Journal of Business Research - Senior Lecturer II Foundation/Diploma School of Pre ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Journal of Business Research 132 (2021) 146–157 Contents lists available at ScienceDirect Journal of Business Research journal homepage: www.elsevier.com/locate/jbusres “Stop the unattainable ideal for an ordinary me!” fostering parasocial relationships with social media influencers: The role of self-discrepancy Eugene Cheng-Xi Aw a, *, Stephanie Hui-Wen Chuah b, c a UCSI Graduate Business School, UCSI University, Kuala Lumpur, Malaysia b Institute of Innovation and Circular Economy, Asia University, Taichung, Taiwan c Department of Business Administration, Asia University, Taichung, Taiwan A R T I C L E I N F O A B S T R A C T Keywords: The unprecedented interactivity of social media has empowered social media influencers to develop close re Social media influencers lationships with their followers, and such relationships carry important marketing implications for social media Parasocial relationships influencers and brands. The present study examines the antecedents and outcomes of followers’ parasocial re Self-serving motive lationships with social media influencers. Drawing upon the theoretical lens of parasocial interaction theory, Self-discrepancy Gender influence framework, and attribution framework, the study proposes and empirically tests the effects of social Number of followers media influencers’ influence attempts on parasocial relationships, and the subsequent downstream outcomes of perceived endorser motives and consumer purchase intentions. Using a survey-based approach, the study collected 361 usable responses, and data were analyzed using partial least squares structural equation modeling. The study found that (1) influence attempts (i.e., attractiveness, prestige, and expertise) positively influence parasocial relationships, whereas (2) parasocial relationships negatively influence perceived endorser motive (self-serving), which in turn (3) reduces purchase intention, and (4) self-discrepancy moderates the relationships between influence attempts and parasocial relationships. Two post-hoc exploratory analyses uncovers the im pacts of number of followers and gender in the proposed relationships. Theoretical and practical implications are discussed. 1. Introduction endorsements, whereby brands are currently paying influencers about $25, $20, and $10 for every 1000 followers on Facebook, YouTube, and The rise of social media has opened a plethora of new marketing Instagram and Snapchat, respectively (Johnson, 2020). opportunities, including influencer marketing, which involves promo In today’s marketplace flooded by advertisements and influencer tion of products and services by individuals who hold sway over other endorsements, consumers have become more alert to the motives behind consumers (Ge & Gretzel, 2018). In the present study, social media the persuasive attempts of marketers. The importance of motive infer influencers refer to individuals who have gained popularity due to their ence has been made apparent in prior business literature, where authors social media presence and content, such as bloggers, YouTubers, and of extant studies demonstrate that promotions by firms perceived to be Instafamous individuals. Heralded as the new frontier of influencer public-serving instead of self-serving are more likely to generate pur marketing, social media influencers have started to secure their places in chase intent (Lee, Haley, & Yang, 2013; Reinhard, Messner, & Sporer, the market and are even deemed more effective in product/brand 2006). Often, consumers’ perception and motive inference towards so endorsement than the traditional celebrities (Jin, Muqaddam, & Ryu, cial media influencers are imperiled when they partner with brands for 2019). A recent survey by Twitter revealed that 40% of platform users endorsement (Audrezet, De Kerviler, & Moulard, 2020; Pöyry, Pelkonen, were enticed into purchasing a product due to Tweets from influencers Naumanen, & Laaksonen, 2019). A recent research report published by (Influencer Marketing Hub, 2020). Likewise, Google statistics indicate Roth Capital Partners indicates that one-third of surveyed Millennials do that 60% of YouTube subscribers tend to follow purchase advice given not believe that social media influencers have their best interests in by their favorite influencers (Influencer Marketing Hub, 2020). These mind (Kirkpatrick, 2017). As consumers are active recipients of positive implications have resulted in substantial growth in endorsement content, they seek to evaluate the ulterior motives of * Corresponding author. E-mail address: eugenecx.aw@gmail.com (E.C.-X. Aw). https://doi.org/10.1016/j.jbusres.2021.04.025 Received 21 July 2020; Received in revised form 9 April 2021; Accepted 11 April 2021 Available online 22 April 2021 0148-2963/© 2021 Elsevier Inc. All rights reserved.
E.C.-X. Aw and S.H.-W. Chuah Journal of Business Research 132 (2021) 146–157 endorsers (Shan, Chen, & Lin, 2019). It has been suggested that, when 2. Conceptual development consumers perceive social media influencers’ brand endorsements as publicity stunts motivated by financial gains, the endorsement effec 2.1. Social media influencers tiveness decreases (e.g., purchase intention declines), posing a threat to the endorsing brand as well as the influencer (Shan et al., 2019). Most of the existing research on influencer endorsement either in the Therefore, we contend that consumers’ attribution of the motives is the traditional celebrity or social media context has been limited to the cornerstone for building a successful influencer marketing strategy. investigation of source effects (source credibility and attractiveness), While being relevant and insightful, prior studies on this topic are product/brand-endorser match-up, and endorsement outcomes (e.g., confined to treating motive inference towards companies or endorsers as brand attitude, brand trust, purchase intention) (Bergkvist & Zhou, the determinant of product evaluation (Lee et al., 2013; Shan et al., 2016; Jin et al., 2019; Schimmelpfennig & Hunt, 2020; Weismueller, 2019). There is a paucity of studies that tap into the management of Harrigan, Wang, & Soutar, 2020). These theoretical perspectives have motive inference, especially in the influencer marketing context. We been critiqued for being overly simplistic and incapable of capturing the address this gap by applying the lens of parasocial relationship theory complexity of influencer-consumer interactions (Moraes et al., 2019). (Horton & Wohl, 1956; Labrecque, 2014) to the existing studies on These issues have prompted authors of recent studies to advocated for influencers’ influence attempts (Audrezet et al., 2020; Ki & Kim, 2019) the adoption of the influencer-follower emotional attachment lens, and our survey data to understand the mechanism through which the focusing specifically on the role of pseudo-relationships (i.e., parasocial perception of endorser motive as self-serving can be mitigated. Our relationships) in understanding endorsement effectiveness, which is rationale is supported by available evidence indicating that consumers highly relevant in the realm of social media yet remains under- are more likely to trust influencers with whom they have developed researched (Aw & Labrecque, 2020; Moraes et al., 2019). The afore parasocial relationships (i.e., one-sided relationships that individuals mentioned gap has motivated the present study, the aim of which is to form with mediated personas), and tend to withhold their critical examine the potential antecedents and outcomes of parasocial re evaluation towards influencers’ sponsored commercial activities lationships in the social media influencer context. Moreover, influ (Audrezet et al., 2020; Moraes, Gountas, Gountas, & Sharma, 2019). encers’ impression management has received scant attention in the Apart from investigating the implications of parasocial relationships, literature. One prominent aspect of impression management in relation there is a need to identify the mechanisms by which celebrity–fan re to influencer endorsement is the endorser motive inference by con lationships can be reinforced (Aw & Labrecque, 2020). Content posted sumers (Carlson, Donavan, Deitz, Bauer, & Lala, 2020; Shan et al., by social media influencers provides a platform for influencer-fan in 2019). This strand of research has revealed that self-serving motives teractions that may give rise to parasocial relationships (Aw & Lab evoke distrust and induce negative consumer evaluations towards en recque, 2020; Wellman, Stoldt, Tully, & Ekdale, 2020). However, dorsers and the related endorsement. Unfortunately, it remains un compared to the extensively tested source effects (i.e., influencer traits), known how social media influencers can strategically choreograph the influence attempts (i.e., content) as a promotional strategy by social consumers’ motive attribution for themselves, given that they are media influencers have been largely overlooked in extant studies. As a constantly under pressure to balance the expectations of their followers result, it remains largely unknown how social media influencers’ influ to be authentic and benevolent, while being economically profitable for ence attempts can be managed to bolster parasocial relationships. Under their brand partners. This research gap represents a fruitful yet insuffi these premises, we seek to understand the role of parasocial relation ciently explored research area, as consumers’ perceptions of endorser ships in relation to perceived endorser motive, and the principal traits of motives can pose a great threat to the social media influencers’ social media influencers’ content that fosters parasocial relationships. persuasiveness and endorsement effectiveness (Audrezet et al., 2020; The concept of self-discrepancy (i.e., discrepancy between actual and Shan et al., 2019; van Driel & Dumitrica, 2020). ideal selves) is indispensable in the social comparison process, as it inevitably occurs when one is exposed to influencer marketing (Jin 2.2. Theoretical foundation et al., 2019). To the best of authors’ knowledge, there is no previously published research examining the role of self-discrepancy in relation to Authors of recent influencer endorsement studies concur that no social media influencers’ influence attempts. To address this gap, we aim singular theory is adequate for explaining the effectiveness of influencer to examine the moderating role of self-discrepancy to provide a more endorsements (Aw & Labrecque, 2020; Schimmelpfennig & Hunt, nuanced understanding of the boundary conditions embedded in the 2020). To overcome this issue, we integrate the influence framework relationship between influence attempts and parasocial relationships, as (Scheer & Stern, 1992) and attribution framework (Kapitan & Silvera, consumers’ meaning appropriation from influencers’ influence attempts 2016) to (i) elucidate how social media influencers’ influence attempts is subjected to the appraisal of self-concept (i.e., self-discrepancy) foster parasocial relationships, and (ii) identify the downstream impli (Malär, Herzog, Krohmer, Hoyer, & Kähr, 2018). cations of parasocial relationships (i.e., perceived endorser motives and The present study advances extant knowledge on two hitherto consumers’ purchase intentions). The influence framework pioneered by overlooked topics in the influencer endorsement literature, namely Scheer and Stern (1992) highlights the power dynamic in the influencer parasocial relationships and perceived endorser motives. Through a marketing context, in which social media influencers are in control of series of main and post-hoc analyses, it contributes to the understanding power resources and leverage them trigger desired attitudinal and of the mechanisms through which social media influencers’ influence behavioral responses in their followers. In the present study, the power attempts (i.e., attractiveness, prestige, and expertise) can foster para resources encompass attractiveness, prestige, and expertise (Ki & Kim, social relationships, which in turn determine consumers’ perceptions of 2019; Scheer & Stern, 1992) as influencers’ primary marketing tools, endorser motives and purchase intentions. Additionally, we offer in- which they promote through their social media content. These influence depth insights into the aforementioned relationships by shedding light attempts (i.e., attractiveness, prestige, and expertise) not only determine on the moderating role of self-discrepancy and gender. Based on our the persuasiveness of communication but also the perceived social re findings, we offer recommendations to brand marketers and social lationships between the information source and the receiver (Ismagi media influencers regarding endorser selection as well as impression/ lova, Slade, Rana, & Dwivedi, 2019; Kapitan & Silvera, 2016). relationship management in order to optimize brand endorsement Moreover, by synthesizing the critical theoretical elements of meaning effectiveness. transfer, attribution, source effect, and information processing, the attribution framework allows identification of the linkages between source factors and correspondence bias (Fiske & Taylor, 1991; Kapitan & Silvera, 2016). The theory contends that people tend to make 147
E.C.-X. Aw and S.H.-W. Chuah Journal of Business Research 132 (2021) 146–157 correspondence bias in their attributional processing when they socially social media influencers that is complementary to the traditional source identify with the interacting entities (i.e., influencer endorsers) that effects. In addition, although past research has contended that the halo manifest positive source characteristics, motivating consumers to adopt effect may cloud consumers’ perceptions and render them less critical of internal/dispositional attribution and to believe that the influencer en paid commercial and promotional activities by their favorite celebrities dorsers truly like the products they are promoting while disregarding the (Moraes et al., 2019), the effect that parasocial relationships may exert fact that they are being paid and sponsored (Kapitan & Silvera, 2016; on perceived influencer endorser motives has never been investigated, Kelman, 1961). Such attributions and correspondent inferences endow presenting a gap that is addressed in the present study. unquestioned credibility to the influencer endorsers’ motives, which underlies the success of endorsement marketing. 2.4. Fostering parasocial relationships through influence attempts 2.3. Parasocial relationships We leveraged on the general social media marketing literature to determine the types of influence attempts to be included in our research Parasocial relationships denote one-sided relationships that a person model. Successful social media marketing entails strategically crafted develops with mediated personas, including social media influencers social media content targeting a specific consumer segment (Dolan, (Horton & Wohl, 1956). These relationships create an illusion of real and Conduit, Frethey-Bentham, Fahy, & Goodman, 2019; Swani, Milne, intense interpersonal connections, encompassing the elements of Brown, Assaf, & Donthu, 2017). In that respect, social media content friendship and understanding (Chung & Cho, 2017; Horton & Wohl, conveys functional and hedonic appeals that address specific needs of 1956). Recently, the role of parasocial relationships has been extended consumers (Swani et al., 2017). Here, the term “functional” refers to the beyond the traditional mass media such as television and radio to informativeness or expertise aspects of the content, while “hedonic” include the online environment (Labrecque, 2014). The social media encapsulates affective and transformational components (Dolan et al., environment provides influencers and fans a more conducive platform 2019; Tafesse & Wein, 2018). According to these definitions, social for fostering interactions and connections of an unprecedented intensity. media content appeals largely conform to the core postulates of the Influencers can interact with consumers instantly through social media source effect model frequently adopted in the influencer literature. For and can effortlessly share their everyday lives from a first-person example, functional appeals via social media content are congruent with perspective. As a result, despite not knowing influencers in person, the source expertise attribute (to address need for information), while consumers tend to develop intense intimacy and psychological connec hedonic appeals—comprising of sensory qualities (e.g., aesthetically tions with them (Schnickel, 1985). appealing) and emotional/resonance cues (e.g., ego and social The concept of parasocial relationships has long been a subject of messages)—reflect the concepts of source attractiveness and prestige, academic research, but the focus has recently shifted from traditional respectively (Kim & Yang, 2017). Notably, findings reported in recent celebrities to social media influencers. Findings yielded by these in social media influencer literature validate the content attributes of vestigations indicate that social media interactions foster parasocial attractiveness, prestige, and expertise as the three most common influ relationships with celebrities (Chung & Cho, 2017). For instance, ce ence attempts employed by influencers in their interactions with fol lebrities are able to reduce interpersonal uncertainty and engender more lowers (Ki, Cuevas, Chong, & Lim, 2020; Ki & Kim, 2019), further intense parasocial relationships through higher posting frequency justifying the inclusion of these influence attempts in the present study. (Ledbetter & Redd, 2016). Thus far, researchers tended to examine Interpersonal theories postulate that communication is pivotal to the parasocial relationships through the lens of source characteristics development of interpersonal relationships, and the same principle ap model. These studies have identified influencers’ attractiveness and plies to the formation of parasocial relationships (Altman & Taylor, credibility as salient predictors of parasocial relationships, suggesting 1973; Rubin & McHugh, 1987), even in the new media context. As that influencers who are perceived as good-looking and trustworthy can evidenced by Labrecque (2014), parasocial interactions can be fostered evoke intense illusionary attachment in their fans (Gong & Li, 2017; through social media message content cues (i.e., perceived interactivity Sakib, Zolfagharian, & Yazdanparast, 2020; Sokolova & Kefi, 2020). and openness in communication). According to the theory of parasocial Another major stream of research relates to the impacts of parasocial interactions, parasocial relationships represent an outcome of a need or relationships on influencer- and brand-related outcomes. Authors of motivational state (e.g., expressive and affiliation needs) (Auter, 1992), prior studies in this domain posit that consumer-endorser relationships and these needs are gratified through communication with the media/ are a powerful indicator of endorsement effectiveness as well as influ media figure (Rubin & McHugh, 1987). Followers tend to gravitate to encers’ ability to persuade consumers (Aw & Labrecuqe, 2020; Carlson wards intimate relationships with media figures if their needs are ful et al., 2020; Wellman et al., 2020). For example, parasocial relationships filled (Thomson, 2006). Media personalities, including social media have been found to increase the trustworthiness of celebrities endorsing influencers, deliberately utilize influence attempts (i.e., social media the brand, which would in turn strengthen followers’ intention to pur content) to engender the feeling of closeness and connectedness (Fer chase (Chung & Cho, 2017; Jin & Ryu, 2020; Reinikainen, Munnukka, chaud, Grzeslo, Orme, & LaGroue, 2018; Yuksel & Labrecque, 2016). Maity, & Luoma-aho, 2020; Sokolova & Kefi, 2020). The principal traits of these influence attempts serve as meaning cues Synthesizing the literature reviewed above, it can be concluded that that consumers appropriate for their need fulfillment, and determine social media influencer research has offered ample evidence indicating emotional bonding with social media influencers (Ki et al., 2020). that influencer characteristics are crucial in fostering parasocial re Drawing upon the social psychology literature, desired attributes such as lationships, which can yield benefits to influencers and their endorsing attractiveness, prestige, and expertise are central to interpersonal brands. However, despite recognizing influencers’ social media content relationship-related judgments (Montoya & Horton, 2014). Therefore, as the key to effective communication and formation of influencer- we posit that these content-driven attributes may contribute to the follower relationships, a holistic understanding of its role as a source development of parasocial relationships that are typically seen as similar of influence attempts is lacking (Martínez-López, Anaya-Sánchez, to interpersonal social relationships. Fernández Giordano, & Lopez-Lopez, 2020). Authors of previous studies In the present study, attractiveness refers to the degree to which one in this field concur that content posted by influencers can trigger perceives a social media influencer’s social media content as visually or changes in consumers’ perceptions and behavior towards influencers as aesthetically appealing (Patzer, 1983). The notion of attractiveness well as their endorsing brands (Casaló, Flavián, & Ibáñez-Sánchez, 2020; stereotypes is well-established in the social psychology literature, sug Ki & Kim, 2019). Hence, we seek to identify the effect of social media gesting that exposure to or interaction with attractive targets tends to influencers’ content characteristics as influence attempts in fostering elicit positive emotions and a desire to establish a close relationship parasocial relationships in order to offer a more manageable strategy for (Lemay Jr, Clark, & Greenberg, 2010). Likewise, previous studies 148
E.C.-X. Aw and S.H.-W. Chuah Journal of Business Research 132 (2021) 146–157 grounded in the source attractiveness model demonstrated a link be media influencer is believed to endorse a brand primarily for personal tween source attractiveness and parasocial relationships, suggesting that benefits, such as financial compensation, enhanced image, and fame, when influencers are perceived as physically or socially attractive, these would be deemed self-serving motives (Shan et al., 2019). A consumers tend to devote more attention to and engage more intensively common source of information that consumers draw upon in making with the influencers, thereby nurturing parasocial relationships (Gong & such attributions is the relationship they have established with influ Li, 2017; Lee & Watkins, 2016; Sokolova & Kefi, 2020). Therefore, as encer endorsers (Um, 2013). Authors of extant studies have demon social media content posted by influencers represents an indispensable strated that consumers that have developed a more intense bridge that connects influencers and consumers, it cultivates influencer- identification or parasocial relationship with the influencer endorser fan relationships. Drawing on the social identity theory (Brewer, 1991), tend to arrive at correspondent inferences that an influencer endorser we posit that consumers are likely to identify with influencers who truly likes the endorsed product, rather than promoting it for personal generate attractive content to satisfy their self-definitional needs, as gains, and are thus more likely to adopt favorable attitude toward that indicated in the following hypothesis: product (Farivar, Wang, & Yuan, 2020; Hwang & Zhang, 2018). This biased evaluation is further supported by the amalgamation of para H1: Attractiveness is positively related to parasocial relationships. social relationship theory and the entertainment overcoming resistance model (Moyer-Gusé, 2008) which put forth the notion that parasocial In line with the arguments put forth by Ki and Kim (2019), the relationships with a media figure can mitigate reactance and counter definition of prestige adopted in the present study refers to the extent a arguments towards persuasive attempts, and ultimately trigger greater social media influencer’s social media content manifests the character compliance (Tukachinsky, Walter, & Saucier, 2020). In sum, owing to istics of high status. Moreover, according to the extant brand literature, the parasocial relationships they form with social media influencers, consumers are motivated to identify with prestigious brands as a means consumers are unlikely to perceive their motives as self-serving. to achieve favorable psychological states, such as self-enhancement (Lam, Ahearne, Mullins, Hayati, & Schillewaert, 2013). In a similar H4: Parasocial relationships are negatively related to perceived self- vein, sense of prestige conveyed through content posted by social media serving motives. influencers leads to a more favorable evaluation by consumers (Ki & Kim, 2019). As postulated by the social identity theory, when defining 2.6. Perceived endorser motives and purchase intention their self-identity, people are driven by a desire for self-enhancement (Kunda, 1999). They would therefore identify with influencers who Influencer marketing strategies can be understood as persuasion at present a prestigious identity and image to immerse themselves in their tempts by persons with a considerable clout at changing and molding reflected glory as a way of enhancing their self-esteem (Hu, Min, Han, & consumers’ attitudes and behaviors towards a product or a brand. When Liu, 2020). Notably, pursuing relationships with prestigious others can exposed to endorsing advertisements, consumers tend to question the improve a person’ perception of own social status, due to which prestige motives behind the influencer’s decision to promote a product or a is a significant cue for interpersonal attraction (Lo, 2008). In line with brand; hence, the effectiveness of the endorsement message or adver these arguments, we posit that prestige as an influence attempt can tisement depends on whether consumers deem these motives to be self- foster parasocial relationships between social media influencers and serving or altruistic (Shan et al., 2019). In other words, social media followers, as indicated below: influencers that are perceived as self-serving endorsers would induce distrust in the advertising message, which would ultimately trigger H2: Prestige is positively related to parasocial relationships. negative response towards the endorsement deal (e.g., reduced credi bility, persuasiveness, and purchase intention) (Lee et al., 2013; Rein Expertise refers to social media influencers’ content attributes that hard et al., 2006). imply greater knowledge and/or experience in certain aspects than in others (McQuarrie, Miller, & Phillips, 2013). Expertise is recognized as a H5: Motives that are perceived as self-serving are negatively related vital element in the domains of interpersonal interaction and attraction to consumers’ purchase intention. (Xiang, Zheng, Lee, & Zhao, 2016). For instance, it has been shown in the social commerce context that users tend to interact and form para 2.7. Moderating role of Self-discrepancy social relationships with other users who are believed to possess considerable knowledge of brands and products. Similarly, extant ce The self-discrepancy theory posits that people are driven to achieve lebrity endorsement literature indicates that celebrities that are the state in which their self-concept aligns with personally relevant self- perceived as experts in certain domains are able to induce positive at guides (Higgins, 1989). The theory is based on the premise that people titudes in and form parasocial relationships with consumers (Sakib et al., have multiple selves (e.g., actual self and ideal self), whereby any dis 2020; Yuan, Kim, & Kim, 2016). Accordingly, we posit that social media crepancies among these selves would result in discomfort. According to influencers attempts to convey expertise can shape their perceived Noble, Haytko, and Phillips (2009), to diminish such discrepancy and importance for consumers’ subjective social experience, which would alleviate discomfort, individuals engage in certain behaviors such as entice more intense emulation as well as interaction, and would even consumption. In the influencer context, Jin et al. (2019) contended that tually result in favorable relational outcomes, such as the development people who are at higher positions and/or possess desired qualities, such of parasocial relationships. as celebrities or social media influencers, induce a sense of self- discrepancy in others. The notion of “basking in the reflected glory” H3: Expertise is positively related to parasocial relationships. suggests that consumers appropriate the admired attributes of a rela tionship partner that signal their ideal self (Cialdini et al., 1976). Social 2.5. Parasocial relationships and perceived endorser motives media influencers’ star power derives from their digital content as a key medium for their identity creation and dissemination, as consumers tend The attribution framework offers a theoretical perspective for to appropriate meaning encapsulated in the content (Escalas & Bettman, examining the cognitive processes involved in the formation of causal 2017). For instance, Ki and Kim (2019) found that influence attempts (i. inferences for human behavior (Kapitan & Silvera, 2016; Kelley, 1967). e., content) by social media influencers increase consumers’ desire to In the influencer endorsement context, consumers draw causal in mimic their attitudes and behaviors. When consumers are exposed to ferences from endorsers’ behaviors to determine whether their motives influence attempts by social media influencers, they are inevitably are altruistic or self-serving (Shan et al., 2019). Consequently, if a social involved in social comparisons. 149
E.C.-X. Aw and S.H.-W. Chuah Journal of Business Research 132 (2021) 146–157 Fig. 1. Research model. However, we argue that how meanings and cues from the influence media platforms (e.g., Facebook and Instagram), which deemed attempts are appropriated is unlikely to be uniform across followers, as thematically best suited to the study context. A screening question was it is subjected to followers’ self-concept appraisal. According to Malär presented on the first page asking respondents to indicate one of their et al. (2018), whether an individual assimilates or disapproves of the favorite social media influencers, who they have followed on social attributes conveyed by the object of upward social comparison (i.e., media. Those who failed the screening criteria were discontinued from social media influencer) is determined by self-discrepancy. The participating, and debriefed. We also explicated a brief definition of information-processing literature holds that information is more likely social media influencers to retrieve more accurate responses from the to be assimilated by consumers and is more persuasive when it fits respondents. The social media influencer’s name indicated was consumers’ self-discrepancy-related goal, which is to narrow the gap embedded in the remainder of the questionnaire. After removing between their actual and ideal self (Dijkstra, Schakenraad, Menninga, straight-lining responses and those who did not address the defined Buunk, & Siero, 2009). People with a high degree of self-discrepancy social media influencers, 361 usable responses were collected. tend to emulate their favorite celebrities as a way of shifting their Majority of the respondents were female (65.1%) and students actual self toward their ideal self, and are thus more empathetic toward (68.1%). Most of them (77.3%) were in the age categories between 18 their favorite celebrities (Derrick, Gabriel, & Tippin, 2008). Studies have and 25 years old. In addition, the majority of respondents were Cauca shown that communication from media figures exerts stronger positive sian (54.3%), followed by Asian (24.8%), Hispanics (6.9%) and others. effect on interpersonally affiliative individuals, as manifested in high Half of the respondents (51.0%) indicated Instagram as their most used degree of self-discrepancy (Lee & Jang, 2013; Lee & Oh, 2012). As a social media platform, followed by YouTube (36.6%), and Facebook result, we posit that consumers with high self-discrepancy are more (6.1%). Examples of frequently mentioned social media influencers by likely to assimilate and resonate with the influence attempts (i.e., the respondents were PewDiePie, Chiara Ferragni, Jenna Marbles, and attractiveness, prestige, and expertise), and form affirming parasocial Shane Dawson, among others. relationships with social media influencers in order to close the self- discrepancy gap. Conversely, the impacts of influence attempts on par 3.2. Measures asocial relationships may be relatively weak for consumers with low self-discrepancy as they are less susceptible to external influence (Pen We adapted well-established and validated measures from the liter tina, Taylor, & Voelker, 2009). Hence, we hypothesize the following: ature (see Table A1). The scales for influence attempts (i.e., attractive ness, prestige, and expertise) and purchase intention were adapted from H6: Self-discrepancy moderates the link between attractiveness and Ki and Kim (2019). The scales for parasocial relationships were adapted parasocial relationships, which is stronger for those exhibiting a high from Chung and Cho (2017). Rifon, Choi, Trimble, and Li (2004) scales degree of self-discrepancy. were adapted to assess perceived self-serving motive. The actual-self and H7: Self-discrepancy moderates the link between prestige and par ideal-self concepts were measured using Sirgy et al. (1997) scales. Self- asocial relationships, which is stronger for those exhibiting a high discrepancy score was calculated by subtracting the actual-self score degree of self-discrepancy. from the ideal-self score. All items were measured with seven-point H8: Self-discrepancy moderates the link between expertise and par Likert scales, anchored by “1 = strongly disagree” and “7 = strongly asocial relationships, which is stronger for those exhibiting a high agree”. Two researchers independently reviewed the measures to ensure degree of self-discrepancy. absence of ambiguity in comprehending the questionnaire items. Sub sequently, the questionnaire was pilot-tested with forty respondents. The proposed research model is illustrated in Fig. 1. Cronbach’s α for all variables ranged from 0.74 to 0.95, indicating satisfactory internal consistency (Hair, Black, Babin, Anderson, & 3. Methodology Tatham, 2006). 3.1. Data collection and sample characteristics 4. Data analysis Given that the sampling frame of social media users who followed 4.1. Common method bias social media influencers is inaccessible, we adopted the purposive sampling technique which deemed appropriate in condition where only In order to minimize the threat of common method bias arises in a limited population of people possesses the required information cross-sectional design study, procedural and statistical remedies were (Sekaran & Bougie, 2016). The questionnaire was administered online undertaken (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). In terms of through personal contacts and sharing survey link on different social procedural remedy, respondents were informed there was no right or 150
E.C.-X. Aw and S.H.-W. Chuah Journal of Business Research 132 (2021) 146–157 Table 1 Construct reliability and validity, discriminant validity. Construct CR AVE Attractiveness Purchase intention Parasocial relationships Prestige Expertise Self-serving motive Attractiveness 0.950 0.862 0.929 Purchase intention 0.937 0.831 0.458 0.912 Parasocial relationships 0.922 0.568 0.397 0.484 0.754 Prestige 0.920 0.792 0.628 0.456 0.372 0.890 Expertise 0.935 0.828 0.515 0.367 0.367 0.596 0.910 Self-serving motive 0.788 0.562 − 0.437 − 0.626 − 0.630 − 0.434 − 0.472 0.749 Notes: Diagonal elements present the square root of the AVE (in bold). wrong answers, and their honest answers matter the most. Also, they Table 2 were assured of anonymity (Podsakoff et al., 2003). With regards to Hypothesis testing for direct effects and moderation. statistical remedy, we performed the Harman’s single-factor test. The result showed that the first factor explained 35% of the total variance, Hypothesis Relationship Std. Beta t-value Decision below the recommended 50% threshold. We further employed the full H1 Attractiveness → Parasocial 0.228** 3.534 Supported collinearity test, and found that the VIF values ranged from 1.067 to relationship H2 Prestige → Parasocial 0.124* 1.965 Supported 2.193, which were below the threshold level of 3.3, thus confirming that relationship common method bias is not a significant threat in the present study H3 Expertise → Parasocial 0.176** 2.611 Supported (Kock, 2015). relationship H4 Parasocial relationship → − 0.630** 17.559 Supported Partial Least Squares Structural Equation Modelling (PLS-SEM) Self-serving motive H5 Self-serving motive → − 0.626** 17.654 Supported Purchase intention PLS-SEM was employed to test the proposed research model for H6 Self- − 0.098* 1.712 Unsupported several reasons. First, PLS-SEM is appropriate when the aim of the study discrepancy*Attractiveness is to predict a key endogenous construct (i.e. purchase intention) (Hair, → Parasocial relationship H7 Self-discrepancy*Prestige − 0.111* 1.619 Unsupported Risher, Sarstedt, & Ringle, 2019). Second, PLS-SEM is superior in → Parasocial relationship handling complex structural models as characterized by the present H8 Self-discrepancy*Expertise 0.052n.s. 0.786 Unsupported study, of which the research model involves many latent constructs, → Parasocial relationship indicators, and complex relationships (e.g., moderation) (Hair et al., Note: *p < 0.05, ** p < 0.01, n.s. = not significant. 2019). Third, PLS-SEM is less restrictive with sample size, measurement scales, and data distribution assumptions. negatively influences purchase intention (β = − 0.626, p < .05). Hence, 4.2. Measurement model H1 to H5 were supported (See Table 2). With respect to the moderation, we found significant moderating Four criteria were assessed in the evaluation of measurement model. effects of self-discrepancy in the links between attractiveness (β = First, as shown in Table 1, all constructs exhibited composite reliability − 0.098, p < .05), prestige (β = − 0.111, p < .05), and parasocial re greater than 0.7, indicating strong internal consistency (Hair, Hult, lationships. Subsequently, we created an interaction plot to better Ringle, & Sarstedt, 2017). Second, the convergent validity was exam interpret the nature of moderation detected. As shown in Figs. 2 and 3, ined in terms of factor loadings and average variance extracted (AVE). the links between influence attempts (i.e., attractiveness and prestige) As exhibited in Table 1, all items showed factor loadings above 0.7 and parasocial relationships are stronger when the self-discrepancy is except three items from perceived self-serving motive, which were low. Although these effects were significant, they contradict to our removed. AVE values of all constructs were above the 0.5 threshold, earlier hypotheses, thus H6 to H8 were unsupported. suggesting that all items accounted for most of the variance of their respective constructs (Hair et al., 2017). Discriminant validity was 4.4. Post-hoc analysis: Exploring the role of the number of followers examined by comparing inter-construct correlations with the square roots of AVE for each construct. As shown in Table 1, square root of AVE Signaling theory contends that consumers seek out cues/signals from for each construct exceeded the respective inter- construct correlations, a brand to mitigate the information asymmetry (Spence, 1973), which is indicating the establishment of discriminant validity (Fornell & Larcker, particularly pronounced in the social media environment where infor 1981). mation abounds. In this respect, consumers rely on easily interpretable metrics such as number of views and likes to draw inferences regarding 4.3. Structural model the quality of online content, product/services, and their providers. As the number of followers is one of the distinctive metrics that charac Before examining the structural relationships, we examined multi terize a social media account (Hill, Troshani, & Chandrasekar, 2017), it collinearity issue which potentially bias the regression results. The result signifies social media influencers’ popularity and potential market reach indicated that the highest VIF value was 1.981, well below the recom (De Veirman, Cauberghe, & Hudders, 2017; Weismueller et al., 2020). mended 3.0 threshold (Hair et al., 2019). Our analysis of the structural Available evidence suggests that social media influencers with a high model found R2 of 0.448, suggesting 44.8% of variance in purchase number of followers are generally seen as more socially desirable in intention was explained by the model. In addition, the model exhibited terms of source attractiveness and credibility (De Veirman et al., 2017; satisfactory predictive relevance, with Q2 value greater than zero (0.32) Jin & Phua, 2014), and these perceived characteristics are transferable (Fornell & Cha, 1994). to the products they endorse or review. Correspondingly, Weismueller The study supports the direct effect of influence attempts, namely et al. (2020) recently articulated that social media influencers with a attractiveness (β = 0.228, p < .05), prestige (β = 0.124, p < .05), and high number of followers can generate effective marketing implications, expertise (β = 0.176, p < .05) on parasocial relationships. Parasocial such as evoking consumers’ purchase intention. relationships has a significant negative effect on perceived self-serving Based on the above discussion, in the present study, we test the motive (β = − 0.630, p < .05). Lastly, perceived self-serving motive impact of number of followers on parasocial relationships, perception of 151
E.C.-X. Aw and S.H.-W. Chuah Journal of Business Research 132 (2021) 146–157 Fig. 2. Moderating effect of self-discrepancy in the relationship between attractiveness and parasocial relationships. Fig. 3. Moderating effect of self-discrepancy in the relationship between prestige and parasocial relationships. influencer motives as self-serving, and consumers’ purchase intentions. significant negative effect on parasocial relationships (β = − 0.097, p < To this end, the number of followers was extracted from the Instagram .05) but not the perception of influencer motives as self-serving (β = accounts of social media influencers mentioned by our respondents. 0.004, p > .05) or consumers’ purchase intention (β = 0.005, p > .05). Following the approach taken by Benitez, Henseler, Castillo, and Schuberth (2020), the number of followers was modeled as a single- indicator construct after applying natural logarithm due to the skewed distribution. The results showed that number of followers has a Table 3 Results of partial invariance measurement testing. Constructs Configural invariance Compositional invariance (Correlation = 1) Equal mean Equal variances FMI C=1 5% quantile Dif CI Dif. CI ATTR Yes 1.000 0.999 0.479 [− 0.230; 0.253] − 0.481 [− 0.435; 0.420] No PRES Yes 0.997 0.996 0.249 [− 0.236; 0.231] − 0.327 [− 0.326; 0.334] No EXP Yes 1.000 0.997 0.110 [− 0.233; 0.253] − 0.387 [− 0.431; 0.393] Yes PR Yes 0.997 0.997 − 0.066 [− 0.266; 0.245] − 0.120 [− 0.348, 0.338] Yes SSM Yes 0.998 0.988 − 0.286 [− 0.256; 0.233] − 0.540 [− 0.384; 0.388] No INT Yes 1.000 0.999 0.421 [− 0.229; 0.237] − 0.542 [− 0.371; 0.342] No Notes: CI = Confidence Interval; Dif = Differences; FMI = Full measurement invariance, ATTR = Attractiveness, PRES = Prestige, EXP = Expertise, PR = Parasocial relationships, SSM = Self-serving motive, INT = Purchase intention. 152
E.C.-X. Aw and S.H.-W. Chuah Journal of Business Research 132 (2021) 146–157 Table 4 Result of PLS-MGA for gender. Hypothesis Relationships Path coefficient (Male) Path coefficient (Female) CIs (Bias corrected) CIs Path coefficient differences Male (Bias corrected) Female H1 ATTR → PR 0.285 0.291 0.140; 0.432 0.061; 0.530 − 0.006 H2 PRES → PR 0.320 0.009 0.167; 0.459 − 0.234; 0.185 0.311* H3 EXP → PR 0.126 0.100 − 0.048; 0.291 − 0.117; 0.318 0.027 H4 PR → SSM − 0.731 − 0.648 − 0.806; − 0.632 − 0.716; − 0.544 − 0.083 H5 SSM → INT − 0.720 − 0.494 − 0.787; − 0.631 − 0.597; − 0.357 − 0.226** Post-hoc NOF → PR − 0.153 − 0.117 − 0.296; − 0.009 − 0.282; 0.081 − 0.036 Post-hoc NOF → SSM − 0.061 − 0.030 − 0.179; 0.049 − 0.155; 0.105 − 0.031 Post-hoc NOF → INT − 0.001 0.059 − 0.122; 0.125 − 0.094; 0.227 − 0.059 Notes: p < 0.01**, p < 0.05*, n.s. = not significant, ATTR = Attractiveness, PRES = Prestige, EXP = Expertise, PR = Parasocial relationships, SSM = Self-serving motive, INT = Purchase intention, NOF = Number of followers. 4.5. Post-hoc analysis: Exploring the moderating role of gender through attempts on several important yet overlooked downstream outcomes (i. multigroup analysis e., parasocial relationships, perceived self-serving motives, and pur chase intention). Available evidence indicates that consumption behaviors and eval In the present study, we extended the applicability of influence uative judgements are gender-dependent (Holbrook, 1986). Hence, theory (Scheer & Stern, 1992) to digital marketing context and explored Bergvist and Zhou (2016) argued that gender difference should be the dynamic interactions between social media influencers and con explored in the influencer context. For instance, Bush, Martin, and Bush sumers. We revealed that social media influencers’ content as a form of (2004) revealed that celebrity role models exert greater influence on influence attempt is capable of fostering the illusionary intimacy be female compared to male teenagers. In addition, several studies have tween influencers and consumers. Our findings further revealed that the identified that men and women use media differently in addressing their greater the extent to which a consumer perceives a social media influ emotional needs, and women tend to develop stronger parasocial re encer’s social media content as visually attractive, displaying sense of lationships with media figures than do men (Lather & Moyer-Guse, prestige, and exhibiting expertise, the more likely it is that a parasocial 2011; Wang, Fink, & Cai, 2008). However, men exhibit a greater level relationship will be established. These findings expand upon the results of celebrity worship (McCutcheon, Lange, & Houran, 2002). Thus, it is yielded by prior studies indicating that intensity of social media in reasonable to assume that gender may affect the extent to which con teractions is a driver of parasocial relationships with celebrities (Aw & sumers develop and rely upon parasocial relationships, as well as how Labrecque, 2020; Chung & Cho, 2017) by identifying key social media they perceive endorser motives. In line with these arguments, we content qualities in developing parasocial relationships. explored the potential observed heterogeneity by comparing the results Moreover, our post-hoc analysis indicates that number of followers yielded by the proposed model when applied to male and female par exerts negative influence on parasocial relationships, thus countering ticipants separately. the prevalent view that popularity is commensurate with likability (De First, we followed the approach adopted by Matthews (2018) and Veirman et al., 2017; Jin & Phua, 2014). We contend that parasocial randomly withdrew respondents from the larger subgroup (females) to relationships followers form with social media influencers differ from ensure comparable group size. Second, adhering to Henseler, Ringle, those involving traditional celebrities. As social media influencers are and Sarstedt (2016) suggestion, we assessed the measurement invari “grass-roots” celebrities, they engender a sense of closeness and relat ance for gender using the MICOM procedure. The configural invariance edness which tends to dissipate as influencers amass a significant was established using the same setup (identical indicators, data treat number of followers (Tafesse & Wood, 2021). This argument is in line ment, and algorithm settings) in model estimation for both male and with Djafarova and Rushwort’s (2017), finding that consumers perceive female sub-groups. As shown in Table 3, compositional invariance was lower-end influencers with modest number of followers as compara established but the results indicated significant differences in means and tively more trustful and personally relevant. The null effects of number variances for a few constructs, suggesting partial measurement invari of followers we found in our study on the perception of endorser motives ance. Therefore, the execution of PLS-MGA for structural path compar as self-serving and consumer purchase intentions concur with the ar ison across groups was allowed (Henseler et al., 2016). As shown in guments put forth by Yuan, Moon, Kim, and Wang (2019), implying that Table 4, the PLS-MGA results revealed significant group differences in indication of popularity alone does not offer sufficient product/brand the effects of prestige on parasocial relationships, which were stronger information to aid purchase decisions. for males. Similarly, the effect of perceived self-serving motives on Another important and novel finding that has emerged from our purchase intention was stronger for male participants study is the moderating effect of self-discrepancy in the relationship between influence attempts (i.e., attractiveness and prestige) and par 5. Discussion asocial relationships. We revealed that influence attempts have stronger impacts on parasocial relationships when consumers exhibit low self- The rise of social media influencers is changing the landscape of discrepancy, thus countering our initial hypothesis. This finding sug influencer endorsement. As they exert powerful clout on their followers, gests that low self-discrepancy drives consumers to appropriate and brand marketers are increasingly relying on social media influencers as assimilate the attractiveness and prestige cues exhibited in social media endorsers of their products and services. Despite being heralded as one influencers’ content to a greater extent, whereby the unexpected di of the most widely adopted promotional strategies, social media influ rection of moderating effects could be explained by the assimilation and encer marketing has been given scant attention in academic literature, contrast theory (Mussweiler & Strack, 2000). The theory argues that the where focus is mainly on traditional celebrity endorsements (Ki & Kim, assimilation effect is more likely to occur when consumers perceive their 2019). This gap presents a critical issue for marketers, as lacking solid ideal self as attainable (Malär et al., 2018). According to Malär et al. understanding of how social media influencer marketing is managed can (2018), consumers who exhibit a high degree of self-discrepancy feel far impede the endorsement effectiveness. Towards this end, we developed removed from their ideal self, especially that related to agentic ideals a model to examine the impact of social media influencers’ influence such as beauty and status, which they feel would be impossible to 153
E.C.-X. Aw and S.H.-W. Chuah Journal of Business Research 132 (2021) 146–157 achieve (Sheldon & Cooper, 2008). Hence, they are less likely to limited to the role of influencer traits, including credibility and attrac assimilate attractiveness and prestige cues in developing relationships tiveness (Sokolova & Kefi, 2020; Yuan & Lou, 2020), thus typically with social media influencers. overlooking the importance of content-driven traits (Casaló et al., 2020; The assumption that influencer endorsement can induce consumers’ Ki et al., 2020). This study has extended the theoretical body of belief that the influencer shares traits with the products or likes the knowledge by unveiling the influence attempts (i.e., attractiveness, products is overly simplistic. This is because consumers are aware of prestige, and expertise) as tantamount in explaining parasocial re persuasion tactics used in marketing, and are likely to question the lationships, thereby offering an alternative view for influencer-fan motives of such communication efforts (Arsena, Silvera, & Pandelaere, relationship management. Moreover, by showcasing that negative and 2014; Forehand & Grier, 2003). Specifically, consumers are mindful of null impacts number of followers exerts on the outcome variables (i.e., the financial compensation influencers receive for their endorsements parasocial relationships, self-serving motive, and purchase intention), and are skeptical of their intentions. We contribute to the influencer the present study offers an alternative view to that purported in the endorsement literature by showing that forming parasocial relationships current literature which seems to overstate the prominence of popularity could reduce the likelihood that influencers’ motives will be perceived as an influencer marketing consideration. as self-serving, thus increasing the endorsement effectiveness. Our Third, the present study unpacks the moderating mechanisms un findings evidenced that, just as in real friendships, consumers are likely derpinning the meaning assimilation and appropriation by consumers to align their attitudes with those of the social media influencers with from external information sources. The significant moderating effect of whom they have parasocial relationships. The information processing self-discrepancy that emerged from our analyses suggests that not all theory contends that emotional connections can effectively augment the influence attempts by social media influencers are equally perceived and processing of advertisement and brand messages (MacInnis & Jaworski, harnessed in cultivating parasocial relationships, thereby offering a 1989). Parasocial relationships that are built upon cumulative infor more nuanced perspective than that typically conveyed by influencer mation disclosure and in-depth understanding fortify the perceived literature based on the assumption of direct linear effects of self-concept authenticity while discounting the potential negative information per appraisal on endorsement outcomes. Lastly, the findings yield by our taining to the social media influencer, thereby creating a positive bias in post-hoc analysis indicate that the impacts of influence attempts as well the evaluation of endorser motives, which eventually leads to greater as perceived self-serving motives are gender-dependent, challenging the purchase intention. validity of the homogeneous lens applied in prior social media influ Our exploratory analysis yielded two interesting findings regarding encer research. The examination of self-concept appraisal (i.e., self- the moderating effect of gender, indicating that the effect of prestige on discrepancy) and gender corroborates Hudders, De Jans, and De Veir parasocial relationships as well as perceived self-serving motive on man (2020) assertion that follower/fan characteristics play an impor purchase intention is stronger for males. This is to be expected, as gender tant role in social media influencer marketing, and require further influences social interactions, whereby men are more likely than women research. to pursue respect and social standing (Gefen & Straub, 1997). According to Cross and Madson (1997), they are also more prone to acting ac 5.2. Practical implications cording to patterns of behavior that elevate personal status. Hence, prestige quality as manifested in the social media influencers’ content From a managerial standpoint, our findings present several impor would be more relevant in fostering the parasocial relationships for men. tant implications for marketing practitioners. In the current market In addition, as women are in general more generous, kind, and socially- flooded by influencer endorsement, consumers are more aware of the oriented (Eagly & Crowley, 1986; Kashdan, Mishra, Breen, & Froh, persuasion tactics employed by brands and are more skeptical of the 2009), they are more likely to tolerate social media influencers’ self- motives behind influencers’ endorsements. Hence, brand marketers and serving motives, especially if they perceive them as close friends. endorsers are striving to reduce attribution of self-serving motive. The Moreover, women tend to ponder on their decisions longer than men do present study highlights the importance of quality of influencer–fan (Meyers-Levy & Maheswaran, 1991). As a result, their attitudes toward relationships in endorser selection, given that the general social media an endorsed product or service are less affected by their perception of influencer metrics, such as total number of fans and followers, may not influencers’ motives, as they would consider other aspects, such as the directly translate into audiences’ attitudes (Ki & Kim, 2019). Conse actual product attributes and their relationships with the influencers. quently, instead of selecting social media influencers based on their Hence, the path between self-serving motive and purchase intention is physical appearance, brand marketers should choose those who have or weaker for women. are able to generate high parasocial relationships with followers through relationship-oriented metrics (e.g., sentiment) (Arora, Bansal, Kandpal, 5.1. Theoretical implications Aswani, & Dwivedi, 2019), to induce positive responses towards the endorsement. The theoretical contributions of this study stem from the exploration We further advocate that influencer marketing should focus on of heretofore-untested mechanism that is salient to perceived endorser activating parasocial relationships. To this end, social media influencers motives and the subsequent endorsement outcome (i.e., purchase need to engage in a more interactive manner (e.g., respond to com intention). Even though authors of prior studies have raised concern ments) with their followers through social media (Aw & Labrecque, over the imperilment of social media influencers’ endorsing motive 2020; Chung & Cho, 2017). More importantly, we suggest that brand when they partner with brands (Audrezet et al., 2020; Shan et al., 2019), marketers and social media influencers need to be mindful in crafting little research has been conducted to understand how social media their content, as it is their main bridge of communication with fans. influencers can manage consumers’ inference of motive. Our study Particularly, they could leverage on social media content analysis to contributes to this research strand by shedding light on the captivating identify and learn from the posts that best convey the qualities of role of parasocial relationships as social media influencers’ tool for attractiveness, prestige, and expertise in order to strengthen parasocial counteracting the consumers’ perception of their endorsement motives relationships. Due to the greater importance of attractiveness attributes as self-serving. This finding adds to the existing social media influencer relative to prestige and expertise, understanding how to incorporate this literature that primarily focuses on the impact of parasocial relation element into social media content is paramount. For example, hypo ships on brand-related outcomes (e.g., purchase intention and product thetically, attractive content may embody features such as forms (e.g., interest) (Farivar et al., 2020; Sokolova & Kefi, 2020; Yuan & Lou, vlog), facial expressions (e.g., smile), gaze direction (gaze at the cam 2020). Second, current social media influencer research as a part of era), camera distance (full-body), and creative language. It is also worth which antecedents of parasocial relationships are explored has been noting that number of followers cannot be a sole metric for successful 154
You can also read