Consumer behavior on Facebook

14 downloads 38913 Views 201KB Size Report
the paper can consider consumers' voluntary behaviors on Facebook from participation to ... EuroMed Journal of Business. Vol. ..... My brand gives me everything that I expect ..... advertising The ZARA case”, Online Information Review, Vol.
The current issue and full text archive of this journal is available at www.emeraldinsight.com/1450-2194.htm

EMJB 9,3

Consumer behavior on Facebook Does consumer participation bring positive consumer evaluation of the brand?

252

Ching-Wei Ho Department of Marketing, Feng Chia University, Taichung, Taiwan

Received 10 December 2013 Revised 24 January 2014 Abstract 28 March 2014 Purpose – The purpose of this paper is to demonstrate consumers’ voluntary behaviors on Facebook Accepted 2 May 2014

through exploring how members’ community participation affects consumer citizenship behaviors toward the brand. The study also provided further insight into the mediating effect by considering brand trust and community identification. Design/methodology/approach – This research begins by developing a framework to describe and examine the relationship among Facebook participants, brand trust, community identification, and consumer citizenship behaviors. Furthermore, it tests the mediating effects of brand trust and community identification on the relationship between Facebook participation and consumer citizenship behaviors. The model and hypotheses in this study employ structural equation modeling with survey data. Findings – First, this study reveals consumers’ community participation on Facebook has directly positive and significant effects on brand trust and community identification. Second, this research confirms that brand trust has directly positive and significant effects on community identification. Third, this study found that brand trust and community identification play a mediating role between Facebook participation and consumer citizenship behaviors. Research limitations/implications – The sample comprised primarily young adults, which may not be completely generalizable to the population at large. This study examined a specific form of virtual community, Facebook, so the results cannot be ascribed to other formats of brand community. Originality/value – The issue of consumer’ voluntary behavior on social networking sites has become more and more important. This study proposed an exclusive model of the process by which the paper can consider consumers’ voluntary behaviors on Facebook from participation to consumer citizenship behavior toward the brand. This finding can be viewed as pioneering, setting a benchmark for further research. Keywords Trust, Identification, Community participation, Citizenship behaviors, Facebook community Paper type Research paper

EuroMed Journal of Business Vol. 9 No. 3, 2014 pp. 252-267 r Emerald Group Publishing Limited 1450-2194 DOI 10.1108/EMJB-12-2013-0057

1. Introduction Currently, social networks have become extremely popular; they are defined as networks of friends for social or professional interactions (Trusov et al., 2009). Social networking is considered a tool that supports both electronic marketing and viral marketing and enables the process of building connections to a network or social circle (Zarella, 2010). Social networking enables connections with a network of people who share common interests or goals (Hsu, 2012) and affords companies the possibility of mapping social connections to expand relationships and spread information (Boyd and Ellison, 2007; Cross and Parker, 2004). Of all the social networks, Facebook is the most popular and claims to have attracted more than 751 million active monthly users (as of March 2013) since starting in February 2004 (www.facebook.com); Facebook has become the top social networking site based on number of users and volume of access or use (Hsu, 2012). Facebook has changed consumer behavior; for example, consumers dedicate almost one-third of their time to social networking (Lang, 2010) and half of these active users

log on every day. Thus, companies and brand players find it necessary to maintain a brand presence somewhere on Facebook. Therefore, the Facebook fan page as a brand community on Facebook was established, where fans and consumers can communicate and interact with companies or brands using the “Like” or “Comment” option. According to Hsu (2012), the Facebook community has the following characteristics: shares company, product, or service information; communicates and shares marketing messages; expands networks; and receives feedback updates, which provide members with as many opportunities as possible to become involved and participate in the community. Community participation is an important issue that influences participants’ behaviors (Kaplan and Haenlein, 2010; Ouwersloot and Odekerken-Schro¨der, 2008; Royo-Vela and Casamassima, 2011). However, in previous studies, the consequence of participation has usually been discussed in terms of loyalty (e.g. Casalo et al., 2007); less mentioned is the specific behavioral form, such as voluntary consumer behaviors that will benefit the brand (i.e. consumer citizenship behaviors). Actually, when a member is willing to participate, regardless of passive or active mode, in a Facebook community, it is a kind of voluntary consumer behavior. Thus, would voluntary behavior in participation in a Facebook community affect a consumer’s citizenship behavior and how? This question signals the gap that this study attempts to close. The objectives of the current empirical study are three: first, to enhance and examine the knowledge and the relationships among Facebook participants, brand trust, community identification, and consumer citizenship behaviors. It hypothesizes that the more members participate in a Facebook brand community, the more likely they are to trust this brand and/or identify as a community member, and then to exhibit consumer citizenship behaviors that benefit the brand; second, to examine the mediating effects of brand trust and community identification; third, to propose an model of the process by which we can consider consumers’ voluntary behaviors on Facebook from participation to consumer citizenship behaviors toward the brand. This research begins by developing a framework to describe and examine the relationship among Facebook participants, brand trust, community identification, and consumer citizenship behaviors. Furthermore, it tests the mediating effects of brand trust and community identification on the relationship between Facebook participation and consumer citizenship behaviors. The model and hypotheses in this study employ structural equation modeling (SEM) with survey data. Finally, this paper concludes with a discussion of the marketing significance, theoretical and practical implications, limitations, and suggestions for future research. 2. Literature and hypotheses development 2.1 Facebook participation and trust Facebook community participation can be discussed in terms of interacting and cooperating with community elements and participating in joint activities (Algesheimer et al., 2005; Matzler et al., 2011). Regarding the community element, according to McAlexander et al. (2002), “a community is made up of its entities and the relationships among them” (p. 38). That is, a Facebook community comprises entities (i.e. the brand, products, customers, and the company). Therefore, community participation can be considered the interactions and communications among elements of a brand community. Meanwhile, Facebook fans control their participation level because of the voluntary nature of the brand community. Members participate for free and there are no clear requirements for their contributions to ongoing communication. Some members in the

Consumer behavior on Facebook 253

EMJB 9,3

254

Facebook community get involved with more active communication, while others focus on watching the continuing communication around shared brand interests (Cothrel and Williams, 1999; Ridings et al., 2002). When a customer logs on to a Facebook community to become a member and comments, shares experiences, interacts with marketers, asks questions about the brand or product, or answers comments, that member is participating in the community’s activities. During these interactions, meaningful experiences, useful information, and other valuable resources are being shared among members so that ties are strengthened in such communities (Laroche et al., 2013) and increase individual willingness to participate in the communities. According to Chaudhuri and Holbrook (2001), brand trust is “the willingness of the average consumer to rely on the ability of the brand to perform its stated function” (p. 82). In a situation of uncertainty, brand trust plays an important role in reducing uncertainty. A trusted brand makes consumers feel comfortable (Chiu et al., 2010; Doney and Cannon, 1997; Gefen et al., 2003; Moorman et al., 1992; Pavlou et al., 2007). Holmes (1991) argued that repeated interaction and maintaining long-term relationships are key factors in building trust. Enhancing relationships with customers and elements of the brand community can enhance relationships and increase contacts between the brand and the customers so that brand trust is influenced positively (Laroche et al., 2013). Kang et al. (2014) also approved that active participation on Facebook fan pages has a positive influence on brand trust. Therefore, in terms of the degree to which Facebook members frequently participate and interact within the brand community, trust in the brand makes members feel more comfortable. Based on the above discussion, the following hypothesis is put forward: H1. The greater the level of participation in a Facebook brand community, the more likely it is that members will trust the brand. 2.2 Facebook participation and identification Facebook community participation can be discussed in terms of acting in ways that endorse the community and enhance its value for members and others (Algesheimer et al., 2005; Matzler et al., 2011). Community member participation is commonly classified into two different modes: passive (participation) and active (interaction) (Kozinets, 1999; Qu and Lee, 2011; Wang and Fesenmaier, 2004). Prior research has suggested that even passive participation, which includes lurking, also relates to members’ sense of identification (Arnett et al., 2003). Some academics have suggested that active lurking, which means actively participating but without interacting, enables members to better evaluate online communications and develop a sense of identification (Bhattacharya and Sen, 2003; Carlson et al., 2008). Moreover, several academics, such as Bergami and Bagozzi (2000), Gruen et al. (2000), and McWilliam (2000), have suggested that if members become psychologically attached to the community, they are more likely to behave in accordance with community values. According to Lembke and Wilson (1998), community identification exists when a member feels, thinks, and behaves like a member of the community, which means that the member distinguishes a community identity from a self-identity. Some researchers have proposed that community identification involves both cognitive self-categorization and affective commitment (e.g. Algesheimer et al., 2005). Cognitive self-categorization occurs through consumers’ comparison of their own defining characteristics to those that define the community (Bergami and Bagozzi, 2000). Affective commitment takes this process a step further into feelings of attachment and

belongingness (Algesheimer et al., 2005; McAlexander et al., 2002). Consumers who engage in a variety of social activities have direct access to other members and can mediate the flow of resources in the community. Thus, participation in brand community activities makes them more like insiders (Tsai and Pai, 2012). Therefore, once an individual participates in a Facebook community and becomes a fan of a brand page, no matter whether in passive or active mode, he or she will be affected by the community values and gradually develop identification with this brand community. Consolidating the theoretical arguments reviewed so far, we hypothesize that: H2. The greater the level of participation in a Facebook brand community, the greater the brand community identification. Analysis of the relationship between trust and identification is not found in the extant literature. However, based on the above discussions about brand trust and community identification, we have proposed that a member who trusts and relies on a brand will also be emotionally attached and identify as a part of the brand community. That is, a positive relationship exists between brand trust and community identification. Thus: H3. The higher the level of trust toward a brand, the greater the brand community identification. 2.3 Consumer citizenship behaviors Customer citizenship behaviors comprise voluntary customer behaviors that benefit the firm and go beyond customer role expectations (Gruen, 1995). Customers perform citizenship behaviors at their sole discretion (Bettencourt, 1997; Groth, 2005) and customer citizenship behaviors provide extraordinary value to the firm. The literature suggests various forms of customer citizenship behaviors, such as positive word-of-mouth (WOM) communication, constructive involvement in suggesting service improvements, and other polite and courteous behaviors (Bettencourt, 1997; Rosenbaum and Massiah, 2007). Based on the concept of organizational citizenship behavior (Bateman and Organ, 1983), consumers are more likely to express their support for an organization (e.g. participate in a brand community) by engaging in in-role behaviors like purchasing products from the company (Ahearne et al., 2005) and extra-role behaviors, such as making recommendations to others and engaging in positive WOM (Anderson et al., 2004; Bettencourt, 1997). In most recent researches, customer citizenship behaviors have been discussed and applied in the online behavior context, e.g. Anaza and Zhao (2013) and Anaza (2014). Social exchange theory (Blau, 1964) suggests that the association between the consumer and organization can be seen as social exchanges in which consumers give back a positive gain (e.g. identity or experience) from a sense of personal obligation or gratitude by providing positive feedback to the organization (Lii and Lee, 2012). In the context of the online community (e.g. a Facebook community), social exchange theory can also be applied to demonstrate the relationships between members and other parties. These relationships are seen as exchanges in which a receiver reciprocates a positive personal effect by providing positive outcomes to the other party, such as citizenship behaviors (Chen et al., 2010). When consumers participate and interact within a Facebook community, they may gain more information by sharing or psychologically supporting and gradually identify themselves as part of this

Consumer behavior on Facebook 255

EMJB 9,3

256

community; they will be more likely to respond with reciprocal behavior that may benefit the community. Citizenship behaviors may be one type of benefit. In addition, Blau (1964) pointed out that social exchange is based on the expectation of trust and reciprocation, as the exact nature of the return is left unspecified. Morgan and Hunt (1994) theorized that trust is the key mediating variable between the antecedents and consequences of developing a long-term customer relationship. Consumers’ willingness to exhibit citizenship behavior to a brand presents their intent to maintain a relationship with the brand. Thus, when consumers participate in a Facebook community, they may gain more information or become familiar with the brand and gradually trust the brand; therefore, they are more likely to engage in reciprocal behavior that may benefit the brand. Thus, the following hypotheses are proposed: H4. The greater the brand trust, the more likely it is that consumers will exhibit in-role and extra-role behaviors that support the community. H5. The greater the community identification, the more likely it is that consumers will exhibit in-role and extra-role behaviors that support the company. The integrated theoretical framework as represented by H1H5 is shown in Figure 1. 3. Research method 3.1 Sample To date, social network users are more concentrated in Asia. Currently, almost 90 percent of Asian brands use social networks as a marketing platform, and 75 percent of these brands have developed social networking strategies that have been in use for longer than a year (Pon and Wang, 2012). Therefore, the data used to examine the hypotheses were collected in Taiwan from Facebook community members. As Zhao (2011) suggested, Facebook communities include four forms: public celebrity communities, individual sharing communities, online game or app software communities, and company/brand communities. In the top 100 Taiwanese Facebook communities in 2011, company/brand communities (e.g. seven-Eleven, Starbucks, Rakuten Ichiba Taiwan) had a 33 percent share (Zhao and Yang, 2011). The purpose of this research was to explore whether Facebook participation leads to citizenship behaviors toward the brand; thus, only the community for companies/ brands was considered in this study. Moreover, the sample of respondents was obtained based on the following qualification: Respondents should have been registered and active in at least one Facebook brand community for longer than three

H4a

Trust H1

In-role behavior

H4b Participation

H3 H5a

H2

Figure 1. The conceptual model

Identification

Ex-role behavior H5b

months. According to Chin and Newsted (1999), a sample size of 150-200 is required to attain reliable coefficient values using partial least squares (PLS) analysis (Hur et al., 2011). Hair et al. (2010) suggested that the ratio of observations to the independent variable should not fall below 5 (5:1), although the preferred ratio is 10 respondents for each independent variable (minimum ratio of observation to variables is 10:1) (Yap et al., 2012). Hence, bearing in mind the 15 variables to be used in SEM, this study required a minimum sample size of 150 respondents. 3.2 Data collection method Data were collected by a structured questionnaire developed for the research and adapted from those used in previous studies. As the target population in this study was members of the Facebook brand community, the questionnaire was distributed through several posts on Facebook and PTT. PTT is the local social networking website in Taiwan with the largest scales and longest history. It provides a platform for discussing Facebook (like the “Facebook forum”) and distributes virtual questionnaires. Gathering data through this site can be more efficient than just gathering data through Facebook. We asked participants to keep in mind a company/ brand Facebook community in which they were members and that they followed while answering the questions. Participants were guaranteed confidentiality and anonymity in relation to their returned questionnaires. Self-administered questionnaires with the assistance of a support letter were used to ensure a good response rate and reduce non-sampling bias in the survey process. An effort was made to randomize data collection at different times of the day and week. At the end of the data collection period, 232 questionnaires were collected with 26 missing values. That is, 206 fully completed questionnaires were used for the data analysis. 3.3 Measurement of variables The survey questionnaire was developed by adapting measures from a variety of studies. Participation in Facebook was measured through three items adapted from Qu and Lee (2011) and Tsai and Pai (2012). To measure brand trust, we used the threeitem scale proposed by Laroche et al. (2013), and to measure community identification, we used the three-item scale proposed by Bergami and Bagozzi (2000) and Tsai and Pai (2012). The construct of in-role behavior consisted of three items adapted from Putrevu and Lord (1994). The measures for the extra-role behavior construct consisted of three items adapted from de Matos et al. (2009). All items for assessing the constructs employed a seven-point Likert scale indicating the extent of agreement or disagreement with the item. The items for each construct and their measurement scales are presented in Table I. 3.4 Data analysis This study used PLS to test the hypotheses and analyze the data. The PLS algorithm allows each indicator to vary in terms of how much it contributes to the composite score of the latent variable, instead of assuming equal weight for all indicators of a scale (Chin et al., 2003; Hur et al., 2011). According to Anderson and Swaminathan (2011), PLS path modeling is commonly used in marketing (Henning-Thurau et al., 2007), international business (Henseler et al., 2009), and information systems (Al-Gahtani et al., 2007; Burton-Jones and Hubona, 2006) studies, necessitating simultaneous estimation of the factor loadings of the measurement model and path coefficients of the structural model. This study used PLS rather than other SEM

Consumer behavior on Facebook 257

EMJB 9,3

258

Construct

Measurement items

Participation in Facebook (Par)

I often watch the FB page activities I actively participate in the FB activities I frequently interact with other FB members My brand gives me everything that I expect out of the product I rely on my brand My brand never disappoints me I feel strong ties to this community I see myself as a part of this community I feel emotionally attached to this community It is very possible that I will buy the brand I will consider buying the brand the next time I need this product I will try this brand I will recommend this brand to my relatives and friends I will tell my relatives and friends about the good experience with this brand I am willing to join the activity hold by the brand in the future

Brand trust (Trust) Community identification (CI) In-role behavior (In)

Ex-role behavior (Ex) Table I. Constructs and their measurement items

Loading

a

CR

AVE

0.829 0.838 0.810

0.77

0.87

0.68

0.928 0.927 0.884 0.902 0.921 0.908 0.759

0.90

0.94

0.83

0.90

0.94

0.83

0.72

0.84

0.64

0.68

0.82

0.61

0.832 0.801 0.799 0.871 0.660

methods (i.e. LISREL, AMOS) because the PLS approach places minimal restrictions on sample size and residual distribution (Hur et al., 2011; Phang et al., 2006). 4. Results 4.1 Demographic profile of respondents Of the 206 respondents, 46 percent was male while 54 percent was female. In terms of age, 62 percent was 20-30 years old, and 29 percent was under 20; these two groups accounted for the largest portion of the sample, followed by those aged 31-40 years (6 percent). Most of the respondents (45 percent) had been members of the brand community for one to two years, 28 percent for two years or longer, and 27 percent for less than one year. 4.2 Measurement model We used the two-step approach as described by Anderson and Gerbing (1988). We first assessed reliability and convergent validity as shown in Table I and then discriminant validity as illustrated in Table II. To examine reliability, Cronbach’s a revealed that all constructs showed a value above 0.6 (the bar adopted by Bagozzi and Yi, 1988).

Par Trust CI In Ex Table II. Correlation matrix

Mean

SD

Par

Trust

CI

In

Ex

3.26 2.66 3.51 2.93 3.30

1.273 0.933 1.216 1.153 1.282

0.83 0.39 0.54 0.45 0.58

0.91 0.55 0.62 0.65

0.91 0.59 0.73

0.80 0.71

0.78

Note: Diagonals represent the square root of the average variance extracted while the other entries represent the correlations

To test for convergent validity, composite reliability (CR), factor loading, and average variance extracted (AVE) were examined. The measures are acceptable if an individual item loading is 40.7, CR exceeds 0.7, and AVE is 40.5 (Gefen et al., 2000). To examine the discriminant validity of the constructs, this study used the Fornell and Lacker (1981) criterion whereby the average variance shared between each construct and its measures should be greater than the variance shared between the construct and other constructs. As shown in Table II, the correlations for each construct are less than the square root of AVE for the indicators measuring that construct, indicating adequate discriminant validity.

Consumer behavior on Facebook 259

4.3 Structural model The explanatory power of the structural model is evaluated by looking at the R2 values. From Figure 2, the R2 values range from 0.154 to 0.625, which suggests that the modeled variables explain 15.4 to 62.5 percent of the variance of the respective dependent variables. From Figure 2, Facebook participation exerts a significant and positive influence on both brand trust (H1, b ¼ 0.393, po0.001) and community identification (H2, b ¼ 0.383, po0.001). Therefore, H1 and H2 are both supported. The model predicted the path from brand trust to community identification (H3) and shows a significant and positive relationship between them (b ¼ 0.402, po0.001). Thus, H3 gains supported. In addition, the paths from brand trust have a significant and positive influence on both in-role behavior (H4a, b ¼ 0.418, po0.001) and extra-role behavior (H4b, b ¼ 0.361, po0.001). Meanwhile, the paths from community identification have a significant and positive influence on both in-role behavior (H5a, b ¼ 0.358, po0.001) and extra-role behavior (H5b, b ¼ 0.532, po0.001). Thus, both H4 and H5 are fully supported. The seven paths examined in the structural model are summarized in Table III. The mediating effects of brand trust and community identification were tested. As seen in Figure 3, the direct path between Facebook participation and in-role and extra-role behavior are both significant. After introducing brand trust as a mediator, the indirect path for the effect of participation on in-role behavior is significant and stronger than the direct path (b ¼ 0.3904b ¼ 0.255). Moreover, the indirect path for the effect of participation on extra-role behavior is also significant and stronger than

Trust R 2 =0.154

0.418***

In-role R 2 =0.468

0.393*** 0.358*** Facebook Participation

0.402*** 0.361*** 0.383*** Identification R 2 =0.429

Significance

Note: ***p