Published on December 14, 2015. Views: 18290. Downloads: 1387. Suggestions: 0.
Native Advertising and Endorsement: Schema, Source-Based Misleadingness, and Omission of Material Facts
Chris Jay Hoofnagle and Eduard Meleshinsky
Changing the background of the photo to blue was able to significantly increase how much credibility viewers assigned to the testimonial
Native advertising is the new term for “advertorials,” advertisements disguised as editorial content. Modern native advertising started in the 1950s, but its first uses were clearly signaled to the consumer. This paper explains why consumers might be misled by advertorials—even when labeled as such—when advertising material has elements of editorial content.
Results summary: We surveyed consumers (N=598) with a realistic, labeled advertorial embedded in a blog. We found that just over one-quarter of respondents (27%) thought that the advertorial was written by a reporter or an editor. We find that labeling—even using a “sponsored content” disclosure—is insufficient to disabuse a significant minority of consumers about the provenance of the advertising material. Our findings are not generalizable, since we targeted the survey to internet users who appeared on marketing lists derived from behavioral tracking. However, our findings are compatible with those of other researchers who suggested that in addition to initial disclosures, elements in the advertorial itself must also signal to the consumer that this may be commercial material.
While the advertorial we tested was a story about the potential of abuse of diet pills, the writing made dramatic claims about the effectiveness of named products for weight loss and included a portrait replicated from a real advertisement appearing in a health magazine. We found that merely using a blue background to frame the endorser’s portrait led many respondents to think her to be a medical expert. Traditionally, the appearance of a lab coat or stethoscope has signaled a medical expert endorsement, something subject to greater regulation. Our findings point to consumers using subtle clues about context to associate an endorser with an expert profession. We conclude by discussing regulatory options for the FTC, including a ban on advertorials, enhanced disclosure requirements, and approaches that put the burden on publishers to show that advertorials are not misleading.
In 1951, Atlantic Monthly publisher Donald B. Snyder observed that products had become more complex and that the existing advertising jingles and slogans could not convey the messages that consumers needed to understand in a modern economy. Snyder announced,
“To provide the facility for advertising to convey information, the Atlantic has developed a new form for the expression of business ideas. We call them Advertorials. They will be paid advertisements…They will involve the thoughtful participation of the reader; they are intended to give him pause—and in the pause, compelling facts about the way American business works. They are predicated upon the belief that the free competition of ideas has made this country great .”
Snyder followed this announcement with five pages of advertorials paid for by the American Iron and Steel Institute. Each page was prominently labeled “An Atlantic Public Interest Advertisement .”
Thus began the modern phenomenon of the advertorial . An advertorial is an “advertisement written in the form of an editorial, which purportedly provides objective information about a commercial or industrial subject .” In recent years, marketers have taken up the term “native advertising”  in referring to advertorials. However termed, the practice is controversial because native advertising blurs the lines between editorial and advertising pages and because advertorials seem designed to disarm the reader’s natural defenses against advertising.
There are no reliable statistics on the growth of native advertising online . Yet, with the growth and popularity of “listicles” and media outlets such as Buzzfeed, native advertising attracts more attention. It has long been observed that readers engage more with the material in native advertising than in other forms of advertising . With the proliferation of ad-blocking technology, advertorials may be a way of reaching technologically savvy users. Furthermore, users may simply like advertorials more than traditional advertisements.
At a 2013 Federal Trade Commission workshop convened to examine native advertising, Professor Nicholas Lemann, then Dean of the Columbia Journalism School, explained that advertorials may be more popular today because of the special dynamics of online advertising. In the past, placement of a print advertisement in a major newspaper was expensive, and thus a major print advertisement signaled a company’s legitimacy. If a company could afford a print advertisement in the New York Times, the company probably was trustworthy. This halo effect is less luminous in the offline world. Online advertising can appear anywhere, and even small businesses can have their marketing posted on the websites of prestigious newspapers. As a result, online marketers are desperate to regain the patina of trustworthiness of the print era. An advertorial, because it is presented in the garb of a trusted media outlet, changes the hue of advertising, making it more trustworthy .
A rich literature explores whether advertorial and other embedded advertising erodes liberal values  and whether it corrupts news media reporting . This article focuses on a different issue: to what extent are people misled about the authorship of advertorials and endorsements that appear in them? Recognizing material as advertising is critical to consumer protection. It is well established in the literature that consumers summon their own natural defenses and skepticism when they recognize material as advertising. Thus it is important that consumers recognize sponsorship, and that recognition occurs prior to viewing the content .
Consumer self-help is also necessary because the government cannot possibly police all advertising material. This is why advertorials represent a serious consumer protection problem. Advertorials risk upending consumer self-help mechanisms if advertorials lead the consumer—even for a short time—into believing that material is editorial.
This article first suggests two theories of consumer deception—schema theory and source-based misleadingless—as reasons advertorials mislead consumers. Schema theory suggests that advertorials mislead by causing consumers not to trigger their innate skepticism to advertising. Advertorials also cause source-based misleadingness problems by imbuing advertising material with the authority normally assigned to editorial content. Second, this article describes an experiment where we showed consumers a realistic advertorial, then asked them to tell us who wrote it and to attribute the endorsement that appears in the advertorial. Third, this paper discusses self-regulatory and legal consumer protection responses to the issue of advertorials. It also calls for more research into the subtle inferences that can be drawn from ambiguous endorsements.
The point of advertorials is obvious: to get the consumer to engage with advertising material. But it is not obvious why advertorials may deceive consumers, or why even a truthful advertorial may run afoul of laws on false advertising. This section introduces two theories that explain why advertorials may be misleading. It then presents several legal theories that consumer protection authorities could employ to police advertorials.
Psychological Theories of Deception
Schema Theory and Omission of Material Facts
Omission of material facts refers to the failure to make certain qualifications without which the consumer “may draw broad inferences... based on prior experience or based on the physical appearance of the product .” The psychological root of this category of deception is schema theory . Schemas are cognitive maps that organize one’s knowledge about a particular domain  and thus create expectations with respect to certain domains . Schemas develop through personal experience and socialization . One example of a schema is our concept of the drive-through restaurant. We know to order before we get to the pay window. We know that we need a car to use the drive-through window. And we know to pay when we get to the window rather than when we order. That is all part of the schema of a drive-through restaurant. If there were no sign or instruction to pay at the window, we would still know to do so because our schema tells us what to expect.
We rely on our schema to determine the “default” behavior in a particular domain and to fill in any gaps in a familiar situation . In the drive-through example above, even without a sign at the drive-through instructing drivers to pull up to the window to pay, customers rely on their schema as a guide to the default behavior (to pull up to the window). Similarly, consumer schemas associated with advertisements raise consumer skepticism . Advertisers use native advertising in order to avoid triggering the consumers’ advertisement schema. Advertorials normally disclose in some fashion that the content is an advertisement, but advertorials use the typeface and styling of editorial content . The result is that differentiation between advertisement content and editorial content is difficult .
Source-based misleadingness also explains how advertorials can deceive consumers. Advertorials draw on the authority of their editorial context and thereby cause consumers to give them the same deference as they give to editorial content. The trappings that surround an advertorial are more than mere camouflage. Matching the advertisement’s font, color, logo, and layout to the editorial content around it prompts readers to give the advertisement consideration similar to what they accord the editorial content. This transfer of trust can be explained by reference to source-based misleadingness. Oftentimes, consumers evaluate a claim based on the source of the claim rather than the content of the claim . An example of this general phenomenon is when a consumer gives weight to a doctor’s endorsement of a brand of cereal even if the doctor has not been shown to be a nutrition expert or have any scientific knowledge about the particular cereal brand being endorsed. Similarly, the trappings of editorial content transfer its authority to the advertisement embedded within . Consumers trust editorial content more than advertising content. When consumers view advertisements that have the trappings of editorial content, they perceive the editorial content “as a (co-)sender of the advertised message,”  thereby “imbuing” the advertisement content with some of the trustworthiness of the editorial content .
In order to test the deceptive capacity of advertorials, we presented a simulated blog page to respondents (N=598) that contained three “articles.” One of the three articles—the middle one—was an advertorial we adapted from a popular fitness magazine that advertised diet pills. We labeled it “Sponsored Report” in a font similar to other elements on the page in size and style. We gave respondents time to review all three articles. Then we asked open-ended and closed-ended questions about the articles’ content.
Figure 1. The Advertorial Placed in the Blog
The advertorial in Figure 1 contains unattributed content that regulators would evaluate as an endorsement of the product’s efficacy. Its message is subtle. It both tells the reader that the pills are efficacious for weight loss, and that some people use the pills as a kind of amphetamine. This tactic touts efficacy in two ways: first by an explicit claim of weight loss potential, but also by suggesting its use as a stimulant shows that it has a real effect on the body. Perhaps the point of the content is to make the reader think of phentermine, the “pick me up” element of the notorious fenfluramine/phentermine (“fen-phen”) craze of the 1990s. It was not clear at all who wrote the content, and we used this as an opportunity to see whether respondents believed the author was a medical expert.
Our experiment was part of a larger effort that studies targeting of consumers based on vulnerability. Princeton Survey Research Associates International (PSRAI) conducted the “Consumer Knowledge Survey” on behalf of Berkeley Law from September 12 through November 19, 2013. We selected the 598 respondents based on their presence on one of four behavioral advertising email lists: alcoholic consumers (n=153), consumers interested in debt consolidation (n=174), general Internet users (n=123; our control group), and subscribers to a financial publication (n=148). Thus, this article relies on a non-random survey that is not representative of the general population. In fact, 68% of respondents were female, and the youngest respondent was 34 years old. Because we targeted respondents based on their presence on email lists, and because they self-selected to participate, the dataset cannot be weighted to make the results representative.
Table 1. Overview of the attributes of the 598 participants in study.
We sought to interview 1,200 individuals. However, recruitment was time-consuming and difficult. In order to achieve almost 600 interviews, PSRAI emailed 119,216 individuals. Between 30% and 40% of the solicitation email addresses were undeliverable. Nevertheless, the participation rate was well below one percent.
We asked respondents, “Was the material on diet pills written by journalists and editors working for the website, or by someone else?” We did this to determine whether the study subjects thought the material was editorial, thus written by journalists, or advertorial, thus written by “someone else.” We did not explicitly use the word advertorial or advertiser, so as not to arouse special suspicion of its marketing nature. Twenty-seven percent of respondents thought that journalists and editors wrote the material, 43% chose “someone else,” and 29% said they did not know.
Figure 2. Response from participants after seeing the advertorial in Figure 1.
Recall that our sample was drawn from marketing lists, and that they hailed from four categories: alcoholic, debt consolidation, financial news subscribers, and general Internet users. We hypothesized that the subjects identified as alcoholics and debt consolidators would perform poorly on this test, and that financial news subscribers would do better than general Internet users. However, there were no significant differences among the groups. Finding no relationship on our hypothesis-dependent variables, we compared the advertorial question to all of our demographic variables and found no significant relationship at all .
Embedded in the advertorial in Figure 1 was an unequivocal claim that the named products were effective for weight loss, even though other content warned of abuse. In it, a young, professional-looking woman writes, “While the pills do cause rapid weight loss (no doubt about that), most people buy them as ‘speed’ pills.” The author’s identity and role is unclear. We sought to determine whether the reader would think the author of the advertorial was the person in the image, or someone else, such as a medical expert. The author’s role as expert was ambiguous. She wore a white, Oxford-style shirt, and what could be a lab coat. However, she is not identified as a medical expert nor is she wearing a stethoscope.
Figure 3. In the control condition, the endorser appeared before a plain, white background.
We divided the respondents into two groups. In one, we subtracted the background, so that the endorser simply appeared against a white background. The second group saw the background from the actual advertisement, where the endorser appears in front of what may be a blue-toned wall of products.
Figure 4. In the experimental condition, the endorser appeared before a wall of blue products.
We presented the endorser’s picture again along with the text as presented in Figures 3 and 4, and asked an open-ended question: “What job does the person in the picture have?” The plurality of respondents presented with the condition image (white background, n=290) thought the endorser was a model (32%). Twenty-three percent thought she was a medical expert: 6% indicated a pharmacist, 6% a doctor, and 11% some other medical professional (e.g. dietician) or multiple categories (e.g., “pharmacist or doctor”). Seven percent thought the endorser was a journalist, and 2% indicated that she was probably a model dressed as a medical expert.
Adding the blue background shifted respondents’ perceptions dramatically (x2(9, n=443) = 130.2722; p <0.001). Asked in an open-ended fashion, the plurality of respondents who saw the image with the blue background (n=308) thought the endorser was a pharmacist (52 percent). Eighteen percent thought she was a model, and less than 1% thought she was a journalist. Overall, 60% thought she was a medical expert: 1% a doctor, 7% some other medical professional. Five percent recognized her a model dressed as a medical expert.
Figure 5. Distribution of response from participants to open-ended question on the job of the endorser based on the color of the background.
Age was associated with recognition of the endorser as a medical expert (x2(3, n=443) = 10.4745; p <0.05). Whether presented with the white- or the blue-background endorser, younger respondents were more likely than older ones to conclude that the endorser was a medical expert when asked an open-ended question. We had not hypothesized that age would interact with the ability to determine the endorser’s role. It could be that older respondents were less likely to consider that women are medical experts because of cultural bias and historical discrimination against women in the professions.
We employed a metric developed by Professor Alan Westin to measure intensity of privacy concern. We found that privacy concern also correlated with recognition of the endorser as a medical expert (x2(3, n=443) = 17.5889; p = 0.001). People identified as “privacy unconcerned” were much more likely (60%) than high-privacy-concern consumers (33%) to identify the endorser as a medical expert. In other literature, author Hoofnagle identified people labeled “privacy unconcerned” by Westin as vulnerable consumers significantly more ignorant of consumer protection rules and duties than other consumers.
We then turned to a closed-ended question with seven options: What industry do you think the person in the picture works in? The plurality of respondents who saw the white background placed the endorser in a health-care field or a pharmacy (49 percent). Twenty-five percent thought she was a model or actress, and 12% a journalist.
Respondents presented with the blue-backgrounder endorser were much more likely to think she worked in health care (x2(2, N=598) = 45.7102; p < 0.001). Seventy-five percent identified her as hailing from the health-care field. Fifteen percent thought her a model.
Figure 6. Distribution of response from participants to closed-ended question on the industry of the endorser based on the color of the background.
Privacy concern held its statistically significant status in the closed-ended inquiry (x2(6, N=598) = 16.3459; p < 0.05). Whether presented with a blue- or white-backgrounded endorser, 72% of low-privacy concern respondents thought the endorser was a medical expert, 65% of mid-level concern considered her a medical expert, while 52% of high-privacy concern consumers thought her to be a medical expert. The difference is even more pronounced when comparing privacy concern among the respondents who saw the blue-backgrounded endorser. In that condition, 60% of high-privacy-concern respondents identified her as a medical expert, while 80% of mid-level, and 87% of low-level concern respondents identified her as a medical expert.
Figure 7. Distribution of participants according to level of privacy concern who responded to the closed-ended question on the industry of the endorser as a “medical expert” based on the color of the background.
While age had a statistically significant relationship with the open-ended question, the relationship was not significant in the closed-ended inquiry. Nor were our other demographic variables .
The American Society of Magazine Editors (ASME) released updated guidelines in 2015 reaffirming the need of publishers to distinguish editorial and advertising content. ASME recommended that publishers label advertorials on websites with terms such as “Sponsor Content” or “Paid Post” and that advertorial content have visual elements to clearly distinguish it from editorial .
The ASME standard sets a higher bar than does the Interactive Advertising Bureau (IAB), which released a one-page-long guidance to advertisers on the same day that a FTC workshop on advertorials was held in 2013. The IAB guidance, styled as a recommendation rather than a requirement, simply states that reasonable consumers should be able to distinguish between paid advertising and editorial content. But it provides little guidance on how this is to be done or what constitutes a reasonable consumer. IAB recommends that disclosures be visible in context, and that they use language “that conveys that the advertising has been paid for…” 
We are skeptical that “Sponsor Content,” “Promoted Post,” or “Paid Post” are effective disclosures in themselves, because they are ambiguous. These phrases do not identify the sponsor or payer. The words “Promoted Post” might mean that users “promoted” it by clicking on it frequently.
Nevertheless, our findings suggest that even with a prominent disclosure, a substantial number of consumers are misled about the advertising nature of advertorials. We used the disclosure “Sponsored Report,” which we thought to be clearer than “Sponsor Content.” This raises the issue of what additional regulatory approaches might address the advertorial problem.
At first, native advertising may not appear to fit the rubric of an illegal deceptive trade practice. For instance, consider that an advertorial could be entirely truthful and informative. However, the deception comes from the form of an otherwise truthful advertisement. In fact, even truthful native advertising can be unlawful, because it involves several practices that the FTC has long held to be deceptive . Misleading tactics used to gain a consumer’s attention, known as “deceptive door openers,” can be illegal even if the initial wrong is cured with later truthful disclosures. Also, information misleadingly framed, even if truthful, can give rise to a FTC deception claim.
While European regulators have several prescriptive rules governing advertorial content , the FTC has yet to promulgate a rule or guide specifically addressing native advertising. Thus, the FTC can only police advertorials using its deceptive trade practices authority. Under the FTC’s Policy Statement on Deception, the Agency must show that there is a representation, omission, or practice that is likely to mislead a consumer; the interpretation of that act or practice is considered from the perspective of a reasonable consumer; and the representation must be material . To demonstrate the reasonable-consumer viewpoint, all the FTC needs to show is that the representation or omission has the tendency or capacity to deceive a significant minority of consumers . Materiality can also be demonstrated because all express claims are material, but so are other claims and omissions that affect purchase or conduct decisions surrounding a product. The Commission would likely find a false endorsement or an advertisement camouflaged as news to be a material fact. Both of these techniques give greater credibility to the facts asserted by the advertiser and could affect consumer attention and decision-making.
Banning Native Advertising
If we take consumer protection and advertising law seriously, one rational approach to advertorials—if they can be reliably defined—is to ban them entirely. A ban would be justified under two rationales: first, advertorials are similar to the “deceptive door openers” of door-to-door salesmen and of telemarketers who deceive consumers by introducing a sales call as a survey or public opinion research. The FTC has long held it unfair or deceptive to mislead in order to get information from a consumer or to make a sale. For instance, in the 1951 Gen-O-Pak matter, the FTC brought an administrative proceeding against a company that promised a free gift in exchange for filling out a postcard that solicited personal information. Although the card framed the information collection as necessary to deliver the gift, in reality the company collected the information to help creditors locate debtors. The cards and surveys sent by the company solicited extensive personal information. The FTC found this practice both unfair and deceptive .
In 1971, the FTC filed a complaint invoking both unfairness and deception against a company for generating direct mail lists from consumer questionnaires. The company wrote to consumers promising the chance to win prizes, saying that there was “nothing to buy” and “no salesman will call on you.” 
In 1975, the agency found that survey researchers promised “anonymity” to data subjects but used secret coding and other techniques to track responses to surveys. The Commission wrote that it was deceptive to induce “consumers to provide information about themselves by expressly or implicitly promising that such information is being provided anonymously, when, in fact, a secret or invisible code is used… that allows identification of the consumer who has provided the information.” The Commission argued that the conduct could also be unfair if used to obtain sensitive information or for some other purpose .
These cases suggest that the FTC has avenues to pursue advertorials when they gain consumers’ attention in a misleading way. The point of advertorials is to trick the reader, if even for a few moments, so that the reader engages with the advertising material. The psychological evidence reviewed above provides evidence that the detection of advertising triggers consumers skepticism. Thus, advertorials induce a brief period of credulity in the consumer, undoing a key self-help mechanism.
Self-help is a critical tool in consumer protection. Consumers who can protect themselves need less regulation and government policing of the marketplace. Without self-help, the government may need to be more involved in helping police deceptive practice.
But in addition to deceptively gaining users’ attention, advertorials could be attacked for their overall misleadingness. Since the FTC’s inception, it has policed companies that present information in a misleading way. Among the FTC’s first cases was a 1917 matter where the agency pursued a seller of vacuums who claimed, falsely, to be an independent and impartial expert on the quality of vacuum cleaners in advertisements that appeared in the New York Times  and Life magazine . In reality, the company was the exclusive seller of a certain kind of vacuum, one that it favored in its sales presentations while denigrating competitors’ machines. The FTC concluded that the tactic was unfair competition .
In 1968, the FTC addressed advertorials head-on in an advisory opinion. The matter involved a regular newspaper column that featured a write-up of a local restaurant containing mostly truthful and objective information about the business—whether the restaurant allowed dancing, the price range of meals, the chefs’ names, and so on. The FTC opined that even a truthful presentation of facts was deceptive when presented as a news article. The FTC advised that it would require clear and conspicuous disclosure that the column was an advertisement to avoid Section 5 liability .
Our experiment lends support to the idea that advertorials—even those with disclosures—continue to mislead consumers. In our experiment, which required respondents to concentrate specifically on a suspect blog post, 27% attributed it to a journalist. Because advertorials mislead in how they get the consumers’ attention, and because a substantial minority of consumers never even realize that advertorials are advertisements, the FTC has grounds to expand its general prohibition of advertisements that are presented as news.
Disclosure of Advertorials
For political economy reasons, much of consumer protection law focuses on disclosures. We trust that consumers can fend for themselves when adequately informed about products and their marketing. As our findings indicate, however, many consumers considered an advertorial nested in a simulated blog to be editorial despite a disclosure. Perhaps consumers skipped over this disclosure because of its placement, because it is often in a different, lighter color than the other text, or because it is smaller than the headline. This study suggests that the “sponsored content” disclosure does not cure the problem of misleading advertorials .
Other literature and the ASME guidelines suggest that in addition to labeling, aspects of the advertorial text itself should signal its advertising nature. For instance, in a widely cited 2001 article that studied a small population of students (N=215), Kim et al. found that most respondents could recognize an unlabeled advertorial. The group found that labeling advertorials increased recognition, but not in a statistically significant way. Kim et al.’s findings support the idea that textual clues in the advertorial itself may be more important than labels .
Kim et al.’s findings suggest a return to the elements found in the first advertorials: both initial disclosure and in-text signaling of advertising. In an example from 1914, the Rotarian magazine ran an advertorial that had a labeling disclosure, employed a different font, and presented each new sentence with a paragraph mark (¶).
Figure 8. The Rotarian's 1914 "Advertorial" used paragraph marks and a different font from editorial pages to distinguish it as advertising.
The 1951 Atlantic Monthly advertorial was more subtle, but still used a larger font and a banner across the top of each page disclosing that it was “An Atlantic Public Interest Advertisement.” The FTC could follow the leads from this research and require both labeling of native advertising and in-text signaling of the material’s advertising nature.
Figure 9. The Atlantic Monthly's 1951 advertorial had different text style and a banner atop each page.
Finally, the FTC could release a guide that places the burden of clear disclosures on the advertiser using advertorials. There is some precedent for this, as the FTC’s Guide to “free offers” treats the use of “free” as presumptively deceptive. Advertisers using “free” must act with “extreme care so as to avoid any possibility that consumers will be misled or deceived.” Conditions on free offers must be disclosed “clearly and conspicuously.” 
In the case of advertorials, the FTC could find that they are presumptively deceptive unless accompanied by clear disclosures that disabuse almost all consumers of their non-editorial nature.
The FTC imposes special rules on the use of experts in endorsements. In its 2009 guidelines, the FTC specified that,
“Whenever an advertisement represents, directly or by implication, that the endorser is an expert with respect to the endorsement message, then the endorser’s qualifications must in fact give the endorser the expertise that he or she is represented as possessing with respect to the endorsement .”
A direct representation of expertise is simple: it includes situations where an advertisement makes reference to a “doctor” or similar expert. Indirect representations are more complex. Under what is known informally as the “white coat” rule, the FTC assumes that consumers will interpret endorsers in medical garb as expert doctors or scientists. Thus, advertisers are at risk of creating an implied expert endorsement when they present the consumer with actors wearing lab coats or stethoscopes, or actors in a laboratory .
Our experiment reveals just how complex consumer inferences surrounding expertise can be. The endorser we tested appeared to be wearing a lab coat, but other indicia of scientific expertise, such as lab supplies or a stethoscope, were missing. We show that merely putting the endorser before a blue wall of products caused 60% of respondents to say she was a medical expert, compared to 23% when she was presented before a white background. This finding suggests that relatively subtle signals communicate to consumers and that consumers can derive legally-significant conclusions from them. We also found that low-privacy-concern respondents were more likely to conclude that the endorser was a medical expert than high-concern consumers. This last finding is troubling because author Hoofnagle has hypothesized that low-privacy-concern is a stand-in for consumer vulnerability. That is, consumers labeled as having low concern have a poor understanding of legal rules and consumers’ own responsibility to protect privacy . It may be the case that the most vulnerable consumers would derive the greatest benefit from advertisers’ signaling.
Limitations of Research
This effort would not be complete without emphasizing limitations of the data and methods. First, our sample was a non-random group of Internet users who responded to an email offering them a small payment in exchange to participate in a survey. Our sample is not representative and older users and women were overrepresented. We could not adjust our sample to be representative of the general population. Many online surveys suffer from these problems. Online surveys often use panels of people who regularly volunteer to do surveys for small payments.
Second, we only tested one native advertisement instead of an array of such marketing. We think testing just one advertisement was appropriate, because incorporating additional examples would sensitize the sample to the problem and cause amplification in responses.
Finally, Amar Bakshi wrote that our methods “likely underestimate consumer confusion in the real world, because Hoofnagle’s subjects were specifically instructed to focus on and evaluate the source of a given article, alerting them to the possibility that it had not been written by the health blog reporter .” We agree with the Bakshi critique. We deliberately called respondents’ attention to elements of the advertorial and probably caused more mentation about the material than would take place in a real life scenario.
Recall that the modern advertorial in 1951 was a product of the Atlantic Monthly. In 1951, the magazine published a rather harmless assessment of the country’s capacity to produce steel. Sixty years later, under the label “Sponsor Content,” the Atlantic ran a lengthy advertorial titled, “David Miscavige Leads Scientology to Milestone Year.”  The timing was suspicious, as Miscavige’s organization was facing the release of a critical book . In the span of these 60 years, the advertorial had evolved from an excursus about steel production into a vehicle for whitewashing a controversial religion. The prominent disclosures of the 1951 advertorial became small, highlighted text in later advertorials. Commentators lamented the editorial corruption of the Atlantic.
Yet the incident revealed an important aspect of advertorial confirmed by prior research: that the content of advertorials helps the reader recognize it as advertising. The suspicious pro-Scientology messaging in an otherwise respectable publication was enough to cause a strong reader reaction. We make two observations about the Atlantic’s Scientology advertorial. First, we should resist the temptation to see the Scientology advertorial as evidence that the field can self-regulate . Simply put, most products are not as controversial as religion and are unlikely to foment an organized online rebellion as was the case with the Scientology advertorial. Second, the content of advertorials themselves is important for consumer recognition of their advertiser provenance.
We argue here that banning advertorials could be justified based on consumer psychology, the FTC’s history of policing misleading advertising techniques, and a study of a properly labeled advertorial that many consumers thought was editorial. Still, middle-way approaches could also work, including a requirement that the text of the advertorial itself signal its advertising nature, and a FTC finding that advertorials are presumptively deceptive unless accompanied by clear and conspicuous disclosures of their advertising nature.
Chris Jay Hoofnagle is adjunct full professor at the University of California, Berkeley, School of Information, and a faculty director of the Berkeley Center for Law & Technology. Hoofnagle teaches computer crime law, internet law, privacy law, and seminars on the Federal Trade Commission and on education technology. He is the author of Federal Trade Commission Privacy Law and Policy (Cambridge University Press 2006). Hoofnagle is of counsel to Gunderson Dettmer Stough Villeneuve Franklin & Hachigian, LLP, and a member of the American Law Institute, the San Francisco Electronic Crimes Task Force, and Palantir's Council on Privacy and Civil Liberties.
Eduard Meleshinsky is an associate attorney at Bryan Schwartz Law, a civil rights employment law firm in Oakland, California. He earned both his J.D. and B.A. at the University of California, Berkeley. More information about Mr. Meleshinsky is available at http://www.bryanschwartzlaw.com/eduard-meleshinky/.
The authors are indebted to Dominique Caamano (Berkeley Law ’14) for research and strategic planning supporting this paper, and of her support in the overall Consumer Knowledge Assessment project
Hoofnagle C, Meleshinsky E. Native Advertising and Endorsement: Schema, Source-Based Misleadingness, and Omission of Material Facts . Technology Science. 2015121503. December 14, 2015. https://techscience.org/a/2015121503/
Under review for data sharing classification.
Enter your recommendation for follow-up or ongoing work in the box at the end of the page. Feel free to provide ideas for next steps, follow-on research, or other research inspired by this paper. Perhaps someone will read your comment, do the described work, and publish a paper about it. What do you recommend as a next research step?