INTRODUCTION

 Consumers’ ubiquitous use and growth of digital media has generated vast interest among practitioners and scholars. Especially, the rise of social media use has represented a paradigm shift in marketing communications (Christodoulides, Jevons, & Bonhomme, 2012), and has changed the way to disseminate branded messages. Advertisers are increasingly interested in engaging with customers, helping shape customers’ experiences, and leveraging their relationships for greater marketing impact, via social media. Thus, establishing a social presence has been considered a “must activity”, and all Advertising Age Top 100 Advertisers have Facebook pages for their brands (Lipsman, Mudd, Rich, & Bruich, 2012). Although advertisers are excited about social media’s potential to drive value for brands, they are still anxious about the how to integrate the full potential of this medium in their overall media mix (eMarkerter, 2012).

 Social media practitioners have emphasized the importance of being open, authentic, and honest about their brands when communicating with consumers (Puglisi, 2011). By the same token, the concept of transparency has become more prominent and received growing attention in recent years. Social media, especially, have raised transparency to a new level by providing an opportunity for brands to present themselves as real people and build relationships with consumers, and providing the means for consumers to share their knowledge and opinions with others. Thus, most practitioners believe that social media offer low cost ways to impact brand transparency and social media presence can help brands increase transparency perceptions (DiStaso & Bortree, 2012). The increasing demand for transparency in social media means that advertisers need to contemplate how and what to communicate with consumers. Despite growing interest, few studies in advertising have investigated the concept of transparency as well as factors that influence consumers’ perceptions of brand transparency in social media settings. Current insights are thus limited, which calls for further research in this area.

 The aim of this study is to empirically investigate what factors may drive consumers’ perceptions of brand transparency. More specifically, this study set out to examine the effect of message sidedness(one-sided versus two- sided) on brand transparency. Message sidedness has been employed by advertisers as a key message strategy (Crowley & Hoyer, 1994) and is thus expected to make differential impacts on brand perceptions when they communicate with consumers in social media (Cone Communications, 2009). Persuasion Knowledge Model (hereinafter PKM) has significant implications in understanding consumers’ perceptions about brand transparency and serves as a theoretical framework in this study.

LITERATURE REVIEW AND HYPOTHESES

Brand Transparency: Conceptualization and Operationalization

 Advertisers have acknowledged the enormous benefits of having a reputation of being honest and open. More important is gaining the trust and loyalty of consumers. Accordingly, transparency is one of the buzzwords among practitioners when pitching a social media campaign (Wilms, 2011). Despite growing interest, what transparency really means in social media has been rarely discussed in prior advertising studies. Transparency in communication is frequently referenced with news media, and refers to informing the public on how and why information is gathered, especially when there are many, often competing, sources of information; when much is known about the method of information delivery; and when the funding of media product is publicly available (Phillips & Young, 2009). However, digital media has created increasing expectation and the tools to deliver greater transparency, because of easy access to and delivery of information. Especially, Lord (2006) noted that “the Internet promises the potential for equalizing influence between organizations and individuals,” and suggested that any organization and individual has the potential to influence what people think.

 Academic scholars from other disciplines, such as public relations, economics, and humanities have attempted to define “transparency”, but there has been little agreement on a set of specific conceptual and operational definitions related to it. Simply defined, transparency entails openness and readily available information coming from organizations (Rawlins, 2008). Florini (2000) noted that “transparency is the opposite of secrecy.” However, this definition is too broad in its generalization of the concept and specific application of transparent actions (Rawlins, 2009). Accordingly, Francis et al. (2009), from the information exchange perspective, defined corporate transparency as “the availability of firm-specific information to those outside publicly traded firms” and found that increasing information availability helps make firms more competitive in the market. Disclosure or making information available to public, however, does not encompass the entire concept of transparency, because transparency is only useful when it enhances understanding, not just increases the flow of information (Wall, 1996). In this context, Holtz and colleagues (2008) defined transparency as “the degree to which an organization shares information its stakeholders need to make informed decisions.” Transparency, however, would not meet the needs of key stakeholders such as customers, employees, and investors, unless the organization knows what they want to know. Therefore, stakeholders’ participation is required in identifying the information they need to make informed decisions. Accordingly, Balkin (1999) suggested that transparency is not just a matter of availability of knowledge, but also of various kinds of participation and accountability. In the same vein, Cotterell (1999) defined transparency as a process involving active participation in acquiring, distributing and creating knowledge as well as responsibility for providing personal accounts. He further defined accountability as “the willingness and responsibility to try to give a meaningful and accurate account of oneself, or of circumstances in which one is involved, or of which one is aware” (Cotterrell, 1999).

 In summary, despite varying definitions in prior literature across several academic disciplines, some important and common dimensions - openness, substantial information, participation, and accountability - to conceptualize transparency have emerged. Therefore, brand transparency for the purpose of this study is defined as “consumers’ perceived levels of a brand’s strategic communication effort to make information available - whether positive or negative in nature - for the purpose of enhancing their understanding and making a brand accountable for marketing practices.” This definition contains all four dimensions of transparency discussed in prior literature. Furthermore, it brings consumers’ participation and understanding to the center of transparency conceptualization. Although deliberate attempt to be transparent in social media is the variable that could fall under the control of practitioners, a brand’s self-assessment of transparency has very little value. Thus, transparency should be operationalized in relation to how consumers perceive the brand’s effort to be transparent in social media.

Persuasion Knowledge Model

 PKM, introduced by Friestad and Wright (1994) suggests that consumers develop knowledge about persuasion and use this knowledge to cope with persuasion attempts from advertisers. Persuasion knowledge refers to “consumers’ loose set of beliefs about persuasion motives and tactics” (Friestad & Wright, 1994). Persuasion knowledge performs various functions, including directing consumers’ attention to a certain aspect of an advertisement, providing inference about why an advertiser uses a particular appeal in the advertisement, predicting the effects of the advertisement, and evaluating the advertisement’s effectiveness (Campbell & Kirmani, 2000; Friestad & Wright, 1994). The proposition that consumers use their persuasion knowledge to draw inferences about advertisers’ motives and to manage the persuasion episode appears very convincing, but few studies have tested this proposition in the context of social media.

 Consumers develop persuasion knowledge over time as they are exposed to marketing tactics and come to recognize them as such. Social media provide commercial content (e.g., brand posts), which differs in form as well as delivery method from traditional media. Advertisers’ unpaid messages can appear on the wall of the brand page and may also appear in the Newsfeed of a fan; advertisers’ unpaid messages can appear when a friend actively engages with a brand (e.g., clicking a “like” button, posting comments, etc.) and become visible on a friend’s wall or in the Newsfeed; advertisers’ paid messages can appear based on a social context in the right hand column; advertisers’ paid messages can appear in the right hand column to fans and friends of fans, and they have been actively distributed broadly (Lipsman et al., 2012). Some (e.g., sponsored stories) are pushed upon consumers and are mere adoptions of traditional media to social media, while others (e.g., stories about friends) rely on consumers to pull content and have no parallel in the traditional media. Although all types of message delivery are of importance for advertisers to communicate and to build relationships with consumers, this study focuses mainly on brand posts containing videos, photos, anecdotes, or other materials, which are considered as the fundamental strategy to leverage social media’s promise of inexpensive reach and widespread of customer advocacy.

 PKM serves as a theoretical framework to explain how unpaid messages presented on a brand page in social media may affect consumers’ transparency perceptions. Consumers tend to draw two broad types of marketers’ intents when they encounter a marketing practice, namely manipulative intent and consumer-serving intent (Campbell & Kirmani, 2000; DeCarlo, 2005). In many of the marketing practices, marketers’ manipulative intent is more easily accessible by consumers than consumer-serving intents, and thus, marketers’ manipulative intent is of primary interest in this study. Especially, the “change of meaning principle” (Friestad & Wright, 1994) the process by which a persuasive tactic comes to be perceived as having manipulative intent provides important implications for how consumers respond to branded messages and perceive the brand promoted in the messages. For example, if the messages on the brand page are not perceived by consumers as a tactic, it may result in greater compliance, largely because the persuasion attempt itself does not trigger persuasion knowledge. In contrast, once a marketer’s attempt has been recognized as a tactic, consumers are more likely to cope with and perhaps protect themselves from what they perceive as a tactic to manipulate their behavior. Thus, when the messages on the brand page are perceived as being manipulative or infested with commercial spam, marketers risk negative consequences.

 Prior PKM studies imply that triggering consumers’ knowledge about the presence of persuasive tactics also prompts coping strategies that results in unfavorable responses to embedded brands or sponsors. For example, Campbell (1995) found that consumers’ attitudes about  advertising tactics used in TV commercials could lower advertising persuasion (i.e., attitudes and purchase intentions). Likewise, Kirmani and Zhu (2007) showed that raising suspicion about the sponsor’s motives negatively affected overall brand evaluations in print ads. Applying to the persuasive messages posted on the brand page, we suggest that perceiving manipulative intent or drawing inferences about a marketer’s ulterior motive would help shape consumers’ transparency perceptions toward the brand.

 The effects of perceiving manipulative intent, however, would not be uniform. Kirmani and Zhu (2007) suggested that message cues (e.g., a comparison with the leading brand) are likely to vary in terms of the salience of manipulative intent. Consistently, in the online advertising setting, Yoo (2012) suggested that the types of messages (e.g., goal-relevant messages) played a significant role in how consumers use persuasion knowledge and influence consumers’ click-through behaviors. In sum, this study aims to examine the effects of the two situational factors that consumers may encounter in social media - whether the brand posts present one-sided or two-sided arguments (i.e., message sidedness) and whether the messages are generated by consumers or marketers.

Message Sidedness

 Considering the importance of transparency, one of the marketers’ main concerns appears to be how and what to present in their messages to increase consumers’ perceptions of brand transparency. We suggest that message sidedness be a factor to influence consumers’ brand transparency perceptions, because it makes a marketer’s manipulative intent more or less salient.

 Message sidedness has been considered as one of the principal strategies in marketing communication (Crowley & Hoyer, 1994). In many prior studies (e.g., Eisend, 2006; Kao, 2011), message sidedness has been operationalized as one-sidedness (i.e., one-sided messages incorporate only positive argument) versus two-sidedness (i.e., two-sided messages incorporate both positive and negative arguments). At first glance, two-sided messages which present opposing viewpoints about brands seem likely to yield negative evaluations of the brands. However, a majority of prior research have documented the heightened effectiveness of two-sided messages as opposed to one-sided messages, in particular on advertising attitude (Kao, 2011), advertising truthfulness (Swinyard, 1981), source credibility (Bohner et al., 2003), brand attitude (Etgar & Goodwin, 1982), brand or service quality (Etgar and Goodwin, 1982; Kamins et al., 1989) and purchase intentions of advertised products (Etgar & Goodwin, 1982).

 The collective understanding about such greater persuasiveness of two-sided messages than one-sided messages suggests that adding negative messages pertaining to brands tends to weight the honest approach of marketers to ‘telling truth’ to consumers by enhancing consumers’ belief that the marketers are likely to describe the ‘actual’ characteristics of their brands. Thus, two-sided messages are likely to engender greater credibility of marketers or arguments and attenuate further opposite or suspicious arguments on the attributes (Crowley & Hoyer, 1994; Eisend, 2006, 2007; Hastak & Park, 1990; Swinyard, 1981).

 Furthermore, admitting the negative performance of products or services is not normative in marketing contexts. Consumers tend to be motivated to process such novel messages in a pleasant way with increasing the probability of favorable changes in attitude (Crowley & Hoyer, 1994; Eisend, 2006; Hastak & Park, 1990). In particular, as one of the key conditions to make the message sidedness effect possible, the negative messages should be low to moderate in amount (Crowley & Hoyer, 1994). Indeed, Crowley and Hoyer (1994) attribute the mixed findings about the message sidedness effect to the varying amounts of negative messages used in prior studies.

 The message sidedness seems to be critical for branded messages in social media settings. Social media provide communication platforms in which consumers can raise their voice by being actively involved in brand communications (Lipsman et al., 2012). Consumers are likely to publicize themselves as fans (or anti-fans) by being affilated with brands they like (or dislike) (ExactTarget, 2010). That is, positive or negative messages about a brand, either responded by consumers or initiated by consumers, would be critical components of social media contents. Furthermore, today’s social media environment, which is open 24 hours, everyday, requires marketers to be more transprenct about their products, services and business practices, sometimes by disclosing negative branded messages to consumers (Cone Communications, 2009). For instance, when Sigg, a water bottle company, lacked transparency in admitting the actual use of BPA, a hazardous substance, in their products, social media served as communication centers to set off negative messages about the brand among consumers who felt betrayed (Zmuda, 2009). Therefore, the notion of message sidedness should be a key message strategy practitioners that take into consideration when attempting to communicate with consumers in social media. Despite such importance, few studies have examined this topic of interest in the context of social media. Thus, in order to bridge our knowledge gap, this study focuses on the relative effect of message sidedness, generated by either consumers or marketers, featured on the brand page in social media.

HYPOTHESES AND RESEARCH QUESTIONS

 Based on the above discussion in relation to message sidedness and brand transparency perceptions within the framework of PKM, it is expected that consumers infer marketers’ manipulative intent from branded messages in social media, depending on whether the messages are one-sided or two- sided. We suggest that when consumers are exposed to the branded messages, they may draw inference on whether marketers intend to manipulate them, typically favorably to the brands. Such suspicion will be more pronounced when the consumers encounter one-sided positive messages than two-sided messages. Accordingly, we suggest that one-sided positive messages about brands make marketers’ manipulative intent highly visible, while two-sided messages including a moderate level of negative information about brands make manipulative intent less salient. Thus:

 H1. On a brand’s social media page, two-sided messages are less likely to be perceived as having manipulative intent than one-sided messages.

 We further expect that consumers’ brand transparency perceptions will be different by message sidedness. Consumers may appreciate marketers’ deliberate communication efforts to disclose negative aspects of or opinions about brands; however, such appreciation (or favorable responses) will be alleviated when they encounter a typical attempt of persuasion from marketer-one-sided messages only. Thus, under the two-sided condition, consumers are more likely to perceive the brands to be participatory, informative, accountable, and open. Thus:

 H2: On a brand’s social media page, two-sided messages are more likely to generate favorable brand transparency perceptions [(a) participation, (b) information, (c) accountability, and (d) openness] than one-sided messages.

 In this study, we attempt to clarify the relation between the message sidedness and consumers’ brand transparency perceptions under the theoretical framework of PKM. Thus, the following hypothesis for the mediating role of brand’s manipulative intent is proposed:

 H3: Perceived levels of manipulative intent will mediate the effects of message sidedness on brand transparency perceptions. 

 Testing the above hypotheses will provide insight on how consumers’ brand transparency perceptions may be influenced by the message sidedness. However, in the context of social media, brand-related perceptions are increasingly shaped by consumers, rather than practitioners who traditionally have ultimate control over what to communicate with consumers. The exponential growth of social media has supported the development of user-generated content (UGC), and more than 155 million consumers in the U.S. are expected to generate and share content online by 2013 (Ostrow, 2009). More importantly, a significant amount of UGC deals with brand- related material. For example, about 70 percent of brand-related searches on social media are for UGC (360i, 2009). Despite the rapid increase in UGC and its potential effect on brands, few studies have examined the impact of consumer-generated content on how brands are perceived (Christodoulides et al., 2012).

 For the purpose of this study, the following research questions arise:

 RQs: How are brand transparency perceptions formed differently by message sidedness when encountering consumer-generated mes-sages? Would it be different from when they encounter market-er-generated messages?

 Although some differences are expected, it is too premature to propose specific hypotheses in relation to the difference in forming brand transparency perceptions in the context of social media. For exploratory purposes, we,  rather, will examine the research questions by comparing two sets of data: one for consumer-generated messages, and the other for marketer-generated messages.

METHODS

 An online experiment, featuring a 2 (one- versus two-sided messages) x 2 (consumer- versus marketer-generated messages) between-subjects design with a covariate (i.e., attitude toward the brand), was conducted to test the above hypotheses and to examine the research questions. The dependent variables are four dimensions of brand transparency (i.e., participation, information, accountability, and openness). Perceived levels of marketers’ manipulative intent were measured and served as a mediator. A development of stimuli and pretest preceded the main experiment.

Development of Experimental Stimuli and Pretest

 In 2011, Miracle Whip ran a ‘Take a Side’ campaign, which included TV commercials opening with a bold and negative statement about their product, such as “On a scale from one to 10, I hate Miracle Whip at like 22.” Miracle Whip attempted to grab consumers’ attention and gain their trust through the campaign, and furthermore integrated social media into the campaign by encouraging consumers to post video messages featuring their opinions about the brand to their Facebook page and Youtube channel (Wong, 2011). These video messages provided a good test bed for the pres-ent study, especially because many branded messages took either a positive or negative side. Thus, prior to the pretest, a pool of video messages about Miracle Whip were collected, and selected based on the following criteria: (1) they should include either consumers’ positive opinions only (one-sided) or both positive and negative (two-sided) opinions about the brand; (2) the length of video should be consistently 30 seconds, similar to the standard ad unit for TV commercials; and (3) it should mention a key attribute of the product. The selected videos were further modified to make them con-sistent in terms of length and disclosure of the brand name, which was add-ed at the end of video messages.

 A pretest, in the form of paper-and-pencil survey, was conducted with 54 college students, (1) to select a product attribute featured as the major content of the messages and (2) to select one- and two-sided messages which were appropriate for the main experiment. Survey participants were asked to list three important attributes of salad dressing products they considered most important. The results showed that taste(or flavor) was considered the most important attribute mentioned in the pretest, accounting for 34% of all 137 answers. Thus, taste-related messages were chosen as the key content of the experimental stimuli for the main experiment.

 A pretest, in the form of paper-and-pencil survey, was conducted with 54 college students, (1) to select a product attribute featured as the major content of the messages and (2) to select one- and two-sided messages which were appropriate for the main experiment. Survey participants were asked to list three important attributes of salad dressing products they considered most important. The results showed that taste(or flavor) was considered the most important attribute mentioned in the pretest, accounting for 34% of all 137 answers. Thus, taste-related messages were chosen as the key content of the experimental stimuli for the main experiment. brand was perceived most positive(M = 5.73) and consequently selected as the one-sided message. Another video clip including both positive and negative messages about the brand(M = 3.50) scored closest to the median point, and thus it was selected as the two-sided message. The amount of negativity in the message was appropriate as being low to moderate, as suggested by Crowley and Hoyer(1994). The paired samples t-test showed a significant mean difference in the participants’ evaluation of message sidedness between these two video clips(t (53) = 8.30, p <.01).

 Next, four Miracle Whip Facebook pages including a video message (2: one- versus two-sided message x 2: consumer- versus marketer-generated message) were created and published online (see Appendix). The mock-up brand pages were consistent with the actual Miracle Whip page on Facebook in terms of content (e.g., written information about the brand, brand pictures) and layout (e.g., menu, color), except for the experimental manipulations (i.e., video messages). Facebook seems an appropriate social media channel for the present study, because it is ranked the number one social media, accounting for 62% of social media market share of visits in 2012 (Kallas, 2012).

 Main Experiment

Samples

 An invitation email was sent to college students at a large southeastern university in the U.S., and students were encouraged to participate in the experiment for a chance to win a gift card. A total of 415 students participated in the online experiment, but after the data cleaning (i.e., we removed those who gave the wrong answer on the question about whether they were exposed to consumer- or marketer-generated, those who had a Facebook membership of the brand examined in this study, and those who saw the stimulus video clips prior to the experiment.), 357 samples were remained. Female outnumbered (67%) male, and subjects averaged 21 years old (s.d. = 4.40). About 90 percent of the subjects have used Facebook for more than three years. The use of college students for the present study seems appropriate, given the fact that college students account for almost a half of Facebook users (Quantcase, 2012).

 Procedure

 Upon clicking the link from the invitation email, subjects were directed to the page informing of purpose of the study (i.e., examining students’ use of social media) and their rights as study subjects. Then, those who agreed to participate in the study were asked to click the link which randomly led them to one of the four conditions. Upon completing the pre-experimental questionnaire, they were given the experimental scenario in which they were asked to imagine a situation where they received an invitation note from their casual friend and visited Miracle Whip Facebook page. They were also asked to turn on the audio volume. The experimental video messages played automatically as the subjects visited the experimental Facebook after clicking the URL on their online questionnaire. Finally, subjects answered the measures for the marketer’s manipulative intent, their perceptions of brand transparency, and demographics (e.g., gender, age, school year etc.).

Experimental Manipulation

 Message Sidedness: Based on the results of the pretest, two video messages were selected and used for the main experiment. The one-sided message showed an adult male and female who addressed their enjoyment of the taste of Miracle Whip at home, also featuring a scene where they ate a snack together with a Miracle Whip topping. Meanwhile, two-sided message showed an adult male and female arguing at home about their different opinions of the taste of Miracle Whip. The adult male verbally explained why he liked Miracle Whip, whereas the female adult suggested opposite opinions about Miracle Whip.

 Consumer-versus Marketer-Generated Messages: For consumer-generated messages, we highlighted that a content generator was a Facebook user by tagging the video with “Crystal Marroquin added one video, 2 hours ago from San Diego, CA” and inserting a written statement, “This video is produced by Crystal Marroquin.” at the beginning of the video. Meanwhile, for marketer-generated messages, we tagged the video with “Miracle Whip added one video, 2 hours ago” and inserted “This video is produced by Miracle Whip.” at the beginning of the video.

Measures

 In the pre-experimental questionnaire, product involvement was measured by a 7-point bipolar scale, using five items (Mittal, 1995) (i.e., unimportant/ important, of no concern/of concern to me, means nothing to me/means a lot to me, does not matter to me/does matter to me, and insignificant /significant) (α = .94). Overall, as expected, subjects tended to have a moderate level of involvement in the product category (i.e., salad dressing) (M = 3.98) and showed no difference in their levels of product involvement across the different experimental conditions (F (3, 353) = 2.01, p = .11). Attitude toward the brand was measured by three 7-point bipolar scales (MacKenzie et al., 1986) (i.e., bad/good, favorable/unfavorable, and unpleasant/pleasant) (α = .98). Similarly, subjects showed relatively low levels of attitude toward the brand (M = 2.88), which were not different across the experimental conditions (F (3,353) = 1.01, p = .39). However, attitude toward the brand had significant effects on some of the dependent variables, which warranted the need to control this variable as a covariate in the main study.

 After completing the experimental task, subjects were asked to answer their perceived levels of manipulative intent, which were assessed by two 7-point bipolar scales (i.e., not manipulative/manipulative and not pushy/ pushy; r = .61), modified from Campbell and Kirmani (2000). Brand transparency was measured by 7-point likert-type scales (strongly disagree (1) and strongly agree (7)), modified from Rawlins’ study (2009): participation dimension was measured with four items (i.e., “… encourages consumers like me to provide feedback,” “… involves consumers like me to help identify information,” “… makes it easy for consumers like me to find information,” and “… asks the opinions of consumers like me”: α = .84); information dimension with five items (i.e., “… provides relevant information,” “… provides accurate information,” “… provides reliable information,” “… provides easy-to- understand information,” and “… provides complete information”; α = .88); accountability dimension with four items (i.e., “… is open to criticism by consumers like me,” “… is forthcoming with information that might be damaging,” “… freely admits when consumers do not like the brand,” and “… presents more than one side of information”; α = .86); and openness dimension with three items (i.e., “… fully discloses information,” “… provides information that can be compared to industry standards,” and “… provides detailed information”: α = .86).

RESULTS

 Manipulation Checks

 To check whether subjects correctly recognized the message generator, a multiple choice question with the following options was used: (1) consumers; (2) marketers; (3) don’t know/don’t remember. Those who failed to choose a correct answer were excluded from the final sample.

 To check the manipulation of message sidedness, subjects were asked to indicate the levels of positivity and negativity of the video messages they were exposed to on a 7-point likert-type scale (strongly disagree (1) to strongly agree (2)). The independent samples t-test reveals that one-sided messages (M = 5.42) were perceived as having more positive messages than two-sided messages (M = 4.11) (t (355) = 10.61, p < .001), and two-sided messages (M = 3.81) were considered having more negative messages than one-sided messages (M = 2.13) (t (355) = -13. 77, p < .001). Thus, the manipulation of message sidedness seemed to be successful among the study subjects.

 Effects of Message Sidedness

 A MANCOVA assumption that the covariance matrices of the dependent variables should be homogeneous across the groups of independent variables was violated (Box’s M = 39.53, F = 2.56 at p < .01 for consumer-generated messages; Box’s M = 37.48, F = 2.42 at p < .01 for marketer-generated messages). Instead, a series of ANCOVAs were performed as the alternative analyses to test the hypotheses, for each of the consumer-generated and mar-keter-generated messages respectively.

Consumer-Generated Messages

 H1 suggested the effects of message sidedness on perceived levels of manipulative intent, while H2a-d predicted the effects of message sidedness on the four dimensions of brand transparency perceptions: (a) participation, (b) information, (c) accountability, and (d) openness. As seen in Table 1, the first ANCOVA showed that a two-sided message (M = 3.46) was perceived as having less manipulative intent than a one-sided message (M = 3.90, F (1, 186) = 4.77, p < .05), supporting H1 for consumer-generated messages. Further ANCOVAs revealed that two-sided messages were greater than one-sided messages in all four dimensions of brand transparency perceptions: Mtwo = 3.59, Mone = 2.97, F (1, 186) = 9.02, p < .01 for participation; Mtwo = 3.40, Mone = 2.69, F (1, 186) = 14.39, p < .01 for information; Mtwo = 4.28, Mone = 2.64, F (1, 186) = 77.30, p < .01 for accountability; Mtwo = 3.10, Mone = 2.40, F (1, 186) = 13.09, p < .01 for openness. Thus, H2a-d were supported for consumer-generated messages.

Marketer-Generated Messages

 As shown in Table 1, the result of an ANCOVA showed that two-sided messages (M = 3.31) were perceived as having less manipulative intent of the marketers than one-sided messages (M = 3.93, F (1, 165) = 9. 29, p < .01), supporting H1 for marketer-generated messages. More ANCOVAs demonstrated that two-sided messages (M = 4.47) were greater than one-sided messages (M = 2.85) in the accountability dimension of brand transparency perceptions (F (1, 165) = 56.55, p < .01), whereas there were no significant differences between the two- and the one-sided messages in terms of other dimensions of brand transparency perceptions: participation (n.s.), information (n.s.), and openness (n.s.). Thus, only H2c was supported for marketer-generated messages.

Table 1 . Descriptive Statistics and Results of ANCOVAs

 Mediating Roles of Perceived Levels of Manipulative intent

 H3 posited perceived levels of manipulative intent would mediate the proposed effect of message sidedness on brand transparency perceptions. To test H3, the mediation analysis using Indirect SPSS Macro (Preacher & Hayes, 2008) was employed for the two data sets, respectively. Message sidedness conditions were dummy-coded (i.e., one-sided message = 0, two-sided message = 1), and attitude toward the brand was controlled as a covariate in the analysis.

 Consumer-Generated Messages

 As illustrated in Figure 1, the first indirect analyses showed that two-sided messages (over one-sided messages) had negative influence on perceived manipulative intent (coefficient = -.44, p < .05) and positive influence on the participation dimension of brand transparency perceptions (coefficient = .63, p < .01); perceived manipulative intent, controlling for message sidedness, negatively influenced the participation dimension (coefficient = -.16, p < .05); two-sided messages (over one-sided messages), controlling for perceived manipulative intent, had positive influence on the participation dimension (coefficient = .56, p < .01); thus, the indirect effect of two-sided messages (over one-sided messages) on the participation dimension via perceived manipulative intent was significant (indirect coefficient = .07 with a 95% confidence interval (CI) [.01, .21]).

 Using the same steps, further indirect effect analyses were conducted for other dimensions of brand transparency. The positive effects of two-sided messages (over one-sided messages) on the information dimension (indirect coefficient = .10 with a 95% CI [.02, .24]), accountability dimension (indirect coefficient = .12 with a 95% CI [.02, .28]) and openness dimension (indirect coefficient = .12 with a 95% CI [.02, .28]) were significantly mediated by perceived manipulative intent, respectively. Thus, in case of consumer-generated messages, H3 was supported for all four dimensions of brand transparency perceptions.

Figure 1 . Mediating roles of Perceived Levels of Manipulative Intent for Consumer-Generated Messages

 Marketer-Generated Messages

 As shown in Figure 2, the positive effect of two-sided messages (over one-sided messages) on the accountability dimension of brand transparency perceptions was significantly mediated by perceived manipulative intent (indirect coefficient = .23 with a 95% CI [.07, .44]). However, the mediating role of manipulative intent in the positive effect of two-sided message (over one-sided message) on other brand transparency dimensions (i.e., participation, information, and openness) were not tested, because no significant effects of message sidedness on these variables was found in the previous ANCOVA tests. Thus, in case of marketer-generated messages, H3 was only supported for the accountability dimension of brand transparency perceptions.

Figure 2 . Mediating roles of Perceived Levels of Manipulative Intent for Marketer-Generated Messages

 Summary of Findings

 H1 was supported for both consumer- and marketer-generated messages, suggesting that two-sided messages are less likely to be perceived as having manipulative intent than one-sided messages, regardless of message generators. More importantly, the results support the proposition that message sidedness makes the marketer’s manipulative intent more or less salient, which further highlights the importance of choosing appropriate message strategies in social media settings.

 For consumer-generated messages, H2a through H2d were also supported, suggesting that two-sided messages are better in shaping every facet of  brand transparency perceptions than one-sided messages. However, for marketer-generated messages, two-sided messages are only better in shaping favorable accountability perceptions of the brand than one-sided messages (H2c).

 H3 was supported for consumer-generated messages, demonstrating that perceiving manipulative intent of the marketer mediates consumers’ brand transparency perceptions. More specifically, perceived levels of the marketer’s manipulative intent by being exposed to consumer-generated messages negatively impact every dimension of brand transparency. The results further underscore the importance of PKM in explaining how consumers shape their perceptions of the brand in the context of social media. However, for marketer-generated messages, only the accountability dimension of brand transparency was mediated by perceived levels of manipulative intent, which implies that underlying mechanism to form consumers’ brand transparency perceptions may be conditioned by message cues (i.e., message sidedness and message generators). 

CONCLUSIONS AND DISCUSSION

 Advertisers increasingly consider Facebook and other social media channels as an important part of their marketing programs, and ultimately like to ‘earn’ consumers through their efforts. Meanwhile, consumers have demanded greater transparency from brands for many years, and that demand has become even greater nowadays. This should not be surprising. Because long before the term transparency became a recent buzzword, the advertising industry has, to some extent, advocated for balanced and open communication between the brand and its consumers (Milliman, Fugate, & Rahim, 1997) Milliman, Fugate, & Rahim, 1997). Social media spread information instantaneously to almost everywhere, and thus expand the information flow and render it virtually impossible for any brand to withhold information. Recently, Nestlé has come under heavy criticism via social media, because the brand tried to keep the source of their ingredients a secret and off of labels (Newlands, 2012). Furthermore, Nestlé attempted to manipulate the contents of social media by removing consumers’ negative remarks (McCarthy, 2010). The Nestlé case further highlights the importance of brand transparency, which significantly contributes to maintaining a positive brand image.

 This study contributes to the body of advertising theories. First, this study introduced ‘brand transparency’ as an important construct when studying advertisers’ communication efforts in social media. Although the term transparency has been frequently mentioned and become a buzzword, advertising researchers have rarely attempted to define the construct and investigate how it may be shaped. We suggest that brand transparency is a rather complex construct consisting of the four dimensions by reviewing the literature from other academic disciplines. In short, transparent brands should be seen as respecting their consumers and being open with their communication by including substantial information. It stands to reason that advertisers’ effort to be transparent should lead to a reputation of being transparent. However, we suggest that brand transparency is only as good as how it is viewed by consumers, not by advertisers themselves. This consumer-centric approach to define the construct suggests that transparency only fulfills its moral responsibility when consumers believe that a brand’s effort to be more transparent with consumers scores high on the participation, information, accountability, and openness dimensions. Thus, it is logical that brand transparency would also lead to improved relationship between the brand and consumers, but this assumed relationship has not been tested in this study. Thus, future studies should tie the concept of brand transparency more closely with relationship-building variables, or even brand engagement variables in social media settings.

 Second, this study brings PKM into the context of social media. PKM is a general theory about how consumers respond to marketers’ attempts at persuasion episodes (Kirmani & Zhu, 2007). In this study, we empirically examined how consumers’ brand transparency perceptions may be shaped within the theoretical framework of PKM. More specifically, we focused on the use of persuasion knowledge as a mediating variable to shape consumers’ brand transparency perceptions and examined some conditional factors (e.g., message sidedness) affecting the activation of persuasion knowledge. Our findings indicate that persuasion knowledge may be applied easily when the one-sided positive messages presented on social media make marketers’ manipulative intent highly salient. Despite the positive nature of messages, consumers may raise their suspicion and perceive the brand less transparent.

 However, the consequences of using persuasion knowledge may be different based on who generated the messages. Those who are exposed to consumer-generated messages on a brand page use persuasion knowledge to shape all four dimensions of brand transparency perceptions. More specifically, when consumer-generated messages on a brand page include both positive and negative information about the brand, consumers may consider the brand as promoting consumers’ participation in information creation, being responsible for their account and action, and stimulating open communication between the brand and its consumers. Meanwhile, those who are exposed to marketer-generated messages on a brand page use persuasion knowledge to shape only one dimension of brand transparency perceptions (i.e., accountability). Especially, when marketer-generated messages include both positive and negative information about the brand, consumers may favorably evaluate the brand in terms of accountability, but not of other dimensions. The findings that being exposed to two-sided messages generated by marketers would not foster consumers’ perceptions of participation and openness dimensions of brand transparency were somewhat expected, because of the one-way nature of marketer-generated messages. However, even though marketers’ presentation of two-sided information about the brand led to consumers’ perceptions that the brand provided more detailed and complete information, the mean difference between two-sided messages and one-sided messages was not significant, which is somewhat surprising. Thus, future studies are required to examine this puzzling issue.

 However, the consequences of using persuasion knowledge may be different based on who generated the messages. Those who are exposed to consumer-generated messages on a brand page use persuasion knowledge to shape all four dimensions of brand transparency perceptions. More specifically, when consumer-generated messages on a brand page include both positive and negative information about the brand, consumers may consider the brand as promoting consumers’ participation in information creation, being responsible for their account and action, and stimulating open communication between the brand and its consumers. Meanwhile, those who are exposed to marketer-generated messages on a brand page use persuasion knowledge to shape only one dimension of brand transparency perceptions (i.e., accountability). Especially, when marketer-generated messages include both positive and negative information about the brand, consumers may favorably evaluate the brand in terms of accountability, but not of other dimensions. The findings that being exposed to two-sided messages generated by marketers would not foster consumers’ perceptions of participation and openness dimensions of brand transparency were somewhat expected, because of the one-way nature of marketer-generated messages. However, even though marketers’ presentation of two-sided information about the brand led to consumers’ perceptions that the brand provided more detailed and complete information, the mean difference between two-sided messages and one-sided messages was not significant, which is somewhat surprising. Thus, future studies are required to examine this puzzling issue.

 Limitations and Future Research

 First, although the present study mainly focused on examining whether and how message sidedness shapes brand transparency and yielded valuable insights, it is also important to examine further persuasive effects of message sidedness in social media. As discussed earlier, future studies are encouraged to test other variables certainly more related to the measurement of social media effectiveness (e.g., consumer engagement or return-on-investment) (Singer, 2012).

 Second, the use of college student sample may be said to limit the generalizability of our findings. Even though college students use Facebook most frequently out of all demographic groups, they only make up 40% of Facebook users (Quantcast, 2012). Future studies should replicate our findings with other population groups.

 Third, this study was conducted on one type of social media platforms - a brand page. However, consumers are exposed to branded messages from multiple platforms within social media. For instance, consumers are most frequently exposed to the branded messages on the Newsfeed (Lipsman et al., 2012). Thus, replicating this study in the different contexts of social media such as newsfeeds, blogs, and video-sharing sites is recommended.

 Despite these limitations, our study advances prior social media research by initiating scholastic interest in brand transparency and by contributing to our understanding of how consumers shape brand transparency perceptions when branded messages are present in social media.

APPENDIX

Sample Stimulus Brand Page

15.JPG64.2KB