Sponsorship Disclosures on Instagram: The Effects of Persuasion Knowledge on Brand Responses and the Moderating Role of Product Involvement.
Lubna Rezzoug – 10747621
Master’s Thesis Communication Science Graduate School of Communication Persuasive Communication
Supervisor: Dr. Stephan Winter 30th of June 2017
2 Abstract
The use of native advertising by companies, such as brand endorsements on social media, is gaining popularity. These practices have given rise to concerns, as the basic rights of
consumers online might be violated. In response to these concerns, the Federal Trade
Commission (FTC) introduced online regulations, namely sponsorship disclosures, which are meant to protect consumer by revealing the persuasive intent of messages. The present study compares the effect of three different ‘intensities’ of sponsorship disclosures on brand recall, brand attitude and purchase intention in an online experiment (N=292). Moreover, this study examines if persuasion knowledge has a mediating role between sponsorship disclosures and brand responses and if the effects are moderated by product involvement. The findings indicate that sponsorship disclosures activate persuasion knowledge, but that more extensive disclosures do not necessarily induce higher levels of persuasion knowledge than simple disclosures, nor does persuasion knowledge act as a mediator of the effect on brand responses. It is important to note that this study is one of the first that incorporates the role of product involvement in the field of sponsorship disclosures on social media. An interaction effect of product involvement was not found, however, the current study reveals that advertisement that contains low-involvement products has stronger effects on brand attitude and purchase intention than ads that include high-involvement products, which contradicts the existing literature. Based on its findings, the current study suggests practical implications for online law enforcers and social media marketers.
Keywords: sponsorship disclosures, Instagram, purchase intention, brand attitude,
3
An article in the Economist (2016) suggests that social media endorsements by influencers “are becoming the latest thing in advertising”. Celebrity personalities such as the Kardashians, Cristiano Ronaldo or even the girl next door who has over ‘200K’ followers on Instagram seem to become popular as a means for connecting brands to large networks of potential customers. However, with this development in advertising, “the grey area between voluntary celebrity endorsement and paid advertisement has grown murky” (The Economist, 2016), making it difficult for consumers to imagine that a social media post from their idol is merely a form of covert advertising.
Not surprisingly, concerns have arisen because consumers might be misled by this intrusion of marketers in the social media sphere, as there is no clear distinction between native advertising, messages which contain a hidden commercial intent, and authentic opinions or experiences (Colliander & Erlandsson, 2015; van Reijmersdal et al., 2016). In response to these ethical concerns, regulations have been adopted in the form of sponsorship disclosures, which aim to aid consumers in recognizing the persuasive intent of native advertisement (van Reijmersdal et al., 2016). Moreover, the US government has made alterations in their guidelines for social media endorsements, as individuals are now obliged to reveal connections with brands by means of a disclaimer (Colliander & Erlandsson, 2015).
Nevertheless, these regulations are almost exclusively applied in the US (Colliander & Erlandsson, 2015). Despite increasing criticism towards native advertising on social media, the phenomenon is flourishing as an advertising strategy for brands over the world. The lack of strict regulations outside of the US may be a cause for the prosperity of this advertising trend. However, even in the US total expenses on native advertising by companies are expected to reach an amount of 8.8 billion dollars in 2018 (Campbell & Marks, 2015). Hence, it is necessary to provide insights into disclosures on social media and their effect on
4
consumer’s cognitions, attitude formations and behavioral intentions. Previous research has been conducted on sponsorship disclosures and the effects on brand attitudes and brand recall (Boerman, Reijmersdal, & Neijens, 2012; Dekker & van Reijmersdal, 2013; Matthes & Naderer, 2016). These studies focused on two types of disclosure (disclosure of persuasive intent vs. source disclosure; Dekker & van Reijmersdal, 2013), rather than incorporating different intensities of disclosures. Moreover, the focus in these studies was on television as a medium. Therefore, the effects of sponsorship disclosures on social media are relatively unexplored, while they are of critical importance for advertisers and consumer protection.
The available studies on disclosures in the social media reference frame have researched media such as blogs (Van Reijmersdal et al., 2016). However, this social media type revolves around textual information, whereas Instagram is considered to be a picture application (Carah & Shaul, 2016). Furthermore, experts in the advertising field claim that Instagram is the most dominant amongst the existing social media platforms and has the potential to be the most effective platform for advertising (Delo, 2014). Therefore, it is crucial to attain more theoretical insights into the properties of disclosures and how these affect Instagram users in understanding the persuasive nature of native advertisement and coping with it.
Moreover, this study will offer novel theoretical insights by including a moderator that might have affected results in previous research (Van Reijmersdal et al., 2016), that is product involvement. Against this background, the present study aims to investigate the effects of Instagram posts that contain sponsorship disclosures for low-involvement products versus high-involvement products on persuasion knowledge (PK) and how this (consecutively) affects brand attitude, brand recall and purchase intention.
The present study could be valuable to answer societal questions, such as to which extent the format and use of sponsorship disclosures should be regulated to protect consumers
5
online. On the other hand, this research could also serve as a guideline to which type of disclosure has the most favorable effects for native advertisers on Instagram.
Theoretical Background Native Advertising and Instagram
Companies are becoming increasingly reliant on the incorporation of their brands or products into non-commercial contexts (Boerman, Reijmersdal, & Neijens, 2012; Van Reijmersdal, Neijens, & Smit, 2007), such as brand endorsement online through the social networks of influencers. This advertising strategy is called native advertising or the use of sponsored content (Boerman et al., 2012; van Reijmersdal et al., 2016). Native advertising is a form of covert, “stealth” or “masked” marketing that has been flourishing almost
simultaneously with the increasing popularity of social media (Campbell, Mohr, & Verlegh, 2013).
Although native advertising is popular in practice, there is no clear definition of the meaning or implications of native advertising. Therefore, the term applies to several types of online communication. For the purpose of this study, native advertising is defined as “desired online marketing communications that appear in-stream” (Campbell & Marks, 2015, p. 600). ‘Desired communication’ in this definition refers to consumers who in a sense allow the online marketer to communicate with them by preserving the standard settings for online interest-based ads for Facebook and Instagram. ‘In-stream’ implies that the ad’s format lends itself to facilitate the user-experience of the medium, such as covert formats that blend into a medium.
This study specifically focuses on native advertising on Instagram. This medium is interesting to investigate in light of native advertising because unlike Twitter and Facebook, Instagram has not detached advertising and non-advertising content (Carah & Shaul, 2016).
6
Therefore, advertisers on Instagram can create ads that blend in more easily with other content than on alternative social media. While brands on Facebook need to pay for content to appear on a user’s newsfeed, brands on Instagram can reach consumers by simply incorporating hashtags statements below their posts. These hashtag statements allow for the brand to virtually be found by anyone. On Instagram, brands seem to operate in similar ways as consumers. Unlike Facebook, Instagram does not force companies to create official brand pages via which they can only appear on the newsfeeds of consumers by paying for content. Hence, most advertising activity on Instagram is in a sense encouraged to go native.
Instagram is an online platform and Social Network Service (SNS), which enables users to take photos and videos on mobile phones, with the goal to facilitate photo-sharing and video-sharing online (Carah & Shaul, 2016; Sheldon & Bryant, 2016). When photos and videos are shared on Instagram, hashtag (#) statements are added for other users to find the photos or videos on the platform (hashtags can be compared to search engine optimization functions). Originally Instagram was designed as a SNS for personal use. Nevertheless, it soon caught the attention of companies and today it has grown to a powerful marketing tool for companies.
While this form of online advertising is embraced by marketers, it also attracts criticism, as this marketing tactic is evaluated as being deceiving towards consumers (Campbell & Marks, 2015). Native advertisements that exclude disclosures are accused of being misleading, because the secrecy regarding the publisher’s identity aims to generate higher degrees of trust from consumers (non-advertising content generates more trust than advertising content; Campbell & Marks, 2015)
Sponsorship Disclosures and PK
As a response to the prosperity of native advertising and the growing concerns, the Federal Trade Commission (FTC) has set stricter online disclosure guidelines (Boerman et
7
al., 2012; Campbell & Marks, 2015; van Reijmersdal et al., 2016). These disclosures aim to protect consumers from native advertising by revealing the persuasive intent of the message, they are called sponsorship disclosures (Matthes & Naderer, 2016;Van Reijmersdal et al., 2016).
When individuals become aware of the persuasive intent of a message, cognitive resistance tactics may be activated to protect oneself from persuasion. One of these cognitive resistance strategy that is activated by exposures to sponsorship disclosures is PK (Van Reijmersdal et al., 2016). The persuasion knowledge model (PKM; Friestad & Wright, 1994) proposes that the recognition of the persuasive intent of advertising messages can influence how efficiently consumers cope with the message (Evans & Park, 2015; Wojdynski & Evans, 2015). Moreover, PKM suggests that when individuals have prior knowledge with regard to a message’s persuasive intent, they can activate PK to protect themselves from advertising (Friestad et al., 1994). These suggestions align with prior studies, in which sponsorship disclosures demonstrate to affect an individual’s understanding of the persuasive intent of sponsored content (Boerman et al., 2012; Nelson, Wood, & Paek, 2009; Tessitore & Geuens, 2013; Van Reijmersdal, Lammers, Rozendaal, & Buijzen, 2015; Wei, Fischer, & Main, 2008). In previous studies, the effect of sponsorship disclosures in a variety of media channels (i.e., television, radio, websites, online blogs) has been explored, all these studies found that indeed exposure to sponsorship disclosures activates PK (Nelson et al., 2009; Tessitore & Geuens, 2013; Van Reijmersdal et al., 2016; Wei et al., 2008; Wojdynski, 2016).
Specifically, company video news releases that were labeled with source disclosures caused viewers to change their perceptions of the releases from being ‘news’ to being ‘commercialized’, suggesting that the disclosure activated PK in this study (Nelson et al., 2009). Moreover, product placement in television shows that included the European product placement symbol, only if people were trained to recognize the symbol or already knew the
8
symbol in advance, led to the activation of PK as well (Tessitore & Geuens, 2013). The activation of PK by means of spoken disclosures was also found to occur when applied in different media, such as radio shows (Wei et al., 2008). The results of another study showed that sponsorship disclosures in written text (news stories) may also affect PK, in specific middle or bottom positioning and the use of the words “advertising” or “sponsored” are most effective in activating PK compared to other labels (Wojdynski & Evans, 2016). Van
Reijmersdal et al. (2016) also found that sponsorship disclosures activate PK, however, their study employed a more up-to-date medium, namely the use of sponsorship disclosures in online blogs.
In accordance with the PKM and the literature on sponsorship disclosures and PK, this study predicts that consumers who are exposed to native Instagram advertising that contain disclosures will activate PK. Hence, this study proposes the following hypothesis:
H1a. Exposure to sponsorship disclosures (on Instagram) will activate PK; the sponsorship disclosure conditions will activate PK, while the no disclosure condition will not.
Additionally, this study proposes that the effect of sponsorship disclosures on PK increases as the intensity of the disclosure increases. There is no literature available yet on the effects of disclosure intensities (i.e., simple disclosure vs. extensive disclosure) on brand responses. Nevertheless, a study showed that different disclosure timing, no disclosure
compared to three second and six second disclosure durations, resulted in different degrees of PK and consequently led to different brand responses. Namely, the study revealed that the six second disclosure condition was more likely to activate conceptual and attitudinal PK and that the six second disclosure led to more negative brand attitudes through higher levels of
9
Furthermore, studies have revealed that longer exposure to messages increases the likelihood to notice and to process the message-content (Buijzen, Van Reijmersdal, & Owen, 2010; Janiszewski, 1993). These findings make it plausible to expect that higher disclosure intensities would make the disclosure more noticeable and easier to process than less obvious disclosure types, such as simple ‘#ad’ statements. Moreover, the limited capacity model (Lang, 2000) may also be used to theorize that the effects of a disclosure may change in accordance with the intensity of disclosure. To process a message, individuals must have sufficient cognitive resources or allocate sufficient cognitive resources to the message. As a result of this, individuals can only process a maximum of information at the same time. In other words people have a limited capacity to process information (Boerman, Reijmersdal, et al., 2012; Buijzen et al., 2010; Lang, 2000).
When Instagram users are exposed to a post, they are exposed to a picture, multiple hashtag statements and the sponsorship disclosure simultaneously. The Instagram user has to allocate cognitive resources to multiple objects at the same time, as individuals have limited cognitive capacity, the disclosure might not be processed as there are insufficient cognitive resources to allocate to the disclosure. However, if the disclosure is more distinct (‘this post contains sponsored content’ together with ‘#ad’ rather than only ‘#ad’), the message-receiver is more likely to notice the disclosure and to allocate cognitive resources to it. Hence, this leads to the following hypothesis:
H1b. The effect of sponsorship disclosure on PK will increase with disclosure intensity; the extensive disclosure condition will have a higher effect on PK than the simple disclosure condition.
10
The Effects of Sponsorship Disclosures on Brand Responses
Sponsorship disclosures have a crucial role in the activation of PK. In turn PK affects the way consumers respond to persuasive messages. Specifically, this type of knowledge influences consumers in their decision to resist persuasion or not (Friestad & Wright, 1994; Van Reijmersdal et al., 2016). The reactance theory suggests that individuals respond
negatively to influence attempts, as they become aroused by threats or removal of behavioral freedom (i.e., manipulation of behaviour; Brehm, 1989). Moreover, resistance to persuasion is triggered by forewarning the persuasive intent of a message, as it motivates individuals to put up cognitive defense mechanisms (Petty & Cacioppo, 1977). The literature shows that
disclosures have negative effects on persuasion through the activation of PK (Boerman et al., 2012; Van Reijmersdal et al., 2016). Specifically, PK results in several resistance processes, such as refuting information, counterarguing and negative attitude formation towards messages (Boerman et al., 2014; Van Reijmersdal et al., 2016). Sponsorship disclosures in online blogs seem to affect perceptions of the brand and may result in negative effects on brand attitudes (Campbell et al., 2013; Liljander, Gummerus, & Söderlund, 2015; Van Reijmersdal et al., 2015).
The current literature on the effect of different types of sponsorship disclosures (simple disclosures vs. extensive disclosures) on cognitive and attitudinal responses is scarce. However, previous study has shown that disclosure timing on television is important, as placing disclosures after product placements rather than before placements may lead to lower brand attitude levels (Campbell et al., 2013). Moreover, another study has shown that a higher duration of disclosure on television results in more negative brand attitudes through higher levels of attitudinal PK (Boerman, Reijmersdal, et al., 2012).
11
H2a. Exposure to sponsorship disclosures (on Instagram) increases negative brand attitudes; this effect increases with the intensity of the sponsorship disclosure.
H2b. The effect of sponsorship disclosures on brand attitude is mediated by PK.
Moreover, previous studies have revealed that sponsorship disclosures also affect brand recall. Boerman et al. (2012) found that exposure to disclosures on television directly increases brand memory. In contrary, Campbell et al. (2013) found that brand placements without disclosure led to lower brand recall than placements with disclosure. The explanation for this is that individuals chose to exclude the brand from their memory to limit the
persuasive impact of the message (Campbell et al., 2013). However, the majority of the available literature on sponsorship disclosures and brand recall suggests that disclosures increase brand memory rather than affecting it negatively. A pioneer study on the positive effect of disclosures on brand memory, argued that disclosures serve as a prime and affect information processing accordingly (Bennett, Pecotich & Putrevu, 1999). Furthermore, other literature also showed that disclosures by brands lead to an increase in brand memory, because there is a higher emphasis on the sponsored content (Boerman, Reijmersdal, et al., 2012; Matthes & Naderer, 2016). Another study showed that brand memory particularly increases when the disclosures are shown during the product placement, rather than after the placement (Van Reijmersdal, Tutaj, & Boerman, 2013). A sponsorship disclosure on
Instagram would also be shown during the placement (or underneath the post), which could also result in an increase of brand recall. Based on previous findings, this study proposes the following hypothesis:
12
H3a. Exposure to sponsorship disclosures (on Instagram) increases brand recall; this effect increases with the intensity of the sponsorship disclosure.
Mathes and Naderer (2015) found that recall and recognition can be influenced positively by the activation of PK. Van Reijmersdal et al. (2015) also found that when PK is activated, message-receivers become alert of the brand’s intention, which increases brand memory. This leads to the following hypothesis:
H3b. The effect of sponsorship disclosures on brand recall is mediated by PK.
The activation of PK does not limit itself to influencing cognitive and affective responses, it also has an impact on behavioral intentions (van Reijmersdal et al., 2016). Several studies found that sponsorship disclosures have a negative effect on purchase intention (Liljander, Gummerus, & Söderlund, 2015; Van Reijmersdal et al., 2016).
Specifically, Liljander, Gummerus and Söderlund (2015) found that sponsorship disclosures in online blog posts have negative effects on the purchase intention of young adults.
Similarly, Van Reijmersdal et al. (2016) found that disclosing content in blogs has a negative effect on the purchase intention of consumers. Accordingly, the following hypothesis is proposed:
H4a. Exposure to sponsorship disclosures (on Instagram) decreases purchase intention; this effect increases with the intensity of the sponsorship disclosure.
The formation of a brand attitude precedes and must occur before an actual purchase can take place (Percy & Rossiter, 1992). As mentioned before, the sponsorship disclosure (or
13
revelation that the post is actually advertisement) is predicted to have a negative effect on brand attitude, and as a direct outcome this will affect purchase intention negatively as well (Colliander & Erlandsson, 2015). This leads to the following hypothesis:
H4b. The effect of sponsorship disclosures on purchase intention is mediated by brand attitude and brand recall.
The Moderating Effect of the Level of Product Involvement
Consumer attitudes and purchase intentions are not only influenced by brand traits or characteristics of advertisement (i.e., types of disclosure). Product types can also influence the relationship between a branding strategy and brand outcomes. The type of product, or product involvement, refers to the perceived relevance of an object, based on intrinsic needs, values and interests of an individual. Involvement specifically represents the cognitions and
subjective emotions which a consumer may have for a specific product (Zaichkowsky, 1994). For the purpose of this study, product involvement is conceptualized as the category of the advertised product, rather than looking at the variation in consumer levels of involvement with product classes (Dens & De Pelsmacker, 2010).
The literature on involvement specifies that involvement levels may determine the motivation or cognitive effort that consumers devote to the relevance and usefulness of a product or brand (Brown, Homer, & Inman, 1998; Zaichkowsky, 1985; Lee, Kim & Sundar, 2015). If involvement levels can influence the manner in which brands or products are
perceived, it might also influence brand evaluations and behavioral intentions. Previous study on branding strategies and the effects on brand attitudes and purchase intentions showed that there was an interaction effect between informational branding appeals, which scored better for high-involvement products, whereas emotional appeals were more effective for
low-14
involvement products (Dens & De Pelsmacker, 2010). An explanation for these results, could be that exposure to different product categories triggers different information processing routes. When a consumer is exposed to a high-involvement product, the person will engage in more elaborate cognitive processing than in a low-involvement situation, this in turn will allow for elevated attention towards the product ( Petty, Cacioppo, & Schumann, 1983; Zaichkowsky, 1985; Lee, Kim & Sundar, 2015).
The two levels of product involvement (low vs. high) tend to adhere to the processing routes of the Elaboration Likelihood Model (ELM) by Petty and Cacciopo (1986).
Specifically, exposure to low-involvement products tends to induce peripheral processing, while exposure to high-involvement products engenders central processing (Brown et al., 1998; Dens & De Pelsmacker, 2010). Furthermore, individuals are prone to respond more actively to advertisement that includes a high-involvement product (Lee, Kim & Sundar 2015). Even though there is no literature available on the possible interaction effect between disclosures and the level of product involvement, one could hypothesize that exposure to an ad with sponsorship disclosure and a high-involvement product could cause an intensified negative effect on brand responses. As the high-involvement product in the ad engenders central processing, which elevates attention for the ad and could make the sponsorship disclosures more easily accessible to process. These assumptions lead to the final hypothesis:
H5. Exposure to sponsorship disclosures on native advertising (on Instagram) lead to a stronger decrease in a) brand attitude and b) purchase intention when the advertisement includes a high-involvement product.
This study explores the model that is presented in Figure 1. The model proposes that sponsorship disclosures on Instagram have a negative effect on brand attitude and purchase
15
intention. The effect of sponsorship disclosures on purchase intention is mediated by brand attitude and brand recall. In addition, the expectation is that the effect of sponsorship disclosures on the brand responses is mediated by PK. Lastly, the effect of sponsorship disclosures on PK is moderated by product involvement.
16 Methods
Sample and Procedure
A 3 (sponsorship disclosure: no disclosure vs. ‘simple’ disclosure vs. ‘extensive’ disclosure) x 2 (level of product involvement: low vs. high) between subjects factorial design experiment was conducted amongst a total of 503 participants. An
experimental design was chosen, as it permits to test causal relationships between independent variables and dependent variables (’t Hart & Boeije, 2009). The independent variable of the present study was ‘sponsorship disclosure’, the second independent variable was ‘product involvement’ and the dependent variables were ‘brand recall’, ‘brand attitude’ and ‘purchase intention’. The participants were recruited via the social media network of the researcher by sharing a post with a link to an online survey and by requesting others to share the post to trigger a snowball effect. The majority (25.2 percent) of Instagram users range between 25 and 34 years of age, another 20.6 percent are between 18 and 24 years old, and 20.1 percent are between 35 and 44 years old (Statista, 2016). Even though a convenience sample was employed, the age distribution of Instagram users aligns with the age distribution of the social media network of the researcher, making it a rather realistic sample.
A total of 213 participants were excluded from the experiment, as they did not fully complete the online questionnaire. Moreover, one participant was removed from the data as the participant reported an implausible age (i.e., 101). The age of the remaining participants (N=292) ranged between 17 and 87 (M =32.87, SD =10.31). 65.5 percent of the participants were female, 32.8 percent were male and 1.7 percent were gender neutral or did not reveal their gender. 73.8 percent of the participants have an Instagram account and 70.3 percent actually use their Instagram account.
The participants were randomly assigned to one of the six different conditions, please see table 1 for specifications of the conditions. First, the participants were asked to give an
17
informed consent with regard to their participation in the online experiment. Hereafter, the participants were exposed to the actual stimuli. The participants were asked to take “a good look at the biography and Instagram post below” and that they would be asked questions about it later without being able to go back to the page. A 40-second timer was imposed on the page of the stimuli to prevent participants from directly clicking to the next page.
Thereafter, distraction questions were asked to make the cover story more credible and to enhance the eventual validity of the experiment. Then, the unaided brand recall question, followed by the aided brand recall question was posed. On the following page the correct product was revealed and the other dependent variables were measured in the following order; brand attitude, purchase intention, product involvement and PK. This order was applied deliberately, as the PK questions could have induced biased answers, which would harm internal validity. Specifically, participants that would not have noticed the persuasive intent of the message might have filled out the self-reported measures for brand responses more
negatively if confronted with a question on PK first. At the end questions on Instagram usage frequency and demographic questions were asked and the experiment was debriefed.
Table 1
The number of participants per experimental condition
Experimental condition Frequency
No disclosure x high-involvement 40 Simple disclosure x high-involvement 47 Extensive disclosure x high-involvement 45 No disclosure x low-involvement 45 Simple disclosure x low-involvement 64 Extensive disclosure x low-involvement 49
18 Stimuli
A fictional character, named Zoe Miller, was introduced to the participants by means of a short biography. A hypothetical character was used to enhance the internal validity of the study, as the prior liking or disliking of an existing influential could contaminate the results. The biography stated that Zoe Miller was a renowned model and lifestyle influencer who has over 200,000 followers on Instagram. Subsequently, the participants were presented to one of the six Instagram posts (see Appendix 1), which were supposedly posted by Miller. In each post, Miller was visible and having breakfast in a casual setting with a man who was framed as her boyfriend. The man was visibly holding a product, which will be specified later. This setting was chosen, as native advertising on Instagram is often incorporated in natural-looking photos to mask persuasive intentions. Moreover, the extra character was male rather than female because an ‘all-female picture’ might affect the brand responses of male participants.
The captions underneath the pictures were manipulated to measure the effects of disclosures on Instagram. For the no sponsorship disclosure conditions the captions solely stated “Sunday morning breakfast with my <3 *” together with the name of the brand (i.e., “#Choxphone”) and casual hashtag statements which were inserted in every condition (See Appendix 1). For the simple disclosure conditions a “#ad” statement was added in front of the brand hashtag. For the extensive disclosure conditions besides the “#ad” statement, the
following additional disclosure was inserted right underneath the picture: “This post contains sponsored content.”. The #ad statement is based on the social media regulations that were recently set by the FTC (FTC, 2015).
Besides from the manipulation of the first independent variable (sponsorship disclosure type), the use of a hypothetical influencer and a fictional brand also allowed to manipulate the second independent variable: the level of product involvement. The level of product involvement was manipulated by altering the product that was held by the man in the
19
picture. For the low-involvement conditions, this product was a smartphone. For the high-involvement conditions, the product was a chocolate bar. The products were inspired by Dens and De Pelsmacker (2010) who used a candy bar and a laptop in their study. In order to disable the influence of possible prior brand attitudes and to enhance the internal validity of the experiment, a fictional brand (Chox) was developed. Naturally, the brand hashtags were altered according to the product in the picture (i.e., “#Choxphone” or “#Choxchocolate”). Besides the different products and the brand hashtags, both photos and captions were kept identical to each other to enhance the internal validity of the experiment (’t Hart & Boeije, 2009).
Pretest
A pretest was conducted to verify if the stimuli were realistic and to test if the manipulations were effective. The participants (N=25) were selected from the direct environment of the researcher and were not included in the actual experiment. Participants were asked to fill out an online survey on Qualtrics. First, an open ended question was asked to check if the participants noticed anything peculiar with respect to the Instagram post. This question was inserted because part of the stimuli were professionally edited (i.e., the low-involvement product was photoshopped into the picture) to ensure high degrees of similarity between the different conditions. Some participants had negative comments (see Appendix 2) regarding the stimuli, but they did not notice the editing of the pictures.
Second, a manipulation check was performed to test if participants would consider the chocolate bar as a low-involvement product and the smartphone as a high-involvement product. To measure participant’s involvement levels, Mittals’s Involvement Scale (MIS; 1988) was applied. The MIS was used as a measure because it is considered to be reliable and relatively short (Emmert & Goldsmith, 1991; Mittal & Lee, 1988), which might cause less fatigue amongst the participants during the experiment. The participants were given the
20
instruction to answer four questions (see Appendix 3) on a scale from 1 (I do not care at all; they are alike; very important; I am not at all concerned) to 7(I care a great deal; they are all very different; very unimportant; I am very much concerned; M= 3.86, SD= 1.07). An independent samples t-test indicated that the participants’ involvement ratings in the low-involvement condition (M = 3.48, SD = .83) did not differ significantly from those exposed to the high-involvement condition (M = 4.27, SD = 1.18), t(23) = -1.94, p = .064). The
Levene’s t-test was not significant, so equal variances could be assumed, F=2.40, p= .135. Even though, the low-involvement product was not significantly rated lower than the high-involvement product, the stimuli were not altered because similar products have been used in previous studies successfully and the p value was relatively low.
Measures
Persuasion Knowledge. To measure the PK of the participants, the participants were presented to three statements (see Appendix 4) that were derived from previous studies on sponsorship disclosures and PK (Boerman et al., 2012; Van Reijmersdal et al., 2016; Van Reijmersdal, Smit, & Neijens, 2010). The participants were asked to rate these statements on a scale from 1 (strongly disagree) to 7 (strongly agree; Cronbach’s α = .91, M= 5.03, SD= 1.58).
Cognitive Resistance. To verify the PK measures, cognitive resistance was measured. The participants were asked to which extent they agreed with four statements (see Appendix 4) on a scale ranging from 1 (strongly disagree) to 7 (strongly agree; Cronbach’s α = .77, M= 3.93, SD= 1.11).
Product involvement. The same MIS scale (see Appendix 3) that was used as manipulation check was applied in the actual experiment. The scale proved to be reliable, Cronbach’s α =.73, M= 4.35, SD= 1.32.
21
Brand attitude. The attitude towards the brands was measured by using a 7-point semantic scale, which was based on previous studies on disclosures and brand responses (Boerman et al., 2012; Matthes, Schemer, & Wirth, 2007; Van Reijmersdal et al., 2015; Van Reijmersdal et al., 2016). The participants were asked to answer the questions (see Appendix 4) on 7-point scales (Cronbach’s α =.86, M= 3.35, SD= 1.02).
Brand recall. Unaided brand recall was measured by asking the following open question: “Which brand did you see in the Instagram post?”. The answers were recoded according to the correct (answers containing Chox, Choxphone or Choxchocolate) or incorrect identification of the brand, (0= Incorrect, 1=Correct; M=.14, SD=.35) and 14.1 percent mentioned the right brand.
Aided brand recall was measured by asking the same question, but by offering thirteen answer categories (see Appendix 4). Choxphone or Choxchocolate answers were recoded to 1 (correct), all other answers were recoded as 0 (Incorrect; M=.15 ,SD= .36; Rozendaal et al., 2013) and 14.8 percent mentioned the correct brand. Aided recall was verified after unaided recall to avoid data contamination (Walsh et al., 2014). Unaided recall (0=Incorrect,
1=Correct) and aided recall (0= Incorrect, 1=Correct) were combined into brand recall
(M=.15, SD=.28), which was constructed by computing the means of these two variables. 76.2 percent answered both questions wrong, 18.6 percent answered either aided brand recall or unaided brand recall correctly and 5.2 percent answered both questions correctly.
Purchase intention. Participant’s purchase intention was measured by four items (see Appendix 4) that were ranked on a 7-point Likert scale (1= Strongly disagree, 7=Strongly agree; Cronbach’s α = .90, M= 2.85, SD= 1.24; Baker & Churchill, 1977).
Control variables. Instagram usage was measured to ensure that the effects of sponsorship disclosure were not affected by other differences between the experimental
22
groups. Participants were asked if they owned an Instagram account, 73.8 percent answered “yes”, 26.2 percent answered “no”. Moreover, participants were asked to answer four questions (see Appendix 4) on a 7-point scale (1=Never, 7= Every day; Cronbach’s α=.89, M=3.09, SD=1.78).
Results Manipulation Check
An independent samples t-test indicated that participants’ involvement ratings in the low-involvement condition (M = 4.13, SD = 1.23) differed significantly from those exposed to the high-involvement product condition (M = 4.61, SD = 1.40), t(288) = -3.12, p =.002). The Levene’s t-test was not significant, so equal variances could be assumed, F=2.11, p= .148. This indicates that the product involvement manipulation was successful.
Randomization
To verify if random assignment to the different conditions was successful, first, an ANOVA with the conditions as independent variable and age as the dependent variable was conducted. There was no main effect of condition, F(5,284)= .13, p=.985, η2 = .00 on age, indicating that there was no difference between the conditions on age. Second, a Chi-square test was conducted, which showed that there were also no difference between the
experimental conditions with regard to whether the participant had an Instagram account or not, χ2 (5) = 3.09, p =.686. Lastly, another ANOVA was conducted to verify whether
participants’ Instagram usage differed between the conditions, this was not the case, F(5,284)= .74, p=.595, η2 =.01. Based on these analyses, one can conclude that
23 Analyses
The aforementioned conceptual model was split into several submodels for the purpose of analyses. Table 2 depicts the means and the standard deviations for the experimental conditions with respect to the dependent variables.
Table 2
Descriptive statistics of the conditions on the dependent variables. *p < .05.
Disclosure Type PK Brand Attitude Brand Recall Purchase Intention M SD M SD M SD M SD No disclosure condition (n=85) 4.68* 1.70 3.32 1.08 .13 .25 2.88 1.30 Simple disclosure condition (n=111) 5.23* 1.45 3.37 1.05 .11 .27 2.94 1.23 Extensive disclosure condition (n=94) 5.12 1.56 3.34 .93 .20 .31 2.72 1.19
The effect of sponsorship disclosures on PK. H1a predicted that the disclosure conditions would activate PK, while the no disclosure condition would not. H1b predicted that the extensive disclosure condition would have a higher effect on PK than the simple
disclosure condition. An ANOVA with disclosure as independent variable and PK as dependent variable revealed a significant main effect of s disclosure, F(2,287) = 3.23, p = .041, η2p = .02. This indicates that there is a significant difference between the no disclosure
condition (M = 4.68, SD = 1.70), the simple disclosure condition (n = 111, M = 5.23, SD = 1.45) and the extensive disclosure condition (n = 94, M = 5.12, SD = 1.56) with regard to PK.
24
condition and the extensive disclosure condition than in the no disclosure condition. Hence, H1a was supported. A Fisher Least Significant Difference (LSD) post hoc test was performed because it is considered to be a powerful and progressive method for effective analyses of three means (Hayter, 1986). The LSD indicated that two conditions differed significantly on PK. Participants in the simple disclosure condition (M=5.23, SD=1.45, p=.015) scored significantly higher on PK than participants in the no disclosure condition (M= 4.68, SD= 1.70, p=.015). However, participants that were in the extensive disclosure condition (M=5.12,
SD= 1.56, p= .058) did not score significantly higher on PK than those in the simple
disclosure condition (p=.626). Therefore, H1b was rejected.
The effect on brand responses and PK as a mediator. H2a stipulated that exposure to disclosures on Instagram would increase negative brand attitudes and that this effect would increase with disclosure intensity. An ANOVA with disclosure as independent variable and PK as dependent variable revealed that the main effect of disclosure type was not significant, F(2,287) =.04, p = .952, η2p = .00. This indicates that participants’ brand attitudes did not
differ significantly across the disclosure conditions. Thus, H2a was rejected.
H2b predicted that the effect of disclosures on brand attitude would be mediated by PK. Hayes’s (2013) model four for PROCESS, which estimates the path coefficients in a mediator model with 95 percent bootstrap confidence for total and indirect effects, was used to test mediations in this study. A total of 10,000 bootstrapped samples were applied to generate estimations of the bias corrected and accelerated confidence intervals (BCACI).
Sponsorship disclosure was assigned as independent variable (X), PK as mediating variable (M) and brand attitude as dependent variable (Y; see Appendix 5). Results for path a (X predicts M) revealed that the effect of disclosure on PK was not significant (b=.22,
SE=.12, p=0.066, 95% [ CI - 0.01, 0.45]). This suggests that within the mediation, the level
25
showed that the combined effect of disclosure and PK on brand attitude was not significant (p=.253). The separate effect of PK on brand attitude, path b, (b =.63, SE=.03, p=.098, 95% [ CI - .01, .13]) was not significant either. Path c’, indicating the extent to which the effect of X
on Y has been lessened was not significant either (b = -.01, SE=.07, p=.937, 95% [ CI - .16, .14]), indicating that PK’s effect on brand attitude did not decrease. The total effect of X on Y
(path c) was not significant (b= 0.01, SE=.08, p=.920, 95% [ CI - .14, .16]), suggesting that disclosure type did not affect brand attitude. These results show that the model has not met Baron and Kenny’s mediation criteria (1986), implying that PK does not mediate disclosure effects on brand attitude. Thus, H2b was rejected.
H3a predicted that exposure to disclosures would increase brand recall, this effect would increase with the intensity of disclosure. An ANOVA with sponsorship disclosure as independent variable and brand recall as dependent variable revealed that the main effect of sponsorship disclosure type was not significant, F(2,287) = 2.53, p = .081, η2p = .02. This
indicates participant’s brand recall did not differ significantly across the disclosure conditions. Thus, H3a was rejected. Interestingly, the LSD post hoc test showed a significant difference between the simple condition (M= .11, SD=.03) and the extensive disclosure condition (M=.20, SD=.03, p=.031). This aligns with the prediction that expanding from simple
disclosures to extensive disclosures enhances brand recall. However, the overall ANOVA results were not significant, making this assumption unreliable for conclusions.
H3B stated that the effect of disclosures on brand recall would be mediated by PK. The PROCESS results revealed that the effect of disclosure on PK was not significant (b = .22, SE=.12, p=.066, 95% [ CI - .01, .45]), implying that levels of disclosure did not predict
PK. However, the combined effect of disclosure and PK on brand recall was significant (p <0.001) and the separate effect of PK on brand recall was also significant (b = .04, SE=0.01,
26
was not significant (b= .03, SE=.02, p=.203, 95% [ CI - 0.01, 0.06]), suggesting that the effect of PK on brand recall has not lessened. The total effect of X on Y was not significant (b=.01, SE=.08, p=.920, 95% [ CI - .14, .16]), therefore disclosure did not affect brand recall. The model (see Appendix 5) did not meet the first criteria for mediation; X predicts Y (Baron & Kenny, 1986). Therefore, there was no mediation of the effect of disclosure on brand recall by PK. Thus, H3B was rejected.
H4a predicted that exposure to disclosures would decrease purchase intention and that this effect increases in line with disclosure intensity. An ANOVA revealed that the effect of sponsorship disclosure on purchase intention was not significant, F(2,287)=.80, p = .450, η2p
= .01, indicating that the purchase intention of participants did not differ significantly across
the disclosure conditions. Thus, H4a was rejected.
H4b suggested that the effect of disclosures on purchase intention would be mediated by brand attitude and brand recall. The PROCESS results for this parallel mediation revealed that the effect of sponsorship disclosure on brand attitude was not significant (b= 0.01, SE=0.08, p=0.920, 95% [ CI - 0.14, 0.16]). Moreover, the effect of sponsorship disclosures
on brand recall was not significant either (b= 0.03, SE=0.02, p=0.096, 95% [ CI - .01, .08]). This suggests that disclosures do not affect brand attitude and brand recall. Furthermore, the results revealed that the direct effect of sponsorship disclosure was not significant (b= - 0.07, SE = .08, p = .367 [95% CI - .22, .08) in predicting purchase intention. However, both the effect of brand attitude (b= 0.68, SE = 0.22, p = 0.000 [95% CI 0.55, 0.79) and the effect of brand recall (b= - 0.52, SE = 0.22, p = 0.018 [95% CI -0.96, -0.09) on purchase intention were significant. These results imply that brand attitude has a positive effect on purchase intention, while surprisingly brand recall has a negative effect on purchase intention. As only M1 and M2 proved to be predicting Y, the model (see Appendix 5) did not meet the mediation criteria (Baron & Kenny, 1986). Therefore, H4b was rejected.
27
Lastly, H5 predicted that exposure to disclosures on native advertising would lead to a stronger decrease in brand attitude and purchase intention when the advertisement would include a high-involvement product. A moderation analysis was conducted by means of a two-way ANOVA, which revealed that disclosure did not have a significant effect on brand attitude F(2,284) = 0.00, p = .998, while product involvement (high vs. low) did have a significant effect on brand attitude, F(1,284) = 5.30, p = 0.022. This suggests that brand attitude was higher in the low-involvement condition (M=3.47, SD= 0.96) than in the high-involvement condition (M=3.20, SD=1.07, p=0.022).However, the interaction of both these factors, F(2,284) = 2.08, p = 0.126, η2 = 0.01, did not have a significant effect on brand attitude.
Another two-way ANOVA revealed that the type of disclosure (no vs. simple vs. extensive) did not have a significant effect on purchase intention, F(2,284) = 0.68, p = .509, while product involvement (high vs. low) did have a significant effect on purchase intention, F(1,284) = 20.05, p < .001, η2p = .07. This suggests that purchase intention was higher in the
low-involvement condition (M=3.14, SD=1.22) than in the high-involvement condition (M=2.50, SD=1.17, p < .001). However, the interaction of these factors, F(2,284) = .66, p =
.515, η2p = .01, did not have a significant effect on purchase intention. Hence, H5 was
rejected.
Discussion
The current study examined the effects of different sponsorship disclosures types on brand attitude, brand recall and purchase intention and whether this relationship was mediated by PK. This study also examined if the effect of PK on purchase intention was mediated by brand recall and brand attitude. Additionally, the role of product involvement as a moderator in the relationship between disclosures and brand attitude or disclosures and purchase
28
First the study demonstrated that disclosures can activate PK. Specifically, a simple disclosure condition can induce higher levels of PK than a no-disclosure condition.
Surprisingly, extensive disclosures did not induce higher activation of PK than simple
disclosures. These findings partially oppose the assumptions of the PKM, which suggests that disclosures aid consumer to recognize the persuasive nature of sponsored content, which results in the activation of PK (Friestad & Wright, 1994). In contrary to the existing literature, the findings imply that not all forms of disclosure necessarily activate PK. An explanation for these contradicting results might be that participants failed to recognize the disclosure
because they were not familiar with the disclaimer in the extensive condition, and
familiarization is a crucial component for the activation of PK (Tessitore & Geuens, 2013). In contradiction with the hypothesized model, the different levels of disclosure did not affect brand recall, brand attitude or purchase intention. This opposes the findings from previous studies, where sponsorship disclosures in native advertisement resulted in higher degrees of brand recall (Boerman et al., 2012), lower brand attitudes and lower purchase intention than in no disclosure situations (Boerman et al., 2012; Van Reijmersdal et al., 2016). A possible explanation for these contradicting results might be the unknown brands that were applied in the stimuli. The use of fictional brands might have caused participants to have difficulty to evaluate the brands or to even recall the brands in the first place. Furthermore, fictional brands differ from actual brands and can affect the relationship between attitude formation and behavioral intentions (Stafford & Faber, 2005). Future studies could overcome this limitation, by using real brands.
Additionally, the findings suggested that the relationship between the level of sponsorship disclosure and brand responses (brand attitude and brand recall) was not mediated by PK. These findings oppose the existing literature, where PK proves to be a mediator, as the activation of PK leads to more critical processing of brand information and
29
leads to the formation of resistance strategies (Boerman et al., 2012; Van Reijmersdal et al., 2016). A possible explanation would again be the use of fictional brands, as mentioned before PK was activated by the sponsorship disclosures, however the little prior knowledge with respect to the brands could have made it difficult for participants to evaluate the brands or recall them.
Subsequently, the effect of PK on purchase intention was not mediated by brand attitude and brand recall either. In a sense this is not surprising, as the present study did not find an effect of PK on brand attitude and brand recall, however, this opposes the theory of reasoned action and the theory of planned behavior which suggests that behavioral intentions of individuals are predicted by brand attitude (Fishbein & Ajzen, 1975). It is important to note that although the mediation was not significant, a positive effect of brand attitude and a negative effect of brand recall was found on purchase intention. The effect of brand attitude on purchase intention adheres to the above mentioned theories of Fishbein and Ajzen (1975). The negative relation between brand recall and purchase intention is surprising, as one would expect that recall of a brand would have a positive effect on purchase intention (Senthilnathan & Tharmi, 2012). However, as the mediation did not prove to be significant, these result should be interpreted cautiously.
Finally, this study examined the interaction between sponsorship disclosure and level of product involvement and its effect on brand attitude and purchase intention. Although participants who were assigned to the chocolate bar condition indicated to be lowly involved, and participants who were assigned to the smartphone condition indicated to be highly involved, the findings showed no moderating role of product involvement in the
aforementioned relationship. In contradiction to the prediction of this study, which was based on reasoning that there would be elevated attention for ads that include high-involvement products (Petty, Cacioppo, & Schumann, 1983; Zaichkowsky, 1985; Lee, Kim & Sundar,
30
2015). A possible explanation for the different findings could be the distracting elements that were applied in the stimuli, which were reported as distracting in many comments (See Appendix 6). As every aspect in the pictures, except for the product, was kept identical across conditions, it could be possible that the attention levels for the product in the low and high-involvement condition were similar (i.e., low) because too many cognitive resources were allocated to distracting features of the stimuli.
Practical Implications
The findings of this study extend on previous research on the effect of sponsorship disclosures on the activation of PK and its effect on brand responses. The findings that sponsorship disclosures activate PK were reproduced ( Van Reijmersdal et al., 2016). However, the activation of PK did not apply to the most extensive type of disclosure.
The findings of the present study have significant implications for future applications of sponsorship disclosures on Instagram, as the literature on the effect of different intensities of disclosures on the activation of PK is limited. The findings of the present study indicate that simple disclosures induce a higher activation of PK than no disclosures. However, it does not seem to make a difference when the sponsorship disclosures is becoming more extensive as the activation of PK is not increased compared to a simple disclosure situation. This finding is relevant for marketers, who try to circumvent the activation of PK, and law enforcers who strive to protect the rights of online consumers. As only sponsorship
disclosures that are recognized by consumers will have the ability to activate PK (Tessitore & Geuens, 2013).
Limitations and Implications for Future Research
The present study has several limitations. First, it should be noted that the percentage of participants that recalled the brand from the Instagram post was very low, as only 18.6 percent answered either brand recall or recognition correctly and 5.2 percent answered both
31
questions correctly. This could suggest that people were distracted during the experiment, either by external factors or internal factors of the experiment, such as characteristics of the stimuli. Future research should ensure that the Instagram picture does not contain too many unnecessary distracting features (see Appendix 6) as individuals have a limited capacity to store information (Lang, 2000). Another recommendation would be to include the brand name or logo in the picture, as mentioning the brand in the hashtag statements does not seem to suffice.
Another important limitation of the present study was the use of a fictive brand, as many participants responded by stating that they did not know the brand, therefore they could not make any evaluations on the brand. This might be an explanation for the insignificant results with regard to the effect of sponsorship disclosures on brand attitude and purchase intention. A follow-up study could include an existing brand in the Instagram posts.
Furthermore, the hashtag statement in the extended disclosure condition was less inclined to activate PK than in the other conditions (note: this finding was not significant). This contradicts the existing literature on sponsorship disclosures and PK and it could indicate that other hashtag statements and texts should be tested in pilot studies and used for future research. Future research could also incorporate a training component, where participants are familiarized with the symbol that stands for the sponsorship disclosure. This familiarization proved to be an important component of the activation of PK in previous research (Tessitore & Geuens, 2013).
Conclusion
The present study sheds light on the importance of the use of sponsorship disclosures on Instagram. The findings indicate that simple disclosures induce a higher activation of PK than no disclosures. However, it does not seem to make a difference when the sponsorship disclosures is becoming more extensive as the activation of PK is not increased compared to a
32
simple disclosure situation. In summary, the findings suggest that sponsorship disclosures have the ability to activate PK. However, further research is necessary with respect to the types (symbols, sizes, positioning) of sponsorship disclosures on Instagram that are most effective in activating PK.
33 References
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of personality and social psychology, 51(6), 1173.
Baker, M. J., & Churchill, , Gilbert A. (1977). The impact of physically attractive models on advertising evaluations. Journal of Marketing Research, 14(4), 538
555.doi:10.2307/3151194
Bennett, M., Pecotich, A., & Putrevu, S. (1999). The influence of warnings on product placements. In B. Dubois, T. M. Lowrey, L. J. Shrum, & M. Vanhuele (Eds.), European advances in consumer research (4th ed., pp. 193–200). Provo, UT: Association for Consumer Research.
Boeije, H., & ’t Hart, H. (2009). In H. Boeije, H. ‟t Hart, & J. Hox (Red.), Onderzoeksmethoden (pp.170-173). Amsterdam: Boom.
Boerman, S. C., Reijmersdal, E. A., & Neijens, P. C. (2012). Sponsorship Disclosure: Effects of Duration on PK and Brand Responses. Journal of
Communication, 62(6), 1047–1064. https://doi.org/10.1111/j.1460-2466.2012.01677.x
Boerman, S. C., van Reijmersdal, E. A., & Neijens, P. C. (2012). Sponsorship Disclosure: Effects of Duration on PK and Brand Responses. Journal of
Communication, 62(6), 1047–1064. https://doi.org/10.1111/j.1460-2466.2012.01677.x
Brehm, J. W. (1989). Psychological Reactance: Theory and Applications. Advances in Consumer Research, 16(1), 72–75. https://doi.org/10.1007/BF00290976
Brown, S. P., Homer , P. M. and Inman, J.J. (1998). A meta-analysis of relationships between ad-evoked feelings and advertising responses . Journal of Marketing
34 Research 35(1), 114 – 126 .
Buijzen, M., Van Reijmersdal, E. A., & Owen, L. H. (2010). Introducing the PCMC model: An investigative framework for young people’s processing of commercialized media content. Communication Theory, 20(4), 427–450. https://doi.org/10.1111/j.1468- 2885.2010.01370.x
Cacioppo, J. T., & Petty, R. E. (1983). the Elaboration Likelihood Model of Persuasion. Advances in Consumer Research, 11(1), 673–675.
https://doi.org/10.1558/ijsll.v14i2.309
Campbell, C., & Marks, L. J. (2015). Good native advertising isn’t a secret. Business Horizons, 58(6), 599–606. https://doi.org/10.1016/j.bushor.2015.06.003
Campbell, M. C., Mohr, G. S., & Verlegh, P. W. J. (2013). Can disclosures lead consumers to resist covert persuasion? The important roles of disclosure timing and type of
response. Journal of Consumer Psychology, 23(4), 483–495. https://doi.org/10.1016/j.jcps.2012.10.012
Carah, N., & Shaul, M. (2016). Brands and Instagram: Point, tap, swipe, glance. Mobile Media & Communication, 4(1), 69–84. https://doi.org/10.1177/2050157915598180
Colliander, J., & Erlandsson, S. (2015). The blog and the bountiful: Exploring the effects of disguised product placement on blogs that are revealed by a third party. Journal of Marketing Communications. Taylor & Francis.
Cotton Delo. (2014). After careful test, Instagram takes its pitch to advertisers. Advertising Age, 85(7), 0008.
Dekker, K., & van Reijmersdal, E. A. (2013). Disclosing Celebrity Endorsement in a Television Program to Mitigate Persuasion: How Disclosure Type and Celebrity
35
Credibility Interact. Journal of Promotion Management, 19(1), 224–240. https://doi.org/10.1080/10496491.2013.769473
Dens, N., & De Pelsmacker, P. (2010). Consumer response to different advertising appeals for new products: The moderating influence of branding strategy and product
involvement. Journal of Brand Management, 18(1), 50–65. https://doi.org/10.1057/bm.2010.22
Evans, N. J., & Park, D. (2015). Rethinking the PK Model: Schematic Antecedents and Associative Outcomes of PK Activation for
Covert Advertising. Journal of Current Issues & Research in Advertising (Routledge), 36(2), 157–176. https://doi.org/10.1080/10641734.2015.1023873
Fishbein, M., & Ajzen, I. (1975). Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research. Reading, MA: Addison-Wesley Publishing Co.
Friestad, M., & Wright, P. (1994). The People Persuasion Cope with Knowledge Persuasion Model : How Attempts. Journal of consumer research, 21(1), 1-31.
https://doi.org/10.1086/209380
FTC, (2015b). Native Advertising: A Guide for Businesses | Federal Trade Commission. Ftc.gov. Available: https://www.ftc.gov/tips-advice/businesscenter/guidance/native- advertising-guide-businesses, Last accessed on the 10th of March.
Goldsmith, R.E, & Emmert, J. (1991). Measuring product involvement: A multitrait- multimethod study. Journal of Business Research, 23(4), 363-371.
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. Guilford Press.
36
Difference Test. Journal of the American Statistical Association,81(396), 1000-1004.
Janiszewski, C. (1993). Preattentive Mere Exposure Effects. Journal of Consumer Research, 20(3), 376–392. https://doi.org/10.1016/0148-2963(91)90021-O.
Lang, A. (2000). The limited capacity model of mediated message processing. Journal of Communication, 50(1), 46–70. https://doi.org/10.1111/j.1460-2466.2000.tb02833.x
Lee, S., Kim, K. J., & Sundar, S. S. (2015). Customization in location-based advertising: Effects of tailoring source, locational congruity, and product involvement on ad attitudes. Computers in Human Behavior, 51, 336-343. doi:10.1016/j.chb.2015.04.049
Liljander, V., Gummerus, J., & Söderlund, M. (2015). Young consumers’ responses to suspected covert and overt blog marketing. Internet Research, 25(4), 610–632. https://doi.org/10.1108/IntR-02-2014-0041.
Matthes, J., & Naderer, B. (2016). Product placement disclosures: Exploring the moderating effect of placement frequency on brand responses via PK.
International Journal of Advertising, 35(2), 185–199. https://doi.org/10.1080/02650487.2015.1071947
Matthes, J., Schemer, C., & Wirth, W. 2007. More than meets the eye: Investigating the hidden impact of brand placements in television magazines. International Journal of
Advertising, 26(4), 477–503.
Mittal, B., & Lee, M.-S. (1988). Separating Brand-Choice Involvement From Product Involvement Via Consumer Involvement Profiles. Advances in Consumer Research, 15(1), 43–49.
37
Video News Releases: Audience Beliefs About News and Support for Source Disclosure. Journal of Mass Media Ethics, 24(4), 220–237.
https://doi.org/10.1080/08900520903332626
Percy, L., & Rossiter, J. R. (1992). A model of brand awareness and brand attitude advertising strategies. Psychology & Marketing, 9(4), 263–274.
https://doi.org/10.1002/mar.4220090402.
Petty, R. E., & Cacioppo, J. T. (1977). Forewarning, cognitive responding, and resistance to persuasion. Journal of Personality and Social Psychology, 35(9), 645–655.
https://doi.org/10.1037/0022-3514.35.9.645.
Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Central and peripheral routes to advertising effectiveness: The moderating role of involvement. Journal of Consumer Research, 10(2), 135.
Rozendaal, E., Slot, N., Van Reijmersdal, E., & Buijzen, M. (2013). Children's responses to advertising in social Games. Journal of Advertising, 42(2-3), 142-154.
Sheldon, P., & Bryant, K. (2016). Instagram: Motives for its use and relationship to narcissism and contextual age. Computers in Human Behavior, 58, 89–97. https://doi.org/10.1016/j.chb.2015.12.059.
Stafford, M. R., & Faber, R. J. (2005). Advertising, promotion, and new media. Armonk, NY: M.E. Sharpe.
Statista. (2017). Number of Monthly Active Instagram Users. Available:
https://www.statista.com/statistics/253577/number-of-monthly-active-instagram-users/ Last accessed 15th of March 2017.
38
Senthilnathan, S., & Tharmi, U. (2012). The relationship of brand equity to purchase intention. The IUP Journal of Marketing Management : IJMM,11(2), 7-26.
Tessitore, T., & Geuens, M. (2013). PP for “product placement” or “puzzled public”? The effectiveness of symbols as warnings of product placement and the moderating role of brand recall. International Journal of Advertising, 32(3).
The Economist. (2016). Gold Posts. Available:
http://www.economist.com/news/business/21708717-social-media-endorsements-are- latest-thing-advertising-gold-posts. Last accessed 10th of March 2016.
Van Reijmersdal, E. A., Fransen, M. L., van Noort, G., Opree, S. J., Vandeberg, L., Reusch, S., … Boerman, S. C. (2016a). Effects of Disclosing Sponsored Content in Blogs: How the Use of Resistance Strategies Mediates Effects on Persuasion. American Behavioral Scientist, 60(12), 1458–1474. https://doi.org/10.1177/0002764216660141.
Van Reijmersdal, E. A., Fransen, M. L., van Noort, G., Opree, S. J., Vandeberg, L., Reusch, S., … Boerman, S. C. (2016b). Effects of Disclosing Sponsored Content in Blogs: How the Use of Resistance Strategies Mediates Effects on Persuasion. American Behavioral Scientist, 60(12), 1458–1474. https://doi.org/10.1177/0002764216660141
Van Reijmersdal, E. A., Lammers, N., Rozendaal, E., & Buijzen B, M. (2015). Disclosing the persuasive nature of advergames: Moderation effects of mood on brand responses via PK. International Journal of Advertising, 34(1), 70–84.
https://doi.org/10.1080/02650487.2014.993795.
Van Reijmersdal, E. A., Tutaj, K., & Boerman, S. C. (2013). The effects of brand placement disclosures on skepticism and brand memory. Communications, 38(2), 127–146. https://doi.org/10.1515/commun-2013-0008.
39
Van Reijmersdal, E., Smit, E., & Neijens, P. (2010). How media factors affect audience responses to brand placement. International Journal of Advertising, 29(2), 279–302. https://doi.org/10.2501/S0265048710201154.
Walsh, P., Clavio, G., Mullane, S., & Whisenant, W. (2014). Brand awareness and attitudes towards political advertisements in sport video games. Public Organization Review, 14(2), 127-138.
Wei, M.-L., Fischer, E., & Main, K. J. (2008). An Examination of the Effects of Activating PK on Consumer Response to Brands Engaging in Covert
Marketing. Journal of Public Policy & Marketing, 27(1), 34–44. https://doi.org/10.1509/jppm.27.1.34.
Wojdynski, B. W. (2016). The Deceptiveness of Sponsored News Articles: How Readers Recognize and Perceive Native Advertising. American Behavioral Scientist, 60(12), 1475–1491. https://doi.org/10.1177/0002764216660140
Wojdynski, B. W., & Evans, N. J. (2015). Going native: Effects of disclosure position and language on the recognition and evaluation of online native advertising. Journal of Advertising, 3367(April), 1–12. https://doi.org/10.1080/00913367.2015.1115380
Wojdynski, B. W., & Evans, N. J. (2016). Going Native: Effects of Disclosure Position and Language on the Recognition and Evaluation of Online Native Advertising. Journal of Advertising, 45(2), 157–168. https://doi.org/10.1080/00913367.2015.1115380
Zaichkowsky, J. L. (1985). Measuring the involvement construct. Journal of Consumer Research. JCR: an Interdisciplinary Quarterly, 12(3), 341-352.
Zaichkowsky, J. L. (1994). The Personal Involvement Inventory: Reduction, Revision, and Application to Advertising. Journal of Advertising, 23(4), 59–70.
40
Appendix 1: Stimuli
Condition 1: No disclosure x high-involvement.
Condition 2: Simple disclosure x high-involvement.
Condition 3: Extensive disclosure x high-involvement.
41
Figure 3. Stimulus material for the low-involvement conditions. Condition 4: No disclosure x
low-involvement.
Condition 5: Simple disclosure x low-involvement.
Condition 6: Extensive disclosure x low-involvement.
42
Appendix 2: Skeptic Comments from Pretest
Question: “Did you notice anything peculiar about the Instagram post?” Answer A: “Peculiar Hashtag choxphone.”
Answer B: “It seems a bit fake/unnatural.”
43
Appendix 3: MIS Scale
MIS Scale (Emmert & Goldsmith, 1991; Mittal & Lee, 1988)
(MIS1) “When you buy this product in a store, to which extent do you care about the purchase decision?”
(MIS2) “Do you think that the different brands of this product that are available in the market are all very alike or all very different?”
(MIS3) “How important is it to you to make the ‘right’ (e.g. fair price-quality ratio, high quality product) purchase decision of this product?”
(MIS4) “When you purchase this product, how concerned are you with the consequences of this purchase decision?”
