i
‘Native Advertising on Social Media: the effect of
heuristics on attitudes and purchase behavior’
Author: Jekaterina Smakova Student number: 11139366
Programme: MSc Business Administration- Marketing Thesis supervisor: Tina Dudenhöffer
ii
Statement of Originality
This document is written by Jekaterina Smakova who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
iii
Acknowledgements
I would like to thank my supervisor, Tina Dudenhöffer, for taking the time to discuss my ideas, provide helpful recommendations and support. Without her direction and feedback this project would not have been successfully completed.
I am also grateful for the support of my friends and family, who helped me to stay focused and offered moral support when I experience some setbacks in the process of conducting this study.
Moreover, I would like to thank all the participants that completed the survey and made this research study possible.
1 TABLE OF CONTENT:
Abstract……….…….2
Chapter 1. Introduction……….……….…....3
Chapter 2. Literature Review……….…..…5
2.1 Advertising industry trends ………..………...5
2.2 Native advertising ……….……...6
2.3 Benefits of native advertising………..…..….7
2.4 Challenges for native advertising. ……….……….………....9
2.5 Research gap……….….11
2.5 Heuristics……….………...12
a. Source credibility……….. ...……….…14
b. Brand familiarity………...16
c. Bandwagon effect……….……….…….…17
d. Interaction effects among source credibility…..….18
brand familiarity and bandwagon effect 2.6 Conceptual Framework………19
Chapter 3. Theoretical Framework……….………...20
3.1 Research design……….………..20 3.2 Measures……….……….20 3.3 Stimuli development………..……..22 3.4 Subjects………..23 3.5 Procedure……….24 Chapter 4. Results………..25 4.1 Frequency check………..25 4.2 Manipulation check………25 4.3 Reliability Analysis………..26 4.4 Hypotheses testing……….26
Chapter 5. Discussion and conclusions………..……..34
5.1 Limitations and future research……….36
5.2 Conclusion……….37
References………...…39
Appendix A. Questionnaire………...47
2
Abstract
It is becoming increasingly difficult for marketers to attract consumer’s attention in an
overcrowded on-line environment and native advertising is left as one of the few bright spots in the advertising landscape (Marvin, 2013). While the industry has embraced this new tool, the academic literature has been quite limited in the area of native advertising. This study, therefore, contributes to the existing literature by exploring different factors that affect the way consumers evaluate native adverts and to what extent are these factors important in producing attitude and behavior change.
A 2 x 2 x 2 on-line experiment systematically varied two levels of the source credibility, two levels of brand familiarity and two levels of perceived popularity (bandwagon effect) in order to investigate the effect of those cues on consumer evaluations and purchase behavior. The findings show that all three cues- source credibility, brand familiarity and bandwagon cues had a significant effect on consumer’s evaluations of the advertisement as well as the brand, but did not have a significant effect on purchase intent or intention to share the advertisement.
Moreover, the results support previous research showing that there is a cue-cumulation effect, suggesting that these cues can work together to influence consumer evaluations.
3
Chapter 1. Introduction
It is not a secret that advertising is vital for any company that wants to be successful. Recent economic climate and technological changes, however, have been making it increasingly difficult for marketers to find ways to attract consumers’ attention with traditional marketing techniques. Native advertising has been gaining recognition over the past few years as one of the most promising and powerful new techniques in marketers’ tool kit. As a term it can generally refer to any paid advertising that takes the specific form and appearance of editorial content from the publisher itself (Wojdynski & Evans, 2015). Even though the term itself is quite recent, the concept is not new. Advertisers disguised promotional messages to look like regular content since the early days of newspapers and magazines (Pollitt, 2014). Those kind of adverts are known as advertorials, and they would be placed alongside the content simulating publishers’ editorial style but would feature an ‘advertisement’ label to distinguish from other content (Van Reijmersdal et al., 2005). Developments in technology, however, lead to more opportunities for marketers to take this type of advertisement even further. While some may be similar to advertorials in their execution, others may take completely different forms. For example, some native advertising may appear in the form of a paid search listings,
recommendation units or sponsored social media posts (IAB, 2013).
Consumers in general find native ads less irritating than banner advertising (Becker-Olsen, 2003), have more positive attitudes towards the brand (Wojdynski and Evans, 2015) and native ads tend to generate greater engagement among consumers (Wojdynski, 2016; Wang, 2006; Konrad, 2015). This resulted in 9% higher brand lift, 18% higher purchase intent and 21% higher personal identification with the brand (Sharethough, 2013) that consumers experienced
compared to traditional banner ads.
With significant increase in internet and social media use among consumers, the importance of those channels for advertisers has become undeniable (Swedowsky, 2009). Social media is particularly attractive to advertisers due to its wide reach (eg. Facebook has 1.28 billion daily active users (Statista, 2017), Instagram- over 500 million and Snapchat over 150 million
4
relationships with consumers through two-way interaction (Hensel & Deis, 2010). Due to its non-intrusive nature and generally more positive attitudes of consumers, native advertising has taken over the social media platforms (Dix & Phau, 2009 in Lee et al., 2016), representing 38.8% of entire social media spending in 2014 and it predicted to reach 42.4% of the total spending in 2017 (eMarketer, 2015; Lee et al., 2016).
Despite growing popularity, the research in the area of native advertising has been fairly limited. For example, it is still not completely clear how exactly consumers process native advertising and which factors might moderate the effectiveness of those ads in leading to attitude and behavioral change. It is largely presumed that individuals generally evaluate information in a cognitively effortful fashion (Metzger et al., 2010). There has been significant evidence, however, that in everyday situations involving efforts to persuade, consumers are more likely to use quick judgement rules or heuristics rather than systematically processing the message (Iyengar & Valentino, 2000; Metzger et al., 2010; Sundar, 2008; Go et al., 2014). There is a number of potential heuristics that consumers might employ when processing persuasive messages, from specific homepage presentation factors such as use of photos and graphics, headline characteristics or source credibility to pre-existing attitudes towards the advertiser, brand or product and many others (Wojdynski, 2016; Sharethrough, 2013; Keller 1993; Wei et al., 2008; Petty and Cacioppo 1983). Currently, nonetheless, little research has been done to examine how much influence those different cues might have on consumers’ processing of a persuasive message.
This research aims to contribute to the existing literature by examining the effects of three separate heuristics- brand familiarity, source credibility and bandwagon effect on the consumer attitudes and purchase behavior in the context of native advertising. Additionally, it aims to fill the gap in the existing research regarding the potential interaction effects among these cues.
The main research question of this paper is:
To what extent source credibility, brand familiarity and the bandwagon effect affects consumers’ evaluations and purchase behavior in the native advertising context?
5
The rest of the paper is divided into four main chapters. The next chapter outlines the main concepts, introduces a review of existing literature and introduces the proposed hypotheses. The subsequent two chapters go into explaining the methodology used in this study to test the hypotheses and then provides the results of the data collection. Lastly, the paper will present the discussion of the findings, including theoretical and managerial implications, conclusions and directions for further research.
Chapter 2. Literature Review
2.1 ADVERTISING INDUSTRY TRENDSIt is an undeniable fact that marketing is an essential activity for any successful organization to communicate value and attract consumers (Wright et al. 2010). It is estimated that global advertising industry is worth over $600 billion and is experiencing a steady growth of about 6% annually (EMarketer, 2014). To illustrate this Johnson (2006) estimated that an average
American is exposed to 5.000 ads a day and considering how much the industry has grown since then this number is likely to have increased significantly (Wright et al. 2010).
There have also been some notable shifts in the industry itself- for example over 40% of all advertising spending is going into digital and mobile advertising (around 33.5% in 2015). This is equal to the share of TV advertising and far surpasses print advertising (Carlier et. al. 2016). These numbers are predicted to continue changing with more and more budget directed into mobile and digital, while print and TV are likely to continue shrinking (Carlier et al, 2016; Sweney, 2016), therefore giving publishers and advertisers no choice but to embrace digital (Wright et al. 2010).
Such rapid expansion, however, is making it increasingly difficult for marketers to attract consumers’ attention in cluttered on-line space and engage consumers to respond positively to line advertising messages (Becker-Olsen, 2003). Advertisers are generally attracted to
on-6
line advertising due to the increases in on-line population, generally lower costs and wide reach of such ads. This, however, lead to ever increasing number of Internet ads consumers are exposed to which makes it more challenging for advertisers to connect with on-line consumers (Ha & McCann, 2008; Becker-Olsen, 2003).
Internet advertising has been continuously evolving with new forms appearing to accommodate for changing consumers’ attitudes. On-line advertising can take forms of banner advertising, pop-ups, classified advertising, keyword search, paid text links, sponsorships, e-mail ads, and so forth (Cho & Cheon 2004; Moore et al. 2005; Zeff & Aronson 1999).
Research is consistently showing that consumers find display advertising annoying and actively avoid looking at online banners (Goldfarb & Tucker, 2011; Benway, 1998; Goldstein et al., 2014). Click-through rates (CTR) have been consistently falling from 2% in 1995 when banner ads were first introduced to about 0.05% on average nowadays (Chaffey, 2016). The constantly increasing number of people installing AdBlock software is also a clear demonstration of how much people want to avoid being exposed to online advertisement (Goldstein et al. 2014; Carlier et al., 2016). A recent Interactive Advertising Bureau (IAB) report found that 26 percent of desktop users and 15 percent of mobile consumers use AdBlockers to remove ads, and a PageFair report found around 615 million devices now using some type of ad blocking software (Cortland, 2017; Johnson, 2016).
This in combination with shrinking revenues from traditional forms of advertising is forcing media companies to explore new areas for monetization (Sahni & Nair, 2016) and this is where native advertising comes into the spotlight.
2.2 NATIVE ADVERTISING
Native advertising has been gaining recognition as a potential response to some of the issues that advertisers are experiencing. It has initially been popularized by digital natives such as Buzzfeed, Mashable and Huffington post but recently legacy brands such as the New York Times and the Wall Street Journal also started to join in seeing its potential (Carlson, 2015; Sebastian, 2013). This lead to a significant growth that native is experiencing, with spending on it
7
forecasted to reach$21 billion in 2018, which is around 600% growth since 2014 (Fullerton, 2017; BI Intelligence, 2015).
Native advertising as a term refers to a broad spectrum of different on-line advertising forms which share a common focus on minimizing disruption to user experience (Campbell & Marks, 2015). There is no one set definition for native advertising but the majority of definitions include elements of integration into the design of the publisher’s website, include content produced in conjunction with or created on behalf of advertisers and aims to minimize interruption to user experience (Hallahan, 2014; IAB, 2013).
This paper is going to adopt one of the most popular definitions provided by Interactive
Advertising Bureau. They essentially define native as a type of paid ad designed to blend in the page content, assimilated into the design, and consistent with the platform behavior that the viewer simply feels that they belong (IAB, 2013). The IAB also separated native ads into a range of categories that include in-feed units, paid search, recommendation widget, promoted listings and custom formats.
The undisrupted user experience is at the core of the native advertising and there are several ways native advertising may try to do so. This might be done by optimizing placement for increased relevance for viewers- this strategy is often used for paid search units as those are based on keywords user put in the search engine or in recommendation widgets based on the content user is viewing. Another way native might reduce disruption is by making sure native ad blends with the surrounding content- this approach is generally used in the case of in-feed units. In-feed unit are similar to advertorials in a sense that they blend in with surrounding content and are meant to be viewed more as part of the content rather than an ad (Campbell & Marks, 2015; Hallahan, 2014).
2.3 BENEFITS OF NATIVE ADVETRISING
There is a number of reasons why publishers, advertisers and consumers might be attracted native advertising. First, as it was already noted it is less disruptive for the user and fits better with the initial consumer motivation to come to the publisher’s website. Those motivations are
8
much more likely to be information seeking, entertainment or socializing rather than viewing display ads trying to sell a product (Sonderman &Tran, 2013; Van Reijmersdal et al. 2005). Kim and Sundar (2010) found that when ads provide information unrelated to Internet users’ interests, it leads to interruption of users’ specific goals and, therefore, often causes annoyance. Native advertising has been found effective in minimizing this annoyance. For example, Becker-Olsen (2003) found that native adverts were rated more positively and perceived as less annoying by respondents; Schreiber (2016) found that 80% of millennial consumers perceived native ads as good user experience and a study by HubShout (Pophal, 2014) found that 72.8% of participants believed sponsored content had equal or greater value as non-sponsored content on the same website.
Secondly, publishers interest in native is another big reason of its increasing growth. The majority of publishers are experiencing tough times with subscription revenues plummeting and banner advertising prices going down due to their ineffectiveness (Bakshi, 2014; Ponsford, 2014). Native advertising offers several opportunities to overcome those challenges. For
example, native in general attracts higher rates than other forms of on-line advertising (Benton, 2014). Publishers can earn more per impression as these ads are by design guaranteed to attract users’ attention, leading to advertisers valuing those impressions two to four times more than traditional ads (Bakshi, 2014; Sonderman &Tran, 2013). For publishers those premium rates are justified by promise of greater quality content, and consequently greater interaction and engagement on the part of consumers (Wojdynski, 2016; Wang, 2006; Carlier, 2016). Additionally, many of these publishers also started to employ in-house studios
facilitating content with advertisers, which allows them to capture an even greater share of revenue and further institutionalize advertising–editorial collaboration (Carlson, 2015; Marvin, 2013).
Furthermore, what publishers can offer to advertisers apart from producing and/or placing content on their website is their own brand associations that could potentially transfer to the advertised brand. Brand associations is essentially any information, images or symbols
regarding the product/brand stored in the consumers’ mind. Positive brand associations are essential for creating brand equity and differentiation for the product from competing brands (Aaker, 1991; Keller, 2001). Therefore, some advertisers might choose a particular publisher to
9
add some desired brand associations, for example, when GE places sponsored content on sites like Quartz and The Economist it should attach a more innovation-friendly feeling to the brand (Benton, 2014).
Lastly, native advertising is recognized as a promising way to “close the mobile revenue gap” (Sternberg, 2013). As it has been noted earlier mobile is experiencing tremendous growth, with over 60% increase between 2014 and 2015 (EMarketer, 2014). Mobile became so ubiquitous that in fact more than half of all digital time spent online happens using a mobile device (Fulgoni & Lipsman, 2014). Publishers and advertisers, however, are experiencing some challenges with integrating effectively into the mobile environment. Due to the limited screen environment it is difficult to make adverts effective and non-intrusive (Sonderman & Tran, 2013). Additionally, on the publisher’s side mobile ads are worth a small fraction of what desktop ads are worth, making them not a very attractive choice (Mobilecore, 2015). Native advertising can help overcome those challenges- is by definition supposed to blend in with other types of content on the platform making it a non-intrusive part of an organic content flow. Moreover, many mobile platforms utilize infinite scrolling, which provides the opportunity for advertisements consuming more real estate on the screen than standard mobile-advertising units (Sahni & Nair, 2016; Sonderman & Tran, 2013; Fulgoni & Lipsman, 2014).
2.4 CHALENGES FOR NATIVE ADVERTISING
Despite having a lot of potential benefits, there is also a number of issues that might arise in the area of native advertising. Firstly, some critics argued that native advertising is blurring
boundaries between editorial and advertorial content, which in turn violates readers’
expectations of journalistic autonomy (Marvin, 2013; Bakshi, 2014). There is an expectation of journalists to be objective, neutral and independent, and native advertisement could be perceived as directly infringing those principles. This in turn could affect the credibility of the publisher and consequently advertiser/brand attitudes (Carlson, 2015). Wojdynski and Evans (2015) found that consumers’ reported more negative attitudes towards the company after consumers realized that native advertisement content was promotional in nature.
10
Another important issue is the fact that native advertising might raise some ethical concerns regarding source-attribution confusion and consumer deception. A properly disclosed native ad should not be harmful; but properly disclosed native advertising is very often a contradiction in terms as a native ad by definition is supposed to look like normal editorial content (Bakshi, 2014). Several studies frond that people often have trouble recognizing the fact that native ads are in fact advertisement (Howe & Teufel, 2014). For example, a study by Hoofnagle and Meleshinsky (2015) found that even when containing a disclosure that a native ad is a sponsored content 27% still attributed it to a journalist instead. In two experiments by
Wojdynski and Evans (2015) only 7% and 18.3% of respondents understood that the article was a paid advertising.
This can lead to consumers processing the piece of content differently, most likely in a less critical manner as their persuasion knowledge has not been activated. Friestad and Wright (1994) proposed a Persuasion Knowledge Model to explain how presence or absence of advertisement cues might affect consumers’ attitudes. Generally, individuals tend to process persuasive messages more critically when they successfully recognize the persuasive attempt (Boerman et al. 2014; Wu et al. 2016). When consumers realize that a communication is an advertisement it leads to activation of protective mechanisms such as increased skepticism, that in turn influence attitudes toward and perceptions of advertising content in a negative manner (Wojdynski & Evans, 2015; Boerman et al., 2015; Friestad & Wright 1994). As consumers are often unaware of the persuasive intent of native advertising they might not, therefore, experience increased cognitive resistance to a native advertisement.
Some previous research, however, found in some situations people may find it easier to identify native advertisement as an ad. For example, it was found that in general people were better able to identify native adverts on social media. Lazauskas (2016) found that consumers were much more likely to be able to identify native advertisement on Facebook compared with the homepage of the publisher. This might be due to consumers being more accustomed to seeing ads in the Facebook News Feed as well as being more familiar with the advertisement label Facebook uses. Based on this, this research is going to use Facebook as the main platform to test the effectiveness of native advertising to minimize the potential confusion over whether a native advertisement is in fact promotional content.
11 2.5 REASERCH GAP
As native advertising only recently started to attract marketing scholars’ attention the research in the area has been limited. So far a large majority of studies focus on understanding the conditions under which consumers perceive native advertising as paid content rather than independent editorial content focusing on secrecy (Campbell & Marks, 2015), disclosure
position and language (Wojdynski & Evans, 2016; Hoofnagle & Meleshinsky, 2015; Sahni & Nair, 2016), disclosure timing (Boerman et al.,2014), and need for regulation (Bakshi, 2014). In terms of how native advertisement is perceived most empirical research focuses on comparing native advertisement and banner advertisement (Tutaj and van Reijmersdal, 2012;Becker-Olsen, 2003). Not many studies, however, focus on differences between native adverts and how to maximize the effectiveness of those ads in creating attitude change among consumers. For advertisers it is vital to be able to identify how to make their ads more effective, what characteristics of native advertising content would make consumers more likely to view and engage with it (Wojdynski, 2016), therefore, making this an important area to look into.
To begin with it is essential to understand how people consume and process native advertising in general. It has been found that people often use mental shortcuts or heuristics to make their decision, due to lack of motivation or capacity (Chaiken, 1980). This is especially relevant
regarding the way people process native advertisement on Facebook as they are faced with hundreds of posts on the News Feed and would likely take only few seconds to make an evaluation of the material they are seeing. Therefore, the cues that trigger these heuristics to be activated are likely have an important effect on the evaluations and its crucial to understand them better. These cues could include content related factors such as use of graphics, headlines and story leads (Knobloch‐Westerwick & Johnson, 2014) and homepage presentation factors such as popularity indicators (Wojdynsky, 2016). Additionally, consumer evaluations can be externally influenced by mood, involvement, perceived credibility of the publisher, age or gender (Knobloch‐Westerwick & Johnson, 2014; Go et al., 2014; Howe & Teufel, 2014). Even though it is known that those factors likely have a significant effect on the attitudes of consumers, it is still largely unclear to what extent those have an influence and which factors are the most important for marketers and advertisers to pay attention to.
12
The research reported in this paper, therefore aims to bridge several gaps in academic
literature. Firstly, it aims to contribute to the existing literature by examining several different factors that might influence the overall effectiveness of native advertising. Based on the previous research three cue - source credibility, brand familiarity and perceived popularity (bandwagon effect) were chosen to be tested to see how much they influence consumers’ attitudes. Next sections will go into more detail why those factors were chosen and how they might be influential in creating attitude change. Secondly, the current study aims to test the cumulative influence of these cues and examine what combinations of these cues would be most effective.
2.6 HEURISTICS
The main goal of any advertising message is persuasion (Zhu & Tan, 2007). In order to be able to effectively persuade consumers it is important to understand how they process advertising and which factors are affect the creation of attitudinal and behavioral change (Wojdynski, 2016).
Some of the most popular models that are used to explain attitude change are the heuristic-systematic model (Chaiken, 1980) and the elaboration-likelihood model (Petty & Cacioppo, 1983) that differentiate between systematic or central route processing from heuristic (peripheral route) processing. The former type implies that people form or update their attitudes by cognitively elaborating persuasive argumentation, while the latter implies that people formed their attitudes by relying on more accessible information such as the source or other non-content cues (Chaiken & Maheswaran, 1994; Sundar, 2008).
Systematic or central route processing is generally more effortful and has a more limited capacity than heuristic or peripheral processing (Chaiken & Maheswaran, 1994; Petty & Cacioppo, 1983) The evidence from attitude change research indicates that in everyday situations involving efforts to persuade, the latter processing type predominates (Iyengar & Valentino, 2000).
13
When processing advertising messages on-line, consumers often don’t have enough time, motivation or capacity due to their sheer volume to process them systematically. For example, whenever someone logs into Facebook, there is an average of 1500 stories from friends and pages they follow for them to see (Backstrom, 2013), and they tend to only spend about 1.7 seconds on any one piece of content they see there (Facebook Business, 2015). This means that is likely that people would not carefully process each post they see on their news feed and would likely engage in heuristic processing to determine the value of different information they see on the page.
There is a number of heuristic cues that consumers might use when evaluating of information online. Those can might include: positioning of the advertising content within the rest of the page area, text and image sizes, text and image content, and wording of any labeling/headings (Wojdynski, 2016), the name of the online agency posting the information, the time the
information was posted, the number of viewers, the characteristics of viewers who recommended a certain story, and other viewers’ comments can all affect consumers’ evaluations (Go et al. ,2014; Sundar et al., 2007).
As this paper is going to focus on Facebook as the platform for native advertisement, it is important to note that due to the specific layout of the website the cues that might trigger heuristic processing are somewhat limited. Picture 1. below shows an example of native advertisement as it would show up in the News Feed of a Facebook user. The top of the post includes the name of the publisher and the advertised brand, date when published, a
sponsorship disclosure (sometimes is presented in a separate line instead of the publication date), followed by a short text summary of the native advertisement, a link to the article and then number of comments, likes and shares on the post.
14
Picture 1. Example of a native advert on Facebook (screenshot from VICE Nederland from December 2016)
Based on this the available cues on Facebook can include: the perceived credibility of the publisher, familiarity/ credibility of the advertiser, story related cues such as headline or use of pictures/graphics, and popularity indicators in the form of likes, shares and comments.
Unfortunately, due to the scope of the study not all of the potential cues can be identifies so three cues were chosen based on the previous research findings. The cues that received most attention and support in the previous research were chosen to be examined in further detail, these included publisher credibility (source credibility), brand familiarity and popularity indicators that can trigger bandwagon effect.
a. SOURCE CREDIBILITY
Perceived source credibility has been found to be one of the important factors that affects individuals’ attitudes after persuasive message exposure (Wu et al., 2016; Wilson & Sherrell, 1993). Source credibility refers to the extent to which the message source is perceived as competent, trustworthy and able to provide correct information without bias (Greer, 2003;
15
Flanagin & Metzger, 2007; Go et al., 2014). In the advertising context, source can be operationalized as the publisher that posts the advertisement or endorsers in the
advertisement itself (Wu et al., 2016; Goldsmith et al., 2002). Previous studies found a strong positive relationship between both the credibility of a media outlet or the endorser’s credibility and consumers’ evaluations of the advertisement.
For example, Hirschman and Thompson (1997) found that consumers who think poorly of a Web site or believe its content lacks credibility will use these evaluations when forming
attitudes toward companies that sponsor the Web site. Additionally, a study by Go at al. (2014) found that the same article published by a credible source is perceived as more credible and having better quality than being published by a less credible source (Wu et al., 2016). Overall, information coming from a high-credibility source leads to greater attitude change compared to low-credibility source, which were found to produce limited change in attitudes (Greer, 2003; Milburn, 1991; Greenburg & Miller, 1966).
The same relationship is likely to work in native advertisement context. A native advertisement is usually posted by publishers unrelated to the advertised brand and, therefore, the publisher would be perceived as an endorser or the source of the native advertisement. This means that such ad is likely to be perceived as more trustworthy and produce greater attitude change when the publisher associated with is perceived as more credible and the opposite is likely to occur with low-credibility publishers. Hence the following hypotheses are proposed:
Hypothesis 1a: Native advertisement from a more credible publisher would result in more positive attitudes towards the advertisement, towards the brand, greater purchase intent and intention to share the advertisement.
Hypothesis 1b: Native advertisement from a less credible publisher would result in
comparatively less positive attitudes towards the brand, towards the advertisement, lower purchase intent and lower intention to share the advertisement.
16 b. BRAND FAMILIARITY
Brand familiarity was identified as another factors that could have an effect on consumers’ response to native advertising. Brand familiarity can be defined as a continuous variable reflecting consumers’ brand knowledge structures or associations that exist within a
consumers’ memory based on the level of direct or indirect experience consumer has with the product (Campbell & Keller, 2003; Alba & Hutchinson, 1987; Kent & Allen, 1994). Brand
experiences that would increase brand familiarity include exposure to various media
advertisements for the brand, exposure to the brand in a store, and purchase or usage of the brand (Alba & Hutchinson, 1987; Park & Stoel, 2005). Previous research has continuously found links between brand familiarity and both purchase decisions, and brand attitudes.
Consumers were found to have a greater motivation to designate attention to product information in advertisements for familiar brands rather than unfamiliar brands (MacInnis et al., 1991) and well-known brands were more likely to achieve better recall and be better able to be protected from competitive advertising interference than less familiar brands (Kent & Allen, 1994; Park & Stoel, 2005). Research has also consistently supported the link between brand familiarity and the intent to purchase. For example, Arora and Stoner (1996) found that purchase intentions were significantly affected by greater name familiarity. Additionally, Hoyer and Brown (1990) found that when consumers were more likely to select the known brand even if it was relatively lower in quality than the unknown counterpart (Sundaram & Webster, 1999).
In the context of native advertising, Wojdynski and Evans (2015) found that when a real brand (Dell) was used instead of a fictitious one attitudes towards the company, perceived story quality and intention to share were not negatively affected after participants realized that the article was a native advertisement. They theorized that brand familiarity might have had a moderating effect on this relationship, but this has not been specifically examined in their study. Based on this the following hypotheses is proposed for this study:
Hypothesis 2a: Native advertisement for familiar brands would produce greater purchase intention, more positive attitudes towards the advertisement and the brand and greater intention to share.
17 Hypothesis 2b: Native advertisement for unfamiliar brands would result in comparatively lower purchase intention, less positive attitudes towards the advertisement and the brand, and lower intention to share.
c. BANDWAGON EFFECT
Due to the social nature of Facebook, popularity indicators such as likes, comments and shares are very prominent. These indicators are often used by consumers to make assumptions about how others feel about on-line content and in turn could influence their personal evaluations though bandwagon effect (Chaiken,1987). The bandwagon effect is a type of heuristic
processing that involves a mental shortcut stating that “if others think something is good, then I should, too”, or in other words that if an opinion is supported collectively it must be credible (Go et al. 2014; Sundar et al. 2008). The effects of this heuristic on users’ perceptions of online content have been widely supported by previous studies (Go et al., 2014; Kim & Sundar, 2010; Sundar & Nass, 2001; Sundar et al., 2008).
In a study by Sundar and Nass (2001) participants were given an identical news story and were told that it was either selected by a news editor, the computer, or other users; the story chosen by other users was found to be the favorite out of the three types showing the bandwagon effect in action. Another study tested the importance of the bandwagon heuristic in the e-commerce context. It was found that higher star ratings as well as sales rankings (both
bandwagon cues) led to significantly higher levels of bandwagon perception, i.e the perception that other people are more likely to buy this product, higher perceived product credibility, perceived product quality, as well as to more favorable attitudes and higher levels of purchase intent (Go et al., 2014; Sundar et al., 2008).
As it has been already noted, in the context of social media there is a limited amount of
editorial cues available for consumers to form a basis for their evaluative judgements (Messing & Westwood, 2014). The cues that are available on Facebook and might potentially trigger bandwagon heuristic are the number of likes and comments. Therefore, the following is hypothesized:
18 Hypothesis 3: Greater amount of likes and comments on a native advertisement story (high bandwagon effect cue) would result in more positive attitudes towards the advertisement, more positive attitudes towards the brand, greater purchase intent and greater intention to share.
d. INTERACTION EFFECTS AMONG SOURCE CREDIBILITY, BRAND FAMILIARITY AND
BANDWAGON EFFECT
In addition to evaluating the effect of each individual heuristic on consumer attitudes and behaviors, it is important to examine whether there is an interaction between them as those would not appear in isolation in real life situations. Sundar et al. (2007) proposed
cue-cumulation effect based on logic of dual-process persuasion models. They argue that positive effects of one cue might be enhanced further by the presence of other positive cues. They also provide some empirical evidence for this. They found that when news reported by a low-credibility source were evaluated, if number of related articles (bandwagon cue) and the recency cue were both at the highest level it led to people evaluating the news story as more credible. Additionally, a study by Go et al. (2014) found that when expertise and bandwagon cues were both high, the perception of news credibility was enhanced. Kim and Sundar (2011) also found some interaction effects between the expertise and bandwagon cues. When high levels of the expertise cue were paired up with high levels of the bandwagon cue, participants rated a website’s content more positively (Go et al. 2014).
The same relationship found between different heuristic triggering cues in the context of news credibility perception might also apply in the context of native advertising. Due to the similarity of native advertising to other on-line content including news stories it might be plausible to assume that similar mechanism would work when evaluating the quality or credibility of a native advertisement. Therefore, the following hypothesis has been proposed:
Hypothesis 4: There is going to be a significant interaction effect between source credibility, brand familiarity and the bandwagon effect on consumers’ evaluations and purchase behavior.
19 2.6 CONCEPTUAL FRAMEWORK
The following diagram graphically illustrates the hypotheses discussed earlier in the paper:
20
Chapter 3. Method
This section is going to outline the experimental design used in this study including stimuli design, sample and measures employed.
3.1 RESEARCH DESIGN
The aim of the study was to test the effect of several heuristic cues on consumers’ evaluations of a native advertisement in the social media environment. To test the hypotheses, a 2 (high source credibility v. low source credibility) x 2 (familiar brand v. unfamiliar brand) x 2 (number of comments and likes: high v. number of comments and likes: low) factorial between-subjects design was adopted. The effects of the three independent variables manipulated in the study were measured on the three dependent variables, attitudes towards the advertisement, attitudes towards the brand and purchase intent. To test this a cross-sectional questionnaire was developed and participants were randomly assigned to eight different treatments.
Figure 2. 2 x 2 x 2 factorial design
High source credibility Familiar Brand Unfamiliar Brand Number of comments and likes: high Condition A Condition B
Number of comments and likes: low Condition C Condition D
Low source credibility Familiar Brand Unfamiliar Brand Number of comments and likes: high Condition E Condition F
Number of comments and likes: low Condition G Condition H
3.2 MEASURES
In order to test the effectiveness of native advertising it is important to outline key
performance indicators. Behavioral intent has been found to be one of the key indicator of the effectiveness of an advertisement. The ultimate goal of any persuasive message is to produce
21
behavior change (Zhu & Tan, 2007) and behavioral intention has been found to be the closest predictor of an actual consumer behavior (Ajzen & Fishbein, 1970; in Zhu & Tan, 2007). Hence purchase intention is often used to anticipate actual consumer behavior and is a common effectiveness measure (Li et al., 2002).
Another important indicator that can predict the effectiveness of an advertisement to produce behavior change is consumer attitudes. Previous research suggested that both brand and advertisement attitudes play an important role in creating purchase intention among
consumers (Laroche et al., 1996; Biehal et al. 1992). Attitudes towards an advertisement (Aad) was found to be one of the factors that influences brand attitude (Ab) and intent to purchase (Phelps & Thorson, 1991; Gresham & Shimp, 1985). It is important to distinguish between Aad and Ab as depending on other factors they might work independently. For example, Aad might be more important for unfamiliar brands as consumers don’t have any Ab yet and would likely base those solely on Aad; while for familiar brands Aad might play a smaller role as consumers already have previous Ab (Phelps & Thorson, 1991). Therefore, all three indicators: Aad, Ab and purchase intent will be used in this study to measure the effectiveness of a native
advertisement. Using attitudinal measures is important in the case of native advertising as it is generally focused more on creating positive attitudes and interest on consumer’s part rather than a direct sale (Carlier et al., 2016; Campbell & Marks, 2015). Additionally, intention to share the advertisement was added as a common measure that was used in relation to content on-line (Wojdynsky and Evans, 2015).
The next section outlines all other variables and measures used to test as well as which measurement scales were used and adapted for the purpose of this paper.
INDEPENDENT MEASURES:
Source credibility. A number of potential publishers were checked in pre-test, two were then
chosen- one perceived as a credible source and one as not credible. In the survey participants rated their perception of credibility on a 7-point scale (ranging between ‘‘strongly disagree’’ and ‘‘strongly agree’’) and using 3 sets of adjectives (trustworthy/untrustworthy,
22 Brand familiarity. Two brands were used in the study- one existing and one fictitious. To double
check participants were asked to indicate whether they were familiar with the brand prior to seeing the advertisement.
Bandwagon effect was manipulated by varying the number of comments and likes on a native
advertisement. To measure the bandwagon effect participants were asked to indicate “how likely are other people to view the full article?” (not likely/ very likely); “Many people recommended this article”, “Other people found this article interesting” (strongly agree/ strongly disagree).
DEPENDENT MEASURES:
Attitude toward the brand (Ab) was measured using four items chosen from frameworks used
in previous research (Spears & Singh, 2004; Wojdynski & Evans, 2015). Participants rated their perception of the brand on a 7-point Likert scale (strongly disagree/strongly agree) rating four sets of adjectives (unappealing/appealing, bad/good, unfavorable/favorable,
unlikeable/likeable).
Attitudes towards the advertisement (Aad) were also measured using a 7-point Likert scale,
rating five sets of adjectives (interesting/uninteresting, useful/useless, entertaining/ not entertaining, positive/negative, appealing/unappealing).
Purchase intention was measured by asking the participants whether they would consider using
the brand next time they book a holiday (definitely yes/ definitely no).
Intent to share the story was measured on 7-point Likert scale where participants rated a
statement “I would recommend this story to a friend” ranging between “strongly disagree to strongly agree”.
3.3 STIMULI DEVELOPMENT
A native advertisement has been adopted for this study from an existing article from Atlantic (Machado, 2014). The content of the advertisement was kept the same in all eight conditions, while the brand name, publisher name and the number of likes/comments were manipulated.
23
Keeping content consistent is important to ensure internal validity and in order to be able to isolate the effects of different heuristic cues being tested.
A screenshot on a Facebook News Feed was used for the stimulus material with one of the stories changed to the native advertisement adopted for the study. To test the effects of credibility heuristic, the source of advertisements was manipulated in the different conditions- for a high credibility source ‘The New York Times’ was used and in low credibility condition ‘Buzzfeed’ was used as a publisher. The publishers were chosen based on a short pre-test survey (N=11) distributed among University of Amsterdam students. To test brand familiarity, two brands were used in the study, one fictitious (Travel First) and one established brand (Booking.com). Lastly, to test the effects of the bandwagon effect likes comments and shares on the native advertisement have been manipulated. Two bandwagon effect conditions had a set amount of likes, comments and shares on the native advertisement story- in all the low bandwagon effect conditions likes (N=83), comments (N=8), shares (N2); in all the high
bandwagon effect conditions- likes (N=6400), comments (N=584) and shares (N=842) as well as a top row of the advertisement showing how many friends liked the publishers page.
3.4 SUBJECTS
As the main focus of the study is social media and on-line advertising it made sense to do all the data collection on-line. Therefore, an on-line survey was developed using Qualtrics Survey Software and then distributed though the University of Amsterdam social network, personal network of contacts (distributed by email and social media) and though Amazon’s Mechanical Turk (MTurk), web-based human intelligence crowd sourcing platform. Overall, 286 participants completed the survey- 150 though MTurk and 134 though social media. Participants ranged between the ages 19-65, 83% of whom held a bachelor’s degree or higher. A slightly higher percentage of males (56.9%) completed the survey compared to the female participants (43.1%). 30% of participants were from the United States of America, 37% from India, 11% for the Netherlands and 8% from the United Kingdom, with the rest spread out between other 21 countries.
24 3.5 PROCEDURE
The questionnaire was created using Qualtrics Survey Software and participants were able to access the survey using an anonymous web link. Firstly, participants were shown an
introduction page that specified for what purposes the data will be used and indicated that all the data will all be kept anonymous and confidential. After that participants were randomly assigned to one of the eight conditions.
On the next page, participants were shown one of the stimulus materials developed and asked to identify any advertisements on the page they could find- this was done to check whether participants recognized whether the native advertisement was in fact a promotional material and not a normal post. After that participants were taken to the next page which shown the same stimulus material- a short paragraph stated that the second post was actually a native advertisement and a quick description of what it is was given to ensure all participants were aware of what it was. After that 12 questions were asked in relation to the native
advertisement- testing participants’ familiarity with the brand, perception of the credibility of the publisher, bandwagon effect perception, attitudes towards the advertisement, brand and purchase intent. On the following page participants were asked a few questions about their background such as age, gender, education level.
25
Chapter 4. Results
4.1 FREQUENCY CHECK
As a first step in the analysis, a frequency check has been run on the data collected to ensure there were no errors or missing data. The analysis showed sowed no errors in the data and a small percentage of missing data cases (N=2) found were excluded from further analysis, therefore making the total sample 284 instead of 286. Three variables in the analysis included counter-indicative items. Those were re-coded to be further used in the analysis.
4.2 MANIPULATION CHECKS
To ensure successful manipulation of the independent variable manipulation checks (t-test) were conducted for all three independent variables.
Source credibility. The result shows a significant difference between the means for high
credibility conditions (M=1.63; SD=0.725) and low credibility conditions (M=3.23; SD=0.926; t (286) = -16.1; p>.001). This shows that participants perceived the publisher in the high
credibility conditions as significantly more credible, showing that the manipulation has been successful.
Brand familiarity. The t-test found that participants in the high familiarity group were much
more familiar with the brand (M=1.39; SD=.854) than those in the low familiarity brand group (M=3.4; SD=1.021; t (286) = -21.2; p>.001). This again shows that manipulation of this variable has been successful.
Bandwagon effect. In the high bandwagon effect condition (M=2.81; SD=1.104) the article was
perceived as more popular than in the low bandwagon effect condition (M=3.15; SD=1.26; t (286) = -2.4; p>.01). The results were significant but the difference between the means was not as large as in the other conditions.
26 4.3 RELIABILITY ANALYSIS
Reliability analysis has been completed to check the consistency of the measurements. It has been completed for all items in the attitudes towards the advertisement variable, attitudes towards the brand variable, source credibility variable and bandwagon effect variable. Almost all variables had high reliability, with Cronbach’s Alpha > 0.7 as shown in the table 1. In brand attitudes variable one item was excluded from the analysis as it significantly affected
Chronbachs Alpha, therefore only three out of four items were used in the further analysis.
The corrected item-total correlations show that all remaining items have a good correlation with the total score of the scale being above 0.30, meaning that none of the items would substantially affect reliability if they were deleted.
Table 1. Cronbachs Alpha
Variable Cronbachs Alpha
Attitudes towards the ad 0.857
Attitudes towards the brand 0.778
Source credibility 0.811
Bandwagon effect 0.824
Brand familiarity N/A*
Purchase Intent N/A*
Intention to share N/A*
* only included one item
4.3 HYPOTHESES TESTING
HYPOTHESES 1-3.
Hypothesis 1a and 1b were examined using independent-sample t-test, which treated source credibility as an independent factor that had two groups: “high credibility” and “low
27
advert, attitudes towards the brand, purchase intent and the intention to share. The means between two credibility groups were compared for each independent variable. The results of the test are presented in the Table 2 below.
Source credibility had a significant effect on attitudes towards the advertisement t (282) = -12.50, p <.001 and attitudes towards the brand t (281.88) = 7.39, p < .001. It did not, however, have a significant effect on purchase intent, t (278.8) = -.895, p > .05 and intention to share, t (282) = -1.32, p > .05. For attitudes towards the advertisement there was a significant
difference in the scores for high source credibility (M=1.7, SD=0.54) and low source credibility (M=2.8, SD=0.88), which suggests that that source credibility does influence how people evaluate the advertisement, i.e. it shows that when the source credibility is high people would have more positive attitudes towards the advert.
A similar relationship is found for attitudes towards the brand, high source credibility (M=1.97, SD=0.61) and low source credibility (M=2.5, SD=0.65). This again shows that source credibility has a positive relationship on consumer evaluations.
Overall, H1a and H1b are both partly supported. Source credibility has a significant effect on Aad and Ab but not on the purchase intent or intention to share.
Table 2. Effects of source credibility on dependent measures Dependent variable Source
credibility
N M SD t Sig.
Attitudes towards the ad High 139 1.70 .54 -12.6 0.001
Low 145 2.80 .88
Attitudes towards the brand High 139 1.97 .61 7.39 0.000
Low 145 2.5 .65
Purchase intent High 139 2.68 1.11 -.89 0.371
Low 145 2.80 1.09
Sharing intent High 139 2.82 1.21 -1.32 0.190
28
An independent-samples t-test was also conducted to compare evaluations in high brand familiarity and low brand familiarity conditions to test hypotheses 2a and 2b. The results are presented in the table 3 below.
For attitudes towards the ad- there was a significant difference in the scores for high brand familiarity (M=2.08, SD=0.87) and low brand familiarity (M=2.45, SD=0.93) conditions; t (275.7) = -3.38, p <.01. These results suggest that brand familiarity does influence how people evaluate the advertisement. Comparing mean scores it can be concluded that there is a higher chance of more positive Aad in the high familiarity condition.
For attitudes towards the brand- there was also a significant relationship between brand familiarity and Ab, t (280.48) = 6.6, p < .001. The mean score for the high familiarity condition (M=2.0, SD=0.64) was significantly higher than the mean score for the low familiarity condition (M=2.5, SD=0.63) suggesting that comparison there are generally more positive brand attitudes in the high familiarity condition.
Interestingly, there a significant relationship has been found between brand familiarity and the intention to share, t (272.7) = 2.80, p < .01. There was a significant difference in the scores for the high familiarity condition (M=3.09, SD=1.1) and low familiarity condition (M=2.7, SD=1.19), this means that people were more likely to want to share the advertisement with a friend if they were not familiar with the brand.
No significant relationship has been found between the brand familiarity and purchase intent, t (280) = 0.356, p > .05.
Again, the hypotheses have been partially supported. A significant relationship has been found between the brand familiarity and attitudes towards the brand as well as attitudes towards the advertisement. A significant relationship has been found between the brand familiarity and the intention to share but in a different direction that that predicted in the hypotheses 2a and 2b. No significant relationship has been found between the brand familiarity and purchase intent.
29
Table 3. Effects of brand familiarity on dependent measures Dependent variable Brand
Familiarity
N M SD t Sig.
Attitudes towards the ad High 148 2.08 .87 -3.38 0.001
Low 136 2.45 .93
Attitudes towards the brand High 148 2.00 .64 -6.63 0.000
Low 136 2.5 .64
Purchase intent High 148 2.76 1.04 .356 0.722
Low 136 2.72 1.18
Sharing intent High 148 3.09 1.90 2.8 0.005
Low 136 2.71 1.20
An independent-sample t-test was performed to test hypothesis 3, that stated that more likes and comments (high bandwagon effect condition) would lead to more positive Aad, Ab, greater purchase intent and intention to share. The findings are summarized in the Table 4.
Bandwagon effect had a significant effect on attitudes towards the advertisement t (274.5) = -3.39, p <.01 and attitudes towards the brand, t (281.6) = 6.39, p < .001. For attitudes towards the advertisement there was a significant difference in the scores for high bandwagon effect condition (M=2.07, SD=0.94) and low bandwagon effect condition (M=2.44, SD=0.85), which suggests that bandwagon effect cue positively affects positive Aad. There was also a significant difference in scores between conditions for Ab- in high bandwagon effect condition (M=1.99, SD=0.64) low bandwagon effect condition (M=2.48, SD=0.65), confirming a positive relationship between the bandwagon effect level and Ab.
No significant relationship has been found between the bandwagon effect and purchase intent, t (279) = -1.68, p > .05 or between the bandwagon effect and the intention to share, t (281.4) = -1.01, p > .05.
Again, the hypothesis has been partially supported. A significant positive relationships have been found between the bandwagon effect, Aad and Ab, while no significant relationships were found between the bandwagon effect and purchase intent or intention to share.
30
Table 4. Effects of bandwagon effect on dependent measures Dependent variable Bandwagon
effect
N M SD t Sig.
Attitudes towards the ad High 138 2.07 .94 -3.39 0.001
Low 146 2.44 .85
Attitudes towards the brand High 138 1.99 .64 6.39 0.000
Low 146 2.48 .65
Purchase intent High 138 2.63 1.09 -1.68 0.093
Low 146 2.85 1.09
Sharing intent High 138 2.84 1.15 -1.01 0.311
Low 149 2.98 1.15
HYPOTHESIS 4.
Hierarchical multiple regression was performed to investigate whether the credibility level, brand familiarity and bandwagon cues predict the level of attitudes towards the advertisement, after controlling for age and gender.
In the first step of hierarchical multiple regression, two predictors were entered: gender and age. The model was not statistically significant F (2, 282) = 0.477, p > .05. The second step introduced the credibility level, brand familiarity and bandwagon effect variables. This time the model was statistically significant with F (3, 281) = 69.56; p <.001 and explained 42.7% of variance in the level of attitudes towards the ad. Credibility was found to be the best predictor of consumer attitudes towards the ad with the highest Beta value (β = .592, p < .001) and when ran separately was found to explain 35% of variance. Familiarity had a Beta value of .199, p < .001 and on its own explained 3.9% of the variance, while bandwagon cue had a Beta value of .175, p < .001 and on its own explained 3.1% of the variance in consumer attitudes toward the ad. The results are presented below in the table 4.
Table 5. Hierarchical Regression Model- Aad
R R2 R2 Change B SE β T Sig.
Step 1 .058 .003 .003 .621
Age -.059 .072 -.049 -.818
Gender -.063 .110 -.034 -.574
31 Age -.036 .056 -.030 -.653 Gender -.039 .085 -.021 -.466 Source Credibility 1.070 .084 .586 12.8 Brand Familiarity .371 .083 .203 4.44 Bandwagon Effect .329 .084 .180 3.90
Additionally, a hierarchical multiple regression was performed to check whether credibility, brand familiarity and bandwagon cues have an effect on attitudes towards the brand. The findings are summarized below in the table 6. Again, only the second model was statistically significant with F (3, 281) = 65.34; p <.001 and explained 41.2% of the variance in the level of brand attitudes among participants. Credibility on its own explained 15.5% of the variance; familiarity- 13.3% and bandwagon cue 12.4%.
Table 6. Hierarchical Regression Model- Ab
R R2 R2 Change B SE β T Sig. Step 1 .087 .008 .008 .341 Age .077 .054 .085 -.1.43 Gender -.023 .083 -.017 -.28 Step 2 .646 .418 .410 .000 Age .056 .042 .061 1.32 Gender -.05 .064 -.036 -.78 Source Credibility .542 .064 .394 8.53 Brand Familiarity .496 .064 .360 7.80 Bandwagon Effect .465 .064 .338 7.30
No significant relationship has been found between the credibility, brand familiarity and bandwagon cue and purchase intent.
Additionally, a three-way ANOVA test (credibility x familiarity x bandwagon) was performed to examine the interaction effects among the three different cues on Aad. Independent variables were all measured at the nominal level (two groups: “high” and “low”), while the dependent variables were all operationalized at continuous interval level. Table 3 below presents the
32
results of the test. All three independent variables separately had a significant effect on the Aad- source credibility, F (1, 276) = 173.49, p <.001, η² = .39; brand familiarity, F (1, 276) = 18.64, p <.001, η² = .06; and the bandwagon effect, F (1, 276) = 15.94, p <.001, η² = .055. Only one two-way relationship between credibility and bandwagon cue was significant F (1, 276) = 3.68, p < .05. This means that only one combination of heuristics, namely high credibility source and high bandwagon perception, lead to significantly more positive attitudes towards the advertisement. No other two-way relationships or a three-way interaction were found to be significant.
Table 6. Factorial ANOVA-Aad
Treatment Variable SS DF MS F η² Sig.
Source Credibility 82.91 1 82.91 173.49 .387 .000
Brand Familiarity 8.91 1 8.91 18.64 .063 .000
Bandwagon Effect 7.61 1 7.62 15.94 .055 .000
Source Credibility * Brand Familiarity 0.002 1 0.002 0.003 .955 .000 Source Credibility * Bandwagon
Effect
1.757 1 1.76 3.676 .013 .056
Brand Familiarity * Bandwagon Effect
0.975 1 0.975 2.04 .007 .154
Source Credibility *Brand Familiarity * Bandwagon Effect
0.511 1 0.55 1.08 .005 .302
Error 131.90 276 .478
Total 1687.38 284
Next another three-way ANOVA was performed to examine the interaction effects among the three different cues on Ab. The Table 7 below summarizes the results obtained. Again all three cues were found to have a significant effect on the Ab independently. There was a significant main effect of source credibility on Ab, F (1, 276) = 78.63, p <.001, η² = .22; a significant relationship between brand familiarity on Ab, F (1, 276) = 64.29, p <.001, η² = .189; and a
33
significant main effect of the bandwagon effect on attitudes towards the brand, F (1, 276) = 56.747, p <.001, η² = .171.
A significant three-way relationship between the three heuristic cues has been discovered F (1, 276) = 5.28, p < .05, η² = .019. Additionally, two significant two-way relationships have been found between credibility and familiarity (F (1, 276) = 3.97, p < .05, η² = .014), and between credibility and bandwagon cue (F (1, 276) = 6.51, p < .01, η² = .023). This means that the
advertisement offered by a high credibility source, sponsored by a familiar brand and having a high level of bandwagon effect cues resulted in more positive brand attitudes. The same was also true for high credibility source with high familiarity but low bandwagon effect; and for high credibility and high bandwagon effect with low brand familiarity.
No significant interaction effect was found between brand familiarity and the bandwagon effect, F (1, 276) = .234, p =.352, η² = .003.
This means that the hypothesized cumulative relationship between source credibility, brand familiarity and the bandwagon effect has been is supported.
Table 7. Factorial ANOVA-Ab
Treatment Variable SS DF MS F η² Sig.
Source Credibility 21.165 1 21.165 78.636 .222 .000
Brand Familiarity 17.306 1 17.306 64.296 .189 .000
Bandwagon Effect 15.274 1 15.274 56.747 .171 .000
Source Credibility * Brand Familiarity 1.069 1 1.069 3.972 .014 .047
Source Credibility * Bandwagon Effect 1.753 1 1.753 6.512 .023 .011
Brand Familiarity * Bandwagon Effect 0.234 1 0.234 .869 .003 .352
Source Credibility *Brand Familiarity * Bandwagon Effect
1.420 1 1.420 5.277 .019 .022
Error 74.287 276 .269
34
Chapter 5. Discussion and conclusions
Overall, the current study confirmed the effects of the different heuristic cues on consumers’ evaluations of the native advertisement. The results supported the influence of the source credibility, brand familiarity and bandwagon effect cues on participant’s attitudes towards the advertisement as well as attitudes towards the brand advertised. Moreover, the findings suggest that there are some interaction effects between the different heuristics in line with Go et al. (2014) and Sundar et al. (2006) results. These findings are important as they help to gain a deeper insight in the effect of different cues in making native advertising more effective in creating attitudinal change. Additionally, they help to understand the interaction between those factors and which conditions lead to the best results. This in turn has important theoretical and practical implications for researchers, marketers and advertisers.
First of all, this study confirmed that source credibility is an important factor in creating positive attitudes among consumers. Source credibility was found to be the most influential factor out of the three different heuristics examined. This means that even when bandwagon effect and band familiarity is low, source credibility would still significantly influence consumers’
evaluations. High source credibility significantly boosted people’s attitudes towards both the advertisement and the brand, while low source credibility had a negative effect on those. This supports the findings from previous studies on source credibility and attitude change such as Greer (2003) and Milburn (1991). Over 86% of respondents stated that the credibility or reputation of the publisher was important to them ad that it affected how they viewed brands that those publishers chose to advertise. Moreover, 70% of the respondents agreed that credible publishers would be more likely to work with credible brands.
These finding also have some important practical implication for advertisers and publishers. As it was found that source credibility has a particularly strong effect on the consumer’s attitudes towards the brand, it means that advertisers need to be particularly careful when choosing the right publisher to work with. If consumers perceive the publisher as credible they are more likely to have more positive attitudes overall. Additionally, particularly for unfamiliar brands
35
consumers would be more likely to assume that the brand is trustworthy if it is advertised by a credible publisher.
Source credibility, however, did not have a significant effect on the purchase intent or on the intention to share the advertisement. In fact, no significant relationship has been found between any of the heuristic cues tested and purchase intent contrary to the predictions from some previous studies, for example, Arora and Stoner (1996). This could, however, potentially be explained by the fact that native advertisement in general is more focused on creating awareness and attitudinal change among consumers through good quality, useful and/or entertaining content rather than trying to sell a product. Overall, it could mean that purchase intent in general might not be an appropriate measure for testing the effectiveness of native advertisement as the purpose of those ads is slightly different to traditional advertisement.
Secondly, this study confirmed that both brand familiarity and bandwagon cues also have an effect on consumer’s evaluations of the advertisement and the brand. There was a significant difference in the consumer evaluations of the advertisement and the brand between the high familiarity (or high bandwagon effect) and low familiarity (or low bandwagon effect) conditions. No relationship has been found between the bandwagon effect and intention to share.
Interestingly, a significant effect was uncovered of brand familiarity and sharing intention, this relationship was, however, the opposite of the one that was predicted in the hypothesis 2. People were more likely to indicate that they would share this article in the low brand
familiarity condition compared to the high brand familiarity condition. One possible explanation for this could be that use of a fictitious brand made the advertising intent less obvious,
therefore, making people more likely to feel comfortable with sharing it.
Additionally, some two-way and three-way relationships were found between the heuristic cues examined, and consumer evaluations. In terms of the attitudes towards the advertisement only one significant two-way interaction was found among the three different heuristic cues, one between credibility and bandwagon effect. It means that an advertisement published by high credibility source and liked/commented on by many other Facebook was rated significantly better by the participants compared to any other combinations of the credibility level and the bandwagon effect.
