Of “Likes” and “Pins”: The Effects of Consumers' Attachment to Social Media

Available online at www.sciencedirect.com

ScienceDirect Journal of Interactive Marketing 32 (2015) 70 – 88 www.elsevier.com/locate/intmar

Of “Likes” and “Pins”: The Effects of Consumers' Attachment to Social Media Rebecca A. VanMeter a,⁎& Douglas B. Grisaffe b & Lawrence B. Chonko b a

b

Department of Marketing, Miller College of Business, Ball State University, Muncie, IN 47306, United States Department of Marketing, College of Business, The University of Texas at Arlington, Arlington, TX 76019, United States

Abstract Marketing researchers and practitioners are showing substantial interest in social media communication, trying to understand the challenges and opportunities associated with this new cultural and social phenomenon. In this research, the authors examine social media as a new attachment phenomenon, positing likely predictive links to marketing-related social media behaviors. Researchers have demonstrated useful applicability of psychological attachment theory to a variety of other marketing contexts, including special possessions, places, brands, and services. Attachment to such varied focal targets has been shown to influence behaviors of interest to marketers. However, research to date has yet to develop a conceptualization or operationalization of attachment in the social media context. The authors seek to contribute to the literature in two primary ways: first, we provide a foundational definition of attachment to social media, and conduct four initial studies to develop a measure that meets desired reliability and validity standards. Secondly, in a fifth study, we use this validated measure to test its empirical usefulness in predicting social media behaviors in an applied retail setting. Taken together, the results are particularly valuable in demonstrating that attachment to social media is a distinct, measurable phenomenon that helps to explain various activities on social media platforms, including C2C advocacy and C2B supportive communication behaviors. Results reveal practical guidance for marketing managers wrestling with developing effective social media marketing strategies. © 2015 Direct Marketing Educational Foundation, Inc., dba Marketing EDGE. All rights reserved. Keywords: Social media; Attachment theory; Scale development; Attachment behaviors; Social media advocacy; Social media support

Introduction Social media has become an important new cultural and social phenomenon, changing the way millions of people and businesses connect and communicate. Academic researchers and practitioners in marketing are showing substantial interest in this new form of communication, trying to understand the challenges and opportunities associated with social media. According to Jim Davis, Sr., Vice President and CMO of SAS, “just a few years ago, we were talking about the information revolution — today, we are witnessing a social media revolution. For business, it's a double-edged sword. On one

⁎ Corresponding author. E-mail addresses: [email protected] (R. VanMeter), [email protected] (D.B. Grisaffe), [email protected] (L.B. Chonko).

hand, one influencer can drive thousands of potential customers (or more) to a website or store. On the other hand, that same influencer can spread his or her dissatisfaction and erode your brand equity and profitability” (Gillan 2010). Social media is thus radically changing the communication landscape. The average American now spends an average of 3.2 hours a day, or 22.4 hours a week, on social networking sites. This is not limited to young users. Those 35–49 spend an average of 21 hours per week, and those 50–64 spend an average of 17 hours a week on social media (Ipsos 2013). Even the first lady of the United States has used social media to engage in “hashtag activism” (e.g., #BringBackOurGirls). Social media has also become a major focus in corporate marketing strategy. According to the 2014 Social Media Industry Report (Stelzner 2014), 97% of companies are using some form of social media to market their business. However, only about a third of them feel like they are doing so efficiently,

http://dx.doi.org/10.1016/j.intmar.2015.09.001 1094-9968/© 2015 Direct Marketing Educational Foundation, Inc., dba Marketing EDGE. All rights reserved.

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

indicating a disconnect between what marketing managers believe they should be doing and knowing exactly what to do to leverage social media. This poses a serious problem since effective marketing strategy depends upon concentrating resources optimally against well-specified activities to increase sales and create sustainable competitive advantage (Aaker 2008). Without a clear course of action to incorporate social media into marketing strategy, most C-suite officers will be left wanting regarding the link between social media marketing efforts and return on marketing investment. Academic research also has begun investigating various social media phenomena including personality traits of those who use social media (Ehrenberg et al. 2008), how use of social media impacts the individual (Valkenburg, Peter, and Schouten 2006), as well as a host of brand-related (Gensler et al. 2013; Hollebeek, Glynn, and Brodie 2014; Labrecque 2014; Naylor, Lamberton, and West 2012) and organization related issues (Blazevic et al. 2014; Rapp et al. 2013; Wang, Yu, and Wei 2012). The motivation of our research adds to this growing literature and stems from an interest in the core ingredient of social media, the individual actor (Peters et al. 2013). Therefore, in our research, we undertake a new direction in marketing-related social media research, seeking a strong theoretical understanding of psychological connections consumers may be forming with social media and how this connection subsequently impacts various consumer behaviors. Are some people forming especially strong bonds with social media itself? If so, does that stronger affinity manifest in social media behaviors that are important to marketers? Prior research has shown that people develop attachments to various “targets” of attachment, including places (e.g., Brocato, Baker, and Voorhees 2014), people (e.g., Bowlby 1979), and brands (e.g., Thomson, MacInnis, and Park 2005). Social media now powerfully enables connection and interaction with these attachment foci (Wilcox and Stephen 2013). It seems plausible that individuals may develop an attachment to the conduit itself, since social media facilitates their interaction with other attachment target(s). As has been historically demonstrated with other attachment research (e.g., Hazan and Zeifman 1999), attachment manifests itself in various behavioral outcomes or consequences. Behavioral manifestation of social media attachment could involve elevated activity in social media dialogues, postings, viewings, sharings, etc., many of which could support important marketing outcomes. The current research makes three key contributions in both attachment theory and social media research. First, we extend attachment theory's role in marketing to the domain of social media by demonstrating that individuals exhibit variation in their attachment to social media. Second, we provide an empirically validated measure that captures quantitative variation in attachment to social media. Third, and most importantly, we empirically demonstrate that ASM has distinct behavioral implications. Attachment to social media more accurately predicts consumer behaviors, advocacy, and supportive communication back to the organization above attitude toward social media and time spent on social media. Beyond their theoretical significance, the results have significant managerial implications,

71

offering that attachment to social media is predictive of consumer's social media behaviors. As far as we are aware, this is the first article to examine attachment to social media, this set of behavioral outcomes in the social media context, and it is the first to demonstrate these effects in relation to attitude toward social media and time spent on social media. Theoretical Background Key Concepts Before presenting our conceptualizations and hypotheses, we define two key components of our research. First, in order to define attachment to social media (ASM) we leverage the existing definitions from prior multi-disciplinary work on attachment (Bowlby 1979; Brocato, Baker, and Voorhees 2014; Park et al. 2010; Thomson, MacInnis, and Park 2005). The common core among these definitions is the strength of the bond between the person and the attachment object. Therefore, we define attachment to social media (ASM) simply as the strength of a bond between a person and social media. Secondly, social media in this paper is defined as an interactive platform that allows social actors to create and share in multi-way, immediate, and contingent communications (Kietzmann et al. 2011; Peters et al. 2013). The focus is on social media as a holistic phenomenon, rather than on one specific social media platform or another (e.g., Facebook, Instagram). Other research, such as Hollebeek, Glynn, and Brodie (2014), has related to the brand in the context of social media. Hollebeek, Glynn, and Brodie 2014 scale is specific to a “brand” of social media, whereas, our scale is a psychological individual-level difference factor with an unbranded medium. Recent research has also taken a holistic approach (e.g., Rapp et al. 2013). Justification for this stance ties to the external reality that social media is continuously evolving — new platforms arise and old platforms become out of vogue. Therefore for our purposes we wish to investigate attachment to social media rather than attachment to a specific social media platform. Our approach could be adapted in future research if researchers seek to study attachment to a specific platform of interest (e.g., Pinterest). With definitions of ASM and social media in hand, we now present the theoretical logic underlying our investigation. Attachment Theory Attachment theory has been particularly useful in psychology and marketing, and we posit that it also offers an excellent framework to investigate peoples' growing psychological connections with social media. Thus in this paper, we use attachment theory to test consumers' attraction to social media and examine behavioral consequences of that attachment such as being an advocate of a brand or company via social media and interacting with brands and companies on social media. Explication of this ASM construct contributes theoretically by extending attachment theory into the domain of social media. Originally, attachment theory described strong “bonds” between mothers and infants attachment that met fundamental

72

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

needs for safety and security through maintenance of proximity (Ainsworth and Bell 1970; Bowlby 1969). This theory has been generalized to other relationships, such as with friends (Trinke and Bartholomew 1997), romantic partners (Hazan and Shaver 1994), and celebrities (Thomson 2006). In marketing, researchers have studied the development of attachments to tangible objects such as gifts (Mick and DeMoss 1990), collectables (Slater 2001), or other types of special or favorite objects (Ball and Tasaki 1992; Kleine, Kleine, and Allen 1995). More recently, research in marketing has focused on emotional attachment to brands (Park et al. 2010; Thomson, MacInnis, and Park 2005), retail places (Brocato, Baker, and Voorhees 2014), as well as attachment in intangible service marketing contexts (Mende, Bolton, and Bitner 2013). In this paper, we extend the trend toward a broader set of potential attachment foci by proposing that consumers also become attached to social media. In that social media prominently facilitates people “connecting” with other people, as well as connections with organizations, causes, companies and brands, it logically could be a new medium by which individuals find relationships that offer comfort, safety and security. As in other domains, we also propose that this psychological attachment then results in a variety of behavioral outcomes, several of which are especially relevant to marketers. For example, attachments have been shown to manifest in a particularly salient classical behavioral consequence, separation distress (e.g., Bowlby 1980; Hazan and Zeifman 1999; Thomson, MacInnis, and Park 2005). This outcome involves the distress that attached individuals feel when there is an actual or threatened separation from the attachment figure (Park et al. 2010). For example, young children may exhibit emotional distress when taken to a daycare facility or when a babysitter arrives. The threat of, or actual separation from, the attachment figure evokes distress. In the same way we would expect individuals who are attached to social media to express psychological distress at the threat of separation from social media. Therefore, our first attachment-focused hypothesis regarding social media follows from this classic outcome of separation distress (Park et al. 2010): H1. Individuals who are more strongly attached to social media will express more distress under threat of separation from social media than those who are less attached to social media. In a marketing context, individuals with strong attachments to social media might experience distress if their lifeline to social reference information about products, services, or organizations was suddenly inaccessible. Marketing-related Outcomes of Attachment to Social Media Social media marketing is expected to make up 21.4% of marketing budgets by 2019 (Soat 2014). However, 63% of marketers are currently unsure of how social media marketing is impacting their ROI (Stelzner 2014). Why are marketers willing to continue investing in social media marketing efforts if the return is so elusive? Part of the issue may be the sense of unrealized potential present in this new form of media.

Marketing researchers in particular are grappling with how social media impacts areas such as brand management (Gensler et al. 2013); consumer brand engagement (Hollebeek, Glynn, and Brodie 2014; Labrecque 2014); consumers' purchase intentions (e.g., Wang, Yu, and Wei 2012); consumer self-control (Wilcox and Stephen 2013); brand perceptions (e.g., Naylor, Lamberton, and West 2012); seller, retailer, and consumer interactions (Blazevic et al. 2014; Rapp et al. 2013); the appropriateness of viral marketing via social media for various product types (Schulze, Schöler, and Skiera 2014); as well as company ROI (e.g., Hoffman and Fodor 2010). According to J. Walker Smith, executive chairman of the London-based marketing consultancy The Futures Company and a Marketing News columnist, “we are still in the learning phase of social media. The presumption across the industry is that social media is the TV of tomorrow, so it is risky to put off getting to scale with digital investments, even if there is still a lot to learn.” Marketers sense that social media is a catalyst to capitalize on customer generated word-of-mouth (WOM); however, very little empirical evidence exists to support this intuition. Positive WOM has a long history of attention as a kind of customergenerated marketing effort by which new customer acquisition is facilitated (de Matos and Rossi 2008). Social media has the potential for exponentially higher degrees of positive WOM within and across large numbers of socially networked individuals. The cost and scope of reach in facilitating WOM on social media becomes very compelling. If attachment to social media is predictive of WOM on social media, there would be great practical value in this insight as a driver of a critical marketing phenomenon of interest. Likewise, relationship marketing emphasizes payoffs for establishing and maintaining relationships with customers. This is no longer just one-way communication from a business to the consumer (B2C). Social media now offers a powerful way for customers to not only proactively and regularly communicate with companies and brands via social media (C2B), but also from consumers to other consumers (C2C). If such C2C and C2B processes occur in higher degrees among individuals who are more strongly attached to social media, the study of social media attachment phenomenon and its resultant behavioral consequences has very important implications for marketing. Thus, we explore ASM in relation to social media: C2C advocacy and supportive C2B communication with organizations. It has long been recognized that WOM advocacy is important in the growth and vibrancy of brands (Brown et al. 2005). This WOM advocacy is affected by emotion (Berger and Milkman 2012), is used to gain personal and social benefits (Alexandrov, Lilly, and Babakus 2013), and when communicated on social media, may be used by consumers to help make their purchasing decisions (Pan and Chiou 2011). Therefore, consumer advocacy is especially relevant and timely in social media research because social media has opened new ways for customers to direct communication with each other (C2C communication) to specifically discuss a company or brand. Part of the potential power here is that social media allows C2C communication to occur in a one-to-many format. One consumer's post can communicate information and sentiment to hundreds of other consumers

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

73

instantaneously. Thus, we consider advocacy via social media to be a very important phenomenon to marketers and posit that higher degrees of social media based WOM are more likely to occur for those who are more strongly attached to social media. Specifically, we hypothesize:

only socially, but also in relation to consumer behaviors. Therefore, we hypothesize the following:

H2. Individuals who are more strongly attached to social media will engage in more C2C advocacy via social media than those who are less attached to social media.

H4b. Individuals who are more strongly attached to social media will engage in notably more consumer-related activities on social media than those who are less attached to social media.

In the same way that social media enhances consumers' ability to communicate with other consumers, it is also increasingly recognized as a way for businesses to listen to consumer feedback (Agnihotri et al. 2012). One way companies can field consumer feedback is when consumers directly and proactively communicate with organizations via social media (C2B). Consumers might do so for the purposes of commenting, complaining, commending, or otherwise supporting the organization. In this paper we focus on support for the organization, because it is a more positive and active form of behavior meant to help the organization (Bettencourt 1997) rather than merely a passive and reactive response by customers. We view consumer support of an organization as a particularly valuable form of C2B communication enabled by social media. Customer supportive behaviors have been examined in contexts involving service organizations (e.g., Bettencourt 1997) and concepts from that line of research are transferrable to social media research. Social media offers a fast and effective means for consumers to communicate to organizations and has a high level of applied marketing relevance. Recent research shows that 83% of marketers are specifically interested in engaging their audience via social media (Stelzner 2014). We posit here that those who are more strongly attached to social media are likely to engage in C2B help and support to a greater degree than those who are less attached to social media. Therefore, we hypothesize the following: H3. Individuals who are more strongly attached to social media will engage in more C2B supportive communication behaviors than those who are less attached to social media. Finally, attachment theory indicates that attached individuals spend significantly more time and effort to be in proximity to the targets of their attachment (Hazan and Zeifman 1999). Those who are more strongly attached to social media are likely to have more activity on social media than those who are less attached. For example, in the social media realm, this could involve more posting, tweeting, reading of others' posts, and other socially-related behaviors than others. From a marketing perspective, those more strongly attached should be more likely to do things like use social media to talk to others about a brand, purchase something because of what they read on social media, or engage in any number of other brand related behaviors via social media. Accordingly, strong attachment to social media likely results in more activity on social media not

H4a. Individuals who are more strongly attached to social media will engage in notably more social-related activities on social media than those who are less attached to social media.

In the remainder of this paper, we test our four key hypotheses across a series of studies. Since application of attachment theory to the social media context is new to the literature, we first operationalize a measure of attachment to social media (ASM). Then, we conduct additional studies to explore ASM as a predictor of marketing-relevant social media behaviors. We examine ASM in relation to C2C advocacy, C2B supportive communication behaviors, and amount of social media activity in social- and consumer-related life domains. Method Study 1 As per classically accepted psychometric procedures (e.g., Churchill 1979; DeVellis 2003), we first sought to generate and test a broad pool of initial items aimed at tapping into the proposed construct of attachment to social media. Methodology Participants and Design. We asked colleagues to contact friends they felt were “frequent users” of social media, and to request participation in a short informal survey. We sought to obtain inputs from approximately twenty respondents in this initial phase, being particularly interested in qualitative descriptions about the role of social media in these individuals' lives. Our aim was to gather elemental descriptions in respondents' own words that would inform item generation. We asked a series of four questions on 7-point Likert-type scales: “Please indicate the degree to which you consider social media to be part of your life; Please indicate how important social media is in your life right now; How much would you agree or disagree with this statement: Social Media brings meaning to my life; and To what extent would you say you are emotionally attached to social media?” Each item was followed by an openended question asking respondents why they gave their particular chosen rating. Ultimately, we acquired rich qualitative responses from a convenience sample of 21 social media users (79% female, average age 36, age range 27–59). Item Generation. After collecting the qualitative responses, thematic analysis was conducted on the textual data to classify verbatims into dimensions based on the natural language of respondents. We generated a set of working items by taking the verbatims of respondents and transforming them into complete

74

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

sentences that could be rated using Likert scales. This served as a starting point for an initial item pool (Richins and Dawson 1992). The pool of items was then compared to extant social media literature (e.g., Fennis, Pruyn, and Maasland 2005; Hollenbeck and Kaikati 2012; Stephen and Galak 2012). Additional items were generated for themes that could be important based on the literature, but which had not been explicitly mentioned in the exploratory research. A preliminary set of 53 items was generated based on this qualitative research and review of literature. Following item purification practices (cf. Bearden, Hardesty, and Rose 2001; Shimp and Sharma 1987) we ran several pretests of these items, examining correlations, descriptive statistics, and item analyses looking for early indications of potentially low communality, ambiguity in the language that was used, and complexities in the wording of our items that might represent more than one underlying concept. Based on these early statistical explorations and conceptual item analyses, improvements were made to produce our initial working set of items aimed at accurately reflecting the domain of interest. We then utilized the key informant technique (Parasuraman, Grewal, and Krishnan 2006) as a means of further refining the initial item pool. We presented the working set of items to two separate industry experts, asking for open-ended critique and feedback. The aim was to get a preliminary check of face validity and content validity, as well as a check of thoroughness of the domain covered by the proposed set as a whole. Based on the key informant expert inputs, the wording of a few items was modified for clarity, and the proposed structure was expanded to capture an additional theme one expert felt was missing.1 In order to incorporate the missing theme, we developed additional items using the language from the expert's description of the missing subdimension.

collection of quantitative data to further explore the structure of the items, and to undertake item purification per accepted measurement development practices (Churchill 1979; DeVellis 2003).

Findings Completion of this process resulted in a revised, starting pool consisting of 45 items. Anticipating a reflective measurement specification according to classical test theory (DeVellis 2003), overlapping statements were designed to reflect higherlevel themes from the item generation phase. For example, the following items centered around the idea of nostalgia: “Using social media makes me feel nostalgic about things that I have done in the past; Sometimes social media reminds me of warm memories from my past; I use social media to reconnect with 'the good old days'." Another example of a set of overlapping items focused on social media's role in helping a person to stay informed: “I use social media to see what other people's opinions are on topics that are important to me; Social media is one of the main ways I get information about others, and Social media is one of my primary sources of information about news.” The ultimate output of Study 1 was an elementally developed, key-informant-refined item pool. This was the starting point for

Measures. The questionnaire first asked participants to indicate the various social media platforms they use. Next, participants responded to the working set of ASM items. Items were randomized in blocks of nine to eleven questions, and blocks were randomized in their presentation to respondents to avoid the threat of order biases. All items were asked on 7-point Likert scales (1 = “strongly disagree” to 7 = “strongly agree”). Respondents also provided basic demographic information as part of the survey.

1

Influence via social media was the proposed missing dimension.

Study 2 Study 2 addressed three objectives: (1) to understand the underlying structure of attachment to social media, (2) to produce a purified set of items to measure those dimensions, and (3) to test the purified subset for replicability and reliability in a new sample. To accomplish these objectives, we collected data from a first sample and conducted exploratory factor analysis and item purification to arrive at a preliminary interpretable set of dimensions. We then collected new data within a non-student validation sample to enact a test of replicability of the derived factor structure. Details of our procedures follow. Methodology Participants and Design. Two hundred thirty-eight undergraduate business students (47% female, average age 24, age range 18–49) completed an online survey in exchange for partial course credit; twelve respondents were dropped because they did not provide complete data. Secondly, we collected a replication and validation sample from a distinct non-student population. As part of an undergraduate marketing research course, about 900 faculty and staff across a southwestern university were contacted to participate in an online survey. Participation was incentivized with a drawing for one of seven $25 gift cards to one of five local restaurants. Of those contacted, 328 began the survey and 226 faculty and staff (67% female, average age 44, age range 18–over 91) provided usable data.

Results Factor Analyses. In the initial analysis sample, the 45 items were subjected to exploratory factor analysis (EFA) with oblique rotation2 using SPSS version 22. Several iterations of item purification and re-analysis addressed issues of low loadings, high cross-loading, low measures of sampling adequacy, a lack of interpretability, or low communalities, to arrive at a solution

2 Both Varimax and Oblique rotations were run with each sample and produced comparable results. The Oblique results are discussed within.

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

with statistical desiderata according to classical test theory (Hair et al. 2010).3 The final working purified set contained 27 items which factored to eight interpretable dimensions (Table 1). These eight dimensions explained 79.6% of the variance in the data with all communalities exceeding the .60 cut off (Hair et al. 2010). We also examined the correlations between the obliquely rotated dimensions in anticipation of a test of a second-order construct specification (see Appendix A). A second-order structure for overall construct measurement has been observed in the brand attachment literature (e.g., Park et al. 2010; Thomson, MacInnis, and Park 2005). Therefore, we believed that significantly correlated subdimensions would support further testing of a single second-order conceptualization. As expected, all obliquely rotated dimensions correlated positively and significantly. Alpha reliability coefficients were computed for each factor and ranged from .82 to .91, which is well within desired guidelines (Nunnally and Bernstein 1994). We also computed an alpha reliability coefficient across summated scales for the eight first-order factors considered together. This preliminary view of a second-order structure also demonstrated adequate reliability (α = .90) (Table 1). Replication and Verification. The factor structure from the initial item purification replicated in the non-student sample (n = 226). The coefficients, communalities, and internal reliabilities of the 27 items with eight subdimensions were thus verified in a more generalizable replication sample (Table 1). The eight dimensions explained 88.2% of the variance in the non-student replication, with all communalities exceeding the .60 acceptability cut off (Hair et al. 2010). The internal consistency of all subdimensions was .80 or higher. A single second-order factor indicated by the first-order dimensions also demonstrated internal reliability with Cronbach's alpha of .93. The results from the factor analyses provided empirical evidence in support of eight first-order factors and collectively one second-order factor.4 The eight first-order subdimensions are named and defined as follows (see Table 1): the first factor, Connecting, reflects respondents' use of social media to stay

We sought “simple structure” in which each variable loaded highly on one and only one factor. Variables that cross-loaded were deleted one at a time and the analysis re-run, per Hair et al. (2010). Besides seeking simple structure, we also sought a solution with a sufficient number of factors to explain a large proportion of variance in the item set. In addition to the variance explained criterion, we also sought a solution that produced communalities of greater than .50 for each retained item, anticipating a more rigorous future test with CFA. In addition, we required interpretability for sets of items loading on retained factors, particularly in light of the knowledge acquired in the item generation phase of the research. We also tested the internal consistency reliability of each face valid item set to ensure the solution produced reliable multi-item sets. The final 8 factor solution had the clearest simple structure, the most interpretable set of factors, strong communalities above the desired threshold, a high percentage of variance explained for the solution as a whole, and internally reliable multi-item sets. Examinations of solutions with more or less factors were less ideal with respect to combined consideration of these criteria. See Appendix B for the full pattern matrix for the non-student sample. 4 In both samples the Bartlett's test was statistically significant, the KMO was above .9, and the individual measures of sampling adequacy for each item were greater than .5 (Hair et al. 2010, see Appendix C). Collectively, these indicate sufficient intercorrelation to justify EFA. 3

75

connected to others, for example, “I use social media because it makes staying in touch with others convenient.” The second factor, Nostalgia, taps into respondents' ability to use social media in order to reconnect with people, places, and events from the past, as when respondents agree with items such as, “Sometimes social media reminds me of warm memories from my past.” The third factor, Informed, deals with social media's role in keeping the respondent informed about news and events, for example, “Social media is one of the main ways I get information about major events.” The fourth factor, Enjoyment, reflects social media's role as a way for the respondent to experience relaxation and enjoyment, illustrated by items such as, “I use social media as a way for me to de-stress after a long day.” The fifth factor, Advice, reflects respondents' ability to seek forms of advice from others via social media. An example item is, “If I'm unsure about an upcoming decision I get input from friends on social media.” The sixth factor, Affirmation, taps into respondents' ability to feel assured and supported through their usage of social media, as demonstrated by the following item, “It makes me feel accepted when people comment on my social media posts.” The seventh factor, Enhances My Life, demonstrates social media's role in making respondents' lives better, for example, “Social media enhances my life.” The eighth factor, Influence, taps into the ability to use social media to encourage, influence, and help others. An example item for this dimension is, “Sometimes I post things just to have a positive effect on other peoples' moods.5” Discussion A refined set of 27 items was thus identified from the original item pool of Study 2. These items reflect a multifaceted structure consisting of eight subdimensions, preliminarily considered as indicative of a second-order construct of attachment to social media (ASM). Using an additional sample, the initial findings replicated well and the dimensionality of ASM was shown to be stable and verifiable in a non-student sample. Study 3 In Study 3, we sought additional psychometric evidence regarding our operationalization while simultaneously starting to focus on various social media outcomes of relevance to marketers. Thus, the objective of Study 3 was threefold: (1) to test the proposed second order structure of ASM with Confirmatory Factor Analysis (CFA), (2) to test the proposed structure against competing measurement models, and (3) to begin to see how ASM relates to marketing relevant outcomes

5 The eight example items presented here may act as surrogates of their respective first order factors, per Hair et al. (2010). Preliminary analyses show these eight items to demonstrate favorable internal consistency as a set and to share substantial variance with the full 27-item scale. As noted in the discussion section, a focus for future research would be development of a short scale version of ASM.

76

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

Table 1 Study 2: Purification & replication sample. Items

1. I use social media to interact with friends. 2. Social media provides a way for me to stay connected to people across distances. 3. I use social media because it makes staying in touch with others convenient. 4. Social media provides a way for me to keep in touch with others that I care about. Connecting 5. Social media allows me to look back at meaningful events, people, and places from my past. 6. Using social media makes me feel nostalgic about things that I have done in the past. 7. Sometimes social media reminds me of warm memories from my past. Nostalgia 8. Social media is one of one of the main ways I get information about major events. 9. Social media allows me to stay informed about events and news. 10. Social media is one of my primary sources of information about news. Informed 11. I use social media as a way for me to de-stress after a long day. 12. I use social media to give myself a break when I've been busy. 13. Social media is an enjoyable way to spend time. Enjoyment 14. I seek advice for upcoming decisions using social media. 15. I get advice about medical questions on social media. 16. If I'm unsure about an upcoming decision I get input from friends on social media. Advice 17. When others comment on my posts I feel affirmed. 18. When people respond to my posts in social media I feel like they care about me. 19. It makes me feel accepted when people comment on my social media posts. Affirmed 20. Social media makes my life a little bit better. 21. Social media enhances my life. 22. My life is a little richer because of social media. Enhances My Life 23. Sometimes I post things just to have a positive effect on other peoples' moods. 24. I post on social media to brighten other peoples' day. 25. I post things on social media that I think will be helpful to my friends' lives. 26. I want to inspire other people with my social media posts. 27. I think it is important to share things on social media so those I care about stay informed. Influence EASM

Purification sample (n = 238)

Replication sample (n = 226)

Mean

Standard deviation

Pattern coefficients

Pattern coefficients

5.31 5.71

1.51 1.41

.82 .80

.81 .96

5.32

1.54

.80

.87

5.42

1.46

.70

.77

5.08

1.55

(.92) .78

(.95) .75

4.09

1.65

.75

.83

4.66

1.58

3.76

1.49

.74 (.82) .88

.71 (.93) .85

3.95 3.40

1.43 1.59

4.29 5.07 4.57

1.77 1.56 1.42

3.10 2.26 3.21

1.63 1.39 1.75

.88 .85 (.88) .73 .65 .44 (.84) .83 .81 .81

.79 .88 (.93) .76 .84 .63 (.92) .77 .89 .84

4.22 3.55

1.62 1.36

(.83) .85 .84

(.91) .85 .90

4.16

1.69

4.09 3.95 3.78

1.50 1.50 1.66

4.07

1.79

.82 (.91) .75 .70 .66 (.89) .88

.91 (.95) .92 .88 .92 (.95) .72

3.89 3.88 3.73 3.84

1.75 1.70 1.75 1.69

.88 .85 .79 .58

.70 .68 .74 .50

(.91) (.90)

(.94) (.93)

Bolded numbers in parenthesis are the Cronbach's alpha. All items had the full range from 1–7.

by formally testing Hypotheses H1 regarding separation distress (Bowlby 1980; Hazan and Zeifman 1999).

dropped because individuals reported not using social media at all. The final sample contained two hundred and nine respondents.

Methodology Participants and Design. Two hundred forty-six undergraduate business students (44% female, average age 24, age range 19–46) completed an online survey in exchange for partial course credit. Of those, five responses were dropped because they did not provide complete data, and thirty-two were

Measures. The 27 items for ASM and a set of 5 items measuring separation distress were randomly presented in blocks of five to ten items. The 27 ASM items (Table 1) were measured on a 7-point Likert scale (1 = “Strongly Disagree” to 7 = “Strongly Agree”). Regarding separation distress, we adapted existing items from the literature (Fraley and Davis

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

77

Influence

Advice

Informed

Connecting

ASM ASM

Enjoyment

Nostalgia

Affirmed

Enhances My Life

a) Second-Order Structure

b) Competing First-Order Structure

Influence Influence

Advice Advice

Cognitive Informed Informed

Connecting Connecting

Enjoyment

Enjoyment

Nostalgia

Affective

Affirmed

EnhancesLife

c) Competing Second-Order Structure

Nostalgia

Affirmed

Enhances My Life

d) Competing 8 Factor First-Order Structure

Fig. 1. Study 3: Competing measurement model specification for ASM. a. Second-order structure. b. Competing first-order structure. c. Competing second-order structure. d. Competing 8 factor first-order structure.

78

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

1997; Hazan and Zeifman 1994) to appropriately reflect a social media context (I would miss social media if I didn't have it, social media would be hard for me to live without, I would be sad without social media). Additional items were adapted from Park et al. (2010) who employed two separation distress items measured on 11-point scales (anchored by 0 = “not at all” and 10 = “completely”) (to what extent would you be distressed if the social media you use were discontinued? and to what extent is it difficult to imagine life without the social media you use?). Results Measurement Model. Following previous second-order conceptualizations in the brand domain (Park et al. 2010; Thomson, MacInnis, and Park 2005), we applied confirmatory factor analysis (CFA) to test the appropriateness of a second-order ASM conceptualization in the social media domain (Fig. 1a). In this model, the 27 items loaded on 8 factors, which in turn loaded on a single second-order factor, our ASM superordinate construct.6 All fit statistics for this model meet standard criteria (e.g., Hu and Bentler 1999): degrees of freedom [df] = 316; chi-squared = 638.91; root mean square error of approximation [RMSEA] = .07; nonnormed fit index [NNFI] = .98; comparative fit index [CFI] = .98; and standardized root mean square residual [SRMR] = .066. Loadings for each of the items on the factors were statistically significant (p b .01). As reported in Table 2, the average variance extracted (AVE) and construct reliabilities (CR) all meet or exceed the minimum accepted standards of .5 and .7 respectively (Fornell and Larcker 1981). Subsequently, three competing measurement models (Hair et al. 2010) were estimated for comparative purposes, similar to the approach taken in the emotional attachment to brands literature by Thomson, MacInnis, and Park (2005) and Park et al. (2010). The first model specified all the items as loading on a single first-order latent factor, with none of the eight subdimensions delineated (Fig. 1b). The second model tested a competing second-order structure (Fig. 1c) in which dimensions of attachment to social media reflected either more cognitive or more affective elements. The third model tested a first-order eight subdimensions latent factor model (Fig. 1d). As the results show (Table 3), the proposed ASM second-order factor model fits the data better than the three competing specifications. This provides confirmatory support for the tenability of a second-order ASM specification, with our eight first-order subdimensions. Convergent and Criterion-related Validity With Separation Distress. Finally, we turn to the predictive ability of our attachment measure against the classic attachment outcome of separation distress. We therefore specified an SEM model to test Hypothesis H1, relating ASM to social media separation distress. Regarding measurement of separation distress, each

Table 2 Study 3: Second-order CFA measurement model results. Loadings Connecting 1. 2. 3. 4. Nostalgia 5. 6. 7. Informed 8. 9. 10. Enjoyment 11. 12. 13. Advice 14. 15. 16. Affirmed 17. 18. 19. Enhances My Life 20. 21. 22. Influence 23. 24. 25. 26. 27. EASM

.79 Interact .85 Connected across distances .78 Convenient staying in touch .93 Keep in touch .81 .88 Look back on past .87 Nostalgic .76 Warm memories from past .87 .71 Main ways get info .88 Stay informed .90 Primary source of info .82 .90 De-stress .81 Give myself a break .85 Enjoyable .85 .66 Upcoming decision .82 Advice about medical .68 Unsure .90 .81 Feel affirmed .87 Feel cared about .91 Feel accepted .92 .86 Little bit better .87 Enhances .89 Little richer .79 .70 Positively effect others' moods .87 Brighten others' day .84 Helpful to friends' lives .89 Inspire others .89 Those I care about stay informed .79

Ave

.88

.71

.83

.70

.87

.75

.83

.70

.78

.65

.91

.81

.85

.72

.91

.73

.93

.63

Numbers in bold are the loadings to the second-order EASM factor.

adapted item had a statistically significant loading on the latent construct of separation distress, with loadings ranging between .76 and .94. The measure of separation distress is internally consistent α = .94. The average variance extracted (AVE) and construct reliabilities (CR) all meet or exceed the minimum

Table 3 Study 3: Assessment of competing models.

(b) One factor (c) Two factors (d) 8 1st order factors (a) 2nd order

χ2

df

RMESA NNFI CFI SRMR

3,163.93 1,746.57 1,542.78 574.13

324 323 324 316

.19 .17 .16 .06

Δχ2

6 Measure of sampling adequacy for this data are KMO = .904 and Bartlett's Test of Sphericity was Approx. chi-square = 4027.367, df = 351, p-value ≤ .000.

CR

(b) vs (a) (c) vs (a) (d) vs (a) ⁎ p-Value b .001.

2,589.8 (8) ⁎ 1,172.44 (7) ⁎ 968.65 (8) ⁎

.80 .90 .91 .98

.82 .91 .92 .98

.11 .01 .38 .07

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

accepted standards of .5 and .7 respectively (Fornell and Larcker 1981). The Fit statistics for the measurement model including separation distress also met all the standard criteria (Hu and Bentler 1999): chi-squared = 899.76 (454), RMSEA = .067, NNFI = .98, CFI = .98, SRMR = .07. With evidence supporting construct measurements, we then examined the predicted associations of ASM with separation distress. The path from ASM to this outcome is meaningfully large and statistically significant, the standardized value is .70 (t-value = 11.04, p-value ≤ .001, Fig. 2). As expected, the context-specific measure of social media attachment predicts the context-adapted attachment outcome of separation distress from social media. This provides further supporting psychometric evidence for criterion-related validity (concurrent and predictive), while at the same time providing an initial empirical test of ASM's ability to predict marketing relevant behavior. For marketing purposes, it is likely that those who do not like to be disconnected from social media (i.e., separation distress), would be more valuable target consumers for things like being reached through social media campaigns and spreading positive word-of-mouth via social media. Discussion Study 3 provides support for the second-order representation of ASM with its eight subdimensions. That conceptualization fits the data better than other model specifications. Further, we provide evidence of support for ASM's criterion-related validity, being strongly related to a key theory-based outcome of attachment as represented in a social media context. Hypotheses H1 is supported. Those who are more strongly attached to

79

social media report higher levels of experiencing distress in the form of negative emotions at the threat of separation from social media. With sufficient confidence in the ASM measure, and initial support for predictive links to a classical attachment outcome, we undertook studies 4 and 5 to examine ASM in relation to additional social media activities that have specific practical and managerial marketing implications.

Study 4 In Study 4, we explore ASM as a predictor of consumer social media behaviors of direct relevance to marketing: C2C advocacy of a company via social media (Hypothesis H2), and C2B supportive communication behavior via social media (Hypothesis H3). Regarding C2C advocacy, social media has revolutionized consumers' power to communicate with and influence others, vastly multiplying the number of people with whom they can share criticism or commendation of products and services in real-time. Regarding C2B supportive communication behaviors, social media has also opened important new channels by which consumers can directly interact with organizations. The trend toward collaborative marketing and branding enables marketing managers to give consumers the opportunity to post both favorable and unfavorable comments, share photos, and provide ideas, feedback, and/or other input to the organization via various social media platforms (Malär et al. 2011). In Study 4, we test ASM as a predictor of C2C and C2B social media behaviors.

Fig. 2. Study 3: Path from ASM to Separation Distress.

80

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

We also address a plausible rival explanatory variable as part of Study 4. We sought to differentiate ASM's predictive power from a general measure of attitude toward social media. A test of this difference has been important in previous marketing-related applications of attachment theory (e.g., Park et al. 2010). This comparison also provides an additional test of discriminant validity for ASM while simultaneously testing for the greater predictive power of a theory-driven attachment scale as compared to a generic measure of favorable attitude.

Table 4 Study 4: (a) Social media advocacy adapted from White and Schneider (2000). (a) Social media advocacy adapted from White and Schneider (2000) When people ask me for a recommendation about a restaurant, I use social media to make a recommendation for this restaurant. When I know someone is looking for a restaurant, I use social media to urge them to check into this restaurant. Without anyone even mentioning restaurants, I find myself telling others on social media about the positive experiences I have had with this restaurant. Because of my experiences with this restaurant, I try to convince friends, family, and coworkers on social media to switch to this restaurant for their dinning needs. (b) Social media supportive behaviors adapted from Bettencourt (1997)

Methodology Participants and Design. Data were collected as part of a larger marketing research study in collaboration with a cluster of restaurants that were newly built in a southwestern city that borders a large urban university and an existing commercial district. Members of several groups from the university and the community were invited to participate in this study. Community group administrators e-mailed a survey invitation to their membership rosters on our behalf. We were not given access to these confidential lists, nor their size. However, from this invitation process, 146 community group members initially responded (58% were female, average age 52, ranging from 20–83). Of these, 120 respondents had top-of-mind awareness of the new restaurant cluster and of those 94 provided complete data. For our purposes in Study 4, we focused on the nonstudent, community member stakeholder group. We infused several construct measures into the data collection effort with this group. Measures. The 27 items for ASM were assessed on 7-point Likert scales (1 = “strongly disagree” to 7 = “strongly agree”). To contrast ASM from a general measure of attitude toward social media, participants also indicated on 6-point semantic differential scales the extent to which they viewed social media as “good” versus “bad,” “positive” versus “negative,” and the extent to which they “like it” versus “dislike it” (Batra and Stayman 1990). Regarding dependent measures, we measured social media based C2C advocacy by adapting four items from White and Schneider (2000, see Table 4), keeping their 7-point rating scales (1 = “describes me very well” and 7 = “does not describe me well at all”). Regarding C2B social media supportive communication behaviors, we adapted Bettencourt's (1997, see Table 4) voluntary participation measures because of their emphasis on active involvement with and support for an organization. We considered these items to be reflective of some of the ways consumers could supportively communicate with organizations via social media. Per Bettencourt (1997), each was measured on a 7-point Likert scale (1 = “very likely” and 7 = “very unlikely”). Analytical Approach. C2C advocacy and C2B supportive communication behavior served as two dependent variables each measured with multiple-item indicators. Attachment to Social Media (ASM) and general attitude toward social media

I would use social media to let this restaurant know of ways that they can better serve my needs. I would use social media to make constructive suggestions to this restaurant on how to improve their service. If I have a useful idea on how to improve service, I would let this restaurant know on social media. When I experience a problem with this restaurant, I would say something to them on social media so they can improve their service. If I notice a problem, I would inform this restaurant on social media even if it does not affect me. If this restaurant's price is incorrect to my advantage, I would still use social media to advise someone at the restaurant. If I had a good experience with this restaurant I would say something to the restaurant on social media.

(ATT) served as two independent variables, each also measured with multiple-item indicators. We allowed ASM and ATT to simultaneously compete for explanatory power in predicting the C2C and C2B outcomes. Our relatively small sample size precluded use of traditional structural equations modeling (SEM) for estimation purposes and led us to select Partial Least Squares (PLS) as a modeling approach better suited to smaller sample size applications (Fornell and Bookstein 1982; Hair et al. 2010). We used SmartPLS 2.0 (Ringle, Wende, and Will 2005) to conduct the estimation. The SmartPLS program generates traditional SEM measurement model statistics and also offers the advantage of a built-in empirical bootstrapping algorithm to produce standard errors and statistical tests of the estimated model parameters. Our estimated model for Study 4 is shown in Fig. 3.7 Results Outer Model Results. PLS-generated measurement model statistics all met acceptable SEM standards. Regarding the

7 We also conducted parallel estimation with Multivariate Multiple Linear Regression (Finn 1974; Lutz and Eckert 1994). While MMLR produced equivalent statistical and substantive results we felt PLS offered superior estimation given its ability to operationalize latent variables as multi-item indicator constructs. Application of MMLR required computation of composite variable indices prior to analysis.

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

81

Fig. 3. Study 4: ASM and ATT simultaneously predicting C2C and C2B social media behavioral intentions.

dependent constructs of C2C advocacy and C2B supportive communication, AVEs were .85 and .73 respectively. AVEs for the independent constructs of ASM and ATT were respectively .52 and .88. Construct reliabilities (CR) were as follows: C2C advocacy = .96; C2B supportive communication behaviors = .95; ASM = .90; ATT = .96. Cronbach's alpha values for C2C advocacy, C2B supportive communication behaviors, ASM, and ATT were .94, .94, .87, and .94, respectively. The squared correlation among each pair of constructs was always less than the AVE values for the given constructs in the pair, indicating reasonable support for discriminant validity. Further, the squared correlation of ASM and ATT was .31, indicating sufficient and substantial unique variance for a competing comparative test of ASM and ATT in the inner model.8 Inner Model Results. Meaningful proportions of variance were explained in both dependent variables of interest. The PLS-reported R-square was .41 for C2C advocacy, and .36 for C2B supportive communication behaviors. Regarding the structural paths of Fig. 3, attachment to social media (ASM) showed sizable and statistically significant effects in predicting C2C advocacy (γ = .588, bootstrap t = 7.05) and in predicting C2B supportive communication behaviors (γ = .509, bootstrap

8 Multicollinearity diagnostics for this level of correlation among ASM and ATT would be TOL = .643, VIF = 1.56. These are well below thresholds at which any concerns arise regarding potential redundancy. ASM is sufficiently distinct from general attitude toward social media. See Appendix D for descriptive statistics and correlations among the variables in the model.

t = 5.09). Further, these ASM-related effects showed dominant predictive power in contrast to general attitude toward social media (ATT). ATT was non-significant in predicting both C2C advocacy (γ = .129, bootstrap t = 1.24) and in predicting C2B supportive communication behaviors (γ = .086, bootstrap t = .710). Discussion In Study 4 we applied the ASM measure in an externally-valid restaurant/retail context using a non-student sample. We predicted marketing-relevant social media outcomes of C2C advocacy and C2B supportive communication behaviors using ASM and a competing explanatory variable of general attitude toward social media. Our theory-driven measure of attachment to social media showed significant and superior predictive power relative to a generic measure of attitude toward social media. ASM significantly predicts C2C and C2B social media behaviors even after taking into account general attitude toward social media. Additionally, general attitude toward social media does not incrementally predict C2C social media advocacy or C2B social media supportive communication above and beyond the substantial variance explained by ASM since results indicated that the attitude measure was not statistically significant in predicting either social media advocacy or supportive communication. Study 4 further supports ASM's applied usefulness in a real marketing context with a non-student sample. Hypotheses H2 and H3 were fully supported. Those who are more strongly attached to social media showed greater propensity to express positive C2C word-of-mouth via social media and higher levels of C2B supportive communication behaviors via social media.

82

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

These effects were not simply attributable to general attitudes toward social media. Study 5 Study 4 demonstrated that those who are more attached to social media are more likely to interact with other consumers via social media about a brand or organization, as well as more likely to use social media to interact with the brand or organization itself. However, the dependent constructs in each case involved scaled intention-type measures. In Study 5 we seek to move past intention measures to outcomes involving respondent self-reports of actual social media behaviors. We developed two self-report behavioral checklists, one for a general social life domain and one for a consumer-related life domain. In each domain, a list of several possible social media actions was provided. Participants were asked to respond with a “check all that you've done in the last two weeks” approach. We chose two substantially varied life domains to further confirm the generalizability of ASM, while showing relevance not only for social media in traditional interpersonal manifestations, but also in consumer-related manifestations. Thus, we formally test H4a and H4b proposing that those who are more strongly attached to social media should be notably more behaviorally active on social media, both in socially- and consumer-related activities. Additionally, we control for another rival predictor of these two domains of outcome behaviors: general amount of time spent on social media. Our reasoning was that more time spent on social media logically could produce more behavioral manifestations, but that more time spent did not necessarily equate to greater affinity or attachment. By statistically including a self-reported measure of time spent on social media, we eliminate a rival hypothesis and model how attachment to social media uniquely translates into social- and consumer-related social media behaviors. Methodology Participants and Design. Two hundred fifty-eight undergraduate business students (47% female, average age 24, age range 18–49) completed an online survey in exchange for partial course credit. Thirty-six respondents were dropped because they did not provide complete data leaving a working sample size of 222. Measures. Each respondent was asked which of several social media platforms they currently use, and for those mentioned, to report the average amount of time they spend per week on each of these social media platforms. The respective amounts of time were summed across all reported platforms for a total average amount of time spent per week on social media. Additionally, the survey contained the items measuring ASM asked with 7-point Likert scales (1 = “strongly disagree” to 7 = “strongly agree”) in the same manner used in our previous studies. Two distinct checklists of social media actions were also presented to respondents asking about their social media behaviors in social-

and consumer-related domains. For each domain, respondents were asked to check all boxes that applied for any behaviors they had engaged in during the previous two weeks (see Appendices E and F). The more boxes a respondent checked in a given life domain, the more behaviorally active in social media they were in that domain. Summated counts of checked boxes for each respondent in each life domain produced two distinct outcome measures of social media activity. It should be noted that this check-box behavioral self-report approach also serves to measure the two distinct dependent variables in a methodologically different way from ASM. Thus, concerns about common method bias in any observed relationships with ASM should be alleviated under this design strategy. Analytical Approach. The number of behaviors checked in a given life domain for any individual produces a count measure for a dependent variable — a nonnegative integer with a restricted range that violates the assumptions of ordinary least squares regression. A Poisson regression model is appropriate for count regression, but assumes equality of variance and mean in the data. When overdispersion occurs with greater variance than the mean, a negative binomial model is more appropriate (Long 1997). Unlike the Poisson model, it does not rely on assumed equivalence between the mean and variance of the number of behaviors, making it a more desirable model in many cases (Long and Freese 2006; Park, Chen, and Gallagher 2002). Tests for overdispersion in both of our models revealed that negative binomial regression was the more appropriate analytical procedure (Long and Freese 2006). Therefore, we applied negative binomial models for the self-reported number of behaviors an individual preformed via social media in two different life domains while testing the rival predictor of time spent on social media. In our Hypotheses H4a and H4b, we are particularly interested in the behavioral activity of those who are strongly attached to social media. We propose that individuals who are more strongly attached to social media will show especially high levels of social media activity in the social- and consumer-related domains. If these relationships are confirmed, marketers have a defined psychographic group to target — individuals strongly attached to social media — who will display significantly more desired social media behaviors, especially of the kind most relevant to marketers. We conducted negative binomial regression with the count variable predicted by the continuous variables of ASM and the rival predictor of time spent on social media. We ran one

Table 5 Study 5: Negative binomial parameter estimates for social media behaviors.

Social behaviors

Brand behaviors

Parameter

B

Std. error

Sig.

Exp(B)

(Intercept) Time spent ASM (Intercept) Time spent ASM

.776 .004 .334 − .240 .000 .321

.155 .002 .038 .273 .004 .066

.000 .067 .000 .379 .950 .000

2.17 1.00 1.40 .79 1.00 1.38

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

83

The results of these models confirm H4a and H4b and are made visually clear by examining the estimated marginal means from the negative binomial models. These are the predicted number of social- and consumer-related behaviors at the mean of ASM, as well as at plus and minus one standard deviation from the mean, after controlling for average time spent on social media. Fig. 4 shows graphs of these estimated marginal means for (a) social-related behaviors and (b) consumer-related behaviors. The graphs demonstrate generally increasing patterns of behavioral activity via social media. As one becomes more attached to social media, significant gains in social behaviors are observed. Regarding consumer behaviors, we see similar results. Stronger attachment relates to a sharp increase in the number of observed consumer-related behaviors on social media.

Fig. 4. Study 5: Estimated marginal means of behavior counts as a function of ASM levels. (a) Social-related behaviors. (b) Consumer-related behaviors.

negative binomial model predicting counts of total self-reported social-related actions on social media in the last two weeks and a second model predicting counts of total self-reported consumerrelated actions on social media in the last two weeks in order to specifically address Hypotheses H4a and H4b.

Discussion In Study 5, we extend the findings from Study 4 on C2C advocacy intentions and C2B supportive communication intentions to focus on self-reported actual social media behaviors in the social and consumer domains. Results indicate support for H4a and H4b, in that individuals who are more strongly attached to social media showed more behavioral activities on social media and specifically, more consumer related behaviors. These included items such as: Shared when I was at a company's location, “Liked” a company or brand's post, Bought something because of what I read on company or brand's social media page, or Advocated for a company or brand on my personal social media page. Additionally, we find that time spent on social media is a marginal predictor of social behaviors via social media; however, the key finding for marketers is time spent on social media is not predictive of interaction or engagement with brands via social media. Therefore, by concentrating on those who are more strongly attached to social media, a company can determine in advance individuals who are more likely to interact and engage with their brand, company, or organization via social media. The same cannot be said for individuals who merely spend greater amounts of time on social media in general. General Discussion

Results The likelihood ratio chi-squares involving tests of the overall model compared to a null model are significant for both dependent variables. For the model predicting social behaviors the likelihood chi-square was 95.7(2), p-value ≤ .000 and for the model predicting consumer-related behaviors the likelihood chi-square was 26.1(2), p-value ≤ .000. Additionally, the test of model effects shows that attachment to social media is a significant predictor of social behaviors: Nagelkerke R-square = .35, W = 112.9, df = 1, p-value ≤ .000; and consumer behaviors: Nagelkerke R-square = 0.11, W = 28.6, df = 1, p-value ≤ .000. Additionally, we see that time spent on social media is a marginally significant predictor of social behaviors via social media but is not a significant predicator of consumer behaviors via social media. The results indicate that being more strongly attached to social media compared to being less attached is a significant predictor of both social and consumer behaviors (see Table 5). This result is consistent across models.

Through a programmatic set of five studies, we conceptualized a measure of consumers' attachment to social media. Our operationalization includes eight distinct elements: (1) connecting — use of social media to stay connected to others, (2) nostalgia — the ability to use social media in order to remember things from the past, (3) informed — social media's role in keeping an individual informed, (4) enjoyment — social media's role in helping an individual to experience relaxation and enjoyment, (5) advice — an individual's use of social media to seek advice from others, (6) affirmation — an individual feeling assured and supported from social media usage, (7) enhances my life — social media's role in enhancing a person's life, and (8) influence — the ability to use social media to encourage, influence, and help others. These eight dimensions serve as indicators of a second-order measure of ASM. We provide preliminary empirical evidence for the reliability and validity of the ASM measure, and test its

84

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

usefulness in predicting specific outcomes of interest to marketers. All hypotheses received statistical support in our modeling efforts. Theoretically, our work contributes to an expansion of attachment theory into a new substantive domain within marketing, namely social media. Managerially, our results show that ASM predicts C2C advocacy and C2B supportive communication behaviors via social media, as well as social media behaviors in social- and consumer-related domains. As marketers seek ways to understand and leverage social media phenomenon, at least one trend is to focus on aggregate streams of big data from the social media “fire hose.” Identifiable individual-level data is often limited to geo-demographic characteristics from host sites, or through various scraping, aggregation and analysis tools. However, psychologically-oriented variables capturing psychographic characteristics may be more relevant in relation to consumer behavior. Attachment to social media, as operationalized by the ASM measure, offers a distinct psychologically-based individual difference variable which relates meaningfully to important social media phenomena. Further, the predictive power of ASM is demonstrated to be incremental above other rival measures of general attitude toward social media and time spent on social media. Marketers seeking to cognize consumer behaviors such as advocacy or supportive communication back to the organization via social media can benefit from this quantitative individual-level variable. Those who are more strongly attached to social media would be especially desirable regarding marketing initiatives and campaigns designed for social media. These individuals are not only more likely to appreciate social media based relationships with organizations, but also are more likely to participate with the company, to offer input to the organization via social media, and to advocate on behalf of the company to others via social media. Their more extensive social activities on social media also offer broader and more intensive reach whenever their consumer generated content is used to promote a brand or organization. Thus, by segmenting and targeting individuals who are more strongly attached to social media, companies' social media efforts could become much more efficient and effective. ASM offers managers a measurable way to segment and subsequently demonstrate the impact of marketing efforts in the social media space. ASM offers a mechanism by which to demonstrate how marketing efforts in social media translate into desirable customer outcomes that ultimately impact ROI.

social media work together to drive brand-specific social media activities. For example, is someone's attachment to the brand more of what drives his/her participation and advocacy with the brand via social media, or is it attachment to social media, or is it both, possibly even with one moderating the effects of the other? If both have additive or interactive effects, simultaneous study of the constructs together might shed further light on what is driving various consumer social media activities. Additionally, engagement has emerged as an important marketing objective and performance metric (Mersey, Malthouse, and Calder 2010; MSI 2013). Recently, formal measures of online and social media engagement have emerged (Calder, Malthouse, and Schaedel 2009; de Valck, van Bruggen, and Wierenga 2009; Hollebeek, Glynn, and Brodie 2014; Van der Lans et al. 2010); however, there seems to be two competing definitions of social media engagement. Hollebeek, Glynn, and Brodie (2014) define engagement as having a behavioral component while the root of all Calder and Malthouse's work does not have a behavioral component. The discrepancy in the definition of social media engagement first needs to be addressed then, the relationship between ASM and social media engagement can be investigated. Finally, we have shown that those who are more strongly attached to social media are more likely to positively interact with companies and/or organizations via social media. To that extent, we also only investigated consumer's positive advocacy and supportive communications via social media. Clearly, however, not all communication via social media is positive. Future research should investigate whether the individual trait ASM also predicts greater levels of negativity on social media such as the likelihood to complain about brands, organizations, or even other consumers via social media, as well as the propensity to share negative WOM via social networks. This would extend the negative word-of-mouth literature (Edison and Geissler 2011; Mattila and Wirtz 2004; Zhang, Feick, and Mittal 2014) directly in the social media context (King, Racherla, and Bush 2014). The comparison and investigation of negative WOM are outside the scope of this particular paper but are worthy of future investigation. With the introduction of our research to the literature, a measure of ASM now exists where these and other emerging substantive research questions can be explored. For now, ASM offers promise in marketing contexts in helping to identify which customers are most likely to be listening, engaging, and reacting on social media.

Limitations & Future Research Acknowledgements Future psychometric work would also bolster the case for the attachment to social media measure, and its applicability to other marketing studies. For example, tests of content validity might have raters/judges sort items under subscale definitions. Also, longitudinal data could provide additional evidence for reliability by examining test–retest stability. Finally, it would be helpful to have a short version of the new measure. Certainly, use of ASM in applied and academic studies would benefit from a short scale. Our work on attachment to social media leverages previous work on attachment to brands. A logical next step would be to combine these to see how attachment to brands and attachment to

The authors thank Dr. Dawn Iacobucci (Vanderbilt University) Dr. Gordon Bruner II (Southern Illinois University), Dr. Traci Freling (University of Texas-Arlington), Dr. Wendy Casper (University of Texas-Arlington), Dr. Emily Goad (Illinois State University), Holly Syrdal (University of Texas-Arlington), and Dr. Marcus Butts (University of Texas-Arlington) for their insightful comments on an earlier version of the manuscript. A version of the article was presented at the Society for Marketing Advances Conference in October 2013. The audience comments are gratefully acknowledged.

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

85

Appendix A. Study 2: Factor Intercorrelations and Descriptive Statistics

Connecting Informed Nostalgia Enjoyment Enhances My Life Advice Affirmed Influence Mean Std. deviation

Connecting

Informed

Nostalgia

Enjoyment

Enhances My Life

Advice

Affirmed

Influence

.52 .54 .52 .43 .22 .50 .40 5.44 1.30

.46 .47 .51 .47 .42 .39 3.70 1.37

.50 .49 .37 .56 .44 4.61 1.36

.64 .41 .57 .36 4.64 1.34

.57 .62 .43 3.94 1.39

.45 .43 2.85 1.36

.48 3.98 1.42

3.88 1.52

Appendix B. Study 2: Full Pattern Matrix, Replication (Non-student) Sample

Component

My life is a little richer because of social media. Social media enhances my life. Social media makes my life a little bit better. I use social media to interact with friends. Social media provides a way for me to stay connected to people across distances. I use social media because it makes staying in touch with others convenient. Social media provides a way for me to keep in touch with others that I care about. Sometimes I post things just to have a positive effect on other peoples' moods. I post on social media to brighten other peoples' day. I post things on social media that I think will be helpful to my friends' lives. I want to inspire other people with my social media posts. I think it is important to share things on social media so those I care about stay informed. I seek advice for upcoming decisions using social media. If I'm unsure about an upcoming decision I get input from friends on social media. I get advice about medical questions on social media. Social media allows me to look back at meaningful events, people, and places from my past. Using social media makes me feel nostalgic about things that I have done in the past. Sometimes social media reminds me of warm memories from my past. When others comment on my posts I feel affirmed. When people respond to my posts in social media I feel like they care about me. It makes me feel accepted when people comment on my social media posts. Social media is one of the main ways I get information about major events. Social media allows me to stay informed about events and news. Social media is one of my primary sources of information about news. I use social media as a way for me to de-stress after a long day. I use social media to give myself a break when I've been busy. Social media is an enjoyable way to spend time.

1

2

3

4

5

6

7

8

.064 .037 − .061 .096 − .039 .021 .020 − .068 − .020 .064 .164 .199 .015 − .019 .017 .019 .052 .099 .006 − .017 .047 − .033 .102 .039 .757 .836 .626

.001 − .036 .058 .080 .007 .007 − .062 − .020 − .024 .070 .160 .207 .768 .838 .886 .036 .086 − .025 − .057 − .006 .075 .028 − .050 .122 .133 −.048 − .083

.921 .876 .917 .080 − .005 .011 .062 .106 .155 .034 − .031 .022 .073 .019 .003 − .002 .086 − .013 .071 − .024 .027 − .005 .028 .045 .132 − .047 .148

− .009 − .019 − .020 .079 − .020 − .086 − .031 − .005 − .019 − .226 − .011 − .143 .028 − .036 − .030 − .045 − .069 − .053 − .849 − .900 − .914 − .076 − .037 .023 − .079 − .006 − .049

.010 .028 .025 .807 .956 .872 .767 .001 − .060 .128 .032 .125 .101 .099 − .094 .194 − .059 .107 .058 − .015 − .012 .054 .093 − .052 − .049 .090 .091

.043 − .027 − .033 − .058 .017 .068 − .046 − .720 − .704 − .676 − .737 − .504 − .132 − .097 .087 − .016 − .021 − .113 − .087 − .044 .081 − .029 .002 .000 .073 − .069 − .115

−.027 .013 .008 −.013 .002 −.057 −.091 −.209 −.178 .063 −.067 .162 −.057 −.034 −.008 −.745 −.828 −.709 .025 −.046 −.056 −.038 −.063 .051 −.107 −.045 −.002

−.042 .067 .018 −.020 .009 .034 .079 .137 .150 −.062 −.001 .105 .026 .005 .095 .011 .006 .065 .002 .097 −.029 .853 .790 .881 −.030 .096 .154

Extraction method: Principal Component Analysis. Rotation method: Oblimin with Kaiser Normalization. a. Rotation converged in 14 iterations.

Appendix C. Study 2: KMO and Bartlett's Test

KMO Measure of Sampling Adequacy Bartlett's Test of Sphericity

Approx. chi-square df Sig.

Purification sample

Replication sample

.926 4,553.00 351 .000

.945 7,259.06 351 .000

86

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

Appendix D. Study 4: Descriptive Statistics and Correlations Among Constructs

Correlations among constructs

ASM ATT C2C advocacy C2B support

Mean

Standard deviation

Composite reliability

Average variance extracted

ASM

ATT

C2C advocacy

C2B support

4.62 5.71 3.27 4.05

.95 .93 1.51 1.46

.90 .96 .96 .95

.52 .88 .85 .73

1.00 .56 .70 .53

1.00 .42 .35

1.00 .45

1.00

Appendix E. Study 5: Social-Related Checklist Below we use the word “post” to refer to the general ability to share content and “page” refers to the general social media platform, with the understanding that different platforms refer to the “page” differently (i.e., Twitter feed, Pinterest pinboard, etc.). Listed below are some activities that people do when using social media. Please indicate which of these you have done in the last 2 weeks associated with your social life. Please check all that apply. Looked at what my friends posted. Shared new trend in fashion, music, and etc. social media. Felt better about myself after reading someone else's post. Liked a friends post. Sought advice from my friends. Felt worse about myself after reading someone else's post. Commented on a friends post. Looked at old pictures from my photos. Became mad or frustrated after reading someone else's post. Re-posted/shared a friends post. Looked at old pictures of my friends. De-friended someone. Shared my location. Re-posted a story that touched me. Shared a photo. Looked at someone else's location. Changed my mind based on something I've read on social media. Re-posted a photo a friend posted. Facebook “stalked” a friend. Encouraged a friend. Invited others to events. Looked at posts of someone I'm not friends with. Reconnected with someone from my past. RSVP'd to attend an event. Became friends with someone new. Connected with a celebrity. Looked for new trends in fashion, music, etc. on social media.

None of these in the last 2 weeks. Appendix F. Study 5: Consumer-Related Checklist Now we would like you to think about Brands and Companies in general. Below are some activities that people do when using social media. Please indicate which of these you have done in the last 2 weeks associated with brands and companies. Below we use the word “post” to refer to the general ability to share content and “page” refers to the general social media platform, with the understanding that different platforms refer to the “page” differently (i.e., Twitter feed, Pinterest pinboard, etc.). Please check all that apply. Shared when I was at a company's location. Made a positive commented on a company or brand's social media page. Shared a photo on a company or brand's social media page. Looked at a company or brands' social media page. Made a negative comment on a company or brand's social media page. Changed my mind based on something I saw on a company or brand's social media page. Learned information from a company or brand's social media page. Bashed a company or brand on my personal social media page. “Unliked” a company or brand's social media page. “Liked” a company or brand's social media page. Advocated for a company or brand on my personal social media page. Participated in a company or brand's contest. Read a company or brand's post. Asked a question on a company or brand's social media page. Bought something because of what I read on company or brand's social media page. “Liked” a company or brand's post. Looked for new trends in fashion, music, etc. on a company or brand's social media page. RSVP'd to attend a company or brand's event.

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

Shared a company or brand's post. Looked at pictures on a company or brand's social media page. None of these in the last 2 weeks. References Aaker, David A. (2008), Strategic Market Management. John Wiley & Sons. Agnihotri, Raj, Prabakar Kothandaraman, Rajiv Kashyap, and Ramendra Singh (2012), “Bringing “Social” into Sales: The Impact of Salespeople's Social Media Use on Service Behaviors and Value Creation,” Journal of Personal Selling & Sales Management, 32, 3, 333–48. Ainsworth, Mary D. Salter and Silvia M. Bell (1970), “Attachment, Exploration, and Separation: Illustrated by the Behavior of One-year-olds in a Strange Situation,” Child Development, 41, 1, 49–67. Alexandrov, Aliosha, Bryan Lilly, and Emin Babakus (2013), “The Effects of Social- and Self-motives on the Intentions to Share Positive and Negative Word of Mouth,” Journal of the Academy of Marketing Science, 41, 5, 531–46. Ball, A. Dwayne and Lori H. Tasaki (1992), “The Role and Measurement of Attachment in Consumer Behavior,” Journal of Consumer Psychology, 1, 2, 155–72. Batra, Rajeev and Douglas M. Stayman (1990), “The Role of Mood in Advertising Effectiveness,” Journal of Consumer Research, 17, 2, 203–14. Bearden, William O., David M. Hardesty, and Randall L. Rose (2001), “Consumer Self‐confidence: Refinements in Conceptualization and Measurement,” Journal of Consumer Research, 28, 1, 121–34. Berger, Jonah and Katherine L. Milkman (2012), “What Makes Online Content Viral?” Journal of Marketing Research, 49, 2, 192–205. Bettencourt, Lance A. (1997), “Customer Voluntary Performance: Customers as Partners in Service Delivery,” Journal of Retailing, 73, 3, 383–406. Blazevic, Vera, Caroline Wiertz, June Cotte, Ko de Ruyter, and Debbie Isobel Keeling (2014), “GOSIP in Cyberspace: Conceptualization and Scale Development for General Online Social Interaction Propensity,” Journal of Interactive Marketing, 28, 2, 87–100. Bowlby, John (1969), Attachment and Loss: Attachment, Vol. 1. Reading: Addison-Wesley. ——— (1979), The Making and Breaking of Affectional Bonds. London: Tavistock. ——— (1980), Attachment and Loss: Sadness and Depression, Vol. 3. New York: Basic Books. Brocato, E. Deanne, Julie Baker, and Clay M. Voorhees (2014), “Creating Consumer Attachment to Retail Service Firms Through Sense of Place,” Journal of the Academy of Marketing Science, 1–21. Brown, Tom J., Thomas E. Barry, Peter A. Dacin, and Richard F. Gunst (2005), “Spreading the Word: Investigating Antecedents of Consumers' Positive Word-of-Mouth Intentions and Behaviors in a Retailing Context,” Journal of the Academy of Marketing Science, 33, 2, 123–38. Calder, Bobby J., Edward C. Malthouse, and Ute Schaedel (2009), “An Experimental Study of the Relationship Between Online Engagement and Advertising Effectiveness,” Journal of Interactive Marketing, 23, 4, 321–31. Churchill Jr., Gilbert A. (1979), “A Paradigm for Developing Better Measures of Marketing Constructs,” Journal of Marketing Research, 16, 1, 64–73. de Matos, Celso Augusto and Carlos Alberto Vargas Rossi (2008), “Word-ofMouth Communications in Marketing: A Meta-analytic Review of the Antecedents and Moderators,” Journal of the Academy of Marketing Science, 36, 4, 578–96. De Valck, Kristine, Gerrit H. Van Bruggen, and Berend Wierenga (2009), “Virtual Communities: A Marketing Perspective,” Decision Support Systems, 47, 3, 185–203. DeVellis, Robert F. (2003), Scale Development: Theory and Applications. (Applied Social Research Methods). London: Sage. Edison, Steve W. and Gary L. Geissler (2011), “An Investigation of Negative Word-of-Mouth Communication Among Market Mavens,” Journal of Behavioral Studies in Business, 3, 1–12.

87

Ehrenberg, Alexandra, Suzanna Juckes, Katherine M. White, and Shari P. Walsh (2008), “Personality and Self-esteem as Predictors of Young People's Technology Use,” CyberPsychology & Behavior, 11, 6, 739–41. Fennis, Bob M., Ad Th. H. Pruyn, and Mascha Maasland (2005), “Revisiting the Malleable Self: Brand Effects on Consumer Self-perceptions of Personality Traits,” Advances in Consumer Research, 32, 1, 371–7. Finn, Jeremy D. (1974), A General Model for Multivariate Analysis. Orlando: Holt, Rinehart & Winston. Fornell, Claes and David F. Larcker (1981), “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error,” Journal of Marketing Research, 18, 39–50. ——— and Fred L. Bookstein (1982), “Two Structural Equation Models: LISREL and PLS Applied to Consumer Exit-voice Theory,” Journal of Marketing Research, 19, 4, 440–52. Fraley, R. Chris and Keith E. Davis (1997), “Attachment Formation and Transfer in Young Adults' Close Friendships and Romantic Relationships,” Personal Relationships, 4, 2, 131–44. Gensler, Sonja, Franziska Völckner, Yuping Liu-Thompkins, and Caroline Wiertz (2013), “Managing Brands in the Social Media Environment,” Journal of Interactive Marketing, 27, 4, 242–56. Gillan, Paul (2010), “The New Conversation: Taking Social Media from Talk to Action,” Harvard Business Review, 1–24. Hair, Joseph F., William C. Black, Barry J. Babin, Rolph E. Anderson, and Ronald L. Tatham (2010), Multivariate Data Analysis, Vol. 7. Upper Saddle River: Pearson Prentice Hall. Hazan, Cindy and Debra Zeifman (1994), “Sex and the Psychological Tether,” in Advances in Personal Relationships, Vol. 5. Kim Bartholomew, Daniel Perlman, editors. London: Jessica Kingsley, 151–78. ——— and Phillip R. Shaver (1994), “Attachment as an Organizational Framework for Research on Close Relationships,” Psychological Inquiry, 5, 1, 1–22. ——— and Debra Zeifman (1999), “Pair Bonds as Attachments,” in Handbook of Attachment, Jude Cassidy, Phillip R. Shaver, editors. New York: Guilford, 336–54. Hoffman, Donna L. and Marek Fodor (2010), “Can you Measure the ROI of Your Social Media Marketing?” Sloan Management Review, 52, 1, 41–9. Hollebeek, Linda D., Mark S. Glynn, and Roderick J. Brodie (2014), “Consumer Brand Engagement in Social Media: Conceptualization, Scale Development and Validation,” Journal of Interactive Marketing, 28, 2, 149–65. Hollenbeck, Candice R. and Andrew M. Kaikati (2012), “Consumers' Use of Brands to Reflect Their Actual and Ideal Selves on Facebook,” International Journal of Research in Marketing, 29, 4, 395–405. Hu, Li-tze T. and Peter M. Bentler (1999), “Cutoff Criteria for Fit Indexes in Covariance Structure Analysis: Conventional Criteria Versus New Alternatives,” Structural Equation Modeling: A Multidisciplinary Journal, 6, 1, 1–55. Ipsos (2013), “Social Networking Eats Up 3+ Hours per Day for the Average American User,” Retrieved January 05, 2013 from http://www.marketingcharts. com/wp/interactive/social-networking-eats-up-3-hours-per-day-for-the-averageamerican-user-26049/. Kietzmann, Jan H., Kristopher Hermkens, Ian P. McCarthy, and Bruno S. Silvestre (2011), “Social Media? Get Serious! Understanding the Functional Building Blocks of Social Media,” Business Horizons, 54, 3, 241–51. King, Robert Allen, Pradeep Racherla, and Victoria D. Bush (2014), “What We Know and Don't Know About Online Word-of-Mouth: A Review and Synthesis of the Literature,” Journal of Interactive Marketing, 28, 3, 167–83. Kleine, Susan Schultz, Robert E. Kleine III, and Chris T. Allen (1995), “How Is a Possession“ Me” or “ Not Me”? Characterizing Types and an Antecedent of Material Possession Attachment,” Journal of Consumer Research, 22, 3, 327–43. Labrecque, Lauren I. (2014), “Fostering Consumer–Brand Relationships in Social Media Environments: The Role of Parasocial Interaction,” Journal of Interactive Marketing, 28, 2, 134–48. Long, J. Scott (1997), “Regression Models for Categorical and Limited Dependent Variables,” Advanced Quantitative Techniques in the Social Sciences, 7.

88

R.A. VanMeter et al. / Journal of Interactive Marketing 32 (2015) 70–88

——— and Jeremy Freese (2006), Regression Models for Categorical Dependent Variables Using Stata. College Station: Stata Press. Lutz, J. Gary and Tanya Eckert (1994), “The Relationship Between Canonical Correlation Analysis and Multivariate Multiple Regression,” Educational and Psychological Measurement, 54, 3, 666–75. Malär, Lucia, Harley Krohmer, Wayne D. Hoyer, and Bettina Nyffenegger (2011), “Emotional Brand Attachment and Brand Personality: The Relative Importance of the Actual and the Ideal Self,” Journal of Marketing, 75, 4, 35–52. Marketing Science Institute (2013), “MSI Call for Research Proposals on Social Interactions and Social Media Marketing,” Retrieved from http://www.msi. org/uploads/files/2013-11-07_MSI_Social_Media_Competition.pdf. Mattila, Anna S. and Jochen Wirtz (2004), “Consumer Complaining to Firms: The Determinants of Channel Choice,” Journal of Services Marketing, 18, 2, 147–55. Mende, Martin, Ruth N. Bolton, and Mary Jo Bitner (2013), “Decoding Customer–Firm Relationships: How Attachment Styles Help Explain Customers' Preferences for Closeness, Repurchase Intentions, and Changes in Relationship Breadth,” Journal of Marketing Research, 50, 1, 125–42. Mersey, Rachel Davis, Edward C. Malthouse, and Bobby J. Calder (2010), “Engagement with Online Media,” Journal of Media Business Studies, 7, 2, 39–56. Mick, David G. and Michelle DeMoss (1990), “Self-gifts: Phenomenological Insights from Four Contexts,” Journal of Consumer Research, 17, 3, 322–32. Naylor, Rebecca Walker, Cait Poynor Lamberton, and Patricia M. West (2012), “Beyond the “Like” Button: The Impact of Mere Virtual Presence on Brand Evaluations and Purchase Intentions in Social Media Settings,” Journal of Marketing, 76, 6, 105–20. Nunnally, Jum C. and Ira H. Bernstein (1994), Psychometric Theory. New York: McGraw-Hill. Pan, Lee-Yun and Jyh-Shen Chiou (2011), “How Much Can You Trust Online Information? Cues for Perceived Trustworthiness of Consumer-generated Online Information,” Journal of Interactive Marketing, 25, 2, 67–74. Parasuraman, Arun, Dhruv Grewal, and Ramu Krishnan (2006), Marketing Research. Boston: Cengage Learning. Park, Seung Ho, Roger Rongxin Chen, and Scott Gallagher (2002), “Firm Resources as Moderators of the Relationship Between Market Growth and Strategic Alliances in Semiconductor Start-ups,” Academy of Management Journal, 45, 3, 527–45. Park, C. Whan, Deborah J. MacInnis, Joseph Priester, Andreas B. Eisingerich, and Dawn Iacobucci (2010), “Brand Attachment and Brand Attitude Strength: Conceptual and Empirical Differentiation of Two Critical Brand Equity Drivers,” Journal of Marketing, 74, 6, 1–17. Peters, Kay, Yubo Chen, Andreas M. Kaplan, Björn Ognibeni, and Koen Pauwels (2013), “Social Media Metrics—A Framework and Guidelines for Managing Social Media,” Journal of Interactive Marketing, 27, 4, 281–98. Rapp, Adam, Lauren Skinner Beitelspacher, Dhruv Grewal, and Douglas E. Hughes (2013), “Understanding Social Media Effects Across Seller, Retailer, and Consumer Interactions,” Journal of the Academy of Marketing Science, 41, 5, 547–66. Richins, Marsha L. and Scott Dawson (1992), “A Consumer Values Orientation for Materialism and Its Measurement: Scale Development and Validation,” Journal of Consumer Research, 19, 3, 303–19.

Ringle, Christian M., Sven Wende, and Alexander Will (2005), SmartPLS Release 2.0. Hamburg, Germany: University of Hamburg. Schulze, Christian, Lisa Schöler, and Bernd Skiera (2014), “Not All Fun and Games: Viral Marketing for Utilitarian Products,” Journal of Marketing, 78, 1, 1–19. Shimp, Terence A. and Subhash Sharma (1987), “Consumer Ethnocentrism: Construction and Validation of the CETSCALE,” Journal of Marketing Research, 24, 3, 280–9. Slater, Jan S. (2001), “Collecting Brand Loyalty: A Comparative Analysis of How Coca-Cola and Hallmark Use Collecting Behavior to Enhance Brand Loyalty,” Advances in Consumer Research, 28, 1, 362–9. Soat, Molly (2014), Marketers Forge Ahead on Social Media Spending Despite Uncertainties, Research Shows. Marketing News Weekly (September, Retrieved from https://www.ama.org/publications/eNewsletters/ Marketing-News-Weekly/Pages/cmo-survey.aspx). Stelzner, Michael A. (2014), How Marketers Are Using Social Media to Grow Their Businesses. Research Report. Social Media Examiner (May 19). Stephen, Andrew T. and Jeff Galak (2012), “The Effects of Traditional and Social Earned Media on Sales: A Study of a Microlending Marketplace,” Journal of Marketing Research, 49, 5, 624–39. Thomson, Matthew, Deborah J. MacInnis, and C. Whan Park (2005), “The Ties That Bind: Measuring the Strength of Consumers' Emotional Attachments to Brands,” Journal of Consumer Psychology, 15, 1, 77–91. ——— (2006), “Human Brands: Investigating Antecedents to Consumers' Strong Attachments to Celebrities,” Journal of Marketing, 70, 3, 104–19. Trinke, Shanna J. and Kim Bartholomew (1997), “Hierarchies of Attachment Relationships in Young Adulthood,” Journal of Social and Personal Relationships, 14, 5, 603–25. Valkenburg, Patti M., Jochen Peter, and Alexander P. Schouten (2006), “Friend Networking Sites and Their Relationship to Adolescents' Well-being and Social Self-esteem,” CyberPsychology & Behavior, 9, 5, 584–90. Van der Lans, Gerrit, Ralf Van Bruggen, Jehoshua Eliashberg, and Berend Wierenga (2010), “A Viral Branching Model for Predicting the Spread of Electronic Word of Mouth,” Marketing Science, 29, 2, 348–65. Wang, Xia, Chunling Yu, and Yujie Wei (2012), “Social Media Peer Communication and Impacts on Purchase Intentions: A Consumer Socialization Framework,” Journal of Interactive Marketing, 26, 4, 198–208. White, Susan S. and Benjamin Schneider (2000), “Climbing the Commitment Ladder the Role of Expectations Disconfirmation on Customers' Behavioral Intentions,” Journal of Service Research, 2, 3, 240–53. Wilcox, Keith and Andrew T. Stephen (2013), “Are Close Friends the Enemy? Online Social Networks, Self-esteem, and Self-control,” Journal of Consumer Research, 40, 1, 90–103. Zhang, Yinlong, Lawrence Feick, and Vikas Mittal (2014), “How Males and Females Differ in Their Likelihood of Transmitting Negative Word of Mouth,” Journal of Consumer Research, 40, 6, 1097–108.