Sunteți pe pagina 1din 21

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0309-0566.

htm

Making your online voice loud: the critical role of WOM information
Minxue Huang
Economics and Management School, Wuhan University, Wuhan, Peoples Republic of China

Making your online voice loud 1277


Received August 2008 Revised February 2009 Accepted October 2009

Fengyan Cai
Marketing Department, Antai College of Economics and Management, Shanghai Jiao Tong University, Shanghai, Peoples Republic of China

Alex S.L. Tsang


Department of Marketing, Hong Kong Baptist University, Hong Kong, Peoples Republic of China, and

Nan Zhou
Department of Marketing, City University of Hong Kong, Hong Kong, Peoples Republic of China
Abstract
Purpose This paper seeks to explore factors that can help generate a bigger ripple for diffusive Word-of-Mouth (WOM) in an online environment. Specically, the paper aims to demonstrate how the characteristics of WOM information inuence the ripple effect. Design/methodology/approach The paper used a survey based on real posts in online discussion forums to collect the data. Structural Equation Modelling (SEM) was utilized for analysis. Findings WOM as diffusive communication could generate a ripple effect. All of the WOM characteristics, which were investigated (quality, authority, authenticity and interestingness), have a positive effect on resenders acceptance toward WOM, which, in turn, have a signicant positive impact on resenders resending intention. Furthermore, the positive impacts of WOM characteristics on resenders acceptance exist, regardless of whether the valence of WOM information is positive or negative. Research limitations/implications This study provides insights into the role of WOM information in facilitating the WOM ripple effect. However, the specic characteristics of WOM information that inuence consumers perceptions of WOM are not explored in the present research. Originality/value The results of the study may help practitioners manipulate and use online WOM information in order to make marketing communications more efcient and inuential. Keywords Word of mouth, Diffusive communication, Online environments, Information characteristics, Internet marketing, Marketing communications Paper type Research paper

The authors would like to thank the Editor and the two anonymous reviewers for their helpful comments and suggestions. This study was funded by the National Natural Science Foundation of China (No. 70672067: 70972091).

European Journal of Marketing Vol. 45 No. 7/8, 2011 pp. 1277-1297 q Emerald Group Publishing Limited 0309-0566 DOI 10.1108/03090561111137714

EJM 45,7/8

1278

1. Introduction Word of Mouth (WOM) is informal advice between people about goods, services, and social issues (East et al., 2007). Faced with a rapid decline of consumer trust in traditional advertising, companies are looking for different ways to promote products; consequently, word-of-mouth marketing has gained ground (Verlegh and Moldovan, 2008). As a powerful communication tool, WOM could have two distinctive functions persuasion and the ripple effect caused by WOM diffusion (Weening and Midden, 1991; Hogan et al., 2004; De Bruyn and Lilien, 2004). Most studies in the extant literature have treated WOM as persuasive communication and focused only on the persuasive effect, but these studies have seldom treated WOM as diffusive communication (Weening and Midden, 1991). The diffusion of WOM helps generate a ripple effect for marketing activities, which is also called BUZZ marketing or viral marketing (Phelps et al., 2004; Verlegh and Moldovan, 2008). The ripple effect has been identied as being able to signicantly multiply or extend the effectiveness of advertising (Hogan et al., 2004). Beyond knowing about the existence of the ripple effect, we also need to understand how to generate a bigger ripple effect for marketing activities. This paper aims to provide insights into the generation of a bigger ripple effect for diffusive WOM in an online environment. We selected an online environment for this study based on the following reasons. First, although WOM is an alternative for advertising, it is hard to manage as marketers cannot control consumers oral communication content in an ofine environment. The internet, in which most of the information is in written format, makes it possible to track, copy and analyze WOM content (Godes and Mayzlin, 2004). Therefore, marketers could have more control over WOM in an online environment than in an ofine environment. Second, as shown by Hoffman and Novak (1996), online WOM communication, like WOM in an ofine environment, could be one to one (e.g. e-mail, voice e-mail, etc.) or one to many (e.g. group e-mail). As a new form of interpersonal inuence, online WOM could be a many to many communication process (e.g. online discussion forums) in which the sources and receivers may be unknown to each other. This provides a chance for companies to manage the WOM information. Furthermore, the ripple effect in essence is driven by social interaction among people. In online discussion forums, many-to-many communication signicantly enlarges the scope of social interaction, consequently making the ripple effect bigger. Theoretically, relative to an ofine environment, online discussion forums can provide not only a better environment for social interaction but also a better opportunity for marketers to manage WOM information. In practice, spending on online social networks (e.g. discussion forums) advertising is already at $280 million (about 2 per cent of all online advertising spending) and expected to reach about $2 billion (6 per cent of all online advertising) in 2010 (Hartmann et al., 2008). Therefore, in this paper we ground our study in the context of online discussion forums. This paper suggests that WOM information is very important in order to generate a big ripple effect. Intuitively, the longer the WOM chain, the bigger the ripple effect becomes. The resender, who at rst gets information as a receiver and then transmits it as a sender, plays a critical role in lengthening the chain of diffusive WOM. As shown in the theoretical model of WOM behaviour proposed by Hartmann et al. (2008), WOM information sent by another agent can inuence the focal agents WOM behaviour. As one node in the diffusive WOM chain, the resenders acceptance and resending behaviours could also be inuenced by WOM information. Thus, we expect that

increasing the attractiveness of WOM information is one important way to enlarge the ripple effect of diffusive WOM. In the context of online discussion forums, while the width of the WOM chain depends on the number of its members, the length of the WOM chain depends on the members (here as resender) intention to share/resend the WOM information with/to other discussion forum members. The stronger the resenders willingness to share WOM information with other members, or with other discussion forums, the longer the WOM chain becomes, and the wider the ripple effect of the WOM. Therefore, the more important consideration in widening the ripple effect of online WOM is to understand the factors that inuence members resending intention. In sum, we argue that marketers could generate a big ripple effect for their marketing campaigns by using attractive WOM information to increase resenders acceptance and resending behaviour. This paper is organized as follows: . the notion and importance of ripple effect are discussed, and an introduction of the research context online discussion forums is presented; . a presentation of the proposed synthesized model and the development of three hypotheses according to the proposed model; . a presentation of the methods used to test the hypotheses and a discussion of the results; and . conclusions and limitations of the study are presented as well as directions for future research. 2. Theoretical background 2.1 Ripple effect, resender, and social interaction As an important but hard to manage market force, WOM is believed to be able to complement and extend the effect of advertising (Herr et al., 1991; Hogan et al., 2004). For example, initial marketing activities (advertisement, promotion) trigger initial purchase reactions, and that purchase experience subsequently triggers the spread of WOM, as customers share their experiences with others (Chevalier and Mayzlin, 2006). The extent to which the effect of advertising is extended (or multiplied) by WOM is called advertising ripple effect caused by WOM (Hogan et al., 2004). Another concept, similar to ripple effect, is called the spillover effect of social interaction, which arises when a marketing activity directed toward an agent affects the behaviour of others in the agents group via social interaction (Hartmann et al., 2008). Since WOM, in essence, is a kind of social interaction, the spillover effect of social interaction could be regarded as a ripple effect of WOM. In this paper, we ground ripple effect in the context of WOM rather than advertising, and dene it as the extent to which a piece of WOM information is transmitted by another round of WOM (diffusive WOM). Specically, the ripple effect of WOM is operationalized as the extent to which resenders would like to resend the WOM information, which means that a higher resending intention indicates a bigger ripple effect. The ripple effect is of primary importance to marketers and policy-makers since it allows a stimulus intended for one individual to be magnied by its dispersion through a network. From a normative point of view, the ripple effect of WOM may vastly increase the return-on-investment and, hence, becomes a signicant interest for rms. Scholars have theoretically and empirically demonstrated that WOM can inuence consumer behaviour and, consequently, increase the effectiveness of advertisements

Making your online voice loud 1279

EJM 45,7/8

1280

(Mayzlin, 2006; Hogan et al., 2004). For example, using a customer lifetime value modelling approach, Hogan et al. (2004) found that multiplier coefcients could be three or even greater in common situations. Furthermore, as aforementioned, rms are spending greater amounts of money on social network advertising in online environments (Hartmann et al., 2008). Given the importance of the ripple effect to marketers, widening the ripple effect of WOM is one area of great interest to marketers. In this paper, we try to identify a method for widening the ripple effect by analyzing diffusive WOM mechanism (see Figure 1). As shown in Figure 1, there are two ways to widen the ripple effect. On one hand, customer A could share his/her experience with as many people as possible widening the width of the WOM chain. On the other hand, there could be more nodes in the WOM chain lengthening the WOM chain. As shown in Figure 1, receivers of WOM information could be divided into two groups: resenders (e.g. C, D, E, and F) and pure receivers (e.g. B). Resenders are receivers who search for and get information via one WOM activity then resend it to others via a different WOM activity. A resender could play both roles of receiver and sender someone who rst gets information from a sender and then transmits it as a sender. Pure receivers are receivers who actively search for and receive information, but pure receivers do not retransmit the information as senders. Resenders are not only important for increasing the width of the ripple effect of WOM, but also important for lengthening the chain of WOM. Therefore, exploring factors that inuence consumers resending behaviour is very important for expanding the ripple effect of WOM. Existing literature has explored many factors that inuence consumers WOM behaviour, such as satisfaction, self-condence, product involvement and social presence (Richins, 1983; Lau and Ng, 2001; De Bruyn and Lilien, 2004; Biyalogorsky et al., 2001). According to the theoretical model proposed by Hartmann et al. (2008) WOM information per se is one of the important factors that inuence consumers WOM behaviour. However, except for the valence of information, factors relating to the information per se are rarely considered as saving, and analyzing WOM information in ofine oral communications is a difcult process (Godes and Mayzlin, 2004). The internet makes this possible since most online information is in the written format. As will be explained later, the online environment not only provides a suitable context for us to study WOM information itself, but also makes the information more important. In ofine WOM, senders and receivers know each other and communicate face to face, receivers acceptance towards WOM information could be inuenced by the senders

Figure 1. General ow model of WOM information

individual characteristics, such as authority, expertise and strength of ties (Feick and Price, 1987; Frenzen and Nakamoto, 1993). Since people are usually anonymous in an online environment, the senders individual characteristics are not available, or need to be inferred from the WOM information. Therefore, in an online environment, receivers have to put more weight on WOM information per se when they evaluate one piece of WOM information. Because WOM information in an online environment is quite important, we will explore the characteristics of online WOM information that can enhance peoples acceptance and resending intention. We will elaborate this point in section 3. 2.2 Online WOM: a resenders territory As mentioned previously, online discussion forums are the context of the present study. In this section, we present the major differences between online and ofine WOM in terms of their characteristics and their inuence on the ripple effect (see Table I). Then, on the basis of this comparison, we further discuss the reasons why an online environment was chosen for this study. Ofine WOM is dened as face-to-face communication between participants (Arndt, 1967, 1968). Ofine WOM requires resenders and receivers to interact in real time,
Online WOM Ofine WOM Characteristics Communication medium Form Synchronicity Type of interaction Format Relationship between sender and receiver Ease of transmission Consequence Ripple effect Focus Critical role Talk, telephone, meeting (letter) Oral (written) communication Synchronous communication One-to-one/many diffusion Many-to-many diffusion Discussion forums, blog Written communication Could be asynchronous/ synchronous communication Virtual interaction/ indirect Non-linear communication Generally anonymous virtual social ties/bigger receiver pool Easy to transmit/ forward High ripple/multiplier effect (WOM ow) Focus on both persuasive and diffusive communication Resenders as critical role

Making your online voice loud 1281

E-mail, text chatting (voice chatting) Written (oral) communication Could be asynchronous/ synchronous communication Face-to-face interaction/ Virtual interaction/ direct indirect Linear or non-linear Mostly linear communication communication Know each other Know each other/real (anonymous)/real social ties/limited (virtual) social ties/ receiver pool bigger receiver pool Difcult to transmit Easy to transmit/ forward Low ripple/multiplier effect (isolated WOM) Focus on persuasive communication Opinion leaders as critical role Middle ripple/multiplier effect (WOM ow) Focus on persuasive communication Opinion leaders as critical role

Source: Hoffman and Novak (1996), Table II (p. 56)

Table I. Ofine WOM and online WOM

EJM 45,7/8

1282

which is termed as temporal synchronicity by Hoffman and Novak (1996). These resenders and receivers must know each other, must have some form of social ties and must exchange information using verbal communication. Furthermore, after receiving information, resenders in ofine WOM have to remember the information and then transmit it to others. Thus, the ofine ow path is difcult to analyze, and hence the resenders role is rarely considered. Extant literature considers communication via discussion forums, e-mail and online chatting as a kind of WOM (Dellarocas, 2003; Hennig-Thurau et al., 2004). The basic difference between ofine and online WOM is that online WOM uses the written format (Bickart and Schindler, 2001). Furthermore, it is easier to process (e.g. edit, copy, and paste) online written information than oral information. As such, the online environment makes it possible to study resenders behaviour and the ripple effect. Furthermore, ofine WOM is a form of linear communication because each participant is thought of as being a singular entity, and the sequence of the WOM is chronological (Greimas, 1971). Compared with a linearly constrained ofine WOM model, online WOM could be nonlinear communication. This is due to its interactive character (e.g. directed to multiple individuals and available to other consumers for an indenite period of time) and it treats its audience as co-creators (Gumpert and Cathcart, 1986; Hoffman and Novak, 1996). The interactivity, and nonlinearity of the internet, may widen the ripple effect of WOM, and make it more efcient. In sum, in online discussion forums, many to many communication is asynchronous (interaction among participants does not occur in real time), anonymous, and written information-based, which could signicantly widen the scope of social interaction and, consequently, make the ripple effect bigger. Relative to ofine environments, online discussion forums can provide not only a better environment for social interaction but also a better opportunity for marketers to manage WOM information. Thus, this paper focuses on online WOM and examines hypotheses in the setting of online discussion forums. 3. Model and hypotheses 3.1 Model Researchers have developed a variety of information-processing and response models in their efforts to understand how consumers process and respond to product-related information. Classical models include Strongs (1925) AIDA Model (attention, interest, desire, action) and Rogers (1995) Innovation-Adoption Model (awareness, interest, evaluation, trial adoption). These models conclude that message communication is a persuasive process and inuences a receivers attitude and behaviour. McGuire (2001) identied ve stages of information processing: exposure, attention, comprehension/ evaluation, yielding/acceptance and retention/diffusion without being concerned with the persuasive process. Considering WOM to be diffusive communication, we employed McGuires model to indicate the antecedents that inuence resenders resending intention (see Figure 2). We follow Gershoff et al. (2003) and dene acceptance as to what extent a resender believes the WOM information is true and is likely to accept the advice proposed in the WOM information. Similarly, resending intention refers to the extent to which resenders would like to resend the information they have received to another audience, including people in both online and ofine

environments. As shown in Figure 2, WOM informations characteristics can inuence a resenders acceptance, which in turn, has an impact on his or her resending intention. In an online WOM ow, a resender rst searches for information by using a search engine or familiar web sites (exposure, attention) and then considers whether the information at hand is valuable enough to satisfy not only his or her own needs but also a potential recipients needs (evaluation). As there are less external cues for people to judge WOM information in anonymous online environments, WOM information, per se, gets more important in online environments than in ofine environments. Although WOM information is of primary importance in online environments, studies about online WOM information are limited. Riehs (2002) did a qualitative study to explore the quality and authority of general online information. In Riehs (2002) study, information quality was operationally dened as the extent to which users thought that the information was useful, good, current, and accurate. For the authority of information, Rieh (2002) dened it as the extent to which users thought that the information was credible, trustworthy, reliable, ofcial, authoritative and amateurish. As suggested by Mitra and Watts (2002), the online voice represented three aspects of communication: who is saying, what is being said and how is it said. Consequently, several standards could be used to evaluate an online voice. Among these standards, the standard of authority was used to evaluate the speaker who is saying, while the standard of authenticity was used to evaluate what is being said and how. If we take a closer look at the items mentioned in the operational denition of information authority in Rieh (2002), we may nd that the rst three items (credible, trustworthy and reliable) focus on what is being said and how whereas the last three items (scholarly, ofcial and amateurish) focus on who is saying. Although it is not explicitly extracted, interestingness is another important factor mentioned by Rieh (2002). As the internet provides a vast amount of information, interestingness of information is of primary importance for drawing users attention in online environments (Rieh, 2002). On the basis of these arguments, we propose four dimensions of WOM information (quality, authenticity, authority, interestingness) that could inuence the acceptance of the WOM information (see Figure 2). After accepting (yielding to) the information, a resender may recall the information (retention) or share it with others (diffusion). Therefore, we further propose that these four dimensions would inuence a resenders acceptance which in turn inuences the resenders resending intention. In the following section, we will explain the relationships between each of the four dimensions and resenders acceptance.

Making your online voice loud 1283

Figure 2. Online WOM transmission process

EJM 45,7/8

1284

3.2 Hypotheses According to the Elaboration Likelihood Model proposed by Petty et al. (1983), the quality of an argument is an important factor in determining the persuasive impact of a message. They also found that argument quality inuences a consumers attitude towards a product. Specically, consumers are likely to have a more favourable attitude towards a product when the product-related information has a high quality (strong) argument rather than a low (weak) one. In online environments, Rieh (2002) found that information quality was the most important antecedent for online users when searching for and reading a web site. Gershoff et al. (2003) also demonstrate that web users decided to accept online information by judging its quality. Based on these arguments, we derive H1a: H1a. A potential resenders acceptance level towards online WOM information is positively inuenced by the perceived quality of that information.

Ofine WOM is inuential and believable because receivers know speakers. Compared with ofine WOM, online WOM may be an anonymous and indirectly written communication. Participants need to judge whether the information is believable because they may not know who posted the information. Therefore, Rieh (2002) proposed that another important factor, cognitive authority, could affect online users in their searching for and reading web site information. He found six facets of authority: credible, trustworthy and reliable, scholarly, ofcial and amateurish. As suggested by Mitra and Watts (2002), we divided these six items into two factors. The former one (credible, trustworthy and reliable) refers to authenticity in the sense that the information is true and credible and the latter one (scholarly, ofcial and amateurish) refers to the general authority which means that the information appears to be from an authoritative source. Mitra and Watts (2002) argued that authenticity was one of the most important factors that inuence whether a reader believed online information. Other extant literature also suggests that authority makes consumers trust the information more, and that a senders expertise is one of the most important factors that inuence a receivers acceptance (Bansal and Voyer, 2000). Based on this, we propose H1b and H1c: H1b. H1c. A potential resenders acceptance level towards online WOM information is positively inuenced by its perceived authenticity. A potential resenders acceptance level towards online WOM information is positively inuenced by its perceived authority.

Interestingness (see Table II) is another important factor that affects a resenders evaluation of online WOM information. Since there is so much information available on the internet (e.g. google.com searches more than 4 billion pages by 2007), it is important to attract a resenders attention to WOM information. Riehs (2002) showed that topical interest plays the most important role in attracting attention (more than 40% of online users rate it as a priority). Furthermore, relevant advertising studies (e.g. Olney et al., 1991) have proven that the perceived interestingness of an advertisement directly affects its audiences viewing time. Derbaix and Vanhamme (2003) found that eliciting surprise helps induce WOM activity. Therefore, we predict that interesting online WOM information, would attract more attention, and would more likely be accepted by consumers. This suggests H1d: H1d. A potential resenders acceptance level of online WOM information is positively inuenced by its perceived interestingness.

Dimensions Quality

Items

Source Information quality (Rieh, 2002)

Good: badb *(c21), goodb (c23), excellent, the state of art, cool Accurate: righta (c25), accurate, correct Current: out-of-dateb *(c24), newb (c26), current, up-to-date Useful: informativea (c22), usefula (c27), uselessb * (c29), helpful Important: importanta (c28) Authenticity Credible: crediblea (c31) Trustworthy: trustworthya (c32), I trust, I believe in, seems real, condence that is true, faith in quality Reliable: suspecta *(c33), Reliableb (c34), Reliably done Authority Scholarly: amateurisha *(c35), professional, scholarly, academic Ofcial: ofciala (c36) Authoritative: authoritativea (c37) Interestingness Topical interest: excitinga (c11), dull *a (c13), affectingb (c16) Interestingness: interesta (c12), funnyb (c14), boringb *(c15)

Making your online voice loud 1285

Cognitive authority (Rieh, 2002)

Rieh (2002); Olney et al. (1991)

Note: adenotes the nal scale item; bdenotes the deleted scale item; *denotes reverse item; codes in parentheses denote item number

Table II. Dimensions of information evaluation

As discussed previously, WOM is essentially a kind of social interaction. Though online WOM takes a written format (communication through written messages) as opposed to an oral format like ofine WOM (e.g. online chat), it is in essence a kind of virtual conversation and has similar functions as ofine social interaction. Actually, some scholars have identied that some resenders use the internet as a means of effective interpersonal communication for building and keeping interpersonal relationships. For example, Zinkhan et al. (2003) found companionship was one of the top ve reasons for web-based chatting. Hennig-Thurau et al. (2004) also found that social benet was one of the eight motives for why consumers communicate online. According to Higginss (1981) ndings, there are some implicit communication norms that guide peoples daily conversations, such as being informative, trustful, relevant, polite and modest. As online WOM is in essence a conversation, it must also follow similar communication norms (termed Netiquette a combination of network and etiquette) (Scheuermann and Taylor, 1997). Therefore, when a resender in an online discussion forum plans to resend some relevant information, he/she should evaluate it rst. If the information is acceptable and it has a low risk of offending anyone, i.e. it follows the rules of Netiquette, then the resender would transmit the information. Based on these arguments, we hypothesize the following: H2. The level of a potential resenders intention to resend WOM information is positively inuenced by his or her acceptance level.

EJM 45,7/8

1286

4. Methodology 4.1 Study overview To test our hypotheses in the context of a discussion forum, we not only need to collect data about participants individual characteristics, but also simulate the context of a discussion forum. Therefore, rst, we used a paper-pencil-based survey to get participants individual information. We then embedded a real online post in a simulated context of a discussion forum and nally obtained the participants reactions to the post. In order to simulate the online WOM process as naturally as possible, we did not provide any incentives to encourage participation. We conducted two pilot studies. The rst one was to select real online posts (see pilot study 1). The second pilot study was designed to develop and purify the measurement of the characteristics of WOM information, as no related measurement is available in extant literature (see pilot study 2). After these two pilot studies, we conducted a two-step formal study to test our hypotheses. 4.2 Pilot study 1: choosing posts To simulate an online environment as realistically as possible, we used the observation method to build a pool of popular posts. First, we searched for online posts in popular Chinese C2C communities (taobao.com and ebay.com.cn) and discussion forums of information web sites (mop.com, bbs.sina.com.cn and bbs.sohu.com). Second, using search engines, such as google.com and baidu.com, we checked the dispersion of each post by computing the number of webpages that cited the post. We selected the top 100 popular posts at that time (late 2006) according to the number of webpages that cite the post. After eliminating the posts without essential marketing information (e.g. entertainment posts) and those concerning more than two products, we arrived at a nal pool of 23 posts. Since product involvement can inuence consumers dependence on different kinds of information, we manipulated product involvement in our study. Product involvement was dened as a persons perceived relevance of the product based on inherent needs, values, and interests (Zaichkowsky, 1986). According to Petty et al. (1983), consumers attitude toward a product of low involvement was more likely to be inuenced by the status of the endorser (peripheral route of ELM), whereas consumers attitude toward the product of high involvement was more likely to be inuenced by the cogency of information (central route suggested by ELM model) rather than the status of the endorser. The valence of WOM information is another key factor that may inuence whether consumers believe WOM information (East et al., 2007); however, in present study the impact of this was controlled through the selection of posts. We recruited 15 graduate students (age range: 23-31) majoring in marketing from a business school in a university to judge product involvement and the valence of WOM information because graduates majoring in marketing can easily understand the concept of product involvement, however they were also a convenience sample. For product involvement judgment, the method was adopted from McQuarrie and Munsons (1992) paper. We rst introduced the denition of product involvement (Zaichkowsky, 1986), and then asked the participants to indicate their opinion on a seven-point scale (1 low, 7 high). The item we used was in terms of product involvement, what is your opinion of the product in the post?. Furthermore, we let the 15 graduates classify the valence of WOM information by asking them the question

What is your opinion of this post in terms of the information it provided (positive, neutral, negative, mixed, irrelevant, not sure)? (Godes and Mayzlin, 2004). Finally, based on the participants evaluation, we chose four typical posts with signicantly different product involvements and valences. The products were: a digital camera (high product involvement, negative), cell phone (high product involvement, positive), toothpaste (low product involvement, negative), and pearl powder (low product involvement, positive). 4.3 Pilot study 2: develop and purify the measurement of WOM information Literature (e.g. Rieh, 2002) regarding internal characteristics of information provide a few dimensions for use with qualitative data. However, a mature measurement scale has not been developed. On the basis of relevant literature, we developed a set of constructs to examine the characteristics of WOM information. As shown in Table II, we followed Riehs (2002) classication and denition. We chose 23 frequently mentioned keywords as original items (see Table II), which cover all of the dimensions mentioned in the literature. Then we chose 19 posts as pilot test posts from the pool of posts that we had collected. These posts were not used in the formal survey. In pilot study 2, a new set of 15 graduate students (age range: 20-25) recruited from a business school in a major university were asked to select and evaluate 8 posts out of the 19 provided. They completed a separate survey for each post by rating them against the 23 key words along a seven-point scale (1 totally disagree, 7 totally agree). We ended up with 117 valid surveys. Using Exploratory Factor Analysis (EFA), we found there were four factors (see Table I) which we named: interestingness (a 0:77), quality (a 0:75), authenticity (a 0:86) and authority (r 0:80). The interestingness factor is consistent with Riehs (2002) qualitative studies. The quality factor is somewhat inconsistent with what we mentioned previously, which excluded two facets, good (e.g. good, bad) and current (new, out of date). These items were excluded because they have a similar loading on quality and on other factors such as interestingness and authority. After purication, the quality factor includes three facets: useful (useful), accurate (right) and important (important). Another difference is that the prior authority factor mentioned in Rieh (2002) was divided into two independent factors. This result is consistent with what was suggested by Mitra and Watts (2002). Following Mitra and Wattss (2002) terminology, we named the two factors as authority and authenticity, respectively. A short version of our measurement scale and the results of EFA are listed in Table III. 4.4 Formal study 4.4.1 Respondents and procedure. As a new online market, China currently has the largest internet population and has many well-known e-commerce companies such as baidu.com, alibaba.com, shengda.com and sina.com. Thus, we believe that China should be an appropriate context for our research. Participants, including undergraduates, postgraduates, part-time MBA students and part-time EMBA students were recruited from two universities in one capital city in the middle part of China. To exclude the 30 graduates participating in the pilot studies, no participants were recruited from the three classes from which those 30 graduates were in. Our study is a two-step survey. In the rst step, 393 respondents participated. In the second step,

Making your online voice loud 1287

EJM 45,7/8

1288

Item description 0.77 0.78 0.86 0.55 0.75 0.59 0.83 0.77 0.86 0.86 0.90 0.70 0.80 0.85 0.79 0.87 0.81 0.87 0.82 0.92 0.84 0.96 0.86 0.92 0.88 0.80 0.80 0.86 0.86 0.78 0.78 0.78 0.78

Interestingness C11 Exciting C12 Interesting C13 Dull (reverse) Quality C25 Right C27 Useful C28 Important Authenticity C31 Credible C32 Trustworthy C33 Suspect (reverse) Authority C36 Ofcial C37 Authoritative

Acceptance Likely Probable Inuential Resending Intention Likely Possible Probable

Notes: Principal components analysis with Varimax rotation; X 2 (104) 149.39; RMSEA 0.046; CFI 0.99; NFI 0.96; GFI 0.92

Table III. Item descriptions and measurement model results for latent constructs Lambda loading Formal study (CFA) (n 210) Cronbachs alpha or correlation Construct coefcient reliability Average variance extracted 0.55 0.54 0.68 0.67 0.70 0.79

Pilot study 2 (EFA) (N 117) Factor Cronbachs alpha or correlation loading coefcient

0.86 0.77 0.79

0.65 0.78 0.67

0.81 0.87

0.85

0.78 0.81

210 out of 393 (54 per cent) respondents from the rst step voluntarily participated in the second survey. Table IV shows our participants demographic data, and the data indicates that our participants represent the typical online users in China. In the rst step, each participant was asked to ll in a short paper-pencil questionnaire which includes questions about product involvement, prior buying behaviour for a given product and some demographic indices (e.g. internet use, gender and age). At the end of the rst step, participants were asked to indicate whether they would like to participate in another survey in the future. Then, they were asked to write down their e-mail addresses so that they could receive the results report of the survey and also receive the questionnaire for the new survey. The main objective of rst step is to get demographic indices to check the representativenss of our sample, and also to get participants e-mail addresses for the following online survey. Furthermore, we asked participants to report their product involvement for one product to recheck manipulation of product involvement. As expected, we found that participants reported product involvement was signicantly higher for high involvement products camera and cell phone than for low involvement products toothpaste and pearl power (Meanhigh 4:80, Meanlow 3:70; F 1; 391 90:28, p , 0:001). The second step started two days later. Each participant, who agreed to participate in a new survey, received an e-mail inviting them, to visit a simulated context, of a discussion forum, in which a qualied post was embedded. To simulate the real online environment, we presented the post in a format similar to the original online post. Then the volunteers, those who opted to participate in the survey, were instructed to follow a link at the end of that post to ll in a web questionnaire that opens in a new window. The questionnaire included the measurement of WOM information perception, acceptance, resending intention, and more. 4.4.2 Measurements. We present the CFA results of all nal measurements used in the formal study in Table III. Except for the puried measurement scale for perception of WOM information, all of the other construct scales were adopted from existing literature. The likelihood of accepting WOM information was measured using a three-item seven-point Likert scale adapted from the consumer acceptance of online
General data (Mainland China) * (%) 57.2 42.8 Average: 27.3 years

Making your online voice loud 1289

Facets Gender Male Female Age , 20 20-25 26-30 31-39 Using internet , 1 year 1-2 years 3-4 years . 5 years

Participants descriptive data 118 92 40 101 45 24 4 53 67 86

(%) 56.2 43.8 19 48.1 21.4 11.4 1.9 25.2 31.9 41.0

Average: 4.2 years Table IV. Participants personal data and general data

Source: www.cnnic.net.cn (Year 2007 Report)

EJM 45,7/8

agent advice (Gershoff et al., 2003). The measurement of resending intention used a seven-point Likert scale adapted from the online purchase intention (Schlosser et al., 2006). The results of CFA show that all of the constructs reliability and validity were satisfactory. 5. Results and discussion 5.1 Model test We used a structural equation model to examine the overall model (see Figure 2) through LISREL 8.7. The t indices (x2141 200:07; GFI 0.91; RMSEA 0.042) showed that this proposed model is acceptable. However, we do not know whether this proposed model t our data best. Anderson and Gerbing (1988) suggested that comparing the substantive model of interest with the next most likely constrained or unconstrained alternatives from a theoretical perspective to the substantive model of interest is a good way to assess structural model. In our proposed framework, acceptance towards WOM information is a full mediator for the relationship between the four dimensions of WOM information and resending intention. Therefore, the next most likely unconstrained alternative to this proposed model should be that acceptance towards WOM information is treated as a partial mediator for the relationship between the four dimensions of WOM information and resending intention. As shown in Table V, when we consider acceptance towards WOM information as a partial mediator, only the relationship between interestingness and resending intention is signicant. So in the puried model, we only consider the relationship between interestingness and resending intention. Because the freely estimated path from interestingness to resending intention in the puried partial mediator model is constrained in our proposed model, our proposed model should be said to be nested within the partial mediator model. According to Anderson and Gerbing (1988), we used a Chi-square difference test to compare these two nested models. The results showed that our proposed model t worse than the partial mediator model (Dx 2 141 2 140 200:07 2 193:42 6:65, p , 0:05). Therefore, we chose the puried partial mediator model (see Figure 3) as the best model for our empirical data. As shown in the partial mediator model, the results support all of our H1 and H2). First, all four dimensions of WOM information signicantly inuenced potential resenders acceptance of it and explained 68 per cent of the variance of acceptance. Among these four dimensions, quality was the most important factor, and authority was the second most important. Second, acceptance towards WOM information was an important partial mediator for the relationship between the perception of WOM information (interestingness, quality, authenticity, and authority) and resending intention. Two antecedents, acceptance towards WOM information, and Interestingness, accounted for 33 per cent of resending intentions variance. Acceptance towards WOM information appears to play the most signicant role. Unexpectedly, interestingness showed a direct inuence on resending intention. We will discuss these issues in detail later. We also tested the effect of the two manipulated variables on a resenders acceptance. Resenders acceptance was subject to a two-way ANOVA with WOM valence and product involvement as independent variables. The results showed a signicant main effect of WOM valence, indicating that a consumers acceptance

1290

Model Accept 0.19 *Interest 0.41 *Quality 0.19 * Authent 0.25 *Author
2

Path model equation R 0:70 X 2 (141) 193.66; CFI 0.99; NFI 0.96; GFI 0.91; RMSEA 0.042

Fit indices

Proposed model

Acceptance as a full mediator Original model

Alternative model

Acceptance as a partial mediator

Puried model

(Final model) Error t-value (0.087) (0.086) 4.84 2.60

Error (0.076) (0.095) (0.071) (0.077) t-value 2.54 4.31 2.76 3.31 Resending 0.56 * R 2 0:31 Accept Error (0.075) t-value 7.46 Accept 0.18 *Interest 0.40 *Quality 0.20 * R 2 0:68 X 2 (137) 184.70; CFI 0.99; NFI 0.96; GFI 0.91; Authent 0.25 *Author RMSEA 0.041 Error (0.077) (0.097) (0.072) (0.078) t-value 2.46 3.87 2.77 3.49 Resending 0.29 *Accept 0.19 *Interest 0.16 * R 2 0:33 Quality 2 0.043 *Authent 0.079 *Author Error (0.14) (0.094) (0.13) (0.088) (0.099) t-value 2.11 1.99 1.22 2 0.50 0.80 Accept 0.17 *Interest 0.41 *Quality 0.20 * R 2 0:68 X 2 (140) 187.61; CFI 0.99; NFI 0.96; GFI 0.91; Authent 0.26 *Author RMSEA 0.040 Error (0.077) (0.096) (0.071) (0.078) t-value 2.21 4.29 2.79 3.28 Resending 0.42 *Accept 0.22 *Interest R 2 0:33

Making your online voice loud

Table V. Model comparison

1291

EJM 45,7/8

1292

Figure 3. Hypotheses examined by the structured equation model (nal model)

towards negative WOM information (Meannegative 5:25) was signicantly higher than their acceptance towards positive WOM information (Meanpositive 4:83) (F 1; 206 11:70, p , 0:01). This result is the same as the results of ofine WOM studies (Charlett et al., 1995), indicating that people are more willing to believe negative information than positive information regardless of its environment. Furthermore, the results of the two-way ANOVA also showed that there was no signicant difference in acceptance between high product involvement and low product involvement (F 1; 206 0:262, p . 0:1). We will discuss this result in detail in the next section. To further generalize our results, we used a multiple group SEM to test whether the positive impact of the four dimensions on acceptance differ between WOM information with different valence. In model 0, we set the structural model equal in the two groups (positive WOM vs. negative WOM). In model 1, we set the four paths from characteristics of WOM (quality, authority, authenticity and interestingness) to acceptance free. The non-signicant difference in Chi-square between model 1 and model 0 (Dx 2 4 5:94; p . 0:2) showed that the four paths were equal in the two groups, indicating that the four dimensions have a positive impact on resenders acceptance no matter if the WOM information is positive or negative. 5.2 Discussion All of the four dimensions of WOM information examined in our study have signicant impact on a participants acceptance. Quality is the most important. Therefore, the usefulness and importance of WOM information is critical for enhancing the ripple effect of diffusive WOM communication. Authority is also an important factor. For example, in online forums, we can observe that participants usually evaluate authority by judging the formality of some of its external characteristics such as reputation of the discussion forum. Thus, peripheral cues are also important in increasing a consumers acceptance of the information. Acceptance is more powerful in facilitating resending intention. Therefore, the WOM information itself is the main driver of the ripple effect produced by diffusive WOM communication. Our results also showed that the signicant impact of the four dimensions on acceptance exist for both positive WOM and negative WOM. Therefore, as an agent in online environments, no matter if

you recommend or criticize a product, you should ensure that the WOM information has good quality, high authority, high authenticity and interesting content. Otherwise, online audiences will not accept it. Our results also showed that peoples acceptance towards WOM information was not inuenced by their product involvement in an online environment. We think this result might be caused by the following reasons. In the extant literature, the impact of product involvement on persuasion or consumers decisions usually occurred at the behavioural decision stage (Herr et al., 1991). Because participants in our survey were not told to make a purchase decision, product involvement might not be able to enter into consideration when people just evaluate the WOM information. At the same time, we believe that this result reects the reality in online environments. For example, as you may have noticed, discussions about paper diapers are as hot as discussions about cars in online forums. So we can not expect that peoples acceptance toward information about paper diapers will be signicantly lower than peoples acceptance toward information about cars. A variation from the proposed model emerges when our results identied a new path from Interestingness to Resending Intention. The new path shows that resenders may share the WOM information with others without accepting it. According to the Year 2006 survey of users of Chinese online community conducted by iResearch (www.iresearch.com.cn), the most frequently mentioned intention for taking part in an online community is to share the joy of life (60.7%), the second most mentioned intention is to look for solutions for difcult situations at work (60.0%). Therefore, when a Chinese resender reads an interesting post, s/he may directly shares this fun with others without judging its quality or authority. 6. Implications and future research directions 6.1 Theoretical implications This study enhances our understanding of the online WOM ripple effect as diffusive communication. This has seldom been done in prior studies of WOM. In this study, we identify a systematic framework by using empirical methods. The results indicate that the perception on WOM information can inuence a consumers acceptance of that information. We also identify that the acceptance of online WOM information affects a resenders intention to share WOM information. Our results show that the WOM information itself is the most important factor in facilitating the spread of WOM information. Although prior studies have considered many factors, such as individual characteristics and social structural factors, factors related to the WOM information per se have been relatively neglected. Therefore, to some extent, our study lls this theoretical gap. 6.2 Managerial implications Our study has several important implications for companies, which can help practitioners achieve their communication goals efciently and effectively by making full use of the powerful online WOM communication. A companys successful communication campaign does not only directly inuence an audience but also indirectly inuences an extensive target market. A company should learn how to produce a ripple effect in its communication campaign by creating a long and broad WOM ow, which will multiply the effect of the original communication campaign

Making your online voice loud 1293

EJM 45,7/8

1294

without increasing its cost. To produce a bigger ripple effect, a company should consider designing its communication information so that it is easy and convenient to transmit or forward on the internet. Most importantly, practitioners should design and organize the information from the consumers perspective. High quality communication information should be designed by focusing on consumers perceived quality (usefulness, accuracy and importance) rst and on the perceived authenticity (trustworthiness, credibility and reliability) second. Peripheral perception (authority, interestingness) should also be considered. WOM information that is interesting or from an authoritative source can not only increase an audiences acceptance of the information, but also trigger a ripple effect of WOM. 6.3 Limitations and directions for future research We conducted our survey in China. As suggested by Hofstede (1993), theories developed in one country may not be applied in another country. Therefore our model may not be applicable to other countries due to two reasons. First, since Chinese university students, at present, are typically the main online users in China, we chose them as a convenience sample. However, online users in other countries may be more diversied. Hence, future research needs to consider the diversity of sample. Second, although China is more open than before, it still has retained some unique cultural norms such as collectivism and high power distance. Authority plays an important role in Chinese society; as such Chinese may give more weight on the authority of WOM information than people in other countries. Future research may explore this difference by using a cross-culture perspective. Although we simulated the online WOM transmitting process and explored four dimensions of online WOM information, there are no clear indications of the related characteristics (e.g. source discussion forums; who posted it; how many read it; and so on) of WOM information that can inuence a participants perception of its quality, authenticity, authority and interestingness. In a future study, researchers may want to explore the specic characteristics of WOM information that can inuence a consumers perception. These ndings can help marketers choose better online discussion forums to generate a ripple effect. We believe that studying the previous questions can greatly increase our understanding of WOM in online environments. As argued before, three factors may inuence a focal agents sharing intention the focal agents characteristics, the WOM information, and other agents behaviour. This paper only focused on the second factor, and did not explore the potential inuence of the other two factors. In the future, research about these two factors could be done. For example, who are more likely to be resenders in online environments, opinion leaders or opinion seekers? Exploring this question may help marketers better understand and inuence potential resenders.
References Anderson, J.C. and Gerbing, D.W. (1988), Structural equation modelling in practice: a review and recommended two-step approach, Psychological Bulletin, Vol. 103 No. 3, pp. 411-23. Arndt, J. (1967), Role of product-related conversations in the diffusion of a new product, Journal of Marketing Research, Vol. 4 No. 3, pp. 291-5. Arndt, J. (1968), Selective processes in word-of-mouth, Journal of Advertising Research, Vol. 118 No. 3, pp. 19-22.

Bansal, H.S. and Voyer, P.A. (2000), Word-of-mouth processes within a services purchase decision context, Journal of Service Research, Vol. 3 No. 2, pp. 166-77. Bickart, B. and Schindler, R.M. (2001), Internet forum as inuential sources of consumer information, Journal of Interactive Marketing, Vol. 1 No. 3, pp. 31-40. Biyalogorsky, E., Gerstner, E. and Liai, B. (2001), Customer referral management optimal reward programs, Marketing Science, Vol. 20 No. 1, pp. 82-95. Charlett, D., Garland, R. and Marr, N. (1995), How damaging is negative word-of-mouth?, Marketing Bulletin, Vol. 6, pp. 42-50. Chevalier, J.A. and Mayzlin, D. (2006), The effect of word-of-mouth on sales: online book reviews, Journal of Marketing Research, Vol. 43 No. 3, pp. 345-54. De Bruyn, A. and Lilien, G.L. (2004), A multi-stage model of word-of-mouth through electronic referrals, working paper, eBusiness Research Center, February. Dellarocas, C. (2003), The digitization of word-of-mouth: promise and challenges of online feedback mechanisms, Management Science, Vol. 49 No. 10, pp. 1407-24. Derbaix, C. and Vanhamme, J. (2003), Inducing word-of-mouth by eliciting surprise a pilot investigation, Journal of Economic Psychology, Vol. 24 No. 1, pp. 99-116. East, R., Hammond, K. and Wright, M. (2007), The relative incidence of positive and negative word-of-mouth: a multi-category study, International Journal of Research in Marketing, Vol. 24 No. 2, pp. 175-84. Feick, L.F. and Price, L.L. (1987), The market maven: a diffuser of marketplace information, Journal of Marketing, Vol. 51 No. 1, pp. 83-97. Frenzen, J. and Nakamoto, K. (1993), Structure, cooperation, and the ow of marketing information, Journal of Consumer Research, Vol. 20 No. 3, pp. 360-75. Gershoff, A.D., Mukherjee, A. and Mukhopadhyay, A. (2003), Consumer acceptance of online agent advice: extremity and positivity effects, Journal of Consumer Psychology, Vol. 13 Nos 1/2, pp. 161-70. Godes, D. and Mayzlin, D. (2004), Using online conversations to study word-of-mouth communication, Marketing Science, Vol. 23 No. 4, pp. 545-60. Greimas, A.J. (1971), Narrative grammar: units and levels, Modern Language Notes, Vol. 86 Nos 11/12, pp. 793-806. Gumpert, G. and Cathcart, R. (1986), I am a camera: the mediated self, Communication Quarterly, Vol. 34 No. 2, pp. 89-102. Hartmann, W.R., Manchanda, P., Nair, H., Bothner, M., Dodds, P., Godes, D., Hosanagar, K. and Tucker, C. (2008), Modelling social interactions: identication, empirical methods and policy implications, Marketing Letters, Vol. 19 Nos 3-4, pp. 287-304. Hennig-Thurau, T., Gwinner, K.P., Walsh, G. and Gremler, D.D. (2004), Electronic word-of-mouth via consumer-opinion platforms: what motivates consumers to articulate themselves on the internet?, Journal of Interactive Marketing, Vol. 18 No. 1, pp. 38-52. Herr, P.M., Kardes, F.R. and Kim, J. (1991), Effects of word-of-mouth and product-attribute information on persuasion: an accessibility-diagnosticity perspective, Journal of Consumer Research, Vol. 17 No. 4, pp. 454-62. Higgins, E.T. (1981), The Communication game: implication for social cognition and persuasion, in Higgins, E.T., Herman, C.P. and Zanna, M.P. (Eds), Social Cognition: The Ontario Symposium, Lawrence Erlbaum, Hillsdale, NJ, pp. 343-92.

Making your online voice loud 1295

EJM 45,7/8

Hoffman, D.L. and Novak, T.P. (1996), Marketing in hypermedia computer-mediated environments: conceptual foundations, Journal of Marketing, Vol. 60 No. 3, pp. 50-68. Hofstede, G. (1993), Cultural constraints in management theories, Management Executive, Vol. 7 No. 1, pp. 81-95. Hogan, J.E., Lemon, K.N. and Libai, B. (2004), Quantifying the ripple: word-of-mouth and advertising effectiveness, Journal of Advertising Research, Vol. 45 No. 3, pp. 271-80. Lau, G.T. and Ng, S. (2001), Individual and situational factors inuencing negative word-of-mouth behaviour, Canadian Journal of Administrative Sciences, Vol. 18 No. 3, pp. 163-79. McGuire, W.J. (2001), Attitude change: the information-processing approach, in McClintock, C.G. (Ed.), Experimental Social Psychology, Holt, Rinehart & Winston, New York, NY, pp. 108-41. McQuarrie, E.F and Munson, J.M. (1992), A revised product involvement inventory: improved usability and validity, in Sherry, J. and Sternthal, B. (Eds), Advances in Consumer Research, Association for Consumer Research, Provo, UT, pp. 108-15. Mayzlin, D. (2006), Promotional chat on the internet, Marketing Science, Vol. 25 No. 2, pp. 155-63. Mitra, A. and Watts, E. (2002), Theorizing cyberspace: the idea of voice applied to the internet discourse, New Media & Society, Vol. 4 No. 4, pp. 479-98. Olney, T.J., Holbrook, M.B. and Batra, R. (1991), Consumer responses to advertising: the effect of ad content emotions and attitude towards the ad on viewing time, Journal of Consumer Research, Vol. 17 No. 4, pp. 440-53. Petty, R.E., Cacioppo, J.T. and Schumann, D. (1983), Central and peripheral routes to advertising effectiveness: the moderating role of involvement, Journal of Consumer Research, Vol. 10 No. 2, pp. 135-46. Phelps, J.E., Lewis, R., Mobilio, L., Perry, D. and Raman, N. (2004), Viral marketing or electronic word-of-mouth advertising? Examining consumer responses and motivations to pass along e-mail, Journal of Advertising Research, Vol. 45 No. 4, pp. 333-48. Richins, M.L. (1983), Negative word-of-mouth by dissatised consumers: a pilot study, Journal of Marketing, Vol. 47 No. 4, pp. 68-78. Rieh, S.Y. (2002), Judgment of information quality and cognitive authority in the web, Journal of the American Society for Information Science and Technology, Vol. 53 No. 2, pp. 145-61. Rogers, E.M. (1995), Diffusion of Innovations, 4th ed., Free Press, New York, NY. Scheuermann, L. and Taylor, G. (1997), Netiquette, Internet Research, Vol. 7 No. 4, pp. 269-73. Schlosser, A.E., White, T.B. and Lloyd, S.M. (2006), Converting web site visitors into buyers: how web site investment increases consumer trusting beliefs and online purchase intentions, Journal of Marketing, Vol. 70 No. 2, pp. 133-48. Strong, E.K. (1925), The Psychology of Selling, McGraw-Hill, New York, NY. Verlegh, P.W.J. and Moldovan, S. (2008), What drives word of mouth? A multi-disciplinary perspective, Advances in Consumer Research, Vol. 35, pp. 49-51. Weening, M. and Midden, C. (1991), Communication network inuences on information diffusion and persuasion, Journal of Personality and Social Psychology, Vol. 61 No. 5, pp. 734-42. Zaichkowsky, J.L. (1986), Conceptualizing involvement, Journal of Advertising, Vol. 15 No. 2, pp. 4-14.

1296

Zinkhan, G.M., Kwak, H., Morrison, M. and Peters, C.O. (2003), Web-based chatting: consumer communication in cyberspace, Journal of Consumer Psychology, Vol. 13 Nos 1/2, pp. 161-70. About the authors Minxue Huang is a Professor in the Marketing Department of Wuhan University. Fengyan Cai is an Assistant Professor in the Marketing Department, Antai College of Economics & Management, Shanghai Jiao Tong University. Fengyan Cai is the corresponding author and can be contacted at: fengyan@baf.msmail.cuhk.edu.hk Alex S.L. Tsang is an Assistant Professor in the Marketing Department of Hong Kong Baptist University. Nan Zhou is a Professor in the Marketing Department of City University of Hong Kong.

Making your online voice loud 1297

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

S-ar putea să vă placă și