Selling their souls to the advertisers? How native advertising degrades the quality of prestige media outlets

ABSTRACT Native advertising – disguising sponsored content as journalistic material – is not a new phenomenon, but only recently has it reached news media outlets that stand for high-quality journalism. To investigate the potential impact of native advertising on perceived media quality, a 2 (source: low vs. high quality media outlet) × 3 (content: ad-free article vs. declared native advertising article vs. undeclared native advertising article) factorial between-subjects experiment was conducted. This experimental study found that news media outlets that stand for high-quality journalism damage their quality as perceived by recipients when publishing properly declared native advertising.

Google and Facebook have disrupted the advertising industry on a global scale, and Amazon is on the rise in the digital advertising market. As the saying goes, when elephants fight, it is the grass that suffers. While tech companies' revenues soar into billions of dollars, legacy news media companies in particular are suffering since advertisers have begun to shift their budgets away from them. Consequently, media companies have not only launched cost-cutting programmes but have also looked to exploit other sources of revenue, with native advertisingsponsored content disguised as journalistic materialamong them (Conill, 2016;Foreman, 2016, pp. 175-179). Native advertising, also called sponsored news, waters down the separation between journalistic and advertising content in order to sell exclusive advertising opportunities with the added benefits of a trustworthy, context-sensitive advertising environment and the circumvention of adblocking software (MacNamara, 2016, p. 134;Matteo & Dal Zotto, 2015, p. 177). Native advertising must be declared (Federal Trade Commission, 2015), but its aim is nevertheless, as Evans and Parks (2015) put it, "to make the content appear less like an advertisement and more like the media in which it is placed" (p. 159).
Native advertising is not a new phenomenon, but only recently has it reached media outlets of quality journalism. Given that critics such as Taylor (2017) call native advertising "the black sheep of the marketing family" (p. 207), it appears remarkable that even news outlets that stand for the highest quality, such as The Washington Post, Le Monde, and The New York Times, today offer native advertising models (Matteo & Dal Zotto, 2015, p. 180;Schauster, Ferrucci, & Neill, 2016;Wojdynski, 2016;Wojdisnky & Golan, 2016;Wu et al., 2016Wu et al., , p. 1494. But from their perspective, it can be argued that native advertising enables news outlets to offer more value for recipients by providing advertising with engaging and relevant content rather than plain forms of advertising. For example, in 2014, The New York Times promoted Netflix's second season of "Orange Is the New Black" using an interactive paid post entitled "Women Inmates: Why the Male Model Doesn't Work" (Deziel, 2012; see also DeCarvalho & Cox, 2016).
Nevertheless, advertising under the guise of editorial content not only contradicts journalists' professional ethics (Laitila, 1995, p. 538), but it also puts the media brand at risk because recipients may feel deceived and disappointed when they discover that an article was sponsored (Amazeen & Wojdynski, 2018;Austin & Newman, 2015, p. 103;Federal Trade Commission, 2013;2015;Lee, Kim, & Ham, 2016;Wojdynski & Evans, 2016). In a case study, Carlson (2015) examines the public outcry against The Atlantic caused by an "odd story" on the magazine's website. Although the piece was properly labelled as sponsored content, many readers were indignant because it was written and paid for by the controversial Church of Scientology (p. 860). While this may be an extreme example, it illustrates the importance of examining the possible risks and side effects of native advertising with regard to quality journalism.
The aim of this study is to examine whether media outlets that stand for high-quality journalism are affected by native advertising differently. It is our assumption that when high quality news outlets offer native advertising, they put their most valuable assets at risk: their credibility and quality as perceived by their recipients. This question, which has not previously been analysed in empirical research, is not only relevant for media managers, who need to decide whether or not to offer native advertising, but also for democratic societies in general, as high-quality news media play an essential role in the formation of public opinion (Lacy, Fico, & Simon, 1991).
To investigate the risk of native advertising with regard to perceived media quality, we combined the persuasion knowledge model and the concept of affective resistance strategy into a coherent model. This model was tested by conducting a 2 (source: low vs. high quality media outlet) × 3 (content: ad-free article vs. declared native advertising article vs. undeclared native advertising article) factorial between-subjects experiment. A random sample of 384 persons from the language-assimilated resident population of German-speaking Switzerland completed the questionnaire. Structural equation modelling (SEM) reveals that news media outlets that stand for high-quality journalism damage their quality as perceived by recipients when publishing properly declared native advertising.
Literature review: From advertorials to native advertising Unlike many others, the term native advertising is not a contested one. According to Matteo and Dal Zotto (2015), "[a] native advertising strategy is based on the adaption of the look-and-feel, the visual design, the usability and the ergonomics of the publisher's website" (p. 177). Wojdynski and Evans (2016) define native advertising similarly, but without limiting it to a digital context, explaining it as "a term used to describe any paid advertising that takes the specific form and appearance of editorial content from the publisher itself" (p. 157). Referring to the idea that media companies sell their products in two separate markets, namely the advertising market and the recipient market (Picard, 1989, p. 17), we understand native advertising as a specific product offered by media companies to advertisers. This content is published, aired, or placed in the form of a journalistic product at a media outlet. The masked aim is to gain recipients' attention and trust, thereby promoting and selling the advertiser's product or service.
Attempting to disguise advertising as journalistic material is not a new practice. As long ago as 2 January 1915, Theodore MacManus was subtly advertising Cadillacs on behalf of General Motors in an essay titled "The Penalty of Leadership", published in the Saturday Evening Post (Fox, 1984, p. 72). Since then, many different types of masked marketing practices have evolved, as demonstrated by Petty and Andrew (2008, p. 8), with advertorials being one of the most common. Advertorials were increasingly applied in the 1980s and 1990s, particularly in magazines, thereby attracting the attention of communication scholars (Ju-Pak, Kim, & Cameron, 1995;. In our opinion, the terms advertorials, infomercial, and native advertising stand for the same basic practice: camouflaging advertising as journalistic content. The sole difference is that advertorials refer to printed media, infomercials to television, and native advertising to online media (Foreman, 2016, p. 175).
Most research on advertorials in this period resembles current research on native advertising. Therefore, by adapting a literature review on advertorials from Kim, Pasadeos and Barban (2001, p. 267), we distinguish three main strands of empirical research regarding the practices of blurring the line between editorial and advertising content in a journalistic context: (a) content analyses on the use and abuse of these practices (Cameron, Ju-Pak, & Kim, 1996;Conill, 2016;Ju-Pak et al., 1995;Stout et al., 1989), (b) interviews and surveys on attitudes towards these practices among media managers or advertisers (Cameron & Haley, 1992;Harms, Bijmolt, & Hoekstra, 2017;Poutanen, Luoma-Aho, Suhanko, 2016;Schauster et al., 2016), and (c) (including this present study) experimental studies on the persuasiveness of these practices and how they are perceived by recipients (Amazeen & Muddiman, 2018;Amazeen & Wojdynski, 2018;Boerman, Reijmersdal, & Neijens, 2015;Howe & Teufel, 2014;Kim et al., 2001;Sweetser, Ahn, Golan, Hochman, 2016;van Reijmersdal, Neijens, & Smit, 2005;Wojdynski, 2016;Wojdynski et al., 2017). Native advertising and media quality from the recipients' perspective Native advertising undermines news media quality The practice of native advertising is controversial because the deliberate confusion of advertising and journalistic content compromises media quality (Schauster et al., 2016). Although media quality is not an inherent property of news, there are generally valid criteria used to assess media quality, such as relevance, diversity, contextuality, and professionalism (Medienqualität Schweiz, 2016;Schatz & Schulz, 1992). Native advertising exists in tension with these normative criteria, which are rooted in democratic theory, considering that the content serves the advertiser's rather than the public's interest (relevance), does not convey diverse perspectives and opinions (diversity), does not place facts and opinions in a wider context (contextuality), and is not a result of independent reporting (professionalism). In liberal democracies at least, these criteria of media quality are not only codified in regulatory and self-regulatory frameworks (e.g. press councils, journalistic codes of ethics, industry guidelines); they are also widely known and recognised by the recipients of quality news (Jarren, 2008;Medienqualität Schweiz, 2016;Urban & Schweiger, 2014). Therefore, recipients would not expect native advertising content in high-quality news outlets (Wu et al., 2016(Wu et al., , p. 1493, and even if labelled as "sponsored content", it still might be considered misleading (Hoofnagle & Meleshinsky, 2015;Wojdynski, Evans, & Hoy, 2018).
Therefore, we assume that native advertising undermines recipients' perceptions of news media quality. This assumption is in line with previous experimental studies on native advertising that focus on similar constructs. Although Howe and Teufel (2014) found no effect of native advertising on a news website's credibility, findings from Amazeen and Wojdynski (2018) suggest that those recipients who recognise an online article as native advertising have less favourable opinions of the news media outlet. Moreover, another study by Amazeen and Muddiman (2018) reveals that legacy and especially online news publishers both harm their brand reputations through native advertising. Against this background, this paper's first hypothesis is as follows: H1. When publishing native advertising content, news media outlets damage their quality as perceived by recipients, (a) especially those media outlets that stand for high quality journalism, and (b) even if the content is properly labelled as native advertising.

Recipients' persuasion knowledge and resistance strategies
The assumed impact of native advertising on recipients' perceptions of media quality is certainly not a direct one, but rather mediated by their recognition and processing of this type of persuasive content. Therefore, in keeping with general recommendations on persuasion effects research (O'Keefe, 2003, p. 237), we take mediating factors into account.
In research on persuasion processes, the persuasion knowledge model by Friestad and Wright (1994) has become a basic theory (Ham, Nelson, & Das, 2015, p. 209). Its premise is that recipientslike human beings in generalare knowledgeable agents who know a great deal about their world and what is acceptable and what is not in a social context. The abilities to cope with persuasive advertising attempts are acquired through a lifelong process of media socialisation (Friedstad & Wright, 1994, pp. 3-4;Wright, 2002). Consequently, adult recipients are in general able to defend themselves against persuasive advertising attempts, as empirical studies on the basis of the persuasion knowledge model show (e.g. Rozendaal, Buijzen, & Valkenburg, 2010;Wright, Friestad, & Boush, 2005).
In empirical research into the effects of advertising, various scholars have extended and refined the persuasion knowledge model (e.g. Amazeen & Muddiman, 2018;Amazeen & Wojdynski, 2018;Evans & Park, 2015). With regard to native advertising, conjunctions with both psychological reactance and resistance have proved fruitful (Boerman, van Reijmersdal, & Neijens, 2012;van Reijmersdal et al., 2016). According to this strand of research, recipients not only have "conceptual persuasion knowledge", which is the cognitive ability to recognise when one is confronted with masked advertising, but they also show affective reactions. By using an anti-binge drinking advertisement, Dillard and Shen (2005) show that the construct reactance is empirically best reflected by the dimensions "anger and negative cognitions" (p. 162; see also Quick & Stephenson, 2007, for further evidence). Transferring the assumptions on sponsorship disclosure from Neijens (2012, p. 1050), as well as from van Reijmersdal et al. (2016Reijmersdal et al. ( , p. 1463, the following hypothesis was formulated: H2. Native advertising content (a) activates conceptual persuasion knowledge, (b) which consequently leads to the use of affective resistance strategies.

Recipients' assessments of media quality
According to the persuasion model, the moment after a recipient becomes aware of an unwanted persuasive attempt, she or he reassesses the meaning of the situation. At the end of this process, the recipient may refine her or his attitudes towards the content and the persuader (Friestad & Wright, 1994, p. 13). In the case of native advertising, a recipient is confronted with a persuasion attempt from two allied sources, the advertiser and the media outlet. Therefore, she or he may refine her or his attitudes towards the content, the advertiser, and the media outlet (Wu et al., 2016(Wu et al., , p. 1497. Theories on psychological reactance and resistance shed further light onto this process. Brehm and Brehm (1981) define reactance as "the motivational state that is hypothesised to occur when a freedom is eliminated or threatened with elimination" (p. 37). This includes the freedom of will or choice, which can be eliminated or threatened by a persuasive message. In this case, the theory predicts that the affected person mistrusts the message and its source (Brehm & Brehm, 1981, pp. 121-150;Meirick & Nisbett, 2011;Wu et al., 2016). Jacks and Cameron (2003) distinguish further strategies to cope with persuasion attempts, namely counterarguing, attitude bolstering, message distortion, social validation, selective exposure, and negative affect (p. 146). Referring to van Reijmersdal et al.'s (2016) study on sponsored content in blogs, with regard to native advertising we expect that when recipients become aware of the persuasion attempt, manifestations of negative affectbecoming annoyed, upset, or even angryintensify recipients' mistrust of the message and its source. As a consequence, recipients reassess the quality of the content as well as the media outlet and advertiser. Against the background of psychological reactance and resistance and previous research on native advertising, the following hypothesis was formulated: H3. Conceptual persuasion knowledge and the use of affective resistance strategies (a) decreases the perceived content quality, (b) which in turn leads to a lower perception of the quality of the news outlet. Figure 1 summarises the postulated relationships. The model predicts a negative effect of native advertising on perceived content quality and media outlet quality due to the related psychological states of conceptual persuasion knowledge and affective resistance strategies. It is assumed that the effect is stronger when quality news outlets publish native advertising content and when such content is not labelled properly.

Experimental design and materials
We developed a 2 (source: low vs. high quality media outlet) × 3 (content: ad-free article vs. declared native advertising article vs. undeclared native advertising article) factorial betweensubjects experiment to test the model. An article about sun protection served as the basis for the creation of the stimulus material. The original newspaper article was shortened and modified so that it was suitable for the experiment.
The experimental study was conducted in Switzerland, which is a suitable country for investigating native advertising. The net advertising revenue of the news media sector in Switzerland plummeted from $2.2 billion dollars in 2013 to $1.3 billion dollars in 2017 (Stiftung Werbestatistik Schweiz, 2018). Faced with these circumstances, news media companies have tried to find new services that can be monetised, with native advertising among them. Even the leading Neue Zürcher Zeitung published several pieces about the benefits of electronic cigarettes that were sponsored by the tobacco manufacturing company Philip Morris (NZZ, 2017).
The manipulation of the source was oriented towards the news sites of two media outlets, 20 Minuten and Tages-Anzeiger. Both of them are well established in the German-speaking part of Switzerland and belong to the Tamedia AG, Switzerland's largest private media company. They, therefore, follow a similar native advertising strategy, but they differ with regard to their media quality. While the free daily newspaper 20 Minuten has a reputation for publishing soft news, Tages-Anzeiger is known as a leading newsstand newspaper in terms of quality. The Media Quality Rating Switzerland, which comprises a representative survey, shows that recipients perceive the Tages-Anzeiger as significantly better than 20 Minuten in terms of quality and credibility (Medienqualität Schweiz, 2016.
The content was manipulated as follows. The ad-free stimulus article reported generally and neutrally on the subject of sun protection. It contained facts about sun protection without mentioning a specific brand or manufacturer. The two native advertising articles had the same content with one exception: There were several references to a cosmetic brand named Korres, which is hardly known in Switzerland.

Sample and procedure
A market research company, respondi, was commissioned to conduct an online survey with a random sample of representative selected persons from the language-assimilated resident population of German-speaking Switzerland. The required sample size was estimated using G*Power (version 3.1.9.3) on the basis of the results of a pre-study survey with students. In the main study, a total of 384 participants completed the online questionnaire (see Table 1). This sample is not large enough to draw general conclusions about the German-speaking population of Switzerland as a whole, but it is sufficient to evaluate this study's hypotheses on how native advertisement is perceived and processed by recipients in Switzerland. The study was conducted in April 2018. Each participant was randomly assigned to one of the six stimuli (see Appendix for an example, Figure 1a). Immediately following the completion of the questionnaire, respondents were debriefed about the fictitious character of the stimulus material. You can contact the corresponding author for the questionnaire, data, and stimulus material.

Measurements
All items were judged on 7-point Likert scales (1 = "strongly disagree" and 7 = "strongly agree"). In order to control for possible side effects, the order of items was randomised.
Affective resistance strategy was measured by adapting similar scales from Dillard and Shen (2005, p. 153) as well as Reijmersdal et al. (2016Reijmersdal et al. ( , p. 1464 to the German language. The participants were asked how they felt while reading the article using statements that can be best translated to English as "a little bit angry", "upset", and "enraged" (M = 2.21, SD = 1.29, Cronbach's α = .94). A fourth item ("annoyed") was removed after it was found to have a negative effect on scale reliability.
For a more powerful analysis of the data, we ran structural equation models (SEM) using Mplus 7.4 (see Table 3). Our first hypothesis assumed that the publication of native advertising content damages the perceived quality of news media outlets. This hypothesis was confirmed with regard to the high-quality media outlet (Tages-Anzeiger), provided that the content was properly labelled as native advertising (model 1: β = 0.19, SE = 0.09, p < .05). Hidden native advertising, in contrast, had no significant effect on perceived media quality (model 2: β = 0.15, SE = 0.20, p = .44). The hypothesis was not supported with regard to the low-quality media outlet (20 Minuten), regardless of whether the native advertising content was properly declared (model 3: β = −0.01, SE = 0.09, p = .94) or not (model 4: β = −0.12, SE = 0.19, p = .53). In other words, when publishing properly declared native advertising content, only those news media outlets that stand for high-quality journalism damage their quality as perceived by recipients.
To understand this difference between low-quality and high-quality media outlets, we subsequently tested the hypothesised model as a whole. Accordingly, we took the mediators from the second and third hypotheses into account: conceptual persuasion knowledge, affective resistance strategy, and perceived quality of content. The full SEM (models 6 and 7) initially did not yield satisfactory global fit indices as proposed by Hu and Bentler (1999) or Hooper, Coughlan, and Mullen (2008). On the basis of the modification indices, we fitted the model by removing two cross-loading items that reflected perceived quality of content. The two nested models both fit the empirical data of the experimental online survey (models 7 and 8, see Figure 2). Both final models confirmed our second hypothesis: Native advertising content activated conceptual persuasion knowledge (model 7: β = −0.70, SE = 0.07, p < .05; model 8: β = −0.66, SE = 0.06, p < .05), which consequently led to the use of affective resistance strategies (model 7: β = 0.36, SE = 0.09, p < .05; model 8: β = 0.37, SE = 0.11, p < .05). Both models also  showin line with the third hypothesisthat the use of affective resistance strategies decreased the perceived content quality. This effect was stronger when the news media outlet stood for high-quality journalism (model 7: β = −0.22, SE = 0.11, p < .05; model 8: β = −0.46, SE = 0.09, p < .05). Moreover, conceptual persuasion knowledge only had a significant effect on perceived content quality with regard to high-quality news (model 7: β = −0.20, SE = 0.12, p = .08; model 8: β = −0.37, SE = 0.12, p < .05). In both cases, a more favourable perception of content quality strengthened the perceived quality of the news outlet (model 7: β = 0.50, SE = 0.07, p < .05; model 8: β = 0.51, SE = 0.11, p < .05).

Discussion and conclusion
Quality journalism had been a protected national industry until Google, Facebook, and most recently Amazon disrupted the advertising industry. In times of a disrupted advertising industry, there are compelling reasons for news media outlets to offer native advertising. Gene Foreman (2016), who has played a leading role in American journalism for decades, explains that "[i]n a turbulent era of transition, news companies seek financial stability" (p. 174). However, he warns news companies about the consequences of native advertising: "[N]ative advertisinglike its predecessors, advertorials in print and infomercials on televisionundermines the credibility of the website's real editorial content. Native advertising blurs the distinction between a journalist's effort to inform the public and an advertising copywriter's effort to promote a brand. In the long term, ads camouflaged as news could lead to consumers to stop reading a website at all" (p. 175).
The aim of this study was to examine whether media outlets that offer native advertising put one of their most valuable assets at risk, namely their quality as perceived by their recipients. The findings of this study confirm this assumption. Our results suggest that news media outlets that stand for high-quality journalism undermine their quality when they engage in native advertising. This study has several limitations, most notably that an experimental online study is an artificial situation. Respondents in survey situations pay more attention to persuasive attempts than they do in everyday media situations (Ham et al., 2015). It should also be noted that using a questionnaire to measure the use of affective resistance strategies is less than ideal. Here, psychological measures, especially the measure of electrodermal activity, would have been more valid (Leiner, Fahr, & Früh, 2012). Moreover, the results regarding native advertising of sun protection are, strictly speaking, not transferable to other industries. Native advertising articles on controversial topics or industriesfor example a tobacco company, like in the case of NZZmay have stronger negative side effects. In contrast, it is also possible that recipients of high-quality news outlets might appreciate lavishly produced native advertising content, such as the "Orange Is the New Black" interactive paid post on The New York Times' website.
Despite its limitations, the results of this study allow us to conclude that news media that stand for high quality, such as The Washington Post, Le Monde, The New York Times, and the Neue Zürcher Zeitung, undermine their quality as perceived by their recipients by engaging in native advertising. These publications should, therefore, steer clear of native advertising because recipients are likely to abandon news media outlets if they fall short of quality news standards (Karlsson. Clerwall, & Nord, 2017). While it may be an exaggeration to warn that recipients will stop reading a quality news outlet altogether solely because of its native advertising practice, this study's results do indicate a trade-off between native advertising and perceived quality. These findings are in line with previous studies on native advertising (Amazeen & Muddiman, 2018;Amazeen & Wojdynski, 2018). Therefore, remembering the adage that little strokes fell big oaks, news media outlets that stand for high-quality journalism are advised to rethink whether native advertising is a sustainable business model.
The question, then, is how else can legacy news media companies make money? Highquality media outlets and news reporting are essential for democracies. However, the inconvenient truth is that whichever global tech giant wins the advertising market, legacy news media companies will lose. News media companies must not generate revenue only through advertising, but through paying recipients as well. Studies from media research show that recipients are willing to pay for online journalism behind a paywall if the content is exclusive and of high quality (e.g. Himma-Kadakas & Kõuts, 2015;Lischka, Hutter, & Rademacher, 2012;Vara-Miguel, Sanjurjo, & Diaz-Espina, 2014). Further research could investigate the notion that news media companies should invest in high-quality media, creating content that recipients will pay for rather than economising and resorting to native advertising.