T.S. (1976), “Low-commitment consumer behavior”, Journal of Advertising Research,Vol. 16, pp. 19-24.

Rosenberg, M.J. and Hanland, J.C. (1960), “Low-commitment consumer behavior”, Journalof Abnormal and Social Psychology, Vol. 2, pp. 367-72.

eWOM messageappeal

469

Rossiter, J.R., Percy, L. and Donovan, R.J. (1991), “A better advertising planning grid”, Journal ofAdvertising Research, October/November, pp. 11-21.

Schiffman, L.G. and Kanuk, L.L. (2006), Consumer Behavior, 9th ed., Prentice-Hall, New York, NY.

Sherif, M. and Hovland, C.I. (1961), Social Judgment: Assimilation and Contrast Effects inCommunication and Attitude Change, Yale University Press, New Haven, CT.

Simon, J.L. (1971), The Management of Advertising, 1st ed., Prentice-Hall, Engelwood Cliffs, NJ.

Smith, R.E. and Vogt, C.A. (1995), “The effects of integrating advertising and negativeword-of-mouth communications on message processing and response”, Journal ofConsumer Psychology, Vol. 4 No. 2, pp. 133-51.

Still, R.R., Barnes, J.H. and Kooyman, M.E. (1984), “Word-of-mouth communication in low-riskproduct decisions”, International Journal of Advertising, Vol. 3 No. 4, pp. 335-45.

Syvia, J.K. and Chan-Olmsted, S.M. (2005), “Comparative effects of organization-publicrelationships and product-related attributes on brand attitude”, Journal of MarketingCommunications, Vol. 11 No. 3, pp. 145-70.

Taylor, R.E. (1999), “A six-segment message strategy wheel”, Journal of Advertising Research,Vol. 39 No. 6, pp. 7-17.

Vakratsas, D. and Ambler, T. (1999), “How advertising works: what do we really know?”, Journalof Marketing, Vol. 63 No. 1, pp. 26-43.

Vaughn, R. (1986), “How advertising works: a planning model revisited”, Journal of AdvertisingResearch, February/March, pp. 57-65.

Veale, R. and Quester, P. (2009), “Tasting quality: the roles of intrinsic and extrinsic cues”,Asia Pacific Journal of Marketing and Logistics, Vol. 21 No. 1, pp. 195-207.

Vessey, I. and Galletts, D. (1991), “Cognitive fit: an empirical study of information acquisition”,Information System Research, Vol. 2 No. 1, pp. 63-84.

Voss, K.E., Spangenberg, E.R. and Grohmann, B. (2003), “Measuring the hedonic andutilitarian dimensions of consumer attitude”, Journal of Marketing Research, Vol. 40 No. 3,pp. 310-20.

Wang, C.-C. and Lee, H.-Y. (2005), “The effects of source credibility, affection, and involvementin reducing the belief of internet rumors”, Journal of Management, Vol. 22 No. 3,pp. 391-413.

Wangenheim, F.V. and Bayon, T. (2004), “The effect of word of mouth on services switching:measurement and moderating variables”, European Journal of Marketing, Vol. 38 Nos 9/10,pp. 1173-85.

Wells, W.D. (1980), How Advertising Works, Needham Harper Worldwide, Chicago, IL.

West, D. (1994), “Validating a scale for the measurement of credibility: a covariance structuremodeling approach”, Journalism Quarterly, Vol. 71 No. 1, pp. 159-68.

Wilkie, W. (1986), Consumer Behavior, Wiley, New York, NY.

Zaichkowsky, J.L. (1985), “Measuring the involvement construct”, Journal of Consumer Research,Vol. 12 No. 3, pp. 341-53.

Zaichkowsky, J.L. (1994), “The personal involvement inventory: reduction, revision, andapplication to advertising”, Journal of Advertising, Vol. 23 No. 4, pp. 59-70.

Zhang, Y. and Buda, R. (1999), “Moderating effects of need for cognition on responses topositively versus negatively framed advertising messages”, Journal of Advertising, Vol. 28No. 2, pp. 1-15.

APJML23,4

470

Further reading

Lin, C.-W. and Song, J.-F. (2008), “Consumer response of price-matching refund policies: effect ofperceived credibility”, NTU Management Review, Vol. 18 No. 2, pp. 197-228.

Appendix. Measurements, reliability, and mean of variablesMeasures used in the main studyConstruct (source) (Cronbach’s a and mean: 1 ¼ notebooks, 2 ¼ shampoo).

Message appeal (Kotler and Keller, 2008) (rational appeal: M1 ¼ 7.64, M2 ¼ 8.15; emotionalappeal: M1 ¼ 5.85, M2 ¼ 6.42).

Product involvement (Zaichkowsky, 1994) (a1 ¼ 0.866, a2 ¼ 0.751; M1 ¼ 4.88, M2 ¼ 4.42):. important-unimportant;. boring-interesting;. relevant-irrelevant;. exciting-unexciting;. means nothing-mean a lot to me;. appealing-unappealing;. fascinating-mundane;. worthless-valuable;. involving-uninvolving; and. not needed-needed.

Message source credibility (Ohanian, 1990) (a1 ¼ 0.907, a2 ¼ 0.896; M1 ¼ 4.38, M2 ¼ 3.90).

(1) Expertness:. expert-not an expert;. experience-inexperience;. knowledgeable-unknowledgeable;. qualified-unqualified; and. skilled-unskilled.

(2) Trustworthiness:. dependable-undependable;. honest-dishonest;. reliable-unreliable;. sincere-insincere; and. trustworthy-untrustworthy.

Brand attitude (a1 ¼ 0.954, a2 ¼ 0.956; M1 ¼ 3.83, M2 ¼ 3.81).

(1) Brand trust (Delgado-Ballester, 2004) (a1 ¼ 0.913, a2 ¼ 0.921; M1 ¼ 3.75, M2 ¼ 3.65):. This is a brand that meets my expectations.. I feel confidence in this brand.. This is a brand that will not disappoint me.. This brand guarantees satisfaction.. This brand would be honest and sincere in addressing my concerns.. I could rely on this brand to solve the problem.

eWOM messageappeal

471

. This brand would make any effort to satisfy me.

. This brand would compensate me in some way for the problem with the product.

(2) Brand affection (Chaudhuri and Holbrook, 2002) (a1 ¼ 0.927, a2 ¼ 0.888; M1 ¼ 3.83,M2 ¼ 3.76):. I feel good if I purchase this brand.. This brand makes me happy.. This brand gives me pleasure.

(3) Purchase intention (Dodds et al., 1991) (a1 ¼ 0.934, a2 ¼ 0.965; M1 ¼ 3.95, M2 ¼ 4.08):. The likelihood of purchasing this brand is high.. If I will go into buy this product, I would consider buying this brand.. At the brand, I would consider buying the product.. The probability that I would consider buying the brand is high.. My willingness to buy the brand is high.

About the authorsPaul C.S. Wu is an Associate Professor of Marketing in the Department of BusinessAdministration, Aletheia University, Taiwan. His main research interests are in the area of brandequity and brand extensions, service failures and recovery strategies, and private label brands. Hispapers have been published in Journal of Industrial Marketing and Sun Yat-Sen ManagementReview. Paul C.S. Wu is the corresponding author and can be contacted at: wucs@email.au.edu.tw

Yun-Chen Wang is a Graduate Student at the Graduate School of Management Sciences,Aletheia University, Taiwan.

APJML23,4

472

To purchase reprints of this article please e-mail: reprints@emeraldinsight.comOr visit our web site for further details: www.emeraldinsight.com/reprints