A Study of Measurement and Influence of Wine Blogs As an Information

Journal of Consumer Behaviour, J. Consumer Behav. (2014) Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/cb.1496

Credibility in the blogosphere: A study of measurement and inﬂuence of wine blogs as an information source

TRACY RICKMAN COSENZA1*, MICHAEL R. SOLOMON2† and WI-SUK KWON3‡ 1Department of Marketing and Supply Chain Management, Fogelman College of Business and Economics, University of Memphis, 210 Admin Bldg, Memphis, TN 38152, USA 2Center for Consumer Research, Saint Joseph’s University, Haub School of Business, 257 Mandeville, Philadelphia, PA 19131, USA 3Consumer and Design Sciences, 308 Spidle Hall, Auburn University, Auburn, AL 36849, USA

ABSTRACT

Uncovering what drives the consumer to action is of interest to academia and industry alike, and the advent of social media has added a confounding layer to the consumer decision process. In some ways, it has changed nothing, and in other ways, it has changed everything. Information is at the heart of decision making, and blogs offer a plethora. A positive evaluation of credibility (believability) will ultimately determine its usability for decision making. Using centered resonance analysis of wine blog writers and readers focus group data, this study attempts to address the full range of blog characteristics (subsequent scale items) that can be evaluated for credibility, anchoring them to the relevant research streams of media, website, and word of mouth. Structural equation modeling of viral blog and Twitter survey data is used to test the hypothesized scale with three dimensions—site, message, and source—for validity and also evaluate the credibility influence on behavioral intentions as mediated by trust. Evidence suggests that the author (source) is central to credibility evaluation. The data also support the influence of wine blog credibility perceptions on trust, which leads to following wine blog recommendations and continuing to participate with the wine blog. Copyright © 2014 John Wiley & Sons, Ltd. key words Credibility; Blogs; Blog Credibility; Blog Influence; Social Media; Wine Blog Influence; Wine Blog Credibility

“Google it,” a phrase ubiquitous with consumers’ quest for allowing individuals to freely share information and opinions information, demonstrates the importance of online search on any topic. The power of eWOM lies in its permanence and in a microanalysis of consumer decision making. According delivery mechanism. Interpersonal WOM communication is to Lecinski (2011) in his popular eBook Zero Moment of oral and vanishes as soon as it is uttered except for what is Truth, 70 percent of purchase decisions begin with online remembered by the speaker and the receiver. Whereas the tradi- search – particularly looking for recommendations and prod- tional temporal exchange has no record, the technologies of uct reviews. The Fleishman-Hillard and Harris Interactive eWOM permanently store the communication in cyberspace. Annual Global Study (2012) reveals that 89 percent of con- Permanence is very important for information search. The consumers use Internet search to make purchasing decisions. versation is continuously accumulated on the Web as text and is They further report that Internet search is comparatively available for search. The text is searchable by topics or tags via influential with information from family and friends in the search engines. Most importantly, the Web shifts the geography USA and greater still in Canada, Asia, and India. The Internet of WOM communication from local to global. Formerly, the has simply changed how we decide what we buy. spread of WOM was limited. The Web accelerates eWOM communication, and because it is a virtual conversation ware- Consumer search for information house, the eWOM is available whenever and wherever needed. If a consumer does not possess adequate information to make With the Web, the diffusion of eWOM is limitless. The anytime a decision, an external information search is necessary. availability of persuasive information from fellow consumers Traditional research identifies sources of external informa- on virtually any topic shifts the control of market conversations tion search as follows: (i) interpersonal (word of mouth from business to consumers on sheer volume alone. Phelps (WOM)); (ii) independent (consumer reports); and (c) market et al. (2004) argue that the speed, convenience, one-to-many oriented (advertising). Of these, the influence of WOM has reach, and absence of face-to-face human pressure make online an extensive history in the literature of playing a significant WOM more influential than its traditional offline counterpart, a role in decision choice (e.g., Brown and Reingen, 1987; findingupheldinindustrystudies.Althoughthereareothers Godes and Mayzlin, 2004; Mourali et al., 2005). The Web (Meuter et al., 2013) who would argue the persuasive influence has broadened the WOM communication environment to in- strength of eWOM versus traditional WOM, the overall evi- clude WOM on the Web (electronic WOM (eWOM)). Web dence indicates that e-communication channels shape eWOM tools and social media applications proliferate on the Web, interpersonal communication. Further, the influence of user- generated content, which is the crux of a blog, is more likely to impact the use of a blog as an information source when the *Correspondence to: Tracy Rickman Cosenza, Instructor of Marketing and Coordinator of Social Media Programs, Department of Marketing and Sup- product category is complex (Riegner, 2007). ply Chain Management, Fogelman College of Business and Economics, Blogs, classified as social media and user-generated con- University of Memphis, 210 Admin Bldg, Memphis, TN 38152, USA. tent, cover a variety of topics and represent a virtual ware- E-mail: [email protected] †Professor of Marketing, Director house of searchable information. Blogs are the third most ‡Human Sciences Associate Professor of Retailing likely online platform to influence a purchasing decision

(Technorati, 2013, Digital Influence Report). Created content consumers evaluate the credibility of wine blogs, develop is posted, and interaction is encouraged with readers through an attitude of trust, and consequently use the credibility as- comments. Although the potential of market conversations sessment to make specific decisions (follow blog recommen- initiated by consumers in the blogosphere is widely specu- dations (purchase), spread WOM about a blog, and continue lated, the shift of communication power to the consumer is to participate in a blog). This model is depicted in Figure 1. evident. This access to readily available information from However, there is yet to be a validated measure of blog blogs on the Web produces an immense effect on consumer credibility that captures the eWOM blog genre. As with decision making (Zafiropoulos, 2012). Thus, blogs have the any source of information – the usefulness of blogs is depen- potential to alter how consumers search for decision informa- dent upon how consumers perceive blog credibility and trust tion. Using social network analysis, Ghosh and Lerman them. Therefore, this study focuses on two phases: (2008) suggest that the influence between blogs in the same Phase 1: The development and testing of measurements of community is similar to that which is found among members wine-specific blog credibility. of a close network. Social network analysis is able to identify Phase 2: The structural testing of the model is specified in the influential nodes of the network but is limited in Figure 3 using these credibility measures. interpreting the outcomes of the influence. Interestingly, although social media is a prevalent topic in academic research, there have been few definitive conclusions offered Why study wine blogs? relevant to the influence social media at large or blogs specif- Wine, a social object and product, has become more preva- ically have on behavioral outcomes (Lockshin and Corsi, lent in consumer behavior studies in a variety of social media 2012; Zafiropoulos, 2012) despite being an integral part of contexts (e.g., virtual community, Habel et al., 2010; Thach, nearly every industry marketing strategy. This observation 2010; WOM, Bouzdine-Chameeva and Galam, 2011; blogs, constitutes the research gap this study aims to address. Doyle et al., 2011; search, Atkin and Thach, 2012; Twitter. The proliferation and diversity of blogs makes them im- Wilson and Quinton, 2012; blog influence, Zafiropoulos, possible to study in general. Therefore, successful research 2012), ranging in focus from awareness to purchase. Thach must address them on a more micro level. In addition, to be (2010) purports that wine blogs are clearly a source of infor- influential as a decision information source, blogs must ex- mation for consumers. According to the complete wine blog hibit credibility. Recently, researchers have extended the is- list on Vinography (2014), there are well over 700 wine sue of genre credibility, different criteria for credibility blogs in English, and the list also includes blogs written in assessment of Internet sources of information, to the credibil- 12 other languages. When “googled,” wine blogs as a search ity of blogs. For example, Kaye and Johnson (2011) found term generates over 600 million results. Thus, wine blogs are that different types of blogs were evaluated differently, even chosen as the focus of this study for four important and inter- though they were all offering views of politics or news. related reasons. First, the social context of wine motivates Because evidence indicates the importance of isolating in- people to seek information in order to make a correct choice fluence within a specific genre, this present research seeks to about wine or the lifestyle choices surrounding it. Many con- quantify the influence of blogs in consumer decision-making sumers view wine as a complicated product purchase information by using a more controlled study framework. wrought with uncertainty (Atkin and Thach, 2012). The po- Thus, this study focuses on the product of wine and how tential adverse consequence of making a poor choice leads

Figure 1. Conceptual model of the inﬂuence of wine blog credibility on behavioral intentions as mediated by trust.

Copyright © 2014 John Wiley & Sons, Ltd. J. Consumer Behav. (2014) DOI: 10.1002/cb Credibility first: the influence of wine blogs many consumers to seek risk reduction through repeating the community similar to friends or family, it seems plausible purchase of a successful choice (stick with what you know), that individuals, who use the Internet, will search this online seeking the opinions of friends, or following the advice of surrogate for information before they make a purchase. wine experts (Bouzdine-Chameeva and Galam, 2011). John- Finally, because wine is socially charged, trust is an impor- son and Bruwer (2004) found that when individuals have tant aspect in the behavioral response. We have become accul- more information about their wine purchase, they are better turated toward trusting ratings and reviews when making a able to predict the consequences of their decisions, and their wine decision, for example, sommeliers, point systems, shelf level of anxiety is reduced. talkers, or publications. In common with each of these is some Second, wine is an experience-type product and, as a degree of perceived expertness. Wine blogs offering product category, requires knowledge and information to consumer-generated wine reviews are the most common type purchase (Thach, 2009). There are over 5000 different varie- (Thach, 2010), placing them in a necessary condition of signal- ties of wine grapes. When coupled with the proliferation of ing trustworthiness. Of note, the wine blogging community is labels producing wine from the various varieties, the avail- rather cohesive and has been participating in a Wine Bloggers’ ability of choice is daunting. It is unlikely that individuals Conference since 2008 when the primary focus from the vantage will have the opportunity to taste every wine before a pur- point of wine blogger or wine marketer was “what is the influ- chase; thus, they need to rely on recommendations from ence of wine blogs and are they perceived as credible to the wine others who have had the opportunity. This information consumer?” Wine bloggers desire to be influential and under- ranges from wine evaluation (e.g., ratings) to purchase rec- stand the importance of being perceived as credible. Doyle ommendations (e.g., discussions that are part of the culture et al. (2011) identified the cues that commercial and non- of wine), all of which are used by even the most wine savvy commercial wine bloggers use to signal trustworthiness to their individuals. Because this study focuses on blogs as an infor- readers, finding that commercial bloggers stress product-related mation source, repeated sharing and seeking of information authority cues and non-commercial stress product-related con- is considered a desirable component for study. The commu- sumption experience cues. nity of wine blogs naturally contains this exchange. The premise of this study is that individual wine blog Third, the current study examines the influence of a wine readers evaluate blogs for credibility and trust them to make blog as an information source for decision making. Certain a variety of decisions (e.g., purchase based on recommenda- product categories lend well to recommendation influences, tion). Therefore, the wine blog domain appears to be influential and wine is one of them. Bouzdine-Chameeva and Galam for decision making, providing a good opportunity to examine (2011) offer three competing drivers of consumer choice on the variables proposed in this model (Figure 1). In addition, it wine preference – view of experts, reputation of wine type, is assumed that the preceding justifications could easily transfer and WOM. The social landscape offers various forms of so- to other products that are also experiential and complex. cial media influence for the wine consumer, and several of them have been explored. For example, Habel et al. (2010) found that although individuals participate in virtual commu- KEY LITERATURE FOR THEORETICAL DIMENSIONS nities based on wine as the social object, their main focus is OF BLOG CREDIBILITY AND RELATED HYPOTHESIS on the social aspects of the network itself and less on partic- TESTING ipating to give and receive information within the community. That is not to say that wine-related information will As individuals have greater access to more diverse informa- not be shared and perceived credible as the social needs are tion sources such as blogs, the responsibility of credibility met but rather that individuals may not initially go to the vir- evaluation is shifted to the receiver of the information. Previ- tual community to gain product knowledge. Wilson and ous control mechanisms (like professional gatekeepers and Quinton (2012) found that although Twitter, a microblog, is editorial review) that authenticated credibility may not be used hourly for wine-related information, because of a present or as effective. Further, the content posted by an in- 140-character limit, tweets are relegated to directing traffic dividual blogger is available to anyone who seeks it, making to another source for online or offline media. Thus, influence it an open conversation that is not one to one, but one to any. from another source can be enhanced by the use of Twitter, Blog readers are searching unverifiable media, written by au- making it more useful for soft value as opposed to a hard thors they may not initially know. Consequently, the re- value such as purchase. Additionally, Zafiropoulos (2012) ceivers of information must make a credibility evaluation suggests that the most active and connected wine blogs are of the blog on the basis of observable characteristics before personal and single authored and suggests that their deciding to trust the information. When unknowledgeable networked connectivity gives them the power to be influen- about an information source, current research suggests that tial, further noting that their discussions can be analyzed consumers may rely on a proxy to aid in their vetting such and targeted for promotion. Kozinets (2002) stresses that as search rank algorithms (Weerkamp and de Rijke, 2012) when considering a suitable online community to study, it and system-generated influence scores such as Klout is important for the interaction within the community to pro- (Edwards et al., 2013). The indicators for computer-aided vide the answers to the major research questions. An envi- credibility/influence authentication are relegated to numbers ronmental scan of wine blogs notes multiple behavioral the computer can compile such as the number of comments responses, evidenced within posted comments. Because the and document length with respect to search rank and/or reach interactive social network of wine blogs conveys a sense of and amplification with respect to influence scores. Apart

Copyright © 2014 John Wiley & Sons, Ltd. J. Consumer Behav. (2014) DOI: 10.1002/cb T. R. Cosenza et al. from these proxy indicators, blog readers are left to their own measurement) by comparing travel expert blogs and televi- subjective assessment of blog information. The question sion travel shows via source (trustworthiness, expertise, and arises as to what blog characteristics corroborate the credibil- attractiveness), message (structure and content), and media ity of the blog to the reader. As we will see in the following (trust, accuracy, non-bias and fairness, and completeness). section, to address this question, it is helpful to focus on un- Kang (2009) demonstrated a 14-item reliable measure of derstanding what blogs are. First, they are technically a blog credibility limited to the dimensions of blog source website, although more dynamic and interactive than typical and blog message credibility within the context of public re- websites. Second, as a carrier of information, they are a me- lations. Durukan and Bozaci (2012) used the dual blog dia source, classified as social media. source/message credibility measures proposed by Kang (2009) to study determinants of WOM in social media. Theoretical credibility dimensions of the blog as an infor- Although the results demonstrated support for the hypothe- mation source ses, the results were based on a study of students, and the Three traditional theoretical dimensions of credibility (site or sources they evaluated were their friends. medium, message, and source) are supported in the literature When choosing which measurement items to include in (Eysenbach, 2008). When extending credibility assessment our credibility scale, the focus of measurement must be deter- to the Web environment, Flanagin and Metzger (2007) note mined relative to the genre, the product, and the blog writer/ the complexity of the task, indicating that credibility assess- reader. For example, is the focus on the credibility of the ment of information may be at the level of authorship, mes- source or the credibility of the content that is communicated? sage, site sponsor, or medium. The same is true for the Additionally, information seekers utilize different evaluation dimensions of blog credibility (e.g., Tan and Chang, 2012; criteria depending upon the type of media (newspapers, Weerkamp and de Rijke, 2012). Although they are interre- television, or online) that is assessed (Newhagen and Nass, lated (the blog as the medium, the blogger as the source, 1989). The chosen items to evaluate blog credibility must and the post associated with the blogger as the message), tap into how the users conclude that a particular blog is cred- they have unique functions. For this reason, it is necessary ible. For example, if the blog is perceived to be accurate, to evaluate perceptions of credibility along three distinct what is it about the blog that draws this conclusion? dimensions: (1) source credibility (i.e., the perceived The literature on genre credibility suggests a seemingly credibility of the person (author) of the represented blog); field-specific focus (Kaye and Johnson, 2011; Tan and Chang, (2) message credibility (i.e., the perceived credibility of the 2012). Therefore, for this study, blog credibility assessment content (information) residing on the blog); and (3) site cred- must be developed not only for the study of new media (Wathan ibility (i.e., the perceived credibility of the technical/aesthetic and Burkell, 2002) but also for the study of the specificnature features of the blog website). of product types like wine blogs. Warnick (2004) warns that the “one size fits all” approach on the Web does not work. Measuring blog credibility Although there is a large body of research that looks at the cred- Online trust ibility of various media channels, there is inadequate research Within online trust research, there are two approaches to pertaining to how the credibility of blogs is evaluated by their studying the online relationships between the trustor and users (Kaye and Johnson, 2011). Thus far, studies of blog the object of the trust (Corritore et al., 2003). The first ap- credibility have primarily focused on news or political blogs. proach focuses on computer-mediated communications in Most often, blog credibility researchers compare blogs with person-to-person relationships mediated through technology other sources/mediums of information such as traditional/online such as e-mail messages (Olson and Olson, 2000). The sec- broadcast TV, cable, or newspapers. These research studies ond approach focuses on technology as the object of trust explore blog credibility using traditional measures from the (e.g., trust in a website or software agents; Corritore et al., literature (fairness, accuracy, lack of bias, believability, and 2003). That is, trust must occur toward the technology to depth of information; e.g., Yang, 2007; Kaye and Johnson, be accepted for use. The latter approach is the most prevalent 2011). As a result, researchers are getting mixed results due to in the literature; however, its focus on website infrastructure misleading data (Kang, 2009). Blogs are qualitatively different does not capture the personal nature of the wine blog, which than the media channels with which they are compared, and the includes a human “voice” in content and persona. Yet, what current measures do not capture these differences. can be extrapolated from the website trust literature is that Consideration of assessing credibility specific to blogs is trust in a website, or blog in this instance, occurs through in- not new. Rubin and Liddy (2006) suggested a conceptual teraction with and evaluation of the site (Bart et al., 2005). framework for automation of blog credibility assessment Trust in a wine blog combines both of these approaches. consisting of four dimensions: blogger’s expertise and offline Because the wine blog is the personal voice of the writer, identity disclosure, blogger’s trustworthiness and value sys- trust is established at the individual or person-to-person tem, information quality, and appeals and triggers of a per- level. However, trust also stems from an evaluation of the sonal nature. Weerkamp and de Rijke (2012) proposed wine blog as a whole, which includes an assessment of the adding the textual components of the framework to improve message and the website in addition to the personal source. the retrieval process of blog posts, and the results supported Some of these evaluative characteristics pertain to technolog- the framework indicators. Tan and Chang (2012) explored ical attributes merging both approaches of online relation- travel information credibility (using non-verified credibility ships – technological and interpersonal.

Behavioral intentions credibility in the development of trust (Lee, 2010). Further, The ultimate variable of interest to the consumer behavior re- the predominance of online trust studies also focus on the searcher, the marketer, and even the blogger himself (herself) role of credibility in trust development (e.g., Corritore is the prediction of behavior. Because most studies have et al., 2005; Doyle et al., 2011; Pan and Chiou, 2011), indi- found a positive relationship between intentions and behav- cating the importance of credibility to the development of ior using models such as the theory of reasoned action, the trust on the Web. Therefore, as depicted in H1a, H1b, and theory of planned behavior, and the theory of trying, most H1c, a positive wine blog credibility assessment provides researchers accept that intention is the single best predictor of the reason to trust the wine blog as an information source. behavior and use this proxy research strategy (e.g., de Canniere H1a, b, c: Credibility perceptions (source, site, and message) et al., 2009). positively predict trust in the wine blog. Conceptual framework for model testing Because it is anticipated that the three components of blog The conceptual model presented in Figure 1 reflects the influ- credibility serve different functions in credibility assessment, ence of wine blog credibility (beliefs about source, message, three sub-hypotheses are offered to reflect their independent and site) on behavioral intentions as mediated by an attitude contribution to trust development in the model. of trust. As indicated in the literature and corroborated by the phase 1 credibility scale development, all media channels Blog source credibility (H1a) have source, message, and site formats. However, they differ Perceived source credibility becomes increasingly important considerably and may vary in how they are assessed for cred- as gatekeeping of credibility of information shifts to the re- ibility. Hence, the model further depicts the independent cipient of the information (Haas and Wearden, 2003). Source paths between these dimensions and trust. In addition to the credibility is traditionally defined as the degree to which an prevalent findings in trust studies that support the role of individual perceives a source to be believable (Austin and credibility in trust development, e-commerce studies sub- Dong, 1994), and the research stream boasts a wealth of at- stantiate the importance of trust on behavioral intentions. tention. A vast number of measures for source credibility Consequently, the model (arrows) reflects the influence of have been offered with slightly different item structure for credibility on trust leading to behavioral intentions. each setting (traditional setting, e.g., Markham, 1968; Berlo As depicted in Figure 1, scale development of wine blog et al., 1969; McCroskey and Jenson, 1975; Ohanian, 1990; credibility is identified in phase 1 before the model is tested on the Web, Rieh and Belkin, 1998; Eastin, 2001; Metzger in phase 2. Based upon the literature review and the concep- et al., 2003; and to the context of the blogosphere, e.g., tual model, two main hypotheses are offered. Durukan and Bozaci, 2012; Tan and Chang, 2012; Wang and Chien, 2012). Validating the items needed to assess source credibility has revealed considerable inconsistencies HYPOTHESES FOR THE PREDICTIVE RELATIONSHIP (Metzger et al., 2003). However, the consistency of the per- BETWEEN CREDIBILITY AND TRUST IN THE BLOG suasive influence of source remains undisputed. For example, Kiecker and Cowles (2001), examining online WOM, Individuals seek information for various reasons – knowledge, found that bloggers’ WOM was more persuasive when the empowerment, and even support – but, by and large, meaning- blogger was perceived to be credible. Therefore, it is pre- ful information reduces the uncertainty of decision outcomes dicted that credibility of source in a wine blog will persuade for the seeker (Buckland, 1991). Therefore, individuals look the reader to develop trust in the blog. for good information to make their decisions, that is, informa- H1a: Wine blog readers who perceive the author of the tion they can trust. Good information needs to be filtered out of wine blog to be more credible (source credibility) will all provided information. Individuals filter information by form a more positive attitude of trust toward the wine blog. establishing its credibility (Wathen and Burkell, 2002). As a filter, credibility allows consumers to evaluate the believability of the information based on their personal perspectives Blog site credibility (H1b) (Stanford et al., 2002). Credibility substantiation allows indi- The definition of site credibility is generally accepted as the viduals to trust information enough to actually use it. Because perceived believability of information by the receiver based credibility influences the impact of the message (if and how it on the characteristics of the channel through which the will be used), it is important to understand how people decide source sends the message (Newhagen, 1997). The literature what to believe (Wathen and Burkell, 2002). supports that site credibility is impacted by design and aes- When wine blog credibility is affirmed, an attitude of trust thetics (e.g. Fogg et al., 2002; David and Glore, 2010). One toward the wine blog can be established. Support for this need look no further than Obama Care’s 2013 healthcare. rationale lies in the number of studies that find a positive re- gov website debacle to highlight the importance of website lationship between credibility and trust. A long history of characteristics in communicating credibility. The website source credibility research finds that credible sources of problems were largely back-end, and the site continued to information produce more positive attitudes and behavioral crash because of scale issues. The public was frustrated, and responses than do sources of information that are not deemed the administration was humiliated. As a result, the plan itself as credible (Hovland and Weiss, 1951; Petty et al., 1983). came under fire with a vote of no confidence. The solution This observation particularly supports the critical role of was a total reboot of the account creation process, making it

Copyright © 2014 John Wiley & Sons, Ltd. J. Consumer Behav. (2014) DOI: 10.1002/cb T. R. Cosenza et al. more intuitive to use, simplifying the text, and making the de- an attitude of trust positively influences consumers’ intention sign more responsive to multiple platforms while making the to engage in various types of behaviors. According to the screen more beautiful (Paskin, 2014). Thus, we anticipate an theory of reasoned action, intention indicates a readiness to impact in the model based upon site evaluation as follows: perform a particular behavior and can be considered an antecedent to behavior (de Canniere et al., 2009). H1b: Wine blog readers who perceive a wine blog site to The literature is replete with support for the trust– be more credible will form a more positive attitude of trust intention hypothesis. Studies from the social sciences toward the wine blog. literature support the linkage between the state of trust and the behavior that ensues (e.g., Tanis and Postmes, 2005). Blog message credibility (H1c) Studies from the marketing literature demonstrate the impact Message credibility refers to an evaluation of information that an attitude of trust toward a seller has on intentions to characteristics that produce believability (Slater and Rouner, purchase, make a commitment, or have future interactions 1996). Early research on the dimensions of message credibil- (e.g., Crosby et al., 1990; Morgan and Hunt, 1994; Garbarino ity suggests a multi-dimensional construct and indicates that and Johnson, 1999). The extension of this research stream credibility may be evaluated on information quality, plausi- online (e.g., Komiak and Benbasat, 2004) consistently dem- bility, accuracy, use of evidence, language intensity, and onstrates the critical role of trust on behavioral intentions in message discrepancy (e.g., Slater and Rouner, 1996). The in- e-commerce environments and recommendation systems. formation science literature altered the dimensions somewhat Keeping with this line of reasoning, we posit the following: to include reliability, accuracy, validity, comprehensiveness, H2a, b, c: Wine blog trust will positively predict a wine- and currency (Rieh and Belkin, 1998). related/wine blog-related behavioral intention. A few studies have applied the concept of message credibility to the Web. Alexander and Tate (1999) found that Trusting behaviors are situational and are thus manifested factors such as information accuracy, comprehensiveness, by different decision outcomes. In this present study, wine- and currency are important for enhancing credibility on the related behavioral intentions include the individual’s intention Web. Other researchers (e.g., Sundar and Nass, 2001) have to engage in three behaviors: (i) follow recommendations – relied upon traditional media measures such as accuracy, ob- purchase or other wine-related advice; (ii) share WOM jectivity, bias, believability, fairness, or sensationalism when communication about the wine blog; and (iii) continue partici- evaluating Web credibility. Also using these traditional mea- pation with the wine blog (reading and/or commenting). Each sures, Johnson et al. (2007) found that the information on of these behavioral intentions relates to a behavior that is news blogs is considered more credible than both traditional associated with risk, making trust necessary. Therefore, there news sources and other online news sources by politically are three non-sequential sub-hypotheses to capture the interested users. Interestingly, the survey participants consid- situational outcomes. ered blog information highly credible when judged on the information depth but considered them weaker when evaluated Following recommendations within the wine blog (H2a) on fairness. Saxena (2011) also included the traditional mea- Following recommendations is not a new concept in the surement component of bias to measure message credibility. behavior literature. An influencer sways the outcome of a de- It appears that fairness or lack of bias in the message is not cision by providing input to a decision maker’s evoked set of great importance when blog readers judge credibility. (e.g., personal judgments about product). Although not the Lasica (2002) contends that the aspect thought to be missing only available outcome, purchase intention is typically of in traditional media and most coveted by the blog user is an highest interest. The importance of trust in the influence of opinionated, thoughtful analysis. In fact, a biased message purchase intentions has been widely researched and sup- may be preferred. Further, in consideration of wine, it stands ported in the e-commerce literature (e.g., van der Heijden to reason that a wine blog reader may want an opinionated et al., 2003; Pavlou and Gefen, 2004). It has also recently answer to “what would you do?” Therefore, the difference been studied at the blog level (e.g., Saxena, 2011; Hsu in information types is an important consideration when con- et al., 2012; Wang and Chien, 2012; Wu and Lee, 2012). sidering message credibility. This study specifies the wine Hsu et al. (2012) measured perceived usefulness of recom- blog as a particular information type, and message credibility mendations and trustworthiness of the blogger by modifying is assessed through an evaluation of the characteristics of the past research measures to the online shopping setting. Their blog message. Therefore, we posit the following: adapted measures were not reported. They found that bloggers with high credibility predicted an attitude of trust H1c: Wine blog readers who perceive the posts in a wine and intention to shop online. Additionally, they found that blog to be more credible (message credibility) will form a bloggers with low credibility did not have a direct effect on more positive attitude of trust toward the wine blog. trust and intention but had a significant effect on the perceived usefulness of recommendation before predicting trust HYPOTHESES FOR THE PREDICTIVE RELATIONSHIP and intention to shop. Using measures developed for evaluat- BETWEEN TRUST AND BEHAVIORAL INTENTIONS ing health information, Saxena (2011) found a positive relationship between information credibility, the attitude toward Ultimately, predicting behavior is the crux of studying con- the blog, and the intention to make a purchase. Findings by sumer behavior. Building on the prior theory, we argue that Wang and Chien (2012) seem to suggest positive

Copyright © 2014 John Wiley & Sons, Ltd. J. Consumer Behav. (2014) DOI: 10.1002/cb Credibility first: the influence of wine blogs relationships between the perceived credibility of a product wine blog. A blog’s success, and thus its influence, is depen- blog, the attitude associated with the product, and ultimately dent upon the continued development of participation with having a positive influence on intention to buy. Finally, Wu the wine blog (Hsu and Lin, 2008). Zafiropoulos (2012) spe- and Lee (2012) also found favorable support for purchase in- cifically notes the significance of community connectivity to tention stemming from the blog’s trustworthiness effect on a wine blog’sinfluence. One can extrapolate that community attitude toward the product, although measures were not ref- can be created with one’s own blog when readers choose to erenced in the paper. Because following recommendations, continue to engage with the wine blog as a preferred source. for a particular purchase, is important to the field of social/ Thus, continued participation with a wine blog in this present marketing research (McKnight et al., 2002b), it is included study stems from the importance of recognizing the value as- as an outcome variable (H2a) in the present study. sociated with use of the wine blog and the cost (perceived time risk) associated with finding another information H2a: Positive wine blog trust predicts a behavioral inten- source. Therefore, we posit the following. tion of following the recommendation given within the wine blog. H2c: Positive wine blog trust predicts a behavioral intention of continuing participation with the accessed wine weblog. Recommending the accessed wine blog to others (H2b) Word-of-mouth communication is defined as an exchange of ideas, comments, or opinions between individuals who do not represent a marketing entity (Bone, 1992). This exchange PHASE 1: DIMENSIONALITY/MEASUREMENT DE- may be about a brand, a product, a service, an organization, VELOPMENT OF WINE BLOG CREDIBILITY or even an idea. The spread of WOM communication about the blog is important when considering the reach of user- Statement generation generated content. In the online world, the ranking of a blog, Initial statement generation for the credibility measurement or any website, is most often determined by its traffic. There- scales was initiated through Skype™-hosted online focus fore, the success of the wine blog can be significantly groups. A convenience sample of 25 wine blog readers and impacted by the influence of recommendations. writers was solicited through the independent blog posts of Sharing of posted content as a result of favorable blog two prominent wine bloggers. The sample included a rela- credibility assessment is documented (Durukan and Bozaci, tively even split of writers and readers of blogs, both inside 2012; Wang and Chien, 2012). Wang and Chien (2012) and outside the wine industry. The sample was skewed found that consumers were willing to recommend the heavily toward male readers, a descriptive not uncommon discussed product with others (WOM) when the source was within the blog community in general. found to be credible. Durukan and Bozaci (2012) found that The wine blog participants were invited to discuss their consumers were willing to share WOM in social media when views of what makes a wine blog believable. The inten- the message (information posted) was found to be credible. tion was to isolate the evaluative characteristics of the Support for the positive influence of trust in the WOM be- wine blog and establish a content domain for the con- havior of consumers is found in the relationship marketing struct. To aid discussion, a focus group guide with four literature (e.g., Sichtmann, 2007). In a study of finding the sections of questions was created. The questions in the determinants of eWOM in social networking sites, Chu and first three sections were designed to generate evaluative Kim (2011) found that the higher the level of trust in the con- comments about each of the three theoretical dimensions tact, the greater the likelihood that they would engage in of the wine blog credibility structure (site, message, and eWOM-passing behavior. source) supported in the literature. For example, for blog Therefore, this present study posits (H2b) that an attitude credibility, some of the questions asked and the state- of trust toward a wine blog will positively influence an indi- ments generated were as follows: vidual’s positive WOM behavior about it. 1. When you click on a wine blog site for the first time, what H2b: Positive wine blog trust predicts a behavioral inten- do you initially look for? (Site) tion of recommending the accessed wine blog to others. a. “professionally designed” 2. What are the most important wine blog site elements that Continuing participation with the accessed wine blog would make you stay, read, and/or comment? (Site) (H2c) The value consideration in the calculative component of con- a. “easy to navigate” tinuing participation with the same wine blog is based on the 3. Is good wine information easy to find in wine blogs? perceived benefits of using the wine blog and the cost of (Site) leaving to find another information source. After an individual has determined that a wine blog is a valuable source for a. “searchable” information, the cost of finding another wine-related infor- 4. Do you read the wine blog because of the text, the writer, mation source is associated with time risk. It is important to or both? (Source) the present study to determine whether blog credibility and trust can predict an individual’s decision to return to the same a. “professional experience”

5. How do you know if the writer is knowledgeable regard- traditional media studies, the assessment of message credibil- ing the information they are discussing? (Source) ity was oriented toward capturing the objective, unbiased, just the facts, ma’am nature of the message. Although re- a. “tell me something I do not know” spondents in this study still expressed an interest in content 6. How do you know if the writer is trustworthy regarding that is current, useful, and well written, there also appeared the information they are discussing? (Source) to be much more focus on enjoyment during reading and also content that is opinionated. For example, there was repeated a. “experience with wine, understanding of wine” interest in the entertainment value of the wine blog – wanting 7. How do you perceive the credibility of the message con- entertaining content or liking to laugh while reading a post. tent? (Message) Respondents expressed respect for a writer that could deliver content in an entertaining format. a. “its uniqueness” The respondents also felt that blogs should express the b. “well-written” opinions of the writer. To merely report information without bias suggested to the respondents that the blog was an The final section pertained to overall wine blog credibil- aggregator of information, not a personal voice willing to ity. The participants were asked about the qualities of an take a position. This finding supports Johnson and Kaye’s ideal wine blog, what turns them off, or what makes them (2004) observation that blog readers distrust traditional me- want to return to the wine blog again. An additional section dia and choose blogs as an alternative to traditional informa- was designed with enough freedom in dialog to allow tion sources specifically because blogs do not hide their bias. participants to include what they thought was missed in the In addition, respondents felt that for posted content to pass discussion. muster, it must have a hook. The desire for catchy headlines The intent was to extract the dimensions of blog credibil- to help make a rapid determination of a post’s relevance sug- ity from the perspective of the blog user and also explore gests how Web information seekers look for information – users’ perspectives regarding dimensions from the traditional quickly. These information seekers respect the content that literature. This questioning elicited a set of credibility state- first catches their attention and also helps them make the ments to ensure content validity for the development of the quick determination if the post will be of use or interest wine blog credibility measures. to them. The text transcripts were content analyzed using centering resonance analysis (CRA). The CRA was interpreted on the Statement refinement basis of insights accumulated from the literature, participa- After reducing the statement pool for clarity, redundancy, tion in the wine blogging community, and interaction with and study application, 180 statements remained. The state- participants in the study. The recurring themes within actual ments were grouped according to similar themes, and statements were used to generate an extensive set of items dimensions/sub-dimensions were identified. that most accurately reflect the hypothesized dimensions of To ensure face and theoretical content validity, the re- wine blog credibility (source, message, and site). An initial duced statement pool was sent to five independent judges, pool of 650 statements was developed. three wine blog experts (expertise was subjectively deter- Generally, the findings of the focus groups confirm the mined on the basis of the blogger’s experience with wine importance of the source, message, and site of the wine blog blogging and traffic on the site), and two marketing profes- in its evaluation. The observations in the study coincide with sors specializing in interactive marketing and Web applica- the findings of Fogg et al. (2002). Individuals look at aspects tion. The judges were asked to sort the statements according of the wine blog as a website to determine credibility. Does to the dimensions/sub-dimensions previously identified, al- the site navigate well? Is the layout appealing? Does it appear though they could create dimensions of their own, generically to be professionally designed? Does it have a search function categorize the statements, or not assign the statements at all. or use multimedia? These issues were enumerated in con- From the independent evaluation, 63 statements were retained junction with remarks such as “I’ll move on if.” on the basis of congruent categorization on the dimensions by The preferred source characteristics articulated in this at least four of the five judges. study uphold the general structure of source credibility that prevails in the source credibility literature (Hovland and Weiss, 1951; Berlo et al., 1969; Ohanian, 1990), although Pilot sample the statements reflect the specific character of a wine blog. Early data collection was undertaken for further statement re- For example, demonstration of expertise with wine and/or finement. A questionnaire was developed to include the 63 the industry was important to the user. In addition, the respon- Likert-developed items to validate the credibility scales. dents also desired some level of integrity such as respecting The items were loaded into Questionpro® (online survey comments posted on the blog or revealing conflicts of inter- software) and administered to a sample of five wine bloggers est. In addition, the participants evaluated the appeal of the and five wine blog readers, who were not included in the fi- source and preferred a writer that was like them or shared nal sample of the study. These individuals made comments their viewpoint. on the administration difficulties, question wording, and The message characteristics expressed in this study depart timing of the survey on the online venue. This process re- somewhat from previous credibility scales for media. In sulted in the deletion of six problem items, leaving 57 items

Copyright © 2014 John Wiley & Sons, Ltd. J. Consumer Behav. (2014) DOI: 10.1002/cb Credibility first: the influence of wine blogs for full data collection. The questionnaire was adjusted for 51 percent are over 45 years, 38 percent are inside the wine efficiency and reloaded into the Questionpro® platform. industry, 45 percent earn over 100K, and 67 percent reside in the USA. Sample for credibility scale development After the actual data collection, the data set was examined A viral sampling strategy was implemented over a 40-day to remove any foreseeable inconsistencies in the data that period using all available digital and social media platforms, could add additional error into the subsequent analyses along with an incentive promise of a random drawing for one (Sonquist and Dunkelberg, 1977, p. 41). This editing process of ten e-gift – winners’ choice – certificates valued at $25 reduced the completed survey data set to 263. each. This enhanced practical data collection, completed surveys, and participant representativeness. Scale purification The viral sampling procedure produced 285 completed The original sets of questions were assessed using Likert surveys with a completion rate of 59 percent (285/482) scaling from strongly agre to strongly disagree. Using data reflected by 1618 hits (482 started and 285 completed). The from the random split of the dataset (n = 118), item analysis completion ratio was higher than the average completion ra- was conducted as suggested by a SPSS item analysis white tio found in all of the Questionpro© online surveys. This paper (1998). On the basis of this procedure, 30 items finding is likely due to the sampling procedure of self- remained after the item analysis as seen in Table 1. selected participation. The invitations, initiated from wine blogs, were read by individuals under high involvement with Exploratory factor analysis wine and at least some involvement with wine blogs. Indeed, Using the 30-variable set from the item analysis, the appro- of the 285 completed surveys, 94 percent of participants indi- priateness of factor analysis was satisfied with the Kaiser– cated high involvement with wine. Further, 39 percent have Meyer–Olkin measure for sampling adequacy (0.847) and their own blog, 23 percent have participated with wine blogs Bartlett’s test of sphericity (p < 0.001). An exploratory less than 1 year (32% for 1–2 years, 28% for 3–4 years, and principal component analysis with an oblique promax rota- 17% for five or more years), 32 percent participate over tion was implemented along with a scree test that reduced 2 h/day, 73 percent are men, 80 percent are college graduates, the data to three factors, which explained 46.4 percent of

Table 1. Initial puriﬁed scale of hypothesized credibility structure variable dimension Variable item Variable name Site dimensions (STE CRED) Visual design This wine blog looks professionally designed. sd1 The background colors used on this wine blog are alluring. sd2 Headlines used in this wine blog make it believable. sd3 Functionality This wine blog is easy to navigate. sf1 The use of multimedia makes this wine blog more useful. sf2 This wine blog allows me to search for past content. sf3 Initial appearance I like the layout on this wine blog. sa1 I like the colors used in this wine blog. sa2 Message dimensions (MSG CRED) Content As a reader, I gather useful information from this wine blog. mc1 I do not have to like a topic on this wine blog to read the post. mc2 I like this wine blog to be continually updated. mc3 Style Catchy headlines make the posts on this wine blog more believable. ms1 Catchy headlines add uniqueness to the posts on this wine blog. ms2 I appreciate this well-written wine blog. ms3 Liking I like the entertaining content on this wine blog. ml1 I like to read unbiased wine reviews on this wine blog. ml2 Comments posted on this wine blog make the blog more interesting. ml3 Source (writer) dimensions (SRC CRED) Expertise This wine blogger knows the wine industry. we1 I think this wine blogger is knowledgeable about wine. we2 This wine blogger seems to have a broad knowledge of wine. we3 Trustworthiness I respect the writer of this blog. wt1 I expect this wine blogger to reveal all conﬂicts of interest. wt2 This wine blog writer discloses his or her background. wt3 This wine blogger respects my comments. wt4 Once I established a relationship with this wine blogger, wt5 I never questioned his or her integrity. Appeal I have a lot in common with this wine blogger. wa1 This wine blogger is a creative genius. wa2 Everyone likes this wine blog writer. wa3 I appreciate the effort this wine blog writer puts into his or her posts. wa4 I admire this wine blogger. wa5

Copyright © 2014 John Wiley & Sons, Ltd. J. Consumer Behav. (2014) DOI: 10.1002/cb T. R. Cosenza et al. the variance. Each of these factors related to a hypothetical measurement model. Following the overall model fit assess- dimension of the wine blog – source, site, and message. ment, convergent and discriminant validity was examined A three-factor solution model was retained from the purifica- using multiple measures. tion sample of items, and the factor analysis was again conducted. The overidentified model was examined for offending es- The descriptive variables and the reliabilities for the exploratory timates, and none were found. The CFA indicated that all factor analysis (EFA) are shown in Table 2. The three-factor loadings of indicators were at or above the threshold of solution for wine blog credibility accounted for about 73 percent 0.60. All parameter estimates were significant. The chi- of the total variance. Furthermore, all items in the three-factor square test in the CFA was above the 0.05 minimum solution had 0.70 or greater loadings with no cross loadings on (χ2 = 33.409, df = 24, sig = 0.096). This statistic showed sup- the other factors. This shows evidence for the independence of port for acceptable fit. Additionally, other fit indices chosen the constructs and adequate internal consistency (Gorsuch, 1983). for evaluation were acceptable, indicating a good fit between The resultant attribute structure for each dimension of theory and data. The root mean square error of approxima- seems to be consistent with the original hypothesized structure tion value of 0.056 was below the 0.08 cutoff (MacCallum from the literature. Thus, the remainder of the study uses the et al., 1996). The comparative fit index was 0.980, above following dimensions and the resultant study items (which will the cutoff of 0.950 (Hu and Bentler, 1999). The standardized be measured in a 7-point Likert format) as shown in Table 2. root mean square residual value of 0.034 was below the recommended 0.05 cutoff (Byrne, 2001). Further, the normed chi-square (χ2/df) had a value of 1.392. This value fell under Construct validation 3.0, the recommended range for model parsimony (Hooper The credibility literature review suggests that the new scale et al., 2008). The CFA measurement model of the credibility fi should possess three rst-order factors that correspond to construct is depicted in Figure 2. each of the three dimensions of wine blog credibility (site, message, and source). These factors collectively measure wine blog credibility as a whole but are statistically evaluated as a three-factor correlated model (Tian et al., 2001). Thus, the measurement model was evaluated on the basis of the hypothesized dimensions of wine blog credibility. Confirmatory factor analysis (CFA) was used to assess the reliability and validity of each dimension of the wine blog credibility construct. A measurement model with the three factors was specified, and each of the nine indicators was indicated to a factor according to the results of the EFA. Using the second half of the data (n = 145), the confirmatory factor model was estimated using AMOS 16.0 (SPSS, Inc.), maximum likelihood estimation (Arbuckle, 2009). To obtain multiple indicators for the hypothesized constructs, a more lenient factor loading of 0.60 was expected for retention of the indicators (Hair et al., 1998). In addition to chi-square, the indices of root mean square error of approximation, comparative fit index, standardized root mean square residual, and normed chi-square Figure 2. Confirmatory factor analysis measurement model of (CMIN/df) were evaluated to assess the overall fit of the credibility construct.

Table 2. Descriptive statistics from exploratory factor analysis Component Factor Percentage of Scale name Item loading Eigenvalue variance reliability SRC CRED 3.691 41.014 0.872 I think this wine blogger is knowledgeable about wine (we1). 0.923 This wine blogger knows the wine industry (we2). 0.907 This wine blogger seems to have a broad knowledge of 0.853 wine (we3). STE CRED 1.708 18.976 0.800 I like the layout on this wine blog (sa1). 0.886 This wine blog is easy to navigate (sf1). 0.841 This wine blog looks professionally designed (sd1). 0.806 MSG CRED 1.184 13.155 0.737 Catchy headlines make the posts on this wine blog more 0.868 believable (ms1) I like the entertaining content on this wine blog (ml1). 0.804 Catchy headlines add uniqueness to the posts on this 0.779 wine blog (ms2).

Convergent validity and reliability Table 4. Discriminant validity assessment using average variance Internal consistency is a necessary condition for convergent extracted versus shared variance test validity and was evaluated using Cronbach’s alpha and Variable Items STE CRED MSG CRED SRC CRED composite reliability. All three constructs exceeded the rec- fi STE CRED 3 0.585 0.091 0.132 ommended threshold of 0.70 for coef cient alpha (Nunnally, MSG CRED 3 0.303 0.526 0.275 1978). The composite reliabilities for each construct SRC CRED 4 0.364 0.524 0.711 exceeded the recommended threshold of 0.70 (Churchill, 1979). Convergent validity was indicated by multiple Note: Correlations are below the diagonal, squared correlations are above the diagonal, and average variance extracted estimates are presented on the methods of analysis (Table 3). All factor loadings of individ- diagonal. ual indicators on their respective constructs were positive and significant as suggested by Bagozzi and Yi (1988). Seven out of nine loadings met or exceeded the 0.70 threshold sug- PHASE 2: STRUCTURAL MODEL TESTING FOR gested by Nunnally (1978). The average variance extracted NOMONOLOGICAL VALIDITY (AVE) also exceeded the recommended threshold of 0.50 for each of the constructs (Fornell and Larcker, 1981). Sample The analytics for testing the structural equation modeling (SEM) shown in Figure 1 and specified in Figure 3 utilize Discriminant validity the original combined sample of 263. Given the sample size, Discriminant validity was assessed by multiple measures. A the basic data assumptions of SEM were generally satisfied. comparison between the AVE of each construct and the Univariate normality was satisfied on the basis of Kline’s shared variance between the constructs was conducted (2005) recommendation that only variables greater than |3| (Fornell and Larcker, 1981). As reported in Table 4, the indi- for skewness or |8| for kurtosis indicate concern. However, vidual variance extracted (AVE) for each latent variable the Mahalanobis d-squared observations (from Mahalanobis, exceeded the shared variance (squared correlation) between 1930) indicated 45 outliers for potential deletion. All cases the constructs. Discriminant validity was also tested in a were examined, and 14 cases were dropped because of ex- chi-square difference test of the three paired constructs treme values, leaving 249 cases for analysis. (Anderson and Gerbing, 1988). In one analysis, the correlation between constructs was set at 1.0, whereas the other Measures was permitted to be freely estimated. As noted in Table 5, In addition to the Likert credibility scales, the questionnaire all three comparisons between the constructs were statisti- included multiple-item, 7-point Likert scales for the remain- cally significant. Further, acceptable goodness-of-fit mea- ing variables in the model. These scales were chosen on the sures for the measurement model indicate that the constructs basis of compatibility with theoretical focus and were in the model are different. Thus, all constructs achieved dis- adapted to meet the needs of this study. For example, trust criminant validity. was measured by an adapted scale from Young and Albaum These analyses suggest that the developed scale ade- (2003). The three-item scale was used to measure the central- quately meets or exceeds the empirical requirements for scale ity of emotions in the development of trust. Intention to fol- development. However, it appears that the individual items low recommendations was measured using a three-item scale retained and the number of items retained per factor may adapted from McKnight et al. (2002a). The original scale be inadequate to capture the content validity of the hypothe- was created to measure the behavioral intention-subjective sized dimensional structure. This issue will be further ad- probability of acting on advice. An adapted scale from dressed in the empirical testing of the model, which Zeithaml et al. (1996) was used to measure WOM intentions. examines the causality of the dimensions within the predic- The three items were labeled a priori in the original scale as tive model. WOM intentions. A three-item scale to measure intention to

Table 3. Convergent validity and reliability assessment Items Standardized loading Critical ratio Cronbach’s alpha Scale composite reliability Average variance extracted STE CRED 0.800 0.807 0.585 sa1 0.956 10.766 sd1 0.689 8.098 sf1 0.722 8.575 SRC CRED 0.872 0.880 0.711 we1 0.983 11.263 we2 0.879 12.637 we3 0.750 9.531 MSG CRED 0.737 0.764 0.524 ms1 0.852 9.771 ms2 0.600 6.628 ml1 0.701 7.921

Table 5. Discriminant validity assessment using χ2 difference for paired constructs Construct pairing χ2 free χ2 ﬁxed χ2 difference χ2 signiﬁcance STE CRED/SRC CRED 52.935 193.731 140.796 *** SRC CRED/MSG CRED 15.126 102.620 87.494 *** STE CRED/MSG CRED 20.179 170.561 150.382 ***

Note: Signiﬁcance determined by χ2 difference test. *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 3. Structural model for hypothesized causal testing. continue with the wine blog was adapted from multiple Following an evaluation of overall fit, the model was ex- indicators offered in the literature (Allen and Meyer, 1990; amined at the measurement level and then at the structural Morgan and Hunt, 1994; Hsu and Lin, 2008). The amalgam level. This two-step procedure was intended to ensure that of items was chosen on the basis of the premise that an ex- the structural analytics stemmed from a measurement model change partner (wine blog participant) will desire to cultivate with desirable psychometric properties so that the conclu- or maintain a relationship with a wine blog that is perceived sions are meaningful. to be a valuable source of information. Reliability measures of all the scales used in this study along with sample items Assessment of measurement model fit are reported in Table 6. The results of the measurement model assessment are reported in Table 7. The estimated loadings of the indicators all achieved significance at the p < 0.001 level. The scale Structural equation model analytics composite reliability for each construct achieved a composite To test the relationships among the variables, the hypothe- reliability of greater than the 0.70 recommended cutoff (Hair sized model was estimated using maximum likelihood et al., 1998). Finally, the AVE fell short of the recommended method and evaluated for model fit according to the same cutoff of 0.50 (Hair et al., 1998) for two of the constructs, criterion used in the CFA. Although the χ2 was significant, indicating that more than 50 percent of the variances in the indicating an unacceptable fit between the observed data latent constructs was not accounted for. and the proposed model, all other fit statistics were within a range suggesting acceptable, albeit marginal, fit (Figure 3). Assessment of structural model fit A specification search was conducted in an effort to achieve Table 8 depicts the results of the assessment of the structural a better fit for the model if the modifications made strong model, conducted to determine if the model could offer any substantive sense (Jöreskog and Sörbom, 1993). The size of interpretive meaning. the modification indices did not indicate any theoretical sug- The directions of the parameters between the latent gestions for model respecification. variables were all found to be as hypothesized (+). The

Table 6. Measurement item descriptives (means, standard deviations, and reliabilities of model data set) Cronbach’s Scale name Scale/adapted scale items Item names Mean SD alpha STE CRED I like the layout on this wine blog. sa1 5.45 1.047 0.835 This wine blog looks professionally designed. sd1 5.38 0.977 This wine blog is easy to navigate. sf1 5.41 0.996 MSG CRED Catchy headlines make the posts on this blog more believable. ms1 5.35 0.917 0.724 Catchy headlines add uniqueness to the posts on this wine blog. ms2 5.18 1.069 I like the entertaining content on this wine blog. ml1 5.49 0.784 SRC CRED I think this wine blog writer is knowledgeable about wine. we1 6.03 0.904 0.881 This wine blogger knows the wine industry. we2 5.75 1.071 This wine blogger seems to have a broad knowledge of wine. we3 5.84 0.932 TRUST I can rely on this wine blog. tr1 5.48 0.972 0.795 I have conﬁdence in the accuracy of the information on this wine blog. tr2 5.64 0.869 This wine blogger is only concerned with promoting his or her interests. (r) tr3 5.63 1.175 FREC I would feel comfortable acting upon the information given to me frec1 5.65 0.810 0.902 by this wine blog. I would make a wine-related purchase based on the advice I am frec2 5.21 1.173 given on this wine blog. I would follow wine recommendations from this wine blog. frec3 5.34 1.146 I would not hesitate to purchase wine-related products recommended frec4 5.05 1.250 on this wine blog. WOM I would say positive things about this wine blog to other people. wom1 5.92 1.057 0.869 I would recommend this wine blog to someone who seeks my advice. wom2 5.46 1.270 I would encourage friends and relatives to seek wine-related wom3 4.98 1.372 information from this wine blog. CBP I will return frequently to make comments in this wine blog. cbp1 4.73 1.424 0.765 I will return frequently to read this wine blog. cbp2 5.76 1.051 I have no loyalty to this wine blog. (r) cbp3 4.02 1.756 I feel that there are no beneﬁts associated with participating in this wine blog. cbp4 5.37 1.181

Table 7. Unidimensionality and reliability of measurement model Construct Item Standardized loading Mean SD Average variance extracted Scale composite reliability STE CRED sa1 0.829*** 16.03 2.819 0.629 .837 sd1 0.780*** sf1 0.770*** MSG CRED ml1 0.613*** 17.62 2.230 0.489 0.739 ms1 0.838*** ms2 0.626*** SRC CRED we1 0.847*** 16.23 2.613 0.720 0.887 we2 0.942*** we3 0.757*** TRUST tr1 0.913*** 16.74 2.530 0.599 0.812 tr2 0.805*** tr3 0.562*** FREC frec1 0.756*** 21.25 3.875 0.707 0.906 frec2 0.840*** frec3 0.883*** frec4 0.877*** WOM wom1 0.790*** 16.36 3.297 0.695 0.872 wom2 0.923*** wom3 0.780*** CBP cbp1 0.591*** 19.88 4.142 0.444 0.760 cbp2 0.763*** cbp3 0.659*** cbp4 0.641***

Notes: ***P< 0.001. magnitudes of the estimated parameters were evaluated for assess the accuracy of the prediction in the structural paths, strength of the hypothesized relationships. Five of six param- the amount of variance in each of the endogenous variables eters were statistically significant as indicated by a critical ra- (as indicated by the squared multiple correlations (R2)) was tio in excess of |1.96|. Three parameters failed to achieve examined (Figure 3). In this present model, a strong effect practical significance (structural coefficients above 0.30), in- size was reported for trust (0.49), following recommenda- dicating that the paths were not supported by the data. To tions (0.62), and continued blog participation (0.64), whereas

Table 8. Parameter estimates for hypothesized causal paths Hypothesis Path Structural coefﬁcient estimate Standard error Critical ratio H1a (+) STE CRED → TRUST 0.19 0.068 2.894** H1b (+) MSG CRED → TRUST 0.14 0.122 2.099* H1c (+) SRC CRED → TRUST 0.56 0.062 8.873*** H2a (+) TRUST → FREC 0.78 0.048 11.380*** H2b (+) TRUST → WOM 0.03 0.066 0.394 H2c (+) TRUST → CBP 0.80 0.090 8.438***

*p < 0.05, **p < 0.01, ***p < 0.001. a negligible effect size was reported for WOM (0.00). To are statistically and practically predictive of an attitude of further evaluate the influence of credibility on behavioral trust (TRUST) in the wine blog. intentions, the indirect effect of each of the credibility H1b postulated that there would be a positive relationship constructs on the dependent variables was examined. Table 9 between the credibility beliefs of the wine blog site (STE reports the results of this evaluation. Two of the three credi- CRED) and an attitude of trust (TRUST). The direction of bility dimensions produced weak paths, suggesting that the path coefficient between the two constructs supported a source credibility was the greatest contributor to the propor- positive relationship. As expected, the results indicated that tion of variance accounted for in both following recommen- site characteristics had a significant relationship with the for- dations (R2 = 0.615) and continued wine blog participation mation of trust attitude as evidenced by a critical ratio of (R2 = 0.635). greater than |1.96| (2.90). Therefore, the findings provide support for H1a. However, the magnitude of the estimated parameter lacked practical significance, falling below the cut- MAIN STUDY FINDINGS off of β ≥ 0.3 (0.190). Although the positive direction of the relationship was as hypothesized and the path was statisti- Role of credibility beliefs on trust cally significant, its interpretation is somewhat meaningless The first set of hypotheses stated that source, site, and mes- because of the effect size indicated by the regression sage credibility beliefs would be positively related to an atti- coefficient. tude of trust. The rationale for the general hypothesis was H1c suggested that there would be a positive relation- based on the predominance of trust studies that focus on ship between the credibility beliefs of the wine blog mes- the role of credibility in the formation of trust. The general sage (the posted content; MSG CRED) and an attitude of hypothesis yielded to three sub-hypotheses that reflect the in- trust (TRUST). The positive relationship between the dependent contribution of characteristics of the three theoret- constructs was supported by the direction of the path co- ical dimensions of the wine blog (site, message, and source). efficient. The critical ratio of 2.31 indicated a significant The positive influence of the independent dimensions of relationship between message characteristics and the credibility on trust is statistically supported by the data; how- development of trust attitude. Therefore, the findings ever, the dimensions are not equal in their predictive power. provide support for H1b. However, the magnitude of the H1a proposed that there would be a positive relationship estimated parameter lacked practical significance, falling between the credibility beliefs of the source (author) of the below the cutoff of β ≥ 0.3 (0.144). The positive relation- wine blog (SRC CRED) and an attitude of trust (TRUST). ship of the path was in the hypothesized direction and sta- The direction of the path coefficient was found to be in the tistically significant, but its interpretation is somewhat direction of the hypothesized relationship. Further, the results meaningless given the effect size indicated by the regres- indicated a significant relationship between SRC CRED and sion coefficient. TRUST, evidenced by a critical ratio of 8.998. The magni- Before reflecting on the potential reasons for the preced- tude of the estimated parameter also achieved practical sig- ing results, it may be helpful to consider the items retained nificance (β = 0.571). Thus, the findings provided strong for wine blog credibility measurement. The characteristics support for H1c. Positive source characteristics (SRC CRED) of source credibility (SRC CRED) unilaterally merged on items of expertise. Other items pertaining to trustworthiness or appeal had substantial cross loadings on at least three Table 9. Standardized indirect path effects of credibility other components and were not retained for further analysis. dimensions on outcome variables The items retained for site characteristics (STE CRED) could Indirect effects be superficially evaluated and captured the design, function, Credibility and appeal of the wine blog as a website. The characteristics dimension FREC WOM CBP of message credibility (MSG CRED) in the retained items SRC CRED 0.438 0.015 0.445 reflected the believability, uniqueness, and entertainment MSG CRED 0.109 0.004 0.111 value of the message content. STE CRED 0.150 0.005 0.152 The explanation for the aforementioned dimensional Bold emphasis indicates strength of source credibility paths. influence in the findings may be related to the type of

Copyright © 2014 John Wiley & Sons, Ltd. J. Consumer Behav. (2014) DOI: 10.1002/cb Credibility first: the influence of wine blogs information in the wine blog, the level of analysis used in as- Predictive relationship between trust and behavioral sessment, or an issue of measurement. The message of the intentions wine blog is communicated at the interpersonal level and is The second hypothesis stated that an attitude of trust with a conceptualized as an electronic type of WOM information. given wine blog would be positively related to behavioral in- Consumers are increasingly accepting online opinions tentions. The rationale for this general hypothesis was based (WOM) from people that they do not know (Godar, 2005), on the importance of trust in predicting behavioral intentions and the same seems to be true for wine blogs. The WOM demonstrated in online recommendation and e-commerce literature provides evidence for the influence of a personal studies (Gefen et al., 2003; Pavlou and Gefen, 2004). The exchange of information among acquaintances on attitude general hypothesis was further subdivided to reflect a hypo- change and decision-making behavior (Herr, et al., 1991). thetical range of behavioral acts that may transpire from trust Research studies also indicate that the influence of the in an attitude object (wine blog) – to follow recommenda- WOM communication is based on the perceived credibility tions given in the wine blog, to spread WOM regarding the of the source (Bansal and Voyer, 2000). Following the said wine blog, and to continue participation with the wine blog. logic, it is the credibility of the personal source that drives H2a postulated that there would be a positive relationship the use and influence of the information on the wine blog. between TRUST and FREC, the intention to follow recom- Therefore, evaluation of the source dimension of the wine mendations offered in the blog post. The direction of the path blog is preeminent when evaluating the wine blog as a coefficient supported the hypothesized relationship. Further, whole. a critical ratio of 11.8 indicated a significant relationship Newhagen and Nass (1989) demonstrated that the criteria between the constructs. In addition, the magnitude of the re- used to assess credibility are related to the source level of lationship achieved practical significance (β = 0.801). Thus, analysis that varies with the type of media evaluated. They the findings provided strong support for H2a. A positive rela- reported that television credibility was assessed at the indi- tionship exists between TRUST and FREC, an intention to vidual level as an aggregate of on-camera personalities be- follow the recommendations set forth in the wine blog. cause of the dynamic and immediate nature of television Although this finding is not surprising, it is nevertheless the news delivery. Newspaper credibility was evaluated at the in- most important finding with respect to wine blogs’ potential stitutional level because of the perceived distance between impact on marketing activity. Do wine blogs influence peo- readers and the people that produce the papers. This finding ples’ decisions to purchase? They certainly do. is important to consider with respect to the wine blog. Like H2b suggested that there would be a positive relationship television, the writer of the wine blog is a personality. The between TRUST and the intention to spread WOM about the wine blog is dynamic, and content changes at the hands of wine blog. Although the direction of the path coefficient of- the author. Thus, it is expected that the credibility of the wine fered support for the hypothesized relationship between the blog is at the individual level and the focus is on source two constructs, the magnitude of the estimated parameter characteristics. did not achieve statistical significance (CR = 1.107, Further, the wine blog is an individual voice, and credibil- p = 0.269) or practical significance (β = 0.076). Thus, H2b ity is ultimately the responsibility of the author. Regardless was not supported by the data. the dimension, the evaluation is a reflection on the author. The non-significant finding associated with the estimated Does he or she demonstrate expertise in blogs, in writing parameter between the TRUST and WOM suggests irrele- ability, and in wine? Does the site have good design, func- vance of the path in the model. The sample size was adequate tionality, and appeal? It does if the author is an experienced and was not suspected of causing the non-significant param- blogger. Is the content entertaining, original, and believable? eter. In addition, the path did not demonstrate any practical It is if the author is an experienced writer. Is the author an ex- significance, nor did the R2 (0.010) suggest any explanatory pert in wine? He or she is with demonstrated knowledge of power of the independent latent variables that were expected wine and the industry. to impact WOM. Therefore, the path is a prime candidate for Although the previous arguments are intuitive on the basis elimination from the model in the interest of parsimony of the known importance of source credibility in media and (Byrne, 2001). This finding may have resulted for a couple WOM evaluation, it would be inappropriate to minimize of reasons. First, the measurement items adapted for this the importance of content validity in the measurement of a study may have incorrectly tapped into the spread of WOM construct. The qualitative phase supported the hypothesized pertaining to the wine blog. The items refer to spreading dimensions of wine blog credibility and produced solid WOM about the wine blog (the attitude object) as an infor- sub-dimensions among the constructs. However, subsequent mation source instead of the spread of information found analysis eliminated the sub-dimensionality of the source within the wine blog. In retrospect, of importance to the busi- credibility construct. The measurement items focused on ness reader is the estimate of the WOM influence in the reach the assessed expertise of the source. Further, it is not appar- of content found in the wine blog. The focus on the attitude ent how the conclusion of expertise was assessed. It could object instead of the information found on the wine blog have resulted from characteristics of the source directly or in- may have limited a significant, positive impact of TRUST stead from the message content or quality of the blog site. on WOM. Further, it may have been problematic that the The absence of knowledge regarding the potential interaction statements assumed that the peer group of the survey partic- of dimensional evaluation makes the interpretation of results ipants would be candidates for wine blog use (e.g., I would somewhat questionable. recommend this wine blog to someone who seeks my

Copyright © 2014 John Wiley & Sons, Ltd. J. Consumer Behav. (2014) DOI: 10.1002/cb T. R. Cosenza et al. advice). The outcome may have been different had the re- author, from a forum because control of conversation is held spondents been asked about their intention to pass the infor- by the author, from review sites because of the interactive mation from the wine blog to someone seeking their advice. availability with the author, and from other media outlets be- Second, it is likely that the lack of a significant relation- cause of the expected personal voice. It appears that the crux ship between TRUST and WOM is due to a theoretical of the wine blog is centered on the author. misspecification of the model. The relationship between trust Medium theory (McLuhan, 1964) explores the influence in an individual and subsequent positive WOM is supported of communication technologies and may offer further expla- in the employee–customer relationship literature (Gremler nation for the observation of source influence. McLuhan et al., 2001). However, Gremler et al. (2001) focused on in- (1964) stressed the importance of studying the effects of a terpersonal trust developed over time within an ongoing rela- medium as opposed to the message conveyed. He tionship (bank or dentist). This present study specified that distinguished media on the basis of the characteristics of the respondent was to answer the survey with respect to the the medium and the cognitive process each required. Accord- most recently visited wine blog. It is likely that the majority ing to McLuhan (1964), each new channel of media shapes of participants made a decision of initial trust in the wine the way that individuals, people, and society view the world. blog rather than an interpersonal bond with the writer. This Every medium is a language, and new filters are required to level of trust may have been enough to motivate an individ- view and listen to its messages. Because blogs are a new me- ual to follow a recommendation in the wine blog or return to dium, they have their own character, and the purpose of this the wine blog in the future, yet not at a relationship level that study was to examine this character. Blogs are hypothesized would motivate them toward transference of their own cred- as a personal voice on the Web, unlike previous traditional ibility to someone else on behalf of the wine blog. media or other online sources. As it turns out, this dis- H2c proposed that there would be a positive relationship tinguishing characteristic is the filter that encourages interac- between TRUST and CBP, the intention to continue partici- tion with the blog – evaluation, trust, and behavioral pation with the trusted wine blog. The direction of the path intentions. But what does this mean going forward? This coefficient supported the positive relationship between the study suggests that credibility evaluation lies at the overall constructs. The results also indicated that the relationship be- source level but may also suggest that the evaluation might tween TRUST and CBP reached statistical significance as be more dimensionally complex than previously considered. evidenced by a critical ratio of 8.43. In addition, the magni- The current literature does not offer a methodical structure tude of the estimated parameter achieved practical signifi- for analysis. As previously noted, credibility in the cance with a reported path coefficient of 0.790. Thus, the blogosphere is studied in various genres, using various mea- findings provided strong support for H2c. A positive rela- sures, with various outcomes of interest. Without a continu- tionship exists between TRUST and CBP, an intention to ous, methodical research stream, it is difficult to suggest continue participation with the target blog in the future. where generic conclusions reside. However, within the con- Apparently, an attitude of trust in the wine blog creates stick- text of wine blogs and any similar genre (i.e., social context iness. Thus, it can be concluded that when the wine blog is or experiential), this study offers some interesting observa- determined to be a trusted source of information, it will be tions from both the qualitative and quantitative studies. It is retained in an evoked set of preferred information sources. clear that blogs are an accepted source of information for To augment the earlier results, an examination of the the socially driven, experience-laden context of wine, un- indirect influence of the three credibility variables on the out- doubtedly to mitigate some aspect of risk (Atkin and Thach, come variables reveals the same observed disparity between 2012). The qualitative phase revealed comments such as the credibility variables in predictive power on the outcome “wine blog writers help me understand wine,”“I’m looking variables. Although the path coefficients suggest a strong for things that really teach me about wine,” or “I read them predictive relationship between TRUST and the outcome for information about XXX, information about wineries, in- variables following recommendations (FREC) and continued formation about the wine industry.” This study also offers wine blog participation (CBP), the indirect effects from both some support for Doyle et al. (2011). The indicators retained site credibility (STE CRED) and message credibility (MSG in the factor analysis for source credibility centered on exper- CRED) are weak. Again, the strong indirect effects of source tise. Thus, we know that source trust in the wine blog is credibility (SRC CRED) indicate its contribution to the based on the expertise of the wine blogger, as opposed to proportion of variance accounted for in both following their trustworthiness or appeal. However, it is unknown if recommendations (FREC) and continued wine blog partici- the expertise was signaled from commercial product-related pation (CBP). knowledge or from non-commercial product-related consumption experience because the “last blog visited” was Theoretical development of blog credibility research not specified as such in the survey. Further, on the basis of The most interesting finding in this study was the over- the categories of blogs identified by Thach (2010), it is not whelming importance of source in predicting a decision to surprising that both commercial and non-commercial trust and intention to act. The relevance of the source may bloggers signal ability first for trustworthiness. Expertise is stem from what a blog is and what it is not. The blog is sim- what gives them authority to influence the reader. Further, ilar to face-to-face WOM because of the interpersonal-type this study supports the influence of the wine blog in the con- communication delivered. The blog is different from a tradi- sumer decision-making process that has been previously tional website because of the changing commentary from the noted in other blog study product categories such as medical

Copyright © 2014 John Wiley & Sons, Ltd. J. Consumer Behav. (2014) DOI: 10.1002/cb Credibility first: the influence of wine blogs and beauty products (Wu and Lee, 2012), electronics diluted the rationale and focus on the interpersonal nature (Saxena, 2011), or online shopping (Hsu et al., 2012). of the blog relative to trust behavior. An artificial experiment would have given more control but would have sacrificed a rich study of the real character of wine blogs. DISCUSSION Another limitation in this study relates to the method of recruitment in the sample population. Non-probabilistic sam- The research described in this manuscript was undertaken to pling creates sample bias. The respondents were recruited develop and validate a scale for the measurement of the cred- through purposive sampling, and the initial wine blogs and ibility of wine blogs as a new media source of information. It subsequent viral additions may not have tapped the popula- was specifically chosen to address the gap in research regard- tion of wine blog users. Therefore, the sampling procedure ing the credibility of new media channels. The resulting scale limited external validity. measures the three correlated dimensions of wine blog cred- The construct measurements for the individual dimen- ibility. A conceptual model was tested to investigate the im- sions of source and message credibility are another limitation pact of wine blog credibility on behavioral intentions as in this study. It is apparent that the evaluation of wine blog mediated by an attitude of trust. Thus far, little (if any) re- credibility at the structural level is somewhat interrelated. search has attempted to quantify the behavioral effects of This observation is the likely cause for the substantial cross blogs as a domain of marketing activity. This study is a first loading that occurred in the EFA that produced a less than step toward that goal. desirable number of items for each factor. For content valid- The search for a measure of wine blog credibility insti- ity, the measurement of a factor requires multiple items that gated a look at the structure of the blog to ascertain an appro- capture the identified facets of the construct. The qualitative priate structure for the dimensionality of the construct. Thus, phase was conducted to ensure content validity for each of credibility literature pertaining to websites, media, and inter- the credibility dimensions. However, the item analysis and personal communication was reviewed. It was presupposed EFA limited the full scope of measurement. When reduced, that characteristics that directly related to the website as a the items to measure source captured the sub-dimension of platform, message content, and the source as a personal com- expertise alone. This inadequacy of measurement may have municator would be significant in a credibility assessment. inflated the importance of source credibility findings. More- The qualitative portion of this research corroborated this di- over, the items to measure message credibility focused on as- mensional character of wine blogs in credibility evaluation. pects that were less cognitively assessed than hoped. This Although all three credibility dimensions show a significant measurement issue may have also reduced the importance positive impact on the decision to trust the wine blog, only of the dimension in predicting trust in general. Further anal- source credibility indicates practical significance. In addition, ysis is warranted to examine the interrelationships of the sub- the data support the impact of wine blog credibility on two of dimensions that impacted the factors produced in the EFA. the three behavioral intentions as mediated by trust. There Further, as stated in the methodology, respondents were are several limitations that may have compounded some of asked to reference their responses to the wine blog that they the less than desirable results of this first effort. most recently visited. The literature suggests that trust may Although considerable controls were utilized to eliminate develop over time or may occur swiftly depending on con- bias/error in this study, the results might be far from conclu- text. This study had no means to assess the level of activity sive. The single product domain (wine blogs) might not be as respondents had with the wine blog that they last visited. appealing to some researchers as using a cross section of As such, the predictive power of the individual dimensions blog venues. However, it would be difficult to do the follow- may have been diminished. If they had no prior experience ing: (1) intensely study all of these blogs within a reasonable with the wine blog that they last visited, surface characteris- timeframe and (2) qualitatively evaluate the interpersonal nu- tics may have been more critical, or the evidence of wine ances of all blog venues in order to establish measures and blogger expertise may have superseded other characteristics. relationships for the modeling phase of the study. In the same If they had prior experience, it is doubtful that site character- vein, the sample of participants used to measure the concepts istics would have predicted. in the second phase model came from a sample of readers This research may have ultimately raised more questions and participants of only wine blogs. Therefore, it would be than it answered. Little research has explored the characteris- problematic to generalize the results of this study to a broader tics of blogs as an information source. Thus, there are further blog population outside the specific wine venue selected. opportunities for research. The current study needs to expand Although there seems to be parallels between wine and other and include other populations and product categories and social products, caution is suggested in extrapolating the also consider how the perceived credibility of blogs com- results. pares with their traditional counterpart. During the current In Phase 2 of the study, trust behavioral intentions were study, animosity between traditional wine media and wine examined with SEM. Examining a behavioral outcome and bloggers was observed. There seems to be a battle for attributing it to antecedent effects through survey and SEM territory, and credibility is at the center of the argument. might lend rise to criticism as to why experimentation was Traditional media has been forced to downsize. Subscriptions not used as a method of analysis. In order to use experimen- have declined and so have advertising sales. At the same time, tation in this research, the creation of quasi-real blogs that bloggers have vastly increased their readership and profile as would be necessary for the experiment would have at best a viable information source.

This situation is not isolated to the wine industry. A New elaboration process of the individual credibility dimensions York Times article (Wilson, 2009) profiled a young fashion with respect to establishing trust. blogger that was given a front row seat at shows such as Marc Jacobs in Milan. Many designers have adjusted to the new fashion delivery system. Their collections are posted in cyber- ACKNOWLEDGEMENTS space long before the lead time in traditional media can do likewise. Therefore, the fashion industry is a viable next step This research was conducted through the support of the to test the developed blog credibility measures and the rela- wine blogging community. Personal relationships were tionships set forth by the model in this study. In addition, developed with wine blog writers, wine blog readers, wine comparative research between blogs and traditional media is PR professionals, wine makers, wine sellers, and wine blog warranted. Further, because blogs are a global phenomenon, naysayers. Of course research must be grounded in theory, cross-cultural implications should also be investigated. but insights from the trenches give practical insights to Again, this study did not use experimentation in an effort results. From the point of industry, this research stream to more realistically identify characteristics of blogs that help offers some answers to the crux of the social landscape users establish credibility that impacts an attitude of trust. dilemma. Questions arise such as “do I have influence?” Now that a rich pool of characteristics have been identified, or “do they have value?” It all boils down to “what is the the obvious next step is to set up an experiment using influence of wine blog?” That question is only answered quasi-real blogs to better test the relationship between eval- through finding what makes a particular wine blog uation of the blog and an attitude of trust. influential – its credibility. The main objective of this study was to identify the indi- The authors thank the wine blogging community for making cators of the three individual components of wine weblog this research possible. credibility (media, message, and source). Interaction effects between media, message, and source credibility are supported by a variety of traditional and online research findings (e.g., message influences source evaluation, source influ- BIOGRAPHICAL NOTES ences message evaluation, and perceived credibility is a Tracy Rickman Cosenza, Ph.D. (Auburn University), is currently function of both source and channel characteristics; Slater the Coordinator of Social Media Programs and Marketing Instructor and Rouner, 1996). Although it is important to study and in the Department of Marketing and Supply Chain Management in understand the complexity of credibility assessment interac- the Fogelman College of Business and Economics at the University tions, it was beyond the scope of this research. However, fu- of Memphis. Her teaching interests are in social media marketing, in particular, IMC strategy development and control. Tracy’s research ture research should examine more carefully the intertwined involves the use of social media platforms to enhance customer use, functions of blog media, message, and source credibility for satisfaction, loyalty, and lifetime value (continuance). Her research credibility evaluation and their impact on behavioral inten- has been published in the Journal of Fashion Marketing and Man- tions as mediated by trust. agement, The Journal of Transportation Management, and DMEF In terms of WOM communication, additional research is Extended Abstracts. needed that focuses on community in the blogosphere as it Michael Solomon, Ph.D., is a Professor of Marketing and Director of pertains to the diffusion of information. Blogs are part of a the Center for Consumer Research in the Haub School of Business at ’ ’ social network, and information flows through the links. Saint Joseph s University in Philadelphia, USA. Prof. Solomon s fl primary research interests include consumer behavior and lifestyle There are variables that impact the WOM in uence in the issues, branding strategy, the psychology of fashion, and marketing community. Therefore, research is warranted to study the applications of virtual worlds and other new media. He is the author role of blogs’ (eWOM) credibility in the flow of information of several leading textbooks for Pearson Education including Con- in a network model of influence (Watts and Dodds, 2007). sumer Behavior: Buying, Having, and Being 11/e, Marketing: Real Finally, other variables should be examined that are rele- People, Real Choices 7/e, Better Business 3/e, and most recently, Social Media Marketing (the first textbook on this topic). His most vant to the model. In this research, risk was not directly mea- recent trade book, The Truth about What Customers Want, was sured. Instead, we assumed that it is mitigated by trust when published by FT (Financial Times) Press. Prof. Solomon is fre- there is an intention to engage in a behavior. The literature quently quoted in magazines and newspapers, including Newsweek, review offered specific risks associated with wine purchase The New York Times, and The Wall Street Journal. He has served as from the wine-related marketing literature. Therefore, future a consultant to numerous corporations including Calvin Klein, Intel, Procter & Gamble, Microsoft, State Farm Insurance, Under Armour, research should measure the associated risk variables to more and United Airlines on issues relating to consumer behavior, fully examine the relationship between trust in the blog and marketing strategy, advertising, and retailing. risk-laden behavioral outcomes. Propensity to trust is another fi Wi-Suk Kwon is a Human Sciences Associate Professor of Retailing relevant variable to the model. This study speci ed that trust at Auburn University. She holds a Ph.D. in Consumer Sciences is not a personality trait but rather occurs in a specific situa- from The Ohio State University with minors in Marketing and tion. However, some individuals trust others more freely, Quantitative Psychology. Her research revolves around consumer whereas others have more difficulty. It is also likely that a psychology, branding, multichannel retailing, and retail technolo- propensity to trust moderates the relationship between credi- gies. Her research has been funded by the National Science Foundation and US Department of Commerce and published in bility of the wine blog and an attitude of trust and should be Journal of Retailing, Journal of Business Research, Computers in examined in subsequent testing. And it may also prove useful Human Behavior, Journal of Electronic Commerce Research,and to test the moderator effects of wine blog involvement on the Clothing and Textiles Research Journal, among many others.

REFERENCES Doyle JD, Heslop LA, Ramirez DC, Armenakyan A. 2011. Trust building in wine blogs: a content analysis. International Journal of Wine Business Research 24(3): 196–218. Alexander JE, Tate MA. 1999. Web Wisdom: How to Evaluate and Durukan T, Bozaci I. 2012. A survey on determinants of word of Create Information Quality on the Web. Lawrence Erlbaum: mouth in social media. International Journal of Economics 1 Hillsdale, NJ. (7): 36–44. Allen NJ, Meyer JP. 1990. The measurement and antecedents of Eastin MS. 2001. Credibility assessments of online health informa- affective, continuance, and normative commitment to the organi- – tion: the effects of source expertise and knowledge of content. zation. Journal of Occupational Psychology 63:1 18. Journal of Computer-Mediated Communication 6(4): 0–0. Anderson JC, Gerbing DW. 1988. Structural equation modeling in Edwards C, Spence PR, Gentile CJ, Edwards A, Edwards A. 2013. practice: a review and recommended two-step approach. How much Klout do you have … a test of system generated cues Psychological Bulletin 103(3): 411–423. – ™ ’ on source credibility. Computers in Human Behavior 29:A12 A16. Arbuckle JL. 2009. Amos 16.0 User s Guide. Chicago: SPSS. Eysenbach G. 2008. Credibility of health information and digital Atkin T, Thach L. 2012. Millennial wine consumers: risk perception – media: new perspectives and implications for youth. In Metzger and information search. Wine Economics and Policy 1:54 62. M, Flanagin A (eds). Digital Media, Youth, and Credibility. The Austin EW, Dong Q. 1994. Source v. content effects on judgment of – – MIT P: Cambridge, MA; 123 154. news believability. Journalism Quarterly 71(4): 973 983. Flanagin AJ, Metzger MJ. 2007. The role of site features, user attri- Bagozzi RP, Yi Y. 1988. On the evaluation of structural equation butes, and information verification behaviors on the perceived models. Journal of the Academy of Marketing Science 16(1): – credibility of web-based information. New Media and Society 74 94. 9(2): 319–342. Bansal HS, Voyer PA. 2000. Word-of-mouth processes within a Fleishman-Hillard and Harris Interactive. 2012. Annual Global services purchase decision context. Journal of Service Research Study: 2012 Digital Influence Index. Retrieved at http:// – 3: 166 177. fleishmanhillard.com/2012/01/news-and-opinions/2012-digital- Bart Y, Shankar V, Sultan F, Urban GL. 2005. Are the drivers and influence-index-shows-internet-as-leading-influence-in-consumer- role of online trust the same for all web sites and consumers? A purchasing-choices/ [Accessed on 12 February 2014]. large-scale exploratory empirical study. Journal of Marketing Fogg BJ, Soohoo C, Danielson D. 2002. How do people – Research 69(4): 131 152. evaluate a Web site’s credibility? Results from a large Berlo DK, Lemert JB, Mertz RJ. 1969. Dimensions for evaluating study. Retrieved from http://www.consumerwebwatch.org/ the acceptability of message sources. The Public Opinion dynamic/web-credibility-reports-ealuate-abstract.cfm [Accessed – Quarterly 33(4): 563 576. on 12 September 2013]. Bone PF. 1992. Determinants of word-of-mouth communications Fornell C, Larcker DF. 1981. Evaluating structural equation models during consumption. In Sherry J, Sternthal B (eds). Advances with unobservable variables and measurement error. Journal of in Consumer Research (Vol. 19). Association for Consumer Marketing Research 18(February): 39–50. Research: Provo, UT; 579–583. Garbarino E, Johnson MS. 1999. The different roles of satisfaction, Bouzdine-Chameeva T, Galam S. 2011. Experts versus word-of- trust, and commitment in customer relationships. Journal of mouth in the wine purchasing dynamics: a model from physics. Marketing 63(April): 70–87. Refereed paper – 6th AWBR International Conference, 9–10 Gefen D, Karahanna E, Straub DW. 2003. Trust and TAM in online June 2011, France. shopping: an integrated model. MIS Quarterly 27(1): 51–90. Brown JJ, Reingen PH. 1987. Social ties and word-of-mouth refer- Ghosh R, Lerman K. 2008. Community detection using a measure ral behavior. Journal of Consumer Research 14: 350–362. of global influence. In Proceedings of the 2nd SNA-KDD Work- Buckland M. 1991. Information retrieval of more than text. shop on Social Network Mining and Analysis (20–35) Las Journal of the American Society for Information Science 42: Godar SH. 2005. Worldwide word-of-mouth communication: im- 586–588. pact of a paradigm shift. Paper presented at AIB-SE (USA) Byrne BM. 2001. Structural equations modeling with AMOS: basic Annual Meeting, Charleston, SC. Retrieved from www.aibse. concepts, applications, and programming. Lawrence Erlbaum: org/…/Proceedings%202005/Worldwide%20Mouth.doc London. [Accessed on 12 September 2013]. de Canniere MH, de Pelsmacker P, Geuens M. 2009. Relationship Godes D, Mayzlin D. 2004. Using online conversations to study quality and the theory of planned behavior models of behavioral word-of-mouth communication. Marketing Science 23: 545–60. intentions and purchase behavior. Journal of Business Research Gorsuch RL. 1983. Factor Analysis. Lawrence Erlbaum Associates: 62(1): 82–92. Hillsdale, NJ. Chu S-C, Kim Y. 2011. Determinants of consumer engagement in Gremler DD, Gwinner KP, Brown SW. 2001. Generating positive electronic word-of-mouth (eWOM) in social networking sites. word-of-mouth communication through customer–employee re- International Journal of Advertising 30(1): 47–75. lationships. International Journal of Service Industry Manage- Churchill GA, Jr.. 1979. A paradigm for developing better measures ment 12(1): 44–59. of marketing constructs. Journal of Marketing 16(1): 64–73. Haas C, Wearden ST. 2003. E-credibility: building common ground Corritore CL, Kracher B, Wiedenbeck S. 2003. Online trust: in Web environments. L1 Educational Studies in Language and concepts, evolving themes, a model. International Journal of Literature 3: 169–184. Human Computer Studies 58(6): 737–758. Habel C, Veale R, Lu VN. 2010. I heard it through the Corritore CL, Marble RP, Wiedenbeck S, Kracher B, Chandran A. grapevine! Exploring drivers of participation in virtual com- 2005. Measuring online trust of websites: credibility, perceived munities. Refereed paper – 5th International Academy of ease of use, and risk. AMCIS 2005 Proceedings. Paper 370. Wine Business Research Conference, 8–10 Feb. 2010 Retrieved from http://aisel.aisnet.org/amcis2005/370 [Accessed Auckland, NZ. on 24 August 2012]. Hair JF, Anderson RE, Tatham RL, Black WC. 1998. Multivariate Crosby LA, Evans KR, Cowles D. 1990. Relationship quality in ser- Data Analysis, 5th edn. Prentice Hall: Upper Saddle River, NJ. vices selling: an interpersonal influence perspective. Journal of Herr PM, Kardes FR, Kim J. 1991. Effects of word-of-mouth and Marketing 54(3): 68–81. product attribute information on persuasion: an accessibility- David A, Glore P. 2010. The impact of design and aesthetics on diagnosticity perspective. Journal of Consumer Research usability, credibility, and learning in an online environment. 17(3): 454–462. Online Journal of Distance Learning Administration 13(4). Hooper D, Coughlan J, Mullen M. 2008. Structural equation model- Retrieved from http://www.westga.edu/~distance/ojdla/winter134/ ing: guidelines for determining model fit. Electronic Journal of david_glore134.html [Accessed on 11 April 2012]. Business Research Methods 6(1). Retrieved from http://www.

ejbrm.com/vol6/v6-il/v6-il-art6abstraact.htm [Accessed on 20 McKnight DH, Choiudhury V, Kacmar C. 2002a. Developing June 2013]. and validating trust measures for e-commerce: an integrative Hovland CI, Weiss W. 1951. The influence of source credibility on typology. Information Systems Research 13(3): 334–361. communication effectiveness. Public Opinion Quarterly 15: McKnight DH, Choudhury V, Kacmar C. 2002b. The impact of ini- 635–650. tial consumer trust on intentions to transact with a website: a Hsu C, Lin JC. 2008. Acceptance of blog usage: the roles of trust building model. Journal of Strategic Information Systems technology acceptance, social influence and knowledge sharing 11: 297–323. motivation. Information and Management 45:65–74. McLuhan M. 1964. Understanding Media: The Extensions of Man. Hsu C-L, Lin JC-C, Chiang H-S. 2012. The effects of blogger McGraw-Hill: New York. recommendations on customers’ online shopping intentions. Metzger MJ, Flanagin AJ, Eyal K, Lemus DR, McCann RM. 2003. Internet Research 23(1): 65–74. Credibility for the 21st century: integrating perspectives on Hu LT, Bentler PM. 1999. Cutoff criteria for fit indexes in covari- source, message, and media credibility in the contemporary ance structure analysis: conventional criteria versus new alterna- media environment. In Kalbfleisc PJ (ed). Communication Year- tives. Structural Equation Modeling 11(3): 320–341. book, Vol. 27. Lawrence Erlbaum: Mahwah, NJ; 293–335. Johnson T, Bruwer J. 2004. Generic consumer risk-reduction strat- Meuter ML, McCabe DB, Curran JM. 2013. Electronic word-of- egies (RRS) in wine-related lifestyle segments of the Australian mouth versus interpersonal word-of-mouth: are all forms of wine market. International Journal of Wine Marketing 16(1): word-of-mouth equally influential? Services Marketing Quar- 5–35. DOI: 10.1108/eb008764 terly 34: 240–256. DOI: 10.1080/15332969.2013.798201 Johnson TJ, Kaye BK. 2004. Way the blog: how reliance on tradi- Morgan RM, Hunt SD. 1994. The commitment–trust theory of tional media and the Internet influence credibility perceptions relationship marketing. Journal of Marketing Research 58(3): of weblogs among blog users. Journalism and Mass Communi- 20–38. cations Quarterly 81(Autumn): 622–642. Mourali M, Laroche M, Pons F. 2005. Antecedents of consumer Johnson TJ, Kaye BK, Bichard SL, Wong WJ. 2007. Every blog relative preference for interpersonal information sources in has its day: politically interested Internet users’ perceptions of pre-purchase search. Journal of Consumer Behavior 4(5): blog credibility. Journal of Computer Mediated Communication 307–318. 13(1): 100–122. Newhagen J. 1997. The role of feedback in the assessment Jöreskog KG, Sörbom D. 1993. LISREL8: Structural Equation of news. Information Processing & Management 33(5): Modeling with the SIMPLIS Command Language. Lawrence 583–594. Erlbaum Associates: Hillsdale, NJ. Newhagen J, Nass C. 1989. Differential criteria for evaluating cred- Kang M. 2009. Measuring social media credibility: a study on a ibility of newspapers and TV news. Journalism Quarterly 66(2): measure of blog credibility. Institute for Public Relations. Re- 277–284. < trieved 28 March 2014 from: http://www.instituteforpr.org/ Nunnally JC. 1978. Psychometric Theory, 2nd edn. McGraw-Hill: > wp-content/uploads/BlogCredibility_101210.pdf . New York. Kaye BK, Johnson TJ. 2011. Hot diggity blog: a cluster analysis ex- Ohanian R. 1990. Construction and validation of a scale to measure amining motivations and other factors for why people judge dif- credibility. Journal of Advertising 19(3): 39–52. ferent types of blogs as credible. Mass Communication and – Olson GM, Olson JS. 2000. Distance matters. Human Computer Society 14: 236 263. Interaction 15: 139–179. Kiecker R, Cowles D. 2001. Interpersonal communications and per- fl Pan L-Y, Chiou J-S. 2011. How much can you trust online sonal in uence on the Internet: a framework for examining on- information? Cues for perceived trustworthiness of consumer- Journal of Euromarketing 11 – line word-of-mouth. (2): 71 88. generated online information. Journal of Interactive Marketing Kline RB. 2005. Principles and Practice of Structural Equation 25(2): 67–74. Modeling. The Guilford Press: New York, NY. Bloomberg Businessweek Komiak SYX, Benbasat I. 2004. Understanding customer trust in Paskin J. 2014. Saving Obamacare. March 10: 70–71. agent-mediated electronic commerce, web-mediated electronic commerce, and traditional commerce. Information Technology Pavlou PA, Gefen D. 2004. Building effective online marketplaces with institution-based trust. Information Systems Research and Management (ITM) 5(1&2): 181–207. 15 – Kozinets RV. 2002. The field behind the screen: using netnography (1): 37 59. for marketing research in online communities. Journal of Mar- Petty RE, Cacioppo JT, Schumann D. 1983. Central and keting Research 39(1): 61–72. peripheral routes to advertising effectiveness: the moderating role of involvement. Journal of Consumer Research 10(2): Lasica JD. 2002. Weblogs: a new source of news. In Blood R (ed). – We’ve Got Blog: How Weblogs Are Changing Our Culture. 135 146. Perseus Publishing: Cambridge, MA; 171–182. Phelps JE, Lewis R, Mobilio L, Perry D, Raman N. 2004. Viral Lecinski J. 2011. Winning the Zero Moment of Truth (ebook). marketing or electronic word-of-mouth advertising: examining Retrieved from: http://www.thinkwithgoogle.com/research- consumer responses and motivations to pass along email. Jour- – studies/2011-winning-zmot-ebook.html [Accessed on 12 nal of Advertising Research 44(4): 333 348. February 2014]. Riegner C. 2007. The impact of Web 2.0 on consumer purchase de- Lee T-T. 2010. Why they don’t trust the media: an examination of factors cisions. Journal of Advertising Research 47(4): 437–447. DOI: predicting trust. American Behavioral Scientist 54(1): 8–21. 10.2501/s0021849907070456 Lockshin L, Corsi AM. 2012. Consumer behavior for wine 2.0: a Rieh SY, Belkin NJ. 1998. Understanding judgment of review since 2003 and future directions. Wine Economics and information quality and cognitive authority in the WWW. Policy 1:2–23. Journal of the American Society for Information Sciences MacCallum RC, Browne MW, Sugawara HM. 1996. Power analysis 35: 279–289. and determination of sample size for covariance structure modeling. Rubin VL, Liddy ED. 2006. Assessing credibility of weblogs. In Psychological Methods 1:130–149. Proceedings of the AAAI spring symposium: Computational Mahalanobis PC. 1930. On tests and measures of group divergence. Approaches to Analyzing Weblogs (CAAW). Journal and Proceedings of the Asiatic Society of Bengal 26: Saxena A. 2011. Blogs and their impact on purchase intention: a 541–588. structural equation modelling approach. Paradigm XV(1&2): Markham D. 1968. The dimensions of source credibility of television 102–110. newscasters. Journal of Communication 18:57–64. Sichtmann C. 2007. An analysis of antecedents and consequences McCroskey JC, Jenson TA. 1975. Image of mass media news of trust in a corporate brand. European Journal of Marketing sources. Journal of Broadcasting 19: 169–181. 41(9/10): 999–1015. DOI: 10.1108/03090560710773318

Slater MD, Rouner D. 1996. How message evaluation and source trust perspectives. European Journal of Information Systems attributes may influence credibility assessment and belief 12(1): 41–48. change. Journalism and Mass Communications Quarterly 73 Vinography. 2014. The complete list of wine blogs. Retrieved 20 (4): 974–991. March 2014 from http://www.vinography.com/wine_blog_list. Sonquist JA, Dunkelberg WC. 1977. Survey and Opinion Research: html. Procedure for Processing and Analysis. Prentice Hall: Wang C-C, Chien H-Y. 2012. Believe or skepticism? An empirical Englewood Cliffs, NJ. study on individual’s attitude to blog product review. Interna- SPSS White Paper. 1998. Using SPSS for item analysis: more tional Journal of Innovation, Management and Technology reliable test assessment using statistics. Retrieved from http:// 3(4): 343–348. pages.infinit.net/rlevesqu/Syntax/Item Analysis/UsingSPSSfor Warnick B. 2004. Online ethos: source credibility in an ‘authorless ItemAnalysis.pdf [Accessed on 24 June 2012]. environment’. American Behavioral Scientist 48(2): 256–265. Stanford J, Tauber ER, Fogg BJ, Marable L. 2002. Consumer Wathan CN, Burkell J. 2002. Believe it or not: factors influencing Webwatch. Retrieved 6 July 2007 from http://www.equiden. credibility on the Web. Journal of the American Society for In- no/getfile.php/31086.69/Fogg+-+Experts+view+website+cred- formation Science and Technology 53(2): 134–144. dibility.pdf. Watts DJ, Dodds PS. 2007. Influentials, networks, and public opinion Sundar SS, Nass C. 2001. Conceptualizing sources in online news. formation. Journal of Consumer Research 34(Dec): 441–458. Journal of Communication 51(1): 52. Weerkamp W, de Rijke M. 2012. Credibility-inspired ranking for Tan W-K, Chang Y-C. 2012. Improving users’ credibility percep- blog post retrieval. Information Retrieval 15: 243–277. tion of travel expert blogs: useful lessons from television travel Wilson E. 2009. Bloggers crash fashion’s front row. The New York shows. Information Research 18(2): 1–15. Times. Retrieved from http://www.nytimes.com/2009/12/27/ Tanis M, Postmes T. 2005. A social identity approach to trust: inter- fashion/27BLOGGERS.html [Accessed on 17 January 2013]. personal perception, group membership, and trusting behavior. Wilson D, Quinton S. 2012. Let’s talk about wine: does Twitter European Journal of Social Psychology 35: 413–424. have value? International Journal of Wine Business Research Technorati. 2013. 2013 Digital Influence Report. Retrieved 28 24(4): 271–286. March 2014 from the Technorati website: http://technorati. Wu W-L, Lee Y-C. 2012. The effect of blog trustworthiness, product com/blogging/state-of-the-blogosphere//. attitude, and blog involvement on purchase intention. International Thach L. 2009. Wine 2.0 – the next phase of wine marketing? Ex- Journal of Management & Information Systems 16(3): 265–275. ploring US winery adoption of wine 2.0 components. Journal Yang KCC. 2007. Factors influencing Internet users’ perceived of Wine Research 20(2): 143–157. credibility of news-related blogs in Taiwan. Telematics and Thach L. 2010. Wine blogs: expressing diverse wine opinions in a Informatics 24:69–85. new realm of online wine relationship marketing. Refereed paper Young L, Albaum J. 2003. Developing a measure of trust in – 5th International Academy of Wine Business Research Confer- salesperson/customer relationships in direct selling. Journal of ence, 8–10 Feb. 2010 Auckland, NZ. Personal Selling and Sales Management 24(3): 253–269. Tian KT, Bearden WO, Hunter GL. 2001. Consumers’ need for Zafiropoulos K. 2012. Wine blogs influence and blogs’ community uniqueness: scale development and validation. Journal of connectivity: a social network analysis. European Journal of Consumer Research 28(1): 50–66. DOI: 10.1086/321947 Tourism, Hospitality, and Recreation 3(1): 135–156. Van der Heijden H, Verhagen T, Creemers M. 2003. Understanding Zeithaml VA, Berry LL, Parasuraman A. 1996. The behavioral con- online purchase intentions: contributions from technology and sequences of service quality. Journal of Marketing 60(2): 31–46.