Graduate School of Communication
Faculty of Social and Behavioural Sciences
Master Communication Science – Persuasive Communication
Master Thesis
Influencer marketing: disclosure of branded content on Instagram
Name: Caitlin Verhoeven Student number: 11401540 Supervisor: Sandra Zwier 2nd February 2018
Abstract
The rapid growth of advertising online on social media has led to the arrival of ‘influencer marketing’, especially on Instagram. However, critics are concerned that this new form of marketing is deceptive and does not guarantee fair communication to consumers, as the commercial motive behind the message may not be recognizable as such. Therefore,
regulations require influencers to disclose the commercial purpose of a message, in order to make consumers more aware of this hidden advertising. Nonetheless, since regulations are still developing in Europe and not prescriptive in how disclosure should be effectively
communicated, disclosure of that the content is branded may remain unnoticed. Therefore, the present study investigated the presence of branded content and its different types of disclosure (implicit and explicit disclosure) of branded content on Instagram accounts of Dutch
influencers. The aim was to provide regulators with a better understanding of how the commercial purpose is currently articulated, in order to provide a basis for further
improvements of guidelines regarding the requirements of disclosures. Moreover, the effects of disclosure of branded content were measured on user engagement (likes and comments). The results revealed that branded content was present in large numbers. Moreover, branded content was implicitly disclosed mainly in the form of a tag in the image of the Instagram post. However, explicit disclosure was hardly ever used, meaning that regulations were not adhered and the consumers’ right to know the message has a persuasive intent was violated. Finally, effects of disclosure of branded content on user engagement were not found.
Nonetheless, the results contribute to the current gap in literature on influencer marketing, and suggest some important practical implications for regulators, marketers and influencers,
Keywords: influencer marketing, branded content, Instagram, brand prominence, disclosure, user engagement.
Introduction
The rapid growth of advertising online, especially on social media, has forced marketers to think of new online marketing activities. Social media being overloaded with content nowadays makes it difficult for brands to effectively reach their target group. Therefore, so-called ‘influencer marketing’ has gained popularity among social media marketers
(Jaakonmäki, Müller & Vom Brocke, 2017). To reach a wide audience within online
communities, brands recruit social influencers that receive (in)direct compensation (money or free products) in return for promoting products on social media (Boerman & Van
Reijmersdal, 2016; Brown & Fiorella, 2013). Instagram is one of the main social media platforms on which influencer marketing is employed. Having over 700 hundred million users, and an alleged capacity to trigger 58 times more user engagement than Facebook and 120 times more than Twitter (Tyagi, 2017), the platform has caught the attention of
marketers.
Notwithstanding the increasing popularity of influencer marketing, critics argue that influencer marketing is unethical and deceptive; brands or branded persuasive messages are integrated into editorial content in creative ways (Maheshwari, 2016), whereby the content may not disclose whether it was created for value exchange. To help consumers distinguish commercial content from non-commercial editorial content, regulators such as The Federal Trade Commission (FTC, United States), the Advertising Standards Authority (ASA, United Kingdom) and The Dutch Advertising Code (NRC) have installed guidelines and codes of conduct instructing influencers to disclose the commercial purpose of a message. However, these guidelines are still developing in Europe, which leaves influencers in a big grey area on how to articulate the disclosure in a transparent way. Thus, even when influencers comply with social media advertising guidelines (FTC, 2015; NRC, 2014), a disclosure may remain
of the message (Boerman, Van Reijmersdal, & Neijens, 2012; Boerman, Willemsen, & Van der Aa, 2017).
Therefore, the present study aims to provide regulators with a better understanding of different types of disclosure of branded content that are currently used on social media, in order to provide a basis for further improvements of guidelines regarding the requirements of disclosures. Clear guidelines of disclosures of branded content are helpful for consumers, marketers and influencers alike: making disclosure more noticeable, consumers will be more aware of the branded content. Moreover, marketers and influencers can hereby develop an effective and persuasive communication of branded content without being deceptive.
Although extant research has shown increased interest in effects of disclosure on consumer responses on online blogs (Boerman et. al., 2012; Hwang & Jong, 2006) or Facebook (Boerman et. al., 2017), to our knowledge no previous study has quantitatively investigated whether similar effects occur for branded content and different types of
disclosure on Instagram. Therefore, the present study involved a content analysis of Instagram posts of Dutch influencers. The presence of branded content on Instagram will be analysed through visual features (format and prominence). Next to that that, different types of
disclosure will be analysed through visual and textual features of Instagram posts. Lastly, this study will explore how these different factors of influencer marketing drive user engagement with the Instagram posts.
RQ: How and to what extent is branded content present and disclosed in Instagram posts of Dutch influencers, and how does this affect user engagement?
In the following section, existing literature will be discussed to describe the state of the art of research into branded content on social media and disclosures thereof. Subsequently, a
detailed description of the content analysis of Instagram posts conducted for this study is provided. Finally, the results will be reported, and reflection on the main findings is given in the conclusion and discussion.
Theoretical background
In this section, the theoretical background and main concepts of this study are discussed. This automatically leads to formulation of the hypotheses.
Influencer marketing
Influencer marketing is a renewed form of online celebrity endorsement, where brands indirectly communicate with consumers not by using well-known persons, but social media personalities (Einarsdóttir, 2017). Influencers are chosen to promote a product, aiming to influence the target group’s behaviours or opinions towards this product (Lu, Chang & Chang, 2014).
Influencer marketing is assumed to be effective, being a form of ‘covert’ marketing, masking the commercial source and message (Petty & Andrews, 2008). Therefore, this form is more likely perceived as non-marketing interaction and hence avoids consumer disinterest and scepticism towards traditional marketing messages (Petty & Andrews, 2008). Moreover, brands transfer and capitalize social trust in information coming from consumers by using influencers. Messages transmitted from one individual to another, rather than information coming from brands, are perceived as more reliable and trustworthy (Belch & Belch, 2015; Boerman, et. al., 2017). Hence, consumers are more likely to positively engage with a message received from an influencer as trusted source (Boerman et. al., 2017; Chu & Kim,
2011; Hwang & Jeong, 2016), such as the sharing of product-related information, and opinions, attitudes and purchase and post-purchase experiences (Jaakonmäki et. al., 2017).
Positive effects of influencer marketing are re-enforced by the principles of social influence formulated by Cialdini (1987). One of those is ‘liking’: people are more influenced by people they like. ‘Liking’ on social media compares to ‘following’. If users ‘like’ an influencer, they are interested in the content of his/ her posts and hence more likely to be persuaded by information from this influencer.
Another social influence principle is ‘authority’. Influencers are perceived as role models (Einarsdóttir, 2017), in fashion, beauty, health, travel etc. and therefore have authority in the eyes of target audiences. According to Cialdini (1987), this causes one to feel
obligation towards the influencer to dress like them, to eat as healthy or to travel to places where the influencer went.
Due to high connectedness, influencer marketing is a relatively cost-effective technique to reach a target group. Furthermore, using authority figures could reach target groups with specific interests, lifestyles or demographic characteristics (Belch & Belch, 2015). And when these groups are reached, they are also more likely to be influenced due to the ‘liking’ of influencers and their authority status. In conclusion, influencer marketing is widely seen as an effective strategy to promote products by actively creating branded content, perceived as less obtrusive and more trustworthy than traditional advertising.
Despite proven positive effects of influencer marketing from the perspective of the brands (eMarketer, 2015; Jaakonmäki et. al., 2017; Kirkpatrick, 2016; Millwood, 2016), potential ethical and legal implications apply when it comes to the consumer (Woods, 2016). A potentially unethical implication of this marketing form is that users cannot always
recognize branded content (Maheswari, 2016), since the commercial source and message are masked. Branded content often looks like content posted by friends or other users (Boerman
et. al., 2017). Moreover, expressed feelings and experiences of social influencer could be inauthentic and unreal (Van Reijmersdal, Fransen, Van Noort, Opree, Vandeberg, Reusch, Van Lieshout, & Boerman, 2016). Consumers could thereby be misled, believing the branded content was solely made by genuine enthusiasm for the product (Petty & Andrews, 2008), while the real motivation is most likely monetary compensation. Influencer marketing can therefore be deceptive for consumers. Information whether a message is branded may be absent and may violate consumers’ right to know the commercial motive behind the message (Boerman et. al., 2017).
So far, the important role of influencer marketing in present-day social media content was discussed, as well as ethical considerations. The following sections will address features of branded content in Instagram posts, such as disclosure and the prominence of branded content, and how these features affect signals of user engagement.
Disclosure of branded content
GuidelinesIn the European Union and United States disclosure of branded content in television, movies and on radio is obligatory (Cain, 2011). For social media and online blogs, this is restricted to guidelines and codes of conduct instructing that commercial content on social media should always be recognizable as such (Boerman et. al., 2017). Moreover, the relation with the brand must be described in a clear and conspicuous manner (FTC, 2015a), including that an online influencers has been paid or given something of value in return for promoting a product (Maheshwari, 2016).
On Instagram recommendations for influencers to disclose is using hash tags like #ad, #sp, #spon and #partner. The FTC recently however concluded that these practices were
brands (Spangler, 2017). Influencers should more “explicitly inform the audience when commercial content is integrated into editorial content to guarantee fair communication and avoid persuasion without audience awareness” (Boerman et. al., 2012, p. 1048). Influencers can employ this by both tagging the brand’s Instagram account in the description and using hash tags like #ad (Mediakix Team, 2016). On top of that, Instagram has followed Facebook with a similar tool of adding a “Paid partnership with [Brand name]” tag in June 2017 to increase explicitness and transparency (Stewart, 2017).
However, the regulations regarding disclosure of branded content on social media are still developing in Europe and the “paid partnership” social media tool is still in the early adoption phase. Therefore, the present study will analyse how and to what extent branded content is actually disclosed in Dutch influencers’ Instagram posts. The present study makes a distinction between two main types of disclosures.
First, ‘implicit disclosure’, in which influencers implicitly inform users about the presence of branded content. Here, a third party’s brand name, product, package or logo is clearly visibly or audibly retrievable, or tagged in the pictorial image of the Instagram post (Kulin & Blomgren, 2016). Nevertheless, the relationship between the influencer and the brand is not directly clear, while in both cases this is more general disclosure of potential outside influence without specifically disclosing by what third party the influencer was compensated (Carr & Hayes, 2014).
Secondly, ‘explicit disclosure’, in which influencers explicitly inform users about the presence of branded content. The disclosure is given in the accompanying text of the
Instagram posts with the following options: hash tags with brand name; brand in location tag; brand named or tagged in description; hash tag about the sponsorship (#ad; #spon); paid sponsorship tag. The more influencers explicitly disclose information on the brand, the clearer the content is branded.
It can be expected that Dutch influencers use implicit disclosure more often than explicit disclosures. The reason for this is that when branded content is present, there is always some sort of implicit disclosure present in the image of the Instagram post, or otherwise the branded message would not reach consumers. Second, it can be expected that when Dutch influencers do explicitly disclose branded content is present, they will only use one type of explicit disclosure. This expectation is based on the fact that social media
guidelines are underdeveloped in Europe, which is making that Dutch influencers hardly ever use explicit disclosures in their Instagram posts and when they use it they are not familiar with the guidelines on how to preferably disclose branded content.
H1a: Implicit disclosure is used more often than explicit disclosure in Instagram posts of Dutch influencers.
H1b: Dutch influencers explicitly disclosing branded content present in their Instagram post, use only one type of explicit disclosure.
Consumer responses to disclosure of branded content
The present section addresses the potentially adverse effects of disclosing branded content on consumer responses. Multiple studies have shown that disclosing branded content might affect how consumers process branded content and respond to that. The persuasion
knowledge model of Friestad and Wright (1994) could explain the underlying mechanisms of responses to disclosing branded content.
The persuasion knowledge model explains how people develop knowledge about persuasion tactics over time, and use this to interpret, evaluate and respond to persuasive
people want to have the freedom to enact their behaviours; they don’t like to hear and do what other people tell them (Burgoon, Alvaro, Grandpre & Voulodakis, 2002). Since persuasive communication is often an offensive/ directive/ demanding attempt to change attitudes and behaviours of people, it might be perceived as threatening or eliminating people’s freedom (Burgoon et. al., 2002). As result, people activate cognitive and affective resistance strategies and develop psychological reactance: a negative, aversive motivational state, to cope with and resist the unwanted persuasion attempt (Van Reijmersdal et. al., 2016).
Persuasion knowledge is believed to consist of two dimensions: cognitive and affective (Boerman et. al., 2012). The cognitive dimension is the first step of persuasion knowledge, in which consumers come to recognize information as an advertisement. Sponsorship disclosures thereby function as a warning tool to help consumers recognize branded content as a persuasion attempt and therefore activate the cognitive dimension of persuasion knowledge (Boerman et. al., 2012; Boerman et. al., 2017; Boerman & Van Reijmersdal, 2016; Van Reijmersdal et. al., 2016). For the present research, disclosures on Instagram are assumed to help recognize branded content.
The affective dimension of persuasion knowledge refers to emotional responses, such as feelings as dislike, distrust or scepticism, when confronted with an advertising message (Boerman et. al., 2012; Van Reijmersdal et. al., 2016). These feelings and critical evaluations concern the appropriateness and fairness of the message and its tactics, after the consumer already activated the cognitive dimension of persuasion knowledge (Boerman et. al., 2017; Friestad & Wright, 1994). Disclosure has been shown to have negative effects on the attitude towards the branded content through its recognition as a persuasion attempt (Van Reijmersdal et. al., 2016).
According to the theory of planned behaviour, attitude is directly linked to intention and actual behaviour (Fishbein & Azjen, 1975). Meaning that unwanted persuasive messages
could evoke critical and distrusting beliefs, and negatively influence intention to engage with the message and in turn the behaviour to engage with the message. In the present study the behavioural manifestation of engagement is user engagement: the number of individuals that respond to the online marketing actions, or the number of individuals who participate in electronic word-of-mouth (eWOM) (Boerman et. al., 2017). It is calculated by the percentage of people who respond to a post in some way, including “liking” and commenting
(Jaakonmäki et. al., 2017).
As disclosure has been shown to have negative effects on the attitude towards the branded content through the cognitive dimension of persuasion knowledge (Van Reijmersdal et. al., 2016), it is assumed disclosure also negatively influence intentions to engage in
eWOM with a branded message. Boerman et. al. (2017) have proved this effect on Facebook. Moreover, earlier research showed that type of disclosure has differential effects on intentions to engage in eWOM through the credibility of the influencer (Carr & Hayes, 2014; Chu & Kim, 2011). In the study of Carr and Hayes (2014), in the implicit disclosure type the influencer was perceived as least credible, negatively influencing intentions to engage in eWOM. On the other hand, when the explicit disclosure type was used, the influencer was perceived as the most credible, positively influencing intentions to engage in eWOM
To conclude, research found negative indirect effects of disclosure on intentions to engage in eWOM, especially for implicit disclosure. And, according to the theory of planned behaviour (Fishbein & Azjen, 1975) intention is a key indicator of behaviour. Therefore, the present study assumes that implicit disclosure of branded content negatively affects user engagement more than when both disclosure types are used.
Brand prominence
The present section will address what the role of brand prominence is when presenting branded content on social media.
According to Van Reijmersdal, Rozendaal and Buijzen (2011), the prominence of the brand in a social media post can have the same function as a disclosure as alert to help consumers recognize a message’s persuasive intent. Brand prominence is the extent to which the appearance of the brand’s product or other brand identifiers (name, logo) are visible by virtue of size and position on the screen (Gupta & Lord, 1998). The more the brand is prominently placed in a social media post, the easier the consumer can identify the brand. In the present study the size of the branded content is defined by the camera distance to the brand in the image of the Instagram post; the bigger in size the brand is shown, and the more prominent the branded placement. The position of the brand is the extent to which it is visible in the foreground of the Instagram post; the more it is placed into the foreground, the more prominent the branded placement. In the following section, a review about the extant knowledge on consumer responses to prominently versus subtly placed brands is given.
Consumer responses to brand prominence
In the present study brand prominence in Dutch influencers’ Instagram posts is investigated to analyse user engagement depending on prominently or subtly placed brands.
Earlier research mainly focussed on the effects of brand prominence on cognitive and affective responses to traditional marketing interactions such as television and movies. The effect of brand prominence in non-traditional marketing interactions has not been widely studied so far, expect for advergames. The results of these studies on television, movies and advergames showed that prominently placed brands activate recognition of the brand (Van Reijmersdal, 2009). Consumers realise the brand placement has a persuasive intent and this
triggers cognitive responses to resist the persuasion attempt (Friestad & Wright, 1994). This in turn negatively influences the attitude towards the brand when it is perceived as intrusive, irritating and overly commercial (Van Reijmersdal et. al., 2011). Thus, a highly prominent brand placement appears to have the same adverse effects on consumer responses as disclosure of branded content.
On the other hand, when a brand is subtly placed the consumer is not necessarily aware of branded content and attention is drawn to the editorial content of the social media post (Van Reijmersdal, 2009). Consequently, consumers are influenced by the branded content without knowing, since they do not form cognitive defences against the branded content. On the contrary, consumers’ affective responses towards the branded content are probably favourable. This is why it can be expected that a higher prominence of the brand placement in Instagram posts will negatively affects user engagement.
H2: The higher the prominence of the brand placements in Instagram posts of Dutch influencers, the more negatively it affects user engagement.
Method
In this section, there will be an introduction on the research design first. After that, a detailed description of the sample and how the data for this sample is collected will be given, followed by the description of the variables and the operationalization for the coding. The section will end with a description of the coding procedure including pre-test and inter-coder reliability.
Research design
The research method for this study is a quantitative content analysis in order to examine presence of sponsorship disclosure in Instagram posts of Dutch influencers and thus to what extent influencers are open in informing users about sponsored content.
The reason for using this research method is that content analysis is a way to describe the content of communication in an objective, systematic and quantitative way, according to Krippendorf (2004). The technique helps to identify specific characteristics of what is in a message. Other benefits of this research method are the low costs of conducting the research, unobtrusive data collection, and ease of replication.
Sample and procedure
Relevant content for this research were Instagram accounts of Dutch influencers. The accounts had to be from influencers of Dutch origin. The names of these accounts can be acquired from marketing bureaus (Blyde, Instacreator) and websites
(www.influencerengagementindex.nl), who provide lists of top influencers from the Netherlands.
A subset of the population of top Dutch influencers on Instagram was chosen: 21 accounts of Dutch influencers, which were in the week of 16th to 22nd October top of their category (see Table 1) in terms of number of followers. The categories were chosen based on lifestyle (fashion, beauty, model, fitness, health, travel) and vloggers (people who are
influential on Youtube). These 21 accounts had an average number of 1.715.934,71 followers (SD = 2.314.885,42) and an average number of total posts of 2.377,76 (SD = 1.525,41).
Table 1 Overview of top Dutch influencers on Instagram in week 45 of 2017 (6th – 12th November) based on number of followers in different categories (www.influencerengagementindex.nl). Category Name influencer #Followers #Posts in
total
#Posts in 3 months
Fashion Negin Mirsalehi 4.357.045 6.442 139
Anna Nooshin 567.065 3.099 98
Fred van Leer 442.621 4.326 183
Beauty Nikkie Tutorials 8.600.135 2.948 51
Mascha Feoktistova 565.422 3.323 126
Vera Camilla 232.844 3.881 39
Model Doutzen Kroes 5.496.268 3.318 166
Romee Strijd 3.476.520 1.821 80
Sandra Prikker 3.395.842 549 77
Health Lisa van Cuijk 55.655 1.873 131
Jelmer de Boer 41.982 1.268 35
Annemerel 35.695 4.528 76
Fitness Tavi Castro 2.230.050 703 74
Faya Lourens 164.196 1.680 158
Christel 48.285 751 24
Travel Noor de Groot 418.816 2.477 42
Iris Dijkers 174.858 1.439 96
Tom Grond 126.847 844 53
Vlogger Twan Kuyper 3.780.310 1.087 51
Enzo Knol 1.192.567 1.918 65
Giel de Winter 631.606 1.658 101
Totaal 1865
The unit of analysis were the posts placed by these Dutch influencers in July, August and September 2017 on their Instagram account. A total of 1.865 were posted in these months, with an average of 88,81 (SD = 46,19) posts per account.
Two main elements per post were analysed. Firstly, the image of the Instagram post, including visual, audible, and textual elements. Secondly, the accompanying text of the Instagram post, including the account name, “paid sponsorship” tag, location tag, description, comments, likes or views (if format is audiovisual) and date of posting. The following section will explain this in more detail.
Coded variables
Independent variablesBranded content This variable describes whether there was branded content present in the Instagram posts of the Dutch influencers. Branded content in this case means that a brand is visibly and/ or audibly incorporated in the Instagram post. The influencer can do this by showing the visual logo of the brand, the name of the brand in written or audible form, or tag the brand’s Instagram account, in the pictorial image. The coding for this dichotomous
variable was (1) present or (0) absent. The coder had to indicate for each post whether a brand was visibly and/ or audible incorporated into an Instagram post or not. In the eventual sample of Instagram posts by the Dutch influencers (N = 1.865, 66,1%) was found to contain branded content.
Format type This variable renders insights into the way branded content in Instagram posts of Dutch influencers was made visible. The coding for this variable was adapted from Gupta and Lord’s (1998) modes for product-placement strategies. This has lead to three categories for operationalizing this nominal variable: ‘visual’, ‘audio-visual’ and ‘captioned photo’. An Instagram post having a visual format means the pictorial image of the post is a photo
including all of its visual and textual elements (N = 1.640, 87,9%). An Instagram post having an audio-visual format means the pictorial image of the post is a video including all of its
visual, audible and textual elements (N = 182, 9,8%). An Instagram post having the format of a captioned photo means the pictorial image of the post primarily contains textual elements, such as memes or photos with embedded text (N = 43, 2,3%). The coder had to indicate format type for every Instagram post, whereby only one option was possible.
Brand name This variable lists the names of the brands incorporated into the pictorial image or the audible content, and the accompanied text of Instagram posts of Dutch influencers. In 274 posts (14,7%) there were zero brands incorporated. A total amount of 4738 brands was listed in the whole sample, of which 1563 (33,0%) brands were unique. In case branded content was present, the brand name was listed in the codebook. 76,0% of these listed brands (N = 1.866) were unique. In case branded content was absent in the posts’ pictorial or audible content, there is still the possibility brands are named in the accompanying text of the post, and these were also coded. 49,9% of these listed brands (N = 637) were unique brands (see Appendix II, Table A2). There was also a possibility multiple brands were named in either the posts’ pictorial or audible content or in accompanying text of the post (see Appendix II, Table A1). All these different brand names occurring in a post were listed in alphabetic order.
Product categories This variable categorizes to which type of product category the brands that were present in each Instagram post belong, adapted from a study by Mediakix Team (2017). The product category of every brand name occurring in a post was coded (see Table 2). The coding for this nominal variable was operationalized in ten categories: ‘1 – Fashion’ (clothing, shoes, bags etc.), ‘2 – Beauty’ (make-up, body and hair products), ‘3 – Food and beverages’ (bars, supermarkets etc.), ‘4 – Transport’ (motorized vehicles), ‘5 – Travel’ (cities, hotels, events), ‘6 –Technology’ (gadgets), ‘7 – Media’ (traditional and social media), ‘8 –
People’ (artists, reality stars etc.), ‘9 – Stores’ (department stores, web shops, etc.), and ’10 – Other’ (funds, communities, etc.).
The coder had to indicate for every occurring brand to which product category it belongs. Only one option per brand was possible, but since multiple brands could be present in one post, multiple product categories could occur in one post. Code this variable as ‘0’ if there are no brands present.
Table 2 The frequency in numbers and percentages of the product categories present in the Instagram posts, when branded content was present.
Product category Frequency in image (%) Frequency in text (%) Total (%) Fashion 930 (32,1%) 73 (5,8%) 1.005 (24,1% Beauty 101 (4,7%) 61 (4,8%) 162 (3,4%)
Food & beverages 78 (3,6%) 23 (1,8%) 101 (2,4%)
Transport 42 (1,8%) 3 (0,2%) 45 (1,1%) Travel 229 (7,6%) 665 (52,5%) 894 (2,1%) Technology 41 (2,0%) 11 (0,9%) 53 (1,2%) Media 198 (7,1%) 107 (8,5%) 305 (7,4%) People 1.003 (32,3%) 205 (16,1%) 1.208 (29,0%) Stores 62 (2,7%) 30 (2,4%) 92 (2,2%) Other 219 (5,9%) 87 (6,9%) 306 7,3%) Totaal 2.903 (100%) 1.266 (100%) 4.172 (100%)
Brand prominence This variable describes the extent to which the appearance of the brand was made visible by virtue of size and position on the screen (Gupta & Lord, 1998). When applying this to the present study, this means the extent to which the branded content in the image of the Instagram post was made visible by virtue of size and position. Thus, this ordinal variable consists of two sub-variables (size and position), only coded when branded content was present (see Table 4). For each variable only one option per brand was possible.
The first sub-variable is the ‘Size’ of the branded content, which is defined by the camera distance to the branded human figure(s) and/ or object(s) in the pictorial image of the post. The bigger the branded human figure(s) and/ or object(s) shown are in size, the more prominent the branded content is. This variable was operationalized in eight categories: ‘0 – No shot’, ‘1 – Extreme long shot’, ‘2 – Long shot’, ‘3 – Medium long shot’, ‘4 – Medium shot’, ‘5 – Medium close-up’, ‘6 – Close-up’ and ‘7 – Extreme close-up’. The coder had to make a choice between the categories based on the scale of the branded human figure(s) and/ or object(s), relatively to the background filling up the image. However, ‘No shot possible’ means the tagged brand is not actually incorporated in the image.
The ‘position’ of the branded content being the second sub-variable is the extent to which the branded human figure(s) and/ or object(s) in the pictorial image appeared in the foreground of the post. This variable was operationalized in four categories: ‘0 – Not in pictorial image’, ‘1 – Background’, ‘2 – Intermediary’ and ‘3 – Foreground’. The coder had to make a choice between the categories based on the amount of space in front of the branded human figure(s) and/ or object(s). ‘Not in pictorial image’ means the tagged brand is not actually incorporated in the image.
This eventual variable of ‘Brand prominence’ consisted of a combined score of the sub-variables. ‘Size’ was double weighted relative to ‘Position’, since a closer camera
scores was taken, which resulted in a minimum score of brand prominence of 0 and a maximum of 18,33 (M = 8,16, SD = 3,10).
Table 4 The frequency in numbers and percentages of how brands are portrayed in size and position in the Instagram posts of Dutch influencers.
Size Position
Not in image Background Intermediary Foreground Total No shot 155 (100%) 7 (2,6%) 27 (1,8%) 4 (0,4%) 193 (6,7%) Extreme long shot - 32 (12,7%) 121 (8,0%) 52 (5,3%) 205 (7,1%) Long shot - 85 (32,1%) 631 (42,0%) 232 (23,8%) 948 (32,7%) Medium long shot - 59 (22,3%) 316 (21,0%) 297 (30,4%) 672 (23,2%) Medium shot - 40 (15,1%) 254 (16,9%) 187 (19,2%) 481 (16,6%) Medium close-up - 40 (15,1%) 135 (9,0%) 180 (18,4%) 355 (12,2%)
Close-up - 2 (0,8%) 18 (1,2%) 20 (2,4%) 40 (1,4%)
Extreme close-up - 0 (0,0%) 2 (0,1%) 4 (0,4%) 6 (0,2%) Total 155 (100%) 265 (100%) 1.504 (100%) 976 (100%) 2.900 (100%)
Disclosure This variable describes how branded content was disclosed in the Instagram posts of the Dutch influencers. Disclosing means recognition of the relationship between the brand and the influencer and thus the commercial purpose of the branded content.
‘Disclosure’ as ordinal variable was operationalized as: ‘Implicit disclosure’ and ‘Explicit disclosure’, ordered on level of explicitness of the disclosure, whereas ‘1’ is the lowest level of explicitness and ‘8’ the highest level. The coder had to indicate for every brand shown in the pictorial image or audible content which types of disclosures (type 1 to 8) were used, as well as for the brands disclosed in the accompanying text, whereby more than one option was possible (see Table 5).
Table 5 The frequency in numbers and percentages of how brands are disclosed in the Instagram posts of Dutch influencers. Disclosure Total (%) Audio 54 (0,01%) Logo 770 (11,5%) Tag 2.264 (33,6%) Hash tag 1.077 (16,0%) Location tag 649 (9,6%) Description 1.815 (27,0%) Hash tag sponsorship 102 (1,5%) Paid sponsorhip tag 2 (0,0%)
Totaal 6.733 (100%)
Influencers can use implicit disclosure to inform users about the presence of branded content. There are three options of implicitly disclosing branded content: ‘1 – Audio’, ‘2 – Logo’, ‘3 – Tag’. The coder had to indicate whether the Dutch influencer discloses the brand name by audibly naming it, integrating a visual logo or name in written form, or tagging the brand’s Instagram account.
Furthermore, there are five options of explicitly disclosing the names of brands: ‘4 – Hash tag brand name’, ‘5 – Location tag’, ‘6 – Brand name/ tag’, ‘7 – Hash tag sponsorship’ and ‘8 – Paid sponsorship tag’. The coder had to indicate whether the Dutch influencer discloses the brand name in: a hash tag in the description; in the location tag; in the description by just naming the brand or tagging the brand’s Instagram account; by using words and abbreviations in hash tags to inform users about the presence of sponsored branded content; by tagging a brand in the paid sponsorship tag.
Dependent variables
User engagement. This variable describes the rate at which users behaviourally engage with an Instagram post. The percentage of people who respond to an Instagram post, such as “liking” or commenting (Jaakonmäki et. al., 2017). Therefore, this variable was
operationalized through the following ratio variables: ‘Amount of likes’ (M = 48.153,16, SD = 78.381,51) and ‘Amount of comments’ (M = 370,78, SD = 1.064,71). The sum of these two variables was taken, in which ‘Amount of likes’ was double weighted. This resulted in a total score, which was operationalized as the ratio variable ‘User engagement’.
Control variables The control variables for this research were ‘Amount of account followers’ per the 9th of November 2017 and ‘Amount of account posts in 3 months’ (1 July – 31
September 2017). ‘Amount of followers’ (M = 1.690.939,37, SD = 2.173.819,93) of an
Instagram account was used to control for user engagement in the form of likes and comments per post. Logically, more followers of an Instagram account tend to generate more likes and comments per post. Therefore, these variables have to be measured relative to the amount of followers. Similarly, ‘amount of posts in 3 months’ (M = 108,58, SD = 44,43) has to be controlled for since the more an influencer posted in July, August and September (the sampling period), the more opportunities the influencer had to present branded content. In conclusion, these control variables abrogate the differences between Instagram accounts of Dutch influencer with a large number of followers and posts relative to influencers with a smaller number of followers and posts.
Inter-coder reliability
The author was the main coder. In order to preserve the reliability of the research, a second trained coder also coded 10% of the full sample (N = 189). Since 21 accounts were used for
analysis, 9 posts per account were used for assessing the inter-coder reliability test. Every post on the first date, 15th (or the date nearest to the 15th) and last day of every month was
coded by the second coder to ensure an equal distribution of the posts coded by the second coder.
The inter-coder reliability was calculated for each of the variables. The percentage of agreement between the coders is a “simple, intuitive and easy to calculate” (Lombard, 2002, p. 590) measure that can be used for nominal level variables. Since, the percentage of agreement fails to account for agreement by chance, Cohen’s Kappa (κ) and Krippendorf’s Alpha (α) were also calculated. The latter measure accounts for differences in the distribution of values across the nominal categories for different coders. For the results of the inter-coder reliability test consult Appendix III, Table A4 to A8.
According to Lombard (2002), inter-coder agreement is reliable when the percentage of agreement is 90% or higher. Cohen’s Kappa (κ) en Krippendorf’s Alpha (α) are considered to be reliable when scores of .70 or higher are obtained. Table A4 to A8 (see Appendix III) show that all scores were higher number concerning the percentage of agreement, Cohen’s Kappa (κ) and Krippendorf’s Alpha (α). This indicates that inter-coder agreement was high: the information in the codebook was clear, and the variables are reliable and exhaustive.
Results
The present section is dedicated to the results of the statistical analysis of the study. First, the recoding of the dataset will be discussed. After that, for every hypothesis the results of the statistical analysis will be presented.
Recoding the data
In order to do statistical analyses for the present study, the dataset had to be recoded. The coding categories within both the variables ‘Product category’ and ‘Disclosure’ were recoded into dummy variables, in which each category was coded binary. The recoding on these variables was separately done for the brands present in the image (branded content) and the brands present in the accompanying text of the Instagram post. Refer to Appendix III, Table A9 to A12 for the descriptive statistics on the independent variables.
Furthermore, the dependent variables ‘User engagement’ and its signal variables ‘Amount of likes’ and ‘Amount of comments’ were recoded due to their non-normal
distribution (see Table 6). A log transformation was used to reduce the skewed distribution of these variables, which resulted in a normal distribution for the two variables (see Table 7).
Table 6 Descriptive statistics of the three variables of user engagement before log transformation.
M SD Skewness (SE) Kurtosis (SE)
User engagement 96677.09 157205.65 2.80 (.06) 8.93 (.12)
Likes 48153.16 78381.51 2.80 (.06) 8.86 (.12)
Comments 370.78 1064.71 10.62 (.06) 140.14 (.12)
Table 7 Descriptive statistics of the three variables of user engagement after log transformation.
M SD Skewness (SE) Kurtosis (SE)
User engagement 24.22 5.01 -.32 (.07) -.88 (.13)
Likes 9.67 1.85 -.32 (.07) -1.02 (.13)
Data-analysis
The present section will address to the results study on the effects of the independent variables (implicit and explicit disclosure types, brand prominence, product categories) on user engagement (likes and comments). The following sections will elaborate on each of the independent variables. Since there are many dummy variables, the results are summarized and only some results are highlighted. Refer to Appendix III, Table A13 to A17, for the complete results.
Disclosure of branded content
The following section will address to the hypotheses concerning the disclosure of branded content.
Disclosure types
Based on theory, the expectation was that implicit disclosure was used more often than explicit disclosure in Instagram posts of Dutch influencers (H1a). A paired-sample t-test was conducted to compare the mean of implicit disclosure with the mean of explicit disclosure per Instagram post. The means of both variables were calculated by the sum of either the implicit or explicit disclosure types of the brands present in the image of an Instagram post divided by the amount of brands present in the image of Instagram post. The means of these two types of disclosure illustrated to what extent multiple types of disclosure were used: the higher the sum, the more types of disclosure were given for the brands present in the image of the Instagram post.
The results indicated a significant difference in the scores for implicit disclosure (M = 3.58, SD = 1.03) and explicit disclosure (M = 3.22, SD = 3.68), t(1232) = 3.41, p = .001. These results suggested that implicit disclosure is used .36 times more often than explicit
disclosure in Instagram posts of Dutch influencers (95% CI [.15, .57]). Hereby H1a is accepted.
On top of that, the expectation was that when Dutch influencers explicitly disclose branded content present in their Instagram post, they use only one type of explicit disclosure instead of multiple types of explicit disclosure (H1b). Again, a paired-sample t-test was conducted to compare counts: the amount of times explicit disclosure is used once compared to the amount of times explicit disclosure is used multiple times within an Instagram post.
The results indicated that scores were significantly higher for the amount of times explicit disclosure is used once (M = .51, SD = .90) than for the amount of times explicit disclosure is used multiple times (M = .18, SD = .42) in an Instagram post, t(1590) = -12.77, p = .000. Meaning that influencers generally use only one type of explicit disclosure to inform consumers about the branded content. Therefore, H1b is accepted.
Moreover, the results indicated that branded content was disclosed 4550 times (see Appendix III, Table A3). 67,9% of these disclosures was implicit (N = 3045), in audible form (1,2%, N = 54), a logo (16,9%, N = 770 or tag (49,2%, N = 2264. Thus, the most used type of implicit disclosure was ‘Tag’, in contrary to the type ‘Audio’. This low number of ‘Audio’ disclosers could be accounted to the fact that only 182 posts were audio-visual. Moreover, for 78,3% of the brands (N = 2892) two types of implicit disclosures were used. The possible combinations were: ‘Audio’ – ‘Logo’ or ‘Logo’ – ‘Tag’, since in the audio-visual format brands cannot be tagged.
The results also revealed that branded content was explicitly disclosed 1465 times, in which brands were mostly disclosed by name in hash tags (N = 511) or in the description (N = 784). The ‘Paid sponsorship’ tool, which Instagram introduced in 2017, was only used 2 times. Furthermore, hash tags such as “#ad” or “#spon”, were relatively used not that much (N = 83). The results also indicate that 62,1% of the brands (N = 1800) was not explicitly
disclosed at all and 10,0% (N = 290) of the brands was disclosed with multiple types of explicit disclosures. On the other hand, only one type of explicit disclosure was used in 27,9% (N = 807) of the brands.
The following sections will elaborate on each of the independent variables (product
categories, brand prominence, implicit and explicit disclosure types). The hypotheses were tested through one multiple regression analyses on user engagement (see Appendix III, Table A15), and two on each of the signals of user engagement (likes and comments) to illustrate whether it is predicted by a change in one of the independent variables (see Appendix III, Table A13 & A14)
Model 3 in each multiple regression analysis was found to be well suited to explain the effects on user engagement (likes and comments). A significant regression equation was found for ‘User engagement’ (𝑅2 = .71, F(39,575) = 36.163, p < .001). Regarding ‘Amount of likes’ the regression equation was also found significant, F(35,580) = 42.210, p < .001 with an 𝑅2 of .72. For ‘Amount of comments’, 62,8% of the variance was explained by all the independent variables (𝑅2 = .63, F(35,655) = 31.612, p < .001). In all cases this means the majority of the independent variables could meaningfully predict user engagement (likes and comments) the influencers’ Instagram posts received, in either a positive or negative way.
Consumer responses to disclosure of branded content
Hypothesis 1c was tested to analyse whether implicit disclosure of branded content affected user engagement more negatively than when explicit disclosure was used in Instagram posts of Dutch influencers. The results of multiple regression analyses on the means of both implicit and explicit disclosure were examined to illustrate to what extent these variables
mixed findings on the signals of user engagement (likes and comments) (see Appendix III, Table A13 & A14). Therefore, the results of multiple regression analyses on the means of both implicit and explicit disclosure, and ‘User engagement’ were analysed to test hypothesis 1c (see Table 8)
The results revealed that when an influencer implicitly discloses more, ‘User engagement’ decreased with -.09 (𝐵 = -.09, t(615) = .877). When an influencer explicitly discloses more, ‘User engagement’ increased with .59 (𝐵 = .59, t(615) = .217). Nonetheless, the means of both types of disclosure were not significant predictors of ‘User engagement’. In conclusion, the results suggested that implicit disclosure had a negative effect on user
engagement, in contrary to explicit disclosure, which had a positive effect on user
engagement. Interpreting the results: user engagement of an Instagram post is more negatively influenced when more implicit disclosure used, and user engagement of an Instagram post is more positively influenced when there is explicitly disclosed more. However, since the results are not significant, H1c is rejected.
Brand prominence
This section will discuss the results regarding the prominence of the brands present in the Instagram posts of Dutch influencers, for which was expected that the higher the prominence of the brand placements, the more negatively it affects user engagement (H2). The results of the multiple regression analyses were studied to describe the relationship of brand prominence with user engagement, and whether higher prominent brand placements predicted a change in the amount of likes and comments (see Table 9).
The results revealed that a higher score of ‘Brand prominence’ increased ‘User engagement’ (𝐵 = .15, t(614) = .404). Meaning that when brands were more prominently placed, this generated more user engagement (likes and comments). However, this finding is
Moreover, the results on ‘Brand prominence’ indicated that brands were mostly presented with long/ medium long shot camera distance (55,9%, N = 1620) with a position that was mostly intermediary (51,2%, N = 1504, or in the foreground (33,7%, N = 976), which resulted in brand prominence with M = 8,16, SD = 3,10. Since the maximum brand prominence was 18,33, this suggests that brand placements were slightly subtly placed.
Table 8 Multiple regression analysis on ‘User engagement’, and implicit and explicit disclosure of brands present in the image of the Instagram posts.
B SE B 𝛽 t p Implicit disclosure Mean -.09 .59 -.02 -.16 .875 Audio 15.49 12.09 .04 1.28 .200 Logo 1.29 .87 .10 1.47 .141 Tag 1.28 1.56 .09 .82 .412 Explicit disclosure Mean .59 .53 .42 1.01 .271 Hash tag -2.36 2.12 -.16 -1.11 .267 Location -2.60 2.75 -.08 -.95 .344 Description -3.75 3.13 -.28 -1.20 .231 Hash tag sponsorship -2.88 3.62 -.12 -.80 .426 Paid sponsorship tag 12.33 8.98 .03 1.37 .170 Notes. 𝑅2 = .71.
Table 9 Multiple regression analysis on the three variables of user engagement and ‘Brand prominence’ of the brands present in the image of the Instagram posts.
𝑅2 B SE B 𝛽 t p
User engagement .71 .15 .18 .09 .84 .404
Likes .72 .06 .07 .09 .84 .404
Additional results
In this section, an additional result on the effects of branded content will be presented.
Correlation analyses were pursued to measure the strength and direction of the relationship to understand whether the control variable ‘Amount of accounts posts in 3 months’ abrogated differences between Instagram accounts of Dutch influencers with a large number posts in the sample period of 3 months relative to accounts with a smaller number of posts in that period. The results of the correlation analysis between ‘Amount of account posts in 3 months’ and ‘Branded content’ was not as expected (r(1591) = -.12, p < .001). The negative correlation indicates that the more posts an influencer placed on Instagram, the less branded content was present in these posts. Furthermore, the correlation between ‘Amount of posts in 3 months’ and ‘User engagement’ (r(1427) = -.09, p < .001) was negative, which means that the more an influencer posts on Instagram, the lower the amount of likes and comments per post. In conclusion, control variable ‘Amount of account posts in 3 months’ to some extent negatively explained the variance in user engagement between the Instagram posts of Dutch influencers. Moreover, ‘Amount of account posts in 3 months’ explained that some Dutch influencers, who posted more in the sample period of 3 months, posted less branded content than others in their Instagram posts.
Conclusion
‘Branded content, how and to what extent is it present and disclosed in Instagram posts of Dutch influencers, and how does this affect user engagement’, was the central question guiding the present research.
Firstly, the present study has shown that branded content is present in the majority of Instagram posts of popular Dutch influencers nowadays. Moreover, in more than half of these posts, two or more brands were integrated. Products in the categories ‘Fashion’, ‘Travel’ and
‘People’ were the most popular to brand, and branded content was mostly presented in a visual format where brand placement was slightly subtly placed.
Regarding disclosure of branded content, the present study found no meaningful effects of the disclosure of branded content in Instagram posts by Dutch influencers on user engagement. Nonetheless, explicit disclosure of branded content was scarcely given, proving that regulations are not fully adhered and thus consumers are deceived since the commercial purpose of the Instagram posts cannot be recognized as such.
Altogether, the present study contributed to a better understanding of the presence of branded content in influencer marketing on social media platform Instagram. Especially, the present study proved there seems to be a long way ahead for regulators to expand their
guidelines in order to first get influencers disclosing their branded content articulated in a way that consumers recognize it as such. After that, further research is needed to examine the effects on user engagement of disclosure of branded content on Instagram more closely. And, if the same adverse effects are expected as for online blogs (Van Reijmersdal et. al., 2016), reviews (Carr & Hayes, 2014), advergames (Van Reijmersdal et. al., 2012) and Facebook (Boerman et. al., 2017), influencer marketing will soon suffer under these effects resulting in an ineffective marketing strategy.
Discussion
The findings could either be explained by limitations of the research design or by previous studies. Therefore, theoretical and practical implications are suggested in the present section, resulting in recommendations for future research.
Theoretical implications
findings that were inconsistent with the assumptions. The present section will address the interpretation of these findings, related to theory and previous studies.
The present study investigated whether implicit disclosure influenced user engagement more negatively than explicit disclosure (H1c). Since implicit disclosure is more general disclosure of a third party is potentially influencing the message without specifically disclosing which third party compensated the influencer (Carr & Hayes, 2014), the expectation was that consumers feel deceived, perceiving this as threat to freedom and activate cognitive and affective resistance strategies (Burgoon et. al., 2002, Friestad & Wright, 1994; Van Reijmersdal et. al., 2016). Besides supporting this negative effect of implicit disclosure on user engagement, the results of the present study also indicated that explicit disclosure has positive effects on user engagement. This suggests that influencers should supplement implicit disclosure by explicitly disclosing the relationship with brands in accompanying text of the Instagram post (Tessitore & Geuens, 2013), indicating the
commercial purpose of the brand and reducing feelings of distrust. Thus, explicit disclosure could diminish negative effects of implicit disclosure on user engagement. However, the present study did not find significant results on these effects. This could be accounted to the fact that in most cases implicit disclosure is not accompanied by explicit disclosure at all (H1a), or only one type is used to inform consumers branded content is present in the Instagram post (H1b).
Secondly, results on prominence of branded content in Instagram posts has proved to counter the findings of Van Reijmersdal et. al. (2011). Although highly prominent brand placements are expected to be recognized easily and therefore have a negative influence on consumer responses (H2), the present study proved the opposite: more prominently placed brand generated more user engagement in the form of likes and comments.
Practical and managerial implications
The present study suggests important practical implications for regulators, marketers and online influencers. Even though regulators require influencers to disclose their relation with a brand in a transparent way by naming the brand and using correct hash tags (#ad), influencers are not adhering these regulations. Regulators can therefore be advised to change their
guidelines and be more prescriptive in how disclosure of branded content on social media should be articulated. Moreover, these regulations should be closely monitored and when not adhered to penalties could be issued. However, fulfilling this is restrained to time and money.
Moreover, it is important for marketers and online influencers to understand the negative effects of disclosure on user engagement. Introducing stricter regulations could do more harm to the effectiveness of influencer marketing. At present, this marketing strategy reaches target audiences effectively. Potential long-term consequences arise when more and more consumers will recognize influencer marketing as branded persuasive messages, making them feel deceived. This will eventually hurt the reputation and credibility of both brands and influencers.
Limitations and future research
Although the present study provides important new insights into the effects of branded content in Instagram posts, it does have some limitations. The sample was taken from Instagram from accounts of a relatively small group of top Dutch influencers, chosen based on product categories. According to Boerman et. al. (2017), trust in social media advertising and opinions posted online are generally lower in Europe and the United States than in Asia, Latin America, and Africa (Nielsen, 2015). Moreover, the guidelines are underdeveloped in contrary to the United States. Meaning, effects of branded content and its disclosure could
Moreover, the sample size should be extended with more accounts from different product categories.
Further research could be done to examine the effects of branded content on social media more closely. The data of the present research could be used to fully analyze the
differential effects on user engagement of brand present in the image versus the brands named in the accompanying text of the Instagam posts. Moreover, interaction effects could be
measured between the independent variables to, for example, investigate the effects on user engagement of a highly prominent brand placement being explicitly disclosed.
Moreover, since content analysis is purely descriptive, other research methods should be conducted to examine underlying motives of the effects of disclosure of branded content on user engagement. Firstly, analyzing the cognitive and attitudinal resistance strategies evoked when consumers are confronted with disclosure of branded content on Instagram, in a similar experiment as Boerman et. al. (2017). Secondly, the credibility and trustworthiness of the influencers could be measured through a survey, while Carr and Hayes (2014) state that credibility of influencers is a mediator in the effects of disclosure on intentions to engage in eWOM.
Conducting a survey, could also measure consumer knowledge on the different disclosure types. Findings of this research could advise that consumers are given training in order to create awareness and increase knowledge for different disclosure types being
recognized. Especially, knowledge on the implicit disclosure type ‘Tag’ should be increased. Since users have to click on the pictorial image for the tag to appear, this type of disclosure could remain unnoticeable. As a result, users did not recognize the Instagram post as branded persuasive message and formed negative attitudes towards it. This could have explained why this type of disclosure generated more user engagement.
Furthermore, an experiment could be conducted to make disclosure of branded content more noticeable. Conditions should be created in which disclosure is made noticeable in size, position or wording to a greater or lesser extent. An experiment could be conducted to
understand what the ideal amount of posts per day/ week and ideal number of brands is in one Instagram post. The present study suggests that, the more often an influencer posts and the more brands integrated in an Instagram post, the more negatively it influences user
engagement, in line with (Bakhshi et. al., 2014). Therefore, the findings on such an
experiment could help increase the effectiveness of influencer marketing in Instagram posts. Lastly, to increase an even better understanding of the effects of branded content on social media, these suggestions for future research could be done on different social media platforms such as YouTube, Twitter and Snapchat.
References
Belch, G. E., & Belch, M. A. (2015). Advertising and Promotion: An Integrated Marketing Communications Perspective (10 ed.): New York: McGraw-Hill.
Boerman, S. C., & van Reijmersdal, E. A. (2016). Informing Consumers about “Hidden” Advertising: A Literature Review of the Effects of Disclosing Sponsored Content. In P. De Pelsmacker (Ed.), Advertising in new formats and media: Current research and implications for marketers (pp. 115-146). Retrieved from:
https://pure.uva.nl/ws/files/2606739/178802_Boerman_Van_Reijmersdal_online_versi on.pdf
Boerman, S. C., van Reijmersdal, E. A., & Neijens, P. C. (2012). Sponsorship disclosure: Effects of duration on persuasion knowledge and brand responses. Journal of Communication, 62(6), 1047-1064.
Boerman, S. C., van Reijmersdal, E. A., & Neijens, P. C. (2014). Effects of sponsorship disclosure timing on the processing of sponsored content: A study on the effectiveness of European disclosure regulations. Psychology & Marketing, 31(3), 214-224.
Boerman, S. C., Van Reijmersdal, E. A., & Neijens, P. C. (2015). How audience and disclosure characteristics influence memory of sponsorship disclosure. International Journal of Advertising, 34(4), 576-592.
Brown, D., & Fiorella, S. (2013). Influence marketing: How to Create, manage, and measure Brand Influencers in Social media marketing. Indiana: Que Publishing
Burgoon, M., Alvaro, E., Grandpre, J., & Voulodakis, M. (2002). Revisiting the theory of psychological reactance: Communicating Threats to Attitudinal Freedom. In Dillard, J. P., & Pfau, M. (Ed.), The persuasion handbook: Developments in theory and practice (pp. 213-232). Los Angeles: Sage Publications.
Chu, Shu-Chuan and Yoojung Kim (2011). Determinants of Consumer Engagement in Electronic Word-of-Mouth (eWOM) in Social Networking Sites. International Journal of Advertising, 30(1), 47–75.
Cialdini, R. B. (1987). Influence (Vol. 3). Port Harcourt: A. Michel
Einarsdóttir, V.A. (2017). Influencer marketing with a special focus on the social media platform, Instagram (Bachelor’s thesis, University of Iceland). Retrieved from:
https://skemman.is/bitstream/1946/27493/1/Valger%C3%B0ur%20Anna%20Einarsd C3%B3ttir1.pdf
eMarketer (2015, July 9). Marketers Pair With Influencers – and it Works [Blog post]. Retrieved from: https://www.emarketer.com/Article/Marketers-Pair-Up-with-Influencersand-Works/1012709
Fishbein, M., & Ajzen, I. (1975). Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research. Reading: Addison-Wesley Publishing Co.
Friestad, M., & Wright, P. (1994). The persuasion knowledge model: How people cope with persuasion attempts. Journal of consumer research, 21(1), 1-31.
FTC, (2015a). The FTC’s Endorsement Guides: What People Are Asking | Federal Trade Commission. Ftc.gov. Retrieved from https://www.ftc.gov/tips-advice/business-center/guidance/ftcs-endorsement-guides-what-people-are-asking#social
Gupta, Pola B. and Kenneth R. Lord (1998). Product placements in movies: The effect of prominence and mode on audience recall. Journal of Current Issues and Research in Advertising, 20(1), 47–59.
Hwang, Y. & Jeong, S. (2016). “This is a sponsored blog post, but all opinions are my own”: The effects of sponsorship disclosure on responses to sponsored blog posts. Computers In Human Behavior, 62, 528-535.
Jaakonmäki, R., Müller, O., & vom Brocke, J. (2017). The Impact of Content, Context, and Creator on User Engagement in Social Media Marketing. In Proceedings of the 50th Hawaii International Conference on System Sciences.
Kirkpatrick, D. (2016, April 6). Influencer Marketing spurs 11 times the ROI over traditional tactics: Study [Blog post]. Retrieved from:
https://www.marketingdive.com/news/influencer-marketing-spurs-11-times-the-roi-over-traditional-tactics-study/416911/
Krippendorff, K. (2004). Content analysis: An introduction to its methodology. Los Angeles: Sage.
Kulin, E., & Blomgren, L. (2016). Credible or not: A study on the factors influencing
consumers' credibility assessment of product placements on Instagram (Dissertation). Retrieved from http://urn.kb.se/resolve?urn=urn:nbn:se:hj:diva-30063
Lu, L. C., Chang, W. P., & Chang, H. H. (2014). Consumer attitudes toward blogger’s sponsored recommendations and purchase intention: The effect of sponsorship type, product type, and brand awareness. Computers in Human Behavior, 34, 258-266. Maheshwari, S. (2016, 2 November). Endorsed on Instagram by a Kardashian, but Is It Love
or Just an Ad? Nytimes.com. Retrieved from:
http://www.nytimes.com/2016/08/30/business/media/instagram-ads-marketing-kardashian.html?_r=1
Mediakix Team (2016, September 9). The FTC Social Media Guidlines 2016 [Blog post]. Retrieved from:
http://mediakix.com/2016/09/ftc-social-media-guidelines-2016/#gs.PArpoHE
Mediakix Team (2017, April 8). Understanding YouTube, Instagram & Faceboook’s Branded Content Tools [Blog post]. Retrieved from:
http://mediakix.com/2017/04/branded-content-tools-facebook-youtube-instagram/#gs.Ik3stHg
Mediakix Team (2017, May 31). 93% of Top Celebrity Social Media Endorsements Violate FTC Guidelines [Blog post]. Retrieved from:
http://mediakix.com/2017/05/celebrity-social-media-endorsements-violate-ftc-instagram/#gs.mjHJf9w
Millwood, A. (2016, November 14). Instagram Advertising in 2017: What You Need to Know [Blog post]. Retrieved from: https://www.yotpo.com/blog/advertise-on-instagram/
Muntinga, D. G., Moorman, M., & Smit, E. G. (2011). Introducing COBRAs: Exploring motivations for brand-related social media use. International Journal of
advertising, 30(1), 13-46.
guide for practices that mask marketing messages. Journal of Public Policy & Marketing, 27(1), 7-18.
Riffe, D., Lacy, S. & Fico, F. (2014). Chapter 5: Sampling. In Riffe, D., Lacy, S. & Fico, F. (Ed.), Analyzing Media Messages (pp. 96-122). New York: Routeledge.
Stewart, R. (2017, June 14). Instagram wants to ‘bring transparency’ to influencer and brand content deals with fresh set of tools [Blog post]. Retrieved from:
http://www.thedrum.com/news/2017/06/14/instagram-wants-bring-transparency-influencer-and-brand-content-deals-with-fresh-set
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), 419-442.
Tyagi, R. (2017, August 13). Instagram vs Facebook vs Twitter vs Snapchat: Which is Better for Business Marketing [Blog post]. Retrieved from:
http://www.techmagnate.com/blog/instagram-vs-facebook-vs-twitter-vs-snapchat which-better-business-marketing/
The Dutch Advertising Code (2014). The Dutch Advertising Code. With Information About the Working Procedures of the Advertising Code Committee and the Board of Appeal. Retrieved from: www.reclamecode.nl/ bijlagen/SRC_NRC_Engels_apr14.pdf. Van Doorn, J., Lemon, K. N., Mittal, V., Nass, S., Pick, D., Pirner, P., & Verhoef, P. C.
(2010). Customer engagement behavior: Theoretical foundations and research directions. Journal of service research, 13(3), 253-266
Van Reijmersdal, E. (2009). Brand placement prominence: good for memory! Bad for attitudes? Journal of Advertising Research, 49(2), 151-153.
Content in Blogs How the Use of Resistance Strategies Mediates Effects on Persuasion. American Behavioral Scientist, 60(12), 1458-1474.
Van Reijmersdal, E. A., Rozendaal, E., & Buijzen, M. (2012). Effects of prominence,
involvement, and persuasion knowledge on children's cognitive and affective responses to advergames. Journal of Interactive Marketing, 26(1), 33-42
Wei, M. L., Fischer, E., & Main, K. J. (2008). An examination of the effects of activating persuasion knowledge on consumer response to brands engaging in covert marketing. Journal of Public Policy & Marketing, 27(1), 34-44.
Woods, S. (2016). #Sponsored: The Emergence of Influencer Marketing (Honors Thesis, University of Tennessee). Retrieved from:
http://trace.tennessee.edu/cgi/viewcontent.cgi?article=3010&context=utk_chanhonopr oj
Appendix I – Codebook
The aim of this research is to examine the presence of branded content in Instagram posts of influencers and whether this presence is clearly disclosed or not. A selection is made of Instagram accounts of 21 top Dutch influencers.
Coding
The unit of analysis for media content is each whole Instagram post (see Figure 1), including all of the visual, audible, and textual elements of the pictorial image (see Figure 2), and the accompanied text including the account name, “paid sponsorship” tag, location tag,
description, comments, likes or views (if format is audiovisual) and date of posting (see Figure 3).
C.1 – Step 1: Information on the Instagram account
First, general information of every Instagram account is collected. This information can be found on the overview page of each Instagram accounts of the Dutch influencers (see Figure 27). ‘Amount of posts’ and ‘amount of followers’ are coded to control for measures.
a. Account name
b. Amount of posts in July, August and September c. Amount of followers
C.2 – Step 2: Characteristics of the Instagram post
Second, several general characteristics of every Instagram post are collected. ‘Amount of likes’, ‘amount of comments’ and ‘amount of views’ are coded to control for measures.
a. Account name b. Date of posting c. Amount of likes d. Amount of comments
Total amount of comments on a post, including reactions on comments. Retrieved when moving the mouse onto the Instagram post (see Figure 4).
