Master Thesis in Marketing
Does Sponsorship Disclosure Impact Consumer Responses on Instagram?
An Investigation into Type of Influencer, Brand Familiarity, and Persuasion Knowledge
Author: Karina Rukmini Binol Student number: 5727227
Date of submission: 17-08-2018
Programme: MSc. in Business Administration – Marketing Track Institute: University of Amsterdam
STATEMENT OF ORIGINALITY
This document is written by Karina Rukmini Binol who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
TABLE OF CONTENTS
ABSTRACT ... 1
LIST OF FIGURES AND TABLES... 2
1. INTRODUCTION ... 3 2. LITERATURE REVIEW ... 6 2.1 Instagram ... 6 2.2 Influencer Marketing ... 7 2.3 Type of Influencer ... 9 2.3.1 Celebrities ... 9 2.3.2 Social Influencers ... 11 2.4 Sponsorship Disclosure ... 13
2.4.1 Sponsorship Disclosure on Instagram ... 15
2.5 Persuasion Knowledge ... 16
2.6 Brand Familiarity ... 18
2.7 Consumer Responses ... 19
2.7.1 Brand Evaluations ... 20
2.7.2 eWOM Intention... 21
3. HYPOTHESES & CONCEPTUAL FRAMEWORK ... 24
3.1 Hypotheses ... 24 3.2 Conceptual Framework ... 29 4. METHODOLOGY ... 30 4.1 Research Design ... 30 4.2 Participants ... 30 4.3 Procedure ... 31 4.4 Stimulus Materials ... 32 4.5 Measures ... 33 5. RESULTS ... 35 5.1 Randomization Check ... 35 5.2 Manipulation Check ... 35 5.3 Testing of Hypothesis ... 36 5.3.1 Hypothesis 1 ... 36 5.3.2 Hypothesis 2 ... 37 5.3.3 Hypothesis 3 ... 38 5.3.4 Hypothesis 4 ... 40 5.3.5 Hypothesis 5 ... 42 5.3.6 Hypothesis 6 ... 47 6. DISCUSSION ... 52
7. LIMITATIONS & FUTURE RESEARCH ... 56
8. CONCLUSION ... 58
9. REFERENCES ... 59
ABSTRACT
With the emergence of influencer marketing on social media, the issue of influencers disclosing sponsorship arrangements on social media is gaining importance. Regulatory bodies such as the Federal Trade Commission have set rules and regulations to ensure appropriate sponsorship disclosure practices on social media. Nonetheless, not all influencers adhere to the specific guidelines and without proper disclosure, consumers would fail to notice that a sponsorship arrangement exists and thus end up being misled and deceived. This study is interested in examining the effects of sponsorship disclosure (compared to no disclosure) on persuasion knowledge and consumer responses on Instagram, whilst also looking into possible moderating effects by type of influencer (social influencer vs. celebrity) and brand familiarity (high vs. low). Results indicate that a celebrity Instagram post is more likely to be recognized as advertising compared to a social influencer’s post. Persuasion knowledge was found to mediate the effects of sponsorship disclosure on eWOM intention, but not on brand evaluation. However, brand familiarity, surprisingly did not mitigate the negative effects of persuasion knowledge on eWOM intention as predicted.
LIST OF FIGURES AND TABLES
Figure 1 – Instagram ‘like’ and ‘comment’, ‘like comment’, and ‘reply comment’ functions... 7
Figure 2 – Sponsorship disclosure on Instagram using “paid partnership” feature ... 16
Figure 3 – Instagram ‘tag’ other user function as a form eWOM. ... 22
Figure 4 – Conceptual Framework ... 29
Figure 5 – Effect of sponsorship disclosure (no disclosure vs. disclosure) on persuasion knowledge with respect to the type of influencer (Celebrity vs. Social Influencer) .. 39
Figure 6 – Interaction Effect between Persuasion Knowledge and Brand Familiarity on Brand Evaluations ... 43
Figure 7 – Interaction Effect between Persuasion Knowledge and Brand Familiarity on eWOM intention ... 45
Figure 8 – The mediation effect of persuasion knowledge on sponsorship disclosure to eWOM intention. ... 51
Table 1 – Experimental Conditions ... 30
Table 2 – Effect of Sponsorship Disclosure on Persuasion Knowledge – Regression Analysis .. 36
Table 3 – Effect of Type of Influencer on Persuasion Knowledge – Regression Analysis Results ... 37
Table 4 – Interaction Effect between Sponsorship Disclosure and Type of Influencer on Persuasion Knowledge – Moderation Analysis Results using PROCESS macro ... 39
Table 5 – Effect of Persuasion Knowledge on Brand Evaluation – Regression Analysis Results ... 40
Table 6 – Effect of Persuasion Knowledge on eWOM Intention – Regression Analysis Results 41 Table 7 – Interaction Effect between Persuasion Knowledge and Brand Familiarity on Brand Evaluations – Moderation Analysis Results using PROCESS macro ... 43
Table 8 – Interaction Effect between Persuasion Knowledge and Brand Familiarity on eWOM intention – Moderation Analysis Results using PROCESS macro ... 44
Table 9 – Conditional effects of persuasion knowledge (X) on eWOM intention (Y) at levels of brand familiarity (M) ... 45
Table 10 – Effect of Sponsorship Disclosure on brand evaluation – Regression Analysis Results ... 47
Table 11 – Effect of Sponsorship Disclosure on eWOM intention – Regression Analysis Results ... 48
Table 12 – Mediation Effect of Persuasion Knowledge on Sponsorship Disclosure and eWOM intention – Moderation Analysis Results using PROCESS Macro: Effects on Dependent Variables ... 50
Table 13 – Mediation Effect of Persuasion Knowledge on Sponsorship Disclosure and eWOM intention – Moderation Analysis Results using PROCESS macro: Total, Direct, and Indirect Effects ... 50
1. INTRODUCTION
Nowadays, consumers use multiple smart devices and they can conduct various online activities at the same time: watching a documentary on Netflix, searching for online reviews of a recently released product, and switch seamlessly from one social media application (app) such as Instagram to another app, Facebook. Social media usage is growing significantly and as of June 2018, there are 2.2 billion active monthly users on Facebook and 1 billion active monthly users on Instagram (Facebook, 2018). In the United States (U.S.) alone, “68% of all U.S. adults are Facebook users, while 28% use Instagram, 26% use Pinterest, 25% use LinkedIn and 21% use Twitter” (Pew Research Centre, 2016). With consumer’s increasing usage of social media, marketers are diverting their marketing spending to social media, in the form of influencer marketing campaigns. Influencer marketing entails a sponsorship arrangement between the influencer and the brand on social media. Beauty brands such as L’Oreal, Maybelline, M.A.C and Revlon are launching influencer marketing programs collaborating with influencers in order to engage with their customers and enhance product adoption (Rodulfo, 2017; Hall, 2017; Trefis, 2017; Weinswig, 2016).
Although some influencer marketing programs are announced publicly in the media, other sponsorship (endorsement) arrangements with influencers are often not explicitly revealed by the influencers and/or the marketers themselves (Martens, 2017; Frier & Townsend, 2016). Consumer protection agencies such as the Federal Trade Commission (FTC) of the United States are catching on to these influencer endorsements on social media and the FTC argues that because “consumers put stock in endorsements, [the FTC wants] to make sure [that consumers] are not being deceived” (Frier & Townsend, 2016). In March 2016, the FTC issued a complaint against luxury retailer Lord & Taylor for compensating various Fashion Influencers to generate posts on one of their clothing
items on Instagram without ever disclosing that the retailer had paid them and the dresses were given to the Influencers for free (Frier & Townsend, 2016; Martens, 2017). The FTC argues that any type of compensation, which also includes free products, should be readily disclosed to consumers and for many of its cases the FTC implements a basic test: “If a consumer knew an endorser was compensated in any way, would that alter the view of the endorsement?” (Frier & Townsend, 2016).
This research aims to investigate the effects of disclosing sponsorship on consumer responses in the context of Instagram. It would be interesting to find out whether compensation (sponsorship) of an influencer alters the consumers’ views of the endorsement. Much of the research on sponsorship disclosure have been explored in various contexts such as, television (Boerman, van Reijmersdal & Neijens, 2013; Smink, van Reijmersdal & Boerman, 2017) and radio (Wei, Fischer & Main, 2008). These studies found that sponsorship disclosure activates persuasion knowledge which can negatively impact consumer responses. The effects of activated persuasion knowledge on consumer responses have been explored on social media (Boerman, Willemsen & van der Aa, 2017; Hwang & Zhang, 2018). However, these studies have not examined the comparative effects of different types of influencers on persuasion knowledge and consumer responses. Consequently, this study intends to investigate the effects of the different types of influencer namely, celebrity and social influencer.
Although activated persuasion knowledge has been found to negatively impact consumer responses, Wei, Fischer & Main (2008) found that high brand familiarity could mitigate the aforementioned negative effects of persuasion knowledge in the context college radio program. Could differing levels of brand familiarity impact the relationship between persuasion knowledge
and consumer responses? This study aims to see whether a brand familiarity (high vs. low) would produce the mitigating effects in the context of Instagram.
The primary objective of this study is to investigate the effects of sponsorship disclosure on persuasion knowledge and consumer responses in the context of Instagram and beauty product. Additionally, moderating effects by variables such as type of influencer and brand familiarity will also be examined. Consequently, this study aims to answer the following research questions:
1) How do consumers react to sponsorship disclosure? What is the impact of sponsorship disclosure on consumer responses?
2) What is the role of persuasion knowledge in the relationship between sponsorship disclosure and consumer responses, in what manner does it affect the relationship?
3) How do the various types of influencer activate persuasion knowledge? How does this differ across the sponsorship disclosure conditions?
2. LITERATURE REVIEW 2.1 Instagram
Instagram was launched in 2010 by Kevin Systrom and Mike Krieger. In 2012, Facebook acquired Instagram for around US$ 1 billion in cash and stock (Constine & Cutler, 2012). At the time of acquisition, Instagram was widely known as a photo-sharing application and it had a strong community of photographers and photo lovers (Constine & Cutler, 2012). Over time, Instagram evolved and it has “become the home for visual storytelling for everyone from celebrities, newsrooms and brands, to teens, musicians and anyone with a creative passion.” (“Instagram About Us”, n.d). In September 2017, Instagram announced that they have 800 million users actively using Instagram at least on a monthly basis and 500 million users actively using the platform on a daily basis (Etherington, 2017). As of June 2018 however, there are one billion users actively using Instagram at least on a monthly basis (Constine, 2018). Instagram is gaining tremendous traction amongst the younger demographics, according to Pew Research Center (2018) Americans aged 18 – 24 years old are more likely to use Instagram than their counterparts in mid- to late-20s.
On Instagram, its users would upload photos centering on their daily activities whereby they would share photos of their pets, what they are eating for lunch or what they are wearing. Instagram allows users to manipulate their photos with a selection of filters. Instagram’s main page displays a stream of photos and videos of those you follow. However, Instagram applies the principle of asymmetric social network and thus, a user you follow does not necessarily have to reciprocate and follow you back (Hu, Manikonda & Kambhampati, 2014; De Veirman, Cauberghe & Hudders, 2017). As a consequence, your photos and videos will not be streamed on the main pages of those users who do not follow you back. On Instagram users following another user are
known as “followers”. An Instagram profile would display the number of followers it has (“followers”) and the number of Instagram user account it follows (“following”).
On Instagram, interactivity is built by allowing its users to tap to love and/or comment on a public Instagram post, as illustrated on Figure 1. Instagram has expanded on their interactivity function by enabling its users to reply to the comments and/or like the comment itself. If a user has commented on a post, another user may reply to that comment and/or like the comment. The user is then notified if another has replied to and/or liked their comment.
Figure 1 – Instagram ‘like’ and ‘comment’, ‘like comment’, and ‘reply comment’ functions
2.2 Influencer Marketing
Through the proliferation of smartphones and widespread Internet connectivity many consumers are able to access social media anywhere and anytime. In 2017, a survey of worldwide Internet
users indicated that the average daily time spent on social media is 135 minutes (Statista, 2017). Spending on influencer marketing advertising is expected to reach between $5 billion and $10 billion in 2022 (Gallagher, 2018), whilst spending on influencer marketing for Instagram was projected to reach $1.6 billion in 2018 (MediaKix, 2018). The rise of spending on influencer marketing is caused by a strong decline in TV viewership (replaced with increase in mobile screen time), which causes many brands and marketers to divert their ad spending towards digital advertisements (Maheshwari & Koblin, 2018; Mediakix, 2018).
The core definition of influencer marketing is defined as “promoting brands through use of specific key individuals who exert influence over potential buyers” (Brown & Hayes 2008 as cited in Audrezet, de Kerviler & Moulard, 2018, p. 1). This thesis aims to focus on social media based influencer marketing, whereby brands work with influencers to promote and recommend their products on social media platforms. The beauty industry invest heavily in social media based influencer marketing programs, in 2016, Estée Lauder Companies Inc. acquired a cosmetics company, Too Faced for around US$1.45 billion. Estée Lauder’s President and Chief Executive Officer (CEO), Fabrizio Freda explained that in addition to strengthening Estée Lauder’s leadership position globally, the Too Faced brand “increases our penetration with millennial and Gen Z consumers, who are socially connected beauty enthusiasts.” (Turner & Wong, 2016). As previously mentioned, other beauty brands such as L’Oreal, Revlon, CoverGirl, M.A.C (part of Estée Lauder Companies Inc.) are also increasing their digital investments and launching influencer marketing programs. In early 2017, L’Oreal invests in Tallify, which is “a start-up which connects big brands with influencers to create campaigns” (Trefis, 2017). Maybelline, M.A.C and Revlon are beauty brands famous for past collaborations and marketing campaigns
with traditional celebrities, such as famous supermodels and movie stars but these brands are now working with social influencers (Rodulfo, 2017; Hall, 2017; Trefis, 2017).
2.3 Type of Influencer
There are several types of influencers on social media: celebrities, brand community members and social influencers such as bloggers, vloggers and Instafamous personalities (Djafarova & Rushworth, 2017; Evans, Phua, Lim & Jun, 2017). This study will compare two different types of influencers namely celebrities and social influencers. Celebrities include movie and TV stars, musicians, athletes, famous authors as well as individuals from business, politics, art, fashion and more (McCracken, 1989; Djafarova & Rushworth, 2017). Social influencers are everyday Internet users who have established a significant number of followers on social media and are regarded as trustworthy tastemakers (De Veirman, Cauberghe & Hudders, 2017).
2.3.1 Celebrities
Throughout this study, celebrities will be defined as “individuals who have achieved some level of fame in the entertainment industry, such as in film, television or professional sports” (Frizzel, 2011). Celebrities have been used to endorse products for more than a century (Erdogan, 1999). McCracken (1989) defines the celebrity endorser as “any individual who enjoys public recognition and who uses this recognition on behalf of a consumer good by appearing with it in an advertisement” (p. 310). In their meta-analysis of studies on celebrity endorsements Knoll & Matthes (2017) demonstrated that celebrity endorsements can generate positive effects on consumers’ attitudes compared to no endorsement. The researchers then performed a meta-regression comparing the effects of celebrity endorsements against other types of endorsements on
the attitude towards the endorsed object. The results proved to be intriguing, celebrity endorsements performed worse compared to endorsements with a quality seal or award, or an endorser brand (Knoll & Matthes, 2017).
There is a lack of literature investigating the effects celebrity endorsement on social media platforms. However, a study by Jin & Phua in the context of Twitter found that a celebrity endorser with a high number of followers on Twitter was found to be more credible than a celebrity endorser with a low number of followers (Jin & Phua, 2014). Moreover, when the celebrity has a high number of followers post a positive brand-related tweet, there is a positive effect on purchase intention (Jin & Phua, 2014).
As of August 2018, many of the most followed accounts on Instagram belong to traditional or mainstream celebrities: Selena Gomez (Singer & Actress | 140 million followers), Cristiano Ronaldo (Footballer | 138 million followers), Ariana Grande (Singer | 124 million followers), Beyoncé (Singer & Actress | 116 million followers), Kim Kardashian (Reality TV Star | 115 million followers) and Kylie Jenner (Reality TV Star | 113 million followers). Within the social media beauty scene, no celebrity can compare to the Kardashian-Jenner sisters such as Kim Kardashian West and Kylie Jenner. Both Kim Kardashian West and Kylie Jenner are global social media superstars. In the New York Times, Martin (2016) noted that “social media makeup enthusiasts become facsimiles of another – all some version of Kim Kardashian West”. Her makeup look is emulated frequently and one can find multiple “Kim Kardashian Inspired” makeup tutorial videos on YouTube. Kylie Jenner is Kim Kardashian West’s half-sister and she caused a shortage of MAC Cosmetics Spice lipliner pencil in stores across the U.S. after a mere mention that she used such product to create her fuller lips (Cochrane, 2015).
2.3.2 Social Influencers
As more brands and marketers invest in influencer marketing programs, social influencers have risen to become the new celebrity endorsers (Weinswig, 2016). Social influencers evolve from ordinary consumers and gain a significant number of followers through a megaphone effect, as described by McQuarrie, Miller & Phillips (2013). The authors observed that the fashion bloggers with no professional credentials or family/institutional amass large followings through exhibiting taste leadership and accumulating cultural capital.
Several studies have focused on the identification of key influencers on various social media platforms. Trusov, Bodapati & Bucklin (2010) examined users of social networking sites and developed a methodology in order to identify these individuals. A social network user is considered to be influential when “his or her activity level, as captured by site log-ins over time, has a significant effect on others’ activity levels and, consequently on the site’s overall page view volume” (Trusov, Bodapati & Bucklin, 2010, p. 644). The authors concluded that user profile descriptions such as gender, stated dating objectives, friends count, and profile views are poor predictors of influence and they lack the power to identify influential site members (Trusov, Bodapati & Bucklin, 2010). In reality, technology companies such as Google, Yahoo and Microsoft have patents on how to identify influential users online and in 2015, Facebook also has a patent on a method to “Identify Experts and Influencers in a Social Network” (Yaverbaum, 2015; Katona, Zubcsek & Sarvary, 2011; Trusov, Bodapati & Bucklin, 2010).
DeMers (2016) revealed in Forbes that an ideal social influencer has the following qualities:
1) A large following: Indicated by the number of followers on social media networking sites, such as Twitter, Instagram and Facebook and by the number of subscribers on
YouTube. DeMers (2016) reasons that to have reach, a social influencer must have “an audience of thousands at a minimum, preferably tens of thousands or more”.
2) Regular activity: The ideal social influencer should post or upload regular content and thus remain relatively active on social media to keep her/his followers interested and engaged.
3) Industry relevance: An ideal social influencer should be in the industry relevant to the marketer or brand, in order to connect and relate to the social influencer’s audience. On Instagram, consumers consider social influencers (Instafamous personalities) to be more credible and trustworthy than celebrities and as such the opinions of social influencers are relatively more valuable (Djafarova & Rushworth, 2017). Furthermore, Kapitan & Silvera (2016) proposed that social influencers are considerably more likeable and authentic because they “have developed homegrown audiences” (p. 557). Social influencers demonstrate their authenticity in terms of their passion and transparency (Audrezet, de Kerviler & Moulard, 2018). Specifically, social influencers are driven by their creative intrinsic process to reveal their passionate authenticity, and transparent authenticity as they provide truthful and personal opinions to maintain their integrity (Audrezet, de Kerviler & Moulard, 2018).
Within the beauty industry, Nikkie de Jager is a prominent Dutch social influencer and she is known on her social media platforms as ‘NikkieTutorials’. She posted her first makeup tutorial video on YouTube in 2008 and is considered by Forbes as one of the top social influencers in the beauty category in 2017 (Forbes, 2017). As of August 2018, NikkieTutorials has 10.9 million followers on Instagram and more than 10.6 million subscribers on YouTube. Through influencer marketing programs, Nikkie de Jager has collaborated with brands such as Too Faced Cosmetics, Maybelline, and Ofra Cosmetics.
On Instagram, both celebrities and social influencers often provide their followers with glimpses of their private lives (Abidin, 2016); fans are able to see what they are eating for dinner, what they do on family outings or whilst traveling. By making their Instagram posts more personal, celebrities and social influencers are able to create a sense of familiarity and develop an emotional connection with their followers (Kassing & Sanderson, 2009; De Veirman, Cauberghe & Hudders, 2017). Every now and again, celebrities and social influencers would weave brand-related content on their Instagram posts (Abidin, 2016) and these brand-related posts can be genuine or sponsored. If the post is sponsored, the influencers do not always clarify the sponsorship arrangements explicitly (Martens, 2017; Frier & Townsend, 2016). As such, this has caused much confusion amongst their Instagrams followers to distinguish between genuine and sponsored posts. Hence, recently FTC and other regulatory bodies have taken several actions to ensure sponsored posts are clearly labeled with a sponsorship disclosure in order to alert the consumers that the Instagram post is advertising.
2.4 Sponsorship Disclosure
Sponsorship disclosure functions to heighten the viewers’ awareness of sponsored content (i.e. product placement). Ultimately, the main objective of sponsorship disclosure is to “help viewers access their persuasion knowledge by making it easier to distinguish commercial from editorial content” (Boerman, van Reijmersdal & Neijens, 2014, p. 215). Boerman, van Reijmersdal & Neijens (2013) investigated sponsored content and sponsorship disclosure in television programs. The authors discovered that on average, television viewers are rather neutral about sponsored content (product placement), but they appreciate the existence of a sponsorship disclosure. Essentially, the viewers appreciate the sponsorship disclosure as they are being informed of the
presence of sponsored content in the television program (Boerman, van Reijmersdal & Neijens, 2013).
Boerman & van Reijmersdal (2016) highlighted a crucial development in the formatting of advertising whereby commercial messages are embedded into conventionally non-commercial content. Such advertising formats are also referred to as “as sponsored content, embedded advertising, stealth marketing, covert marketing, branded content, product placement, or native advertising” (Boerman & van Reijmersdal, 2016, p. 3). These advertising formats are blurring the lines between editorial and commercial content and thereby disguising its commercial source and intent (Boerman, van Reijmersdal & Neijens, 2014; Boerman & van Reijmersdal, 2016). Such formats are commonly found within the digital settings (Sahni & Nair, 2017) and on social media they are exemplified through sponsored blog posts and native advertising on social media platforms like Facebook or Instagram.
Sponsorship disclosures are able to help viewers in recognizing sponsored content as advertising and as such allow the possibility that viewers may process the sponsored content critically (Boerman, van Reijmersdal & Neijens, 2013). However, following the recent changes to the advertising format, Boerman, van Reijmersdal & Neijens (2014) remarked that there is a likelihood that consumers are not aware of its persuasive intention and thus, may not access their persuasion knowledge. On a social media platform such as Instagram, celebrities would often reveal particular moments of their private lives that have been traditionally inaccessible (De Veirman, Cauberghe & Hudders, 2017; Boerman, Willemsen & van der Aa, 2017), for example, they would post pictures or videos of what they are cooking for dinner or their family activities. Celebrities or social influencers would also post pictures or videos of themselves using a particular product or service and sharing their product reviews or experiences. However, trouble arises when
celebrities do not disclose sponsorship appropriately. It may confuse the Instagram followers whether the celebrity genuinely uses and recommends the product/service and the followers may fail to recognize the sponsored message as advertisement.
Boerman, Willemsen & van der Aa (2017, p. 83) explained that “when consumers are not able to recognize sponsored content as advertising, they may be persuaded into commercial transactions that they might otherwise avoid”. To avoid confusion amongst consumers, the Advertising Standards Authority for Ireland (ASAI) is setting guidelines for bloggers, influencers and companies so that they disclose any form of marketing communications responsibly:
“To achieve Code compliance, the ASAI is recommending the use of a clearly identifiable hashtag such as #Ad or #SP. The chosen identifiable hashtag must be clear from the beginning of the content. Disclaimers should be visible for consumers to see before they interact with or read the relevant material. A disclaimer that appears below- the-fold on websites, in terms and conditions, or at the end of the marketing communication is not sufficient.” (ASAI, 2017)
2.4.1 Sponsorship Disclosure on Instagram
Influencers on Instagram often uses hashtag to disclose a sponsored post. However, hashtags such as such as #collab, #spon, or #ambassador, are often ambiguous. They are also not always easily visible to Instagram users, for example, when a disclosure only appears when a “click more” link is opened. Following FTC regulations and letter of warnings sent to 90 influencers in April 2017 (FTC, 2017), Instagram has introduced a native tool called ‘paid partnership’ feature for those posts containing sponsored content, as displayed on Figure 2. For the influencer, a sponsored post will contain a tag “Paid partnership with [business partner]”. For businesses, they can collect performance metrics including “likes” and “comments” connected to the tagged post. Although this has brought a much clearer transparency to sponsored contents, the responsibility of disclosing a sponsored content still falls on the influencer and the business, and not on Instagram as a
platform. Ambiguous disclosure, or inappropriate placement of disclosure, will not count as a proper disclosure.
Figure 2 – Sponsorship disclosure on Instagram using “paid partnership” feature
2.5 Persuasion Knowledge
Consumers are continuously bombarded with marketing information on various products and services and they also encounter many influence/persuasive attempts by marketing agents. Throughout their lifespans consumers develop a knowledge structure on marketing and persuasion (Campbell & Kirmani, 2000; Friestad & Wright, 1994). Persuasion knowledge is a highly important resource for consumers and they rely on persuasion knowledge to “identify how, when, and why marketers try to influence them” (Friestad & Wright, 1994, p. 1). As a belief system it informs consumers on these marketers’ persuasion motives and persuasion tactics and how to respond to them (Friestad & Wright, 1994; Campbell & Kirmani, 2000).
Centering on the context of advertising and marketing messages, Friestad & Wright (1994) developed the Persuasion Knowledge Model. Friestad & Wright (1994) discussed how persuasion knowledge “develops from a simplistic set of beliefs into an integrated, complex structure of implicit beliefs, which are automatically activated in everyday contexts” (Wright, Friestad & Boush, 2005, p. 226). The Persuasion Knowledge Model illustrates how a consumer utilizes persuasion knowledge in order to cope with persuasion attempts and how he/she manages the persuasion episode in order to meet his/her own objectives (Friestad & Wright, 1994; Campbell & Kirmani, 2000). It must be noted however that, although persuasion knowledge gradually develops over a lifespan, it is more or less established during adulthood.
Since activated persuasion knowledge increases the consumer’s awareness that they are being influenced, consumers “will be less likely to respond to the message” (Milne, Rohm & Bahl, 2009, p. 108). Many studies have thus investigated the negative effect of persuasion knowledge on consumer responses. In the context of radio college show, a sponsorship disclosure that led to activated persuasion knowledge, reduced the effectiveness of selling tactics (Wei, Fischer, Main, 2008). In social media, Boerman, Willemsen & van der Aa (2017) discovered that when a celebrity discloses sponsorship on their Facebook post, it reduces the likelihood of consumers sharing the post (eWOM). The authors called for further research to investigate whether another type of endorser would also produce similar effects.
Research on persuasion knowledge in marketing has been focusing on identifying factors that could lower the negative effects of persuasion knowledge on consumer responses. In the context of television, Boerman, Reijmersdal & Neijens (2012) has shown that the duration of sponsorship disclosure (6 seconds vs. 3 seconds) leads to different level of activated persuasion knowledge, that resulted in significantly reduced brand evaluation. Furthermore, Hwang & Zhang
(2018) investigated the effect of parasocial relationship between social influencer and its audience on consumer responses. Parasocial relationship is an illusory, one-sided relationship between the audience and the influencer (Horton & Strauss, 1957; Labrecque, 2014), and influencers has been known to exploit this to endorse and to sell products (Lueck, 2015). Parasocial relationship has been shown to mitigate the negative effect of persuasion knowledge on purchase intention and sharing intention (Hwang & Zhang, 2018). Another factor that has been shown to attenuate persuasion knowledge effect is brand familiarity (Wei, Fischer & Main, 2008), as will be described in further details below.
2.6 Brand Familiarity
Brand familiarity is a basic form of consumer knowledge according to Baker, Hutchinson, Burke & Nedungadi (1986) and it is connected to brand awareness (Keller, 1993). Brand familiarity increases with product usage or through exposures to advertising (Keller, 1993). As a variable, brand familiarity indicates the extent of a consumer’s direct and indirect experiences with a brand, for example, consumers or their friends and family may have used the brand, or they may have seen prior advertisements from the brand (Kent & Allen, 1994; Campbell & Keller, 2003). The exposure to the brand may therefore evoke consumer’s memory from past experiences: “brand familiarity captures consumers’ brand knowledge structures, that is, the brand associations that exist within a consumer’s memory.” (Campbell & Keller, 2003, p. 293). As suggested by prior studies, brand familiarity can moderate disclosure as explained by attitude accessibility (Fazio et al. 1998; Smink, van Reijmersdal & Boerman, 2017). Due to a higher attitude accessibility, a familiar brand is expected to command more visual attention than the unfamiliar one (Smink, van Reijmersdal & Boerman, 2017).
Smink, van Reijmersdal & Boerman (2017) studied the effect of sponsorship disclosure on products with high or low familiarity on television. In this study, they found that sponsorship disclosure leads to an increased perceived persuasion in a familiar brand as compared to an unfamiliar brand. Sponsorship disclosure also increases visual attention to the familiar brand than the unfamiliar brand: it can also function as a prime for a brand with high familiarity. In terms of brand evaluation, prior studies demonstrated that sponsorship disclosure could have a negative effect on brand evaluation (Bennett, Pecotich, and Putrevu, 1999; Boerman, van Reijmersdal & Neijens, 2012; Campbell, Mohr, and Verlegh, 2013). Nevertheless, Smink, van Reijmersdal & Boerman (2017) revealed that there is a difference between a subtle and a prominent brand placement. Although familiar brands are easier to recognize as advertisement, sponsorship disclosure does not negatively affect brand attitude when a subtle brand placement is used.
Furthermore, Wei, Fischer & Main (2008) investigated the effect of activating persuasion knowledge, the interaction with brand familiarity, and consumers’ attitude towards a brand, in a college radio program. They demonstrated that activating persuasion knowledge gives an unfavorable effect on brand attitude. However, when brand familiarity is high, the negative effect of activating persuasion knowledge on brand evaluation is diminished. In fact, activating persuasion knowledge can lead to a favorable attitude for brands with high familiarity. These two studies underline a significant difference of persuasion knowledge effect to consumer attitudes toward a brand, when brand familiarity is taken into account.
2.7 Consumer Responses
To measure the success of advertising, marketers rely on several indicators. Amongst these indicators, two key performance indicators: (1) brand evaluation (Spears & Singh, 2004) and (2)
electronic word-of-mouth (eWOM) intention, (Peters et al. 2013; Boerman, Willemsen, & van der Aa, 2017), have been increasingly used.
2.7.1 Brand Evaluations
Brand evaluation, or also referred as brand attitude, has been studied intensively in marketing research for many decades (Spears & Singh, 2004). Brand evaluation is a valuable measure that can predict consumers’ behavior and their purchase intention (Spears & Singh, 2004; Mitchell & Olson, 1981), and it is “ a necessary communication effect if a brand purchase is to occur” (Percy & Rossiter, 1992, p. 266). On social media, in a study amongst Polish Facebook users by Schivinski & Dabrowski (2016) found that brand attitude has a significant positive influence on purchase intention. Similar findings were also observed by Kudeshia & Kumar (2017) for Facebook users in India. Unfortunately, research is currently lacking on the investigation of brand evaluation in the context of Instagram.
Multiple factors can influence brand evaluation. A study by Chiou, Hsu, & Hsieh (2013) has demonstrated that a negative online information can have detrimental impact on brand evaluation. The authors further showed that this negative effect could be attenuated by brand attachment and source credibility (Chiou, Hsu, & Hsieh, 2013). Furthermore, in the context of Instagram, brand uniqueness could also increase brand evaluation (De Veirman, Cauberghe & Hudders, 2017). They found that the positive effect was found to be moderated by the influencer’s number of followers on Instagram (De Veirman, Cauberghe & Hudders, 2017). Another factor is the activation of persuasion knowledge. As mentioned above, in a study of covert marketing on college radio program, Wei, Fischer, & Main (2008) have shown that activated persuasion knowledge could negatively affect brand evaluation by consumers. They discovered that the familiarity of the brand could moderate the negative effect of persuasion knowledge, as a high
familiarity brand was found to be less affected compare to a low familiarity brand. This thesis will investigate whether this effect can also be applied in the context of beauty products and Instagram.
There are multiple measures of brand evaluation that have been used by researchers as summarized by Spears & Singh (2004). In this thesis we will apply measures for brand evalution according to Wei, Fischer, & Main (2008), that will allow us to directly compare to this study.
2.7.2 eWOM Intention
Electronic word-of-mouth (eWOM) is defined by Hennig-Thurau, Gwinner, Walsh & Gremler (2004, p. 39) as “any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet”. This includes liking, sharing, and commenting on a particular post (Boerman, Willemsen & van der Aa, 2017). Social media platforms such as Facebook, Instagram & Twitter have made eWOM communication faster and more convenient (Berger, 2014). Instagram posts that are public can be shared and forwarded to other users through Instagram Direct messaging function or by commenting on the Instagram post itself and alerting the other user(s) via a tagging function. Furthermore, as opposed to traditional WOM, which is mostly one-on-one (King, Racherla & Bush, 2014), social media platforms allow users to communicate eWOM to their social networks, which is a relatively larger group of people (Eisingerich, Chun, Liu, Jia & Bell, 2015). Instagram users are able to interact with each other through liking a post (like function) and commenting on a post (comment function), as illustrated in Figure 1. Within the comment section of an Instagram post, users can tag another Instagram user (@ function) and thus bring the Instagram post into another user’s attention. Essentially, the ‘@ function’ facilitates consumer-to-consumer eWOM communication. Please see below a screenshot of Kylie Jenner’s Instagram post
and a screenshot of comments belonging to that specific post (Figure 3). Kylie Jenner posted a picture on Instagram of her in a dress from the FashionNova brand with the caption “left my designer for this @FashionNova fit”. From the screenshot of the comment section, you can see that user zzc_23 tagged her friend @so_johahn. In her comment, zzc_23 wrote “@so_johahn I want this dress” and by doing so, zzc_23 brought Kylie Jenner’s Instagram post to another user’s (so_johahn) attention.
Figure 3 – Instagram ‘tag’ other user function as a form eWOM.
There is a lack of literature on eWOM intention within the context of Instagram. Boerman, Willemsen & van der Aa (2017) investigated the relationship of sponsorship disclosure on eWOM intention through a mediation of persuasion knowledge in the context of Facebook. The authors
found that when a celebrity discloses sponsorship on a Facebook post, consumers tend to generate critical and skeptical feelings towards the post and consequently are less likely to engage in eWOM about the Facebook post (Boerman, Willemsen & van der Aa, 2017).
It is important to note that eWOM activities have been associated positive consumer responses. eWOM has a positive effect on sales in the context of online book sales, and that consumers read and respond to reviews written by other consumers (Chevalier & Mayzlin, 2006). On social media users, studies have shown the positive effect of eWOM and purchase intention (Erkan & Evans, 2016; Wang, Yu, & Wei, 2012). Thus, eWOM intention is a reliable indicator of consumer response.
3. HYPOTHESES & CONCEPTUAL FRAMEWORK 3.1 Hypotheses
Sponsorship disclosure and Persuasion Knowledge
Sponsorship disclosures clarify the commercial purpose of a message to consumers, they help consumers to recognize these disclosures as advertising, and hence activate their persuasion knowledge (Friestad & Wright, 1994; Boerman, van Reijmersdal, Neijens, 2013; Boerman, Willemsen & van der Aa, 2017). Awareness of the persuasive intent of a message triggers persuasion knowledge and armed with this knowledge a consumer may process the message critically (Boerman, van Reijmersdal & Neijens, 2013).
When a celebrity or social influencer posts a sponsored brand-related content but neglects to reveal the sponsorship arrangement (i.e. sponsorship disclosure on the post), the commercial nature is disguised (Boerman, van Reijmersdal & Neijens, 2014). Consequently, consumers viewing the post are less likely to be aware of the persuasive intent and activate persuasion knowledge (Campbell, Mohr & Verlegh, 2013). With this reasoning, we propose that a sponsorship disclosure on an Instagram post helps to clarify the commercial nature of the nature and its intent to persuade and as such should activate persuasion knowledge. Therefore, the following is proposed for hypothesis 1:
H1: An Instagram post with sponsorship disclosure (“Paid partnership with” & “#Ad”) will be more likely to activate consumers’ persuasion knowledge, than an Instagram post without a sponsorship disclosure
Celebrity endorsers and social influencer on persuasion knowledge
Instagram posts containing endorsement are recently required to explicitly disclose sponsorship. Activation of persuasion knowledge by sponsorship disclosure is known to negatively influence consumer responses (Boerman and Van Reijmersdal 2016; Wei, Fischer & Main, 2008). To gain favourable consumer responses, it is therefore important to identify potential moderating factors of persuasion knowledge in social media endorsements. Recent study by Boerman, Willemsen, and van der Aa (2017) assessed different types of sources, namely celebrity vs brand as the source of a Facebook post. This study found that a Facebook post by a celebrity triggered a significantly lower conceptual persuasion knowledge than a Facebook post by a brand. When the source of the post was a brand, consumers could readily recognize it as an advertisement, but not by a celebrity. This finding suggests that the type of influencer (brand or celebrity) could moderate activation of persuasion knowledge.
Furthermore, a meta-analysis study showed that a celebrity endorsement lead to favorable effects on consumer responses in comparison to no endorsement (Knoll & Mathes, 2017). However, the same study also showed that celebrity endorsement performed worse than other type of endorsements, such as a quality seal or award, or an endorser brand. Friestad and Wright (1994) suggested that consumer would learn endorsement by celebrities as advertisement, as they are exposed to advertisement gradually over time. Therefore, when an Instagram post by celebrity contains a product endorsement, it is expected a high persuasion knowledge is already present.
In contrast to celebrity endorsement, there is lack of studies evaluating whether a social media post by a social influencer also activates persuasion knowledge, and whether they differ with traditional celebrities. Social influencers are considered to be the expert and opinion leaders
in the field, and they are viewed as authentic fellow consumers sharing their opinions and beliefs on the advertised product (Rasmussen, 2018). Additionally, endorsements by social influencer is relatively new to consumers as compared to celebrity endorsements, which make it less likely for consumers to activate persuasion knowledge (De Veirman, Cauberghe & Hudders, 2017). Thus, we predict that compared to the post by a social influencer, a post by a celebrity will elicit a higher persuasion knowledge. Therefore, for H2, we hypothesize that:
H2: An Instagram post by a celebrity will be more likely to activate persuasion knowledge than an Instagram post by a social influencer.
Boerman, Willemsen, and van der Aa (2017) demonstrate that consumers could better recognize a post by celebrity to be an advertisement when sponsorship is disclosed. Evans et al. (2017) have shown that sponsorship disclosure in Instagram post by social influencers also activated persuasion knowledge. However, a comparative study on the effects of sponsorship disclosure on persuasion knowledge between the different types of influencer (celebrity influencer versus social influencer) has never been done before. According to Friestad and Wright (1994), due to being exposed to advertisement gradually over time, consumers would readily associate endorsement by celebrities as advertisement. On the contrary, social influencers are considered by consumers to be credible, trustworthy and authentic sources with no commercial intent (Djafarova & Rushworth, 2017; Kapitan & Silvera, 2016) as such, they are not expected to advertise and this may trigger followers to react more defensively towards the post (Krouwer, Poels, & Paulussen, 2017). We hypothesize that an Instagram post by a social influencer will generate a stronger
persuasion knowledge activation upon disclosure of sponsorship as compared to a post from a celebrity.
H3: The type of influencer will moderate the effects of a sponsorship disclosure on persuasion knowledge. An Instagram post with sponsorship disclosure will have a stronger effect on persuasion knowledge when posted by a social influencer compared to a celebrity
Persuasion knowledge, brand evaluation and eWOM intention
Sponsorship disclosure is known to heighten persuasion knowledge in consumers, resulting into a more critical evaluation of the advertising by the consumers (Boerman, van Reijmersdal & Neijens, 2013). In turn, activated persuasion knowledge have been demonstrated to elicit negative effects on consumer responses, including on brand evaluation and on eWOM intention (Wei, Fischer & Main, 2008; Boerman, Willemsen & van der Aa, 2017; Boerman, Reijmersdal & Neijens, 2012). As measures of consumer responses towards an advertisement, in this study, we focus on two key performance indicators: brand evalution and eWOM intention. We predict that the increased (activated) persuasion knowledge will negatively affect brand evaluation and eWOM intention and for H4, the following is proposed:
H4: Activated persuasion knowledge will negatively affect consumer responses (brand evaluation & eWOM intention)
Brand familiarity and eWOM intention and brand evaluation
The negative effect of persuasion knowledge can significantly impair the effectiveness of an advertising (Milne, Rohm & Bahl, 2009). In order to alleviate the effect of persuasion knowledge on consumer response, multiple factors have been shown to moderate the negative effect of increased persuasion knowledge on eWOM intention and brand evaluation. Brand familiarity, in particular, have been demonstrated to moderate negative effects of persuasion knowledge on consumer responses (Wei, Fischer & Main, 200). In this study, they revealed that persuasion knowledge activation made little difference to brand evaluations when the evaluated brand was highly familiar (Wei, Fischer & Main, 2008). Thus, we hypothesize that brand familiarity will mitigate the negative effect of persuasion knowledge on consumer responses, eWOM intention and brand evaluation.
H5: Brand familiarity will moderate the effect of persuasion knowledge on eWOM intention and brand evaluation. Increasing levels of brand familiarity will mitigate the negative effects of activated persuasion knowledge on consumer responses (brand evaluation & eWOM intention)
Persuasion knowledge mediates the negative effect of sponsorship disclosure on consumer responses
Studies investigating the effect sponsorship disclosure on consumer response have demonstrated a mediation effect by persuasion knowledge. Smink, van Reijmersdal & Boerman (2017) have demonstrated that persuasion knowledge mediates the effect of disclosure of product placement on consumers’ brand evaluation, in the context of television. In social media, Boerman, Willemsen & van der Aa (2017) have shown that the negative effect of sponsorship disclosure of a Facebook
post on eWOM intention is mediated by activation of persuasion knowledge. On Instagram, Evans, Phua, Lim & Jun, (2017) also discovered that persuasion knowledge mediates the negative sponsorship disclosure on brand evaluation and intention to share. Therefore, we hypothesize the sponsorship disclosure, as mediated by increased persuasion knowledge will negatively affecting brand evaluation and eWOM intention.
H6: Sponsorship disclosure mediated through activated persuasion knowledge will negatively affect consumer responses. An Instagram post with sponsorship disclosure will activate persuasion knowledge and consequently reduce both brand evaluation & eWOM intention
3.2 Conceptual Framework
Based on the above discussion of the variables comprised within this study, sponsorship disclosure, type of influencer, persuasion knowledge, brand familiarity, and consumer responses, Figure 4 summarizes the conceptual model and the hypotheses that will be tested in this experimental study.
Figure 4 – Conceptual Framework SPONSORSHIP DICLOSURE TYPE OF INFLUENCER BRAND EVALUATION eWOM INTENTION PERSUASION KNOWLEDGE BRAND FAMILIARITY
4. METHODOLOGY 4.1 Research Design
This study conducted an online experiment with a 2 (disclosure presence: no sponsorship disclosure vs sponsorship disclosure) x 2 (influencer type: celebrity vs social influencer) x 2 (brand familiarity: high vs low) between-subjects design. The research design resulted in eight different experimental conditions as seen on the Table 1 below and each participant was randomly assigned to one of the eight conditions.
Table 1 – Experimental Conditions
Condition Type of Influencer Sponsorship Disclosure
Brand Familiarity
Number of Respondents
1 Celebrity Yes High 49
2 Celebrity Yes Low 48
3 Celebrity No High 48
4 Celebrity No Low 48
5 Social Influencer Yes High 48
6 Social Influencer Yes Low 45
7 Social Influencer No High 47
8 Social Influencer No Low 44
4.2 Participants
A total number of 382 participants were recruited via Prolific, a crowdsourcing platform for research and by means of convenient sampling via the author’s personal communication on Instagram, Facebook and Whatsapp. A large majority of the participants is female (84.3%) and aged between 18 and 34 years old (77.5%). This is in line with Djafarova & Rushworth’s (2017) observation that the most common Instagram users are female and fall between the age of 18 – 30 years old. The survey participants accounted for 40 different nationalities, there are notable
proportions within the sample, with 27% from the United Kingdom, 13.4% from Malaysia, 12% from Indonesia, 10.2% from the Netherlands and 9.9% from the United States of America. Furthermore, the majority of the participants have completed tertiary education, whereby 44% have a bachelor’s degree and 25.9% have a master’s degree. With regards to Instagram usage, 88.8% of the participants stated that they accessed their Instagram account on a daily basis.
4.3 Procedure
The study conducted an online experiment through the use of Qualtrics, the participants were presented with information regarding consent and survey requirements. A general description of the survey was given, that the survey contains questions relating to Instagram usage and responses to an Instagram post. Participants were told that they are free to withdraw from the survey at any point and any information they have filled in will be kept confidential and anonymous. If participants are willing to be part of the study, they have to indicate their agreement and provide informed consent. After providing informed consent, participants were lead to the next page and provided with questions regarding their average Instagram usage and how often they use makeup products.
On the next page, the participants are randomly assigned to eight conditions and a short descriptive text of the respective type of influencer and the stimuli was shown. Participants were exposed to an Instagram post that was posted by either a celebrity or a social influencer, in which sponsorship disclosure was revealed or not revealed and the post included a brand with high familiarity or low familiarity. In viewing the Instagram post, participants were asked to imagine that they are actually following the respective influencer on Instagram and that they came across the post on their Instagram feed. After the participants have looked at the Instagram post, they
proceed to the rest of the survey. Participants were presented with questions about brand familiarity (brand familiarity & brand experience).
This was followed with questions on their consumer responses (brand evaluations and eWOM) and soon after, with questions regarding the participants’ persuasion knowledge. Boerman, Willemsen & van der Aa (2017) suggested that questions on consumer responses should be presented before questions on persuasion knowledge. The authors wanted to ensure that consumer responses would “not be primed by questions that revealed the commercial nature of the [Instagram] post” (Boerman, Willemsen & van der Aa, 2017, p.87). The survey then moved on with questions concerning control variables such as, recall of sponsorship disclosure and demographics (nationality, age, gender and education). When the survey concluded, participants were thanked for completing the survey and they were debriefed with information regarding the purpose of the study and that they are able to contact the researcher – should they have any questions or concerns – through the researcher’s email address.
4.4 Stimulus Materials
The stimulus materials used in this study comprised of eight edited iPhone screenshots of an Instagram post. There are three independent variables being manipulated (sponsorship disclosure, type of influencer and brand familiarity) and all stimuli can be found in the Appendix A. In terms of sponsorship disclosure, it was manipulated in two ways: sponsorship disclosure versus no sponsorship disclosure. The Instagram post with sponsorship disclosure displayed “Paid partnership with [BRAND]” at the top left hand corner of the post and in the caption of the post underneath the influencer’s picture, the hashtags “#Ad” and “#[BRAND]partner”. The paid partnership feature from Instagram and the hashtags are both used to maximize the participants’
recognition of sponsorship disclosure. The Instagram post without disclosure presence, do not have “Paid partnership with [BRAND]” and the aforementioned hashtags shown.
Furthermore, in each condition participants would either be exposed to a celebrity, Kylie Jenner or a social influencer, NikkieTutorials (Nikkie de Jager). The Instagram post displayed the picture of the respective influencer and the influencer’s name is visible as text, both above the picture and below the picture. A brief description of the respective influencer accompanied the stimulus. Brand familiarity was manipulated through exposing participants to different versions of the Instagram post, that is, one would feature a high-familiarity brand and another would feature a low-familiarity brand (Wei, Fischer & Main, 2008). From the pre-test two brands were selected, L’Oréal Makeup (high familiarity) and Bloom Cosmetics (low familiarity). The results of brand familiarity analysis of the pre-test from a separate pool of participants (n = 36) confirmed that L’Oréal Makeup (M = 4.07, SD = 1.01) to be significantly more familiar than Bloom Cosmetics (M = 1.23, SD = 0.504), using independent t-test (p < 0.001). In the sponsorship disclosure condition, the post would display the brand name twice, placed above the picture “Paid partnership with lorealmakeup/bloomcosmetics” and underneath the picture in the caption: “Love my new @lorealmakeup/@bloomcosmetics” lipstick”.
4.5 Measures
Persuasion Knowledge. This study measured persuasion knowledge by asking the participants to indicate on a 7-point Likert scale to what extent they agreed with two statements: “The Instagram post was advertising” adapted from Boerman, Willemsen & van der Aa (2017) and “The brand was mentioned because it paid to be mentioned” adapted from Wei, Fischer & Main (2008). The mean of these two items is used as a measurement of persuasion knowledge
(M = 5.86, SD = 1.27). Reliability analysis indicated that the two items formed a reliable scale (Cronbach’s alpha α = .857).
Brand Familiarity. In order to measure brand familiarity participants were asked to denote their familiarity and experience with the brand on a 7-point Likert scale (1 = “not at all familiar,” and 7 = “very familiar”; 1 = “not at all experienced,” and 7 = “very experienced”). This measure for brand familiarity is adapted from Wei, Fischer & Main (2008). The measure of brand familiarity consisted of the mean score of the two items (M = 3.13, SD = 2.13). The scale is proven to be reliable through a reliability analysis (Cronbach’s alpha α = .873).
Brand Evaluations. Responses to the brand were measured using a semantic differential scale developed from Wei, Fischer & Main (2008) with three items on 7-point scales (1 = “dislikeable” and 7 = “likeable”; 1 = “bad” and 7= “good”; 1 = “unfavorable,” and 7 = “favorable”). The responses from those three items would be averaged to measure brand evaluations and reliability analysis showed the scale to be reliable (M = 4.32, SD = 1.28, Cronbach’s alpha α = .947).
eWOM Intention. In order to measure eWOM intention, this study adapted the four item construct used by Boerman, Willemsen & van der Aa (2017). The first two items measured the intention to share (“I think this Instagram post is worth sharing with others,” and “I will recommend this Instagram post to others”). The other two items showcased different ways to engage in eWOM on Instagram (“I would ‘like’ this Instagram post,” and “I would comment on this Instagram post”). Participants were asked to indicate to what extent they agreed with those four statements on a 7-point Likert scale. eWOM intention was measured by the mean score of those four items (M = 2.90, SD = 1.46) and a reliability analysis revealed the scale to be reliable (Cronbach’s alpha α = .885).
5. RESULTS
5.1 Randomization Check
A randomization check using a Chi-square test was performed in order to check for between-group differences for age, gender, education and nationality. The Chi-square test results indicated that the experimental groups did not differ in terms of age χ2 (42) = 49.108, p = .210,
gender χ2 (21) = 21.124, p = .451, education χ2 (21) = 16.956, p =.714, and nationality
χ2 (273) = 254.272, p =.786.
5.2 Manipulation Check
As previously mentioned, a pre-test on brand familiarity was conducted in order to determine and select the two makeup brands for the high and low brand familiarity conditions. L’Oréal Makeup was selected for high brand familiarity condition and Bloom Cosmetics was chosen for the low brand familiarity condition. To see if the manipulation worked, a One-Way ANOVA test was used to determine the statistical difference between the means amongst the participants in the high familiarity versus low familiarity conditions. Participants exposed to an Instagram post with L’Oréal Makeup indicated higher levels of brand familiarity (M = 4.42, SD = 1.96) in comparison with participants exposed to an Instagram post with Bloom Cosmetics (M = 1.81 , SD = 1.34). The analysis results showed that the difference between the means was statistically significant [F(1, 375) = 226.56, p <0.01].
5.3 Testing of Hypothesis
5.3.1 Hypothesis 1
Hypothesis 1 predicted that sponsorship disclosure would result in the increase of persuasion knowledge. In order to determine whether or not sponsorship disclosure influences persuasion knowledge, we performed a regression analysis with sponsorship disclosure regressed on persuasion knowledge. The regression analysis proved to be significant, R2 = 0.041,
F(4, 377) = 3.994, p < 0.01. As shown on Table 2, sponsorship disclosure has a positive effect on persuasion knowledge (B = 0.370, t = 2.876, p < 0.01) keeping age, gender and education constant. This means that sponsorship disclosure increases persuasion knowledge and thus, H1 is supported. However, the results also imply that gender has an effect on persuasion knowledge controlling for other variables (B = 0.276, t = 1.985, p < 0.05) and this means that on average, women obtain 0.276 points higher level in persuasion knowledge than men.
Table 2 – Effect of Sponsorship Disclosure on Persuasion Knowledge – Regression Analysis
Variables Coefficients (B) S.E
Standardized Coefficients (β) t Sponsorship Disclosure 0.370** 0.129 0.146 2.876 Age 0.088 0.073 0.062 1.211 Gender (Female =1) 0.276* 0.139 0.101 1.985 Education 0.076 0.081 0.049 0.945 Constant 5.013 0.286 17.528 *p<0.05, **p<0.01, ***p<0.001
5.3.2 Hypothesis 2
Hypothesis 2 predicted that the type of influencer had different effects on persuasion knowledge. Specifically, H2 expected that an Instagram post by a celebrity would induce stronger effects on persuasion knowledge compared to an Instagram post by a social influencer. In order to assess whether or not the type of influencer impacts persuasion knowledge, a regression analysis was performed controlling for age, gender and education. The regression analysis proved to be significant, R2 = 0.065, F(4, 377) = 6.56, p < 0.01. The results indicated that the participants had
a significantly greater likelihood to recognize the Instagram post as advertising when it was posted by a celebrity (M = 6.13 , SD = 1.11) compared to when it was posted by a social influencer (M = 5.59 , SD = 1.36). As shown on Table 3, a celebrity induces a stronger effect on persuasion knowledge than a social influencer (B = 0.541, t = 4.284, p < 0.01) keeping age, gender and education constant. Thus, H2 is supported.
Table 3 – Effect of Type of Influencer on Persuasion Knowledge – Regression Analysis Results
Variables Coefficients S.E
Standardized Coefficients
(β)
t
Type of Influencer (Celebrity = 1) 0.541*** 0.126 0.214 4.284
Age 0.087 0.072 0.061 1.211
Gender 0.256 0.137 0.093 1.869
Education 0.108 0.080 0.069 1.358
Constant 4.879 0.285 17.092
5.3.3 Hypothesis 3
The third hypothesis predicted that the type of influencer would moderate the effects of a sponsorship disclosure on persuasion knowledge. The hypothesis expected that an Instagram post with sponsorship disclosure will have a stronger effect on persuasion knowledge when posted by a social influencer compared to a celebrity. To test the moderation effect this study used Hayes’ PROCESS macro (Hayes, 2013) in SPSS. PROCESS is a computational tool that can be used to investigate mediation, moderation and conditional processes.
For the analysis, model 1 was applied (Hayes, 2013) and sponsorship disclosure functioned as the independent variable, type of influencer as the moderator and persuasion knowledge as the dependent variable. Age, gender and education were included as covariates. These variables accounted for 8.7% of the variance of persuasion knowledge, which is statistically significant [F(6, 375) = 5.918, p < 0.001]. The conditional effects of sponsorship disclosure (B = 0.395, p < 0.05) and type of influencer (B = 0.565, p < 0.01) on persuasion knowledge are both found to be statistically significant (Table 4). However, the interaction between the type of influencer and sponsorship disclosure (XM) did not prove to be significant. This indicates that the relationship of sponsorship disclosure on persuasion knowledge does not depend on the type of influencer. Figure 5 illustrates that sponsorship disclosure and type of influencer have significant positive and independent effects on persuasion knowledge and the interaction (XM) does not exist. Therefore, H3 is not supported.
Table 4 – Interaction Effect between Sponsorship Disclosure and Type of Influencer on Persuasion Knowledge – Moderation Analysis Results using PROCESS macro
Coefficient (B) SE t p Intercept i1 4.738 0.287 16.487 <0.001 Sponsorship Disclosure (X) c1 0.395* 0.182 2.167 0.031 Type of Influencer (M) c2 0.565** 0.180 3.145 0.002 Type*Disclosure (XM) c3 -0.043 0.254 -0.171 0.865 Age 0.075 0.072 1.050 0.295 Gender 0.273* 0.136 2.009 0.045 Education 0.088 0.079 1.114 0.266 R2 = 0.087 , p < 0.001 F(6, 375) = 5.918 *p<0.05, **p<0.01, ***p<0.001
Figure 5 – Effect of sponsorship disclosure (no disclosure vs. disclosure) on persuasion knowledge with respect to the type of influencer (Celebrity vs. Social Influencer)
4 4.5 5 5.5 6 No disclosure Disclosure P er su as ion k n ow le d ge Social Influencer Celebrity
5.3.4 Hypothesis 4
Hypothesis 4 expected that activated persuasion knowledge would have negative effects on consumer responses. H4 predicted that when participants recognize the Instagram post as advertising (activated persuasion knowledge) this would reduce their brand evaluations and eWOM intention. With respect to brand evaluation, a regression analysis of persuasion knowledge on brand evaluation was conducted controlling for age, gender and education. The regression analysis proved to be insignificant, R2 = 0.004, F(4, 377) = 0.373, p = 0.828. The Table 5 below
demonstrates that the level of persuasion knowledge does not have a statistically significant effect on brand evaluations (B = -0.013, p = 0.808).
Table 5 – Effect of Persuasion Knowledge on Brand Evaluation – Regression Analysis Results
Variables Coefficients (B) S.E
Standardized Coefficients (β) t Persuasion Knowledge -0.013 0.052 -0.013 -0.243 Age -0.018 0.075 -0.012 -0.239 Gender (Female =1) 0.058 0.143 0.021 0.403 Education -0.088 0.083 -0.056 -1.064 Constant 4.575 0.396 11.566 *p<0.05, **p<0.01, ***p<0.001
With respect to eWOM, a regression analysis was also performed to regress persuasion knowledge on eWOM intention and controlling for age, gender and education. The results of the regression analysis proved to be significant, R2 = 0.102, F(4, 377) = 10.688, p < 0.001. As seen on table 6,
when persuasion knowledge is activated there is a significant negative effect on eWOM intention (B = -0.297, p < 0.001). This implies that when the Instagram post is recognized as advertising, participants are less likely to engage in eWOM. It is worth noting however, that the analysis also
