Abstract
The authors find no evidence that consumers benefit from government-mandated disclaimers in advertising. Experiments and common experience show that admonishments to change or avoid behaviors often have effects opposite to those intended. The authors examine 18 experimental studies that provide evidence relevant to mandatory disclaimers. Mandated messages increased confusion for consumers and were ineffective or harmful in the 15 studies that examined perceptions, attitudes, or decisions. The authors conduct an experiment on the effects of a government-mandated disclaimer for a Florida court case, showing two advertisements for dentists offering implant dentistry to 317 participants. Only one advertiser had implant dentistry credentials. Participants exposed to the disclaimer recommended the advertiser who lacked credentials more often, and women and less-educated participants were particularly prone to this error. In addition, participants drew false and damaging inferences about the credentialed dentist.
Ben-Shahar and Schneider (2011, p. 647) document the “spectacular prevalence” of mandated disclosures.
In this article, we scrutinize the rationale for such restrictions on speech, examine the legal history of disclaimers in the United States, and review prior experimental evidence on the costs and benefits of disclaimers. We then describe an experiment that we conducted for a court case about a disclaimer mandated by the government of Florida.
Economic Rationale for Restrictions on Commercial Speech
The argument for mandatory disclaimers is inconsistent with economic principles and knowledge of the roles of sellers, regulators (who sometimes stand between sellers and buyers), and buyers, as we describe next. For a review of the aspects of the economics of information that are relevant to buyer and seller behavior, see Calfee and Ford (1988).
Sellers
It is in sellers’ economic interests to treat customers well and, especially, to avoid misleading them. They are motivated to tell consumers about the limitations of their products to develop good long-term relationships with them and to avoid the costs of dealing with disgruntled customers and with lawsuits. Unsurprisingly, then, sellers have long used disclaimers in advertising. Prior research on advertisements that tell the bad along with the good has found that they are persuasive when the negative features are important to consumers (Armstrong 2010).
Sellers are motivated to provide warnings with products that may be dangerous in surprising ways or extents—for example, for a clear liquid that is poisonous but not for a knife. Such warnings are usually helpful. In a meta-analysis involving 12 experiments and 3 quasi-experiments involving 79 comparisons, Cox et al. (1997) find the warnings yielded an average gain in compliance of 15.7% compared with no warnings. However, in one-third of the comparisons, the presence of a warning had no effect or reduced safe behavior.
Sellers are also motivated to provide benefits to potential customers and to tell them about those benefits, if they are free to do so. Consider the following examples:
Breakfast cereal companies increased fiber content and introduced advertising of the benefits of fiber when restrictions on advertising health benefits were lifted. Consumers increased their consumption of high-fiber cereals (Ippolito and Mathios 1991).
Women reduced their consumption of saturated fats by 24% in the five years after advertising restrictions were lifted in 1985, a substantially more rapid change than occurred during the preceding eight years (Ippolito and Mathios 1995).
Cigarette companies reduced tar and nicotine levels after the Federal Trade Commission (FTC) prohibition of comparative health claims in cigarette advertisements was lifted (Craswell 1991).
Before mandatory nutrition labeling, sellers were motivated to tell consumers about features of their products that provided health benefits. After the institution of a new mandatory labeling regime, which restricted claims that sellers could make, the share of more healthful cooking oils sold decreased (Mathios 1998).
Buyers
People are accustomed to dealing with biased information in all areas of life, including when making decisions as consumers. When consumers lack expertise in a product category, they are likely to seek out independent information, use trusted suppliers, or buy well-known brands. If customers discover that they have been misled after purchasing a product, they may avoid purchasing the product in the future, demand a refund, tell others not to buy it, post comments on the Internet, or sue.
From their experience, and knowledge of human nature, consumers are also aware that government officials are fallible, sometimes biased, and sometimes duplicitous in the information they provide. In addition, people often attribute higher benefits to products they are told they cannot have (for evidence on this “scarcity principle,” see Armstrong 2010, pp. 71–74). As a consequence, consumers may fail to respond to government-mandated messages in the ways the regulators intend.
Regulators
Although sellers in free markets are typically motivated to look after buyers, some sellers deliberately mislead consumers in the hope of short-term profits. Such exceptions to normal market behavior are proposed as a key rationale for regulation. Market regulators, however, face a complex problem. They must devise, implement, and enforce regulations that increase welfare beyond that which is achieved by many individual buyers and sellers—each with different information, preferences, situations, and trade-offs—who are engaged in many voluntary transactions. They must also do so without violating the property and other rights of citizens. Even with the best of intentions, available evidence implies that it may not be possible to increase welfare by government regulation or information policies (for a review of the evidence, see Winston 2006, 2008).
In practice, the regulatory philosophy that governments adopt may not be one of welfare maximization and may vary, thereby increasing uncertainty for sellers and confusion for buyers (Eggers and Fischhoff 2004). Regulators may also fail to implement the wishes of elected legislators, as Emord (2000, p. 139) describes in relation to the Food and Drug Administration's (FDA's) “arbitrary and capricious” and “virtually unbridled discretion over commercial speech” restrictions on health claims about products. 2
“In particular, Congress condemned the FDA's long delay (until 1996) in authorizing a health claim that associated folic acid with a reduction in the risk of neural tube defects (a claim endorsed by the Centers for Disease Control and Prevention in recommendations to the U.S. public in September of 1992), placing blame for preventable neural tube defect births between 1992 and 1996 squarely on the agency” (Emord 2000, p. 140).
Government officials and judges face neither the direct accountability of sellers nor the search costs and pleasure or regret of buyers. Instead, they face the temptation to impose their own beliefs on others and lobbying from sellers—who would like to restrict their competitors’ ability to communicate benefits—and from organizations with agendas hostile to the seller. The following examples suggest that regulators’ understanding of these complex situations may never be sufficient in practice to ensure that regulations increase welfare.
In 1980, the FDA issued a warning that pregnant women should avoid coffee because of a risk of birth defects (Burros 1982). In 1981, researchers claimed that coffee was responsible for half of all pancreatic cancers. Both claims of harm from coffee consumption were later reversed; the pancreatic cancer claim was reversed by the original researchers (Simon [1996] summarizes three studies on this issue.) Researchers later claimed that coffee has net health benefits (e.g., Larsson and Orsini 2011).
The U.K. Food Safety Act of 1990 effectively outlawed the use of wooden chopping boards and utensils in commercial kitchens in the belief that they were unhygienic. The belief was based on a study that involved the cultivation of scrapings from wooden working surfaces taken from 211 butcher shops and 24 restaurants in London. The researchers found that 4% of the cultivated samples contained salmonella (Gilbert and Watson 1971). Government inspectors vigorously enforced the rule, causing much disruption and upset. Subsequent experimental research in 1993 designed to replicate conditions in kitchens more realistically found that wooden boards have antibacterial qualities, killing 99.9% of bacteria within three minutes, whereas bacteria persisted on the replacement plastic boards. The ban was reversed that year (Booker and North 2007).
The user of a drug developed serious side effects and sued the manufacturer for damages, claiming that the manufacturer knew about mounting evidence of the generic drug's dangers but did not warn consumers. The manufacturer maintained that the company was bound to stick with the mandated labeling. The Supreme Court found in Pliva, Inc. et al. v. Mensing (2011) that pharmaceutical manufacturers could not be held liable under state tort law for insufficiently warning consumers, because changing the warning would have breached the Federal warning label mandate.
In a review of government information policies, Winston (2008) examined three situations that had been proposed in 2007 as examples of mandatory disclosure policies that increased welfare: financial disclosure, mortgage lending, and restaurant hygiene. He found no evidence that the mandated disclosures improved the situations for consumers or that there were problems in the first place. He concluded (p. 174) that “empirical evidence does not persuasively indicate that any information policy has been effective” and proposed benign neglect as the appropriate response by policy makers to alleged information problems.
Economic theory, then, suggests that in free markets, sellers are motivated to treat customers well to make a profit, buyers are motivated to exercise caution, and welfare tends to be maximized. In regulated markets, sellers are restricted in their ability to serve customers; buyers are less cautious; and regulators face temptations, lack knowledge, and, in practice, lack the incentive to obtain useful scientific knowledge on the effects of proposed regulations.
Evidence on Human Behavior Relevant to Mandated Disclaimers
By requiring disclaimers, governments absolve buyers and sellers of responsibility for care and thus encourage irresponsibility. The presence of a government-mandated message suggests that an authority has carefully reviewed the product. The authority of a government-mandated message or product feature might reassure consumers that they are being looked after, causing them to become less vigilant. For example, a study involving 1307 drivers in the State of Washington and 6234 observations of their annual accident frequency from 1992 through 1996 found that drivers who purchased cars with airbags and antilock brakes drove more aggressively, to the extent that the safety benefits were much less than expected (Winston, Maheshri, and Mannering 2006). This type of response is referred to as the risk compensation hypothesis or offset hypothesis in economics literature.
Consider, for example, the effect of a “descriptive-norm” sign posted by the U.S. National Park Service intended to discourage the theft of petrified wood. When the sign was in place, the theft rate was nearly three times higher than when it was not. Why? The sign was a signal to park visitors who would otherwise not have stolen that stealing the petrified wood was a common behavior: In this case, the social proof that fellow visitors stole wood more than outweighed the admonishment from an authority not to steal (Cialdini 2003).
The purpose of government-mandated messages is often to change or discourage specific behaviors—for example, to stop smoking or avoid overconsumption of alcohol. Experimental research on persuasion has shown that it is difficult to change or prevent behavior (a common experience of those with teenage children). Mark Twain (1885, chap. 22, p. 3) recognized that restrictions can make a product more attractive to potential consumers when his character, The Duke, wrote an advertising bill including the lines “For 3 Nights Only!” and “LADIES AND CHILDREN NOT ADMITTED” and then said in reference to the latter, “There, if that line don't fetch them, I don't know Arkansaw.”
Twain's insight is consistent with the evidence Armstrong (2010) summarizes on resistance to persuasion. When consumers are told that they should not or may not do something they are currently free to do, their desire to engage in the behavior increases. For example, when Miami prohibited the sale, possession, and use of laundry detergents containing phosphates, the regulation induced an artificial scarcity and resentment over the loss of freedom to choose. Consumers responded by increasing their ratings of the effectiveness of phosphates in detergent (Mazis 1975). In another example, 64 participants in a laboratory experiment were provided with statements that were said to be from a pornographic book (Zellinger et al. 1975). Half were also told that the book was restricted “to those 21 and over,” which substantially increased their desire to read the book. This phenomenon is widely observed and heavily researched and is referred to elsewhere in the literature as reactance (Ringold 2002).
Disclaimers sometimes conflict with current behaviors or attitudes, such as when consumers are informed of dangerous side effects from smoking. When people are exposed to information that challenges their beliefs or behavior, instead of changing they often react defensively by strengthening their current beliefs. Moreover, contrary to intuition but consistent with evidence from cognitive dissonance studies, when people believe that disconfirming evidence is valid, they tend to strengthen their prior beliefs (see, e.g., Batson 1975).
In a related phenomenon, advertisers sometimes use two-sided arguments. They describe the advantages to create positive beliefs about their product and then describe the problems (e.g., the car has extraordinary performance, but it is only available in manual, and changing gears requires skill). This type of argument increases the believability of their advertisements. Thus, customers exposed to a government-mandated message may believe that because the government has warned of a negative aspect of a product, they do not need to worry because the risk must have been regarded as acceptable by the government's experts; otherwise, the product would not be on the market.
Weak counterarguments are effective at increasing demand when potential consumers are cognitive misers and engage in relatively little effort to process an advertisement. For example, in four experiments involving 555 participants, the participants’ initial positive assessments of products were strengthened when they were exposed to weak negative information (Ein-Gar, Shiv, and Tormala 2012).
Often, mandated disclaimers are irrelevant to consumers. The presence of irrelevant information in messages can distract people and affect their decision making (Osterhouse and Brock 1970).
Much research has been conducted on how to improve readership, and the evidence has been summarized in the form of principles (Armstrong 2010). For effective communication of information, message text should be large enough so that even those with reduced vision can read it, and it should be placed on a white background in columns and in a standard serif typeface. Although presenting text in all capital letters and a bold sans-serif typeface might intuitively seem likely to emphasize a message, it actually reduces readability and readership. Thus, disclaimers are commonly presented in ways that violate the principles and thus discourage readership. For audio advertisements, disclaimers are presented using fast talkers, who sound authoritative and help save on media costs, but they are also not effective for conveying information.
Drafters of disclaimers, whether sellers or regulators, are subject to a handicap: Negative arguments and words are more difficult to understand than positive ones. Disclaimers increase the amount of text in an advertisement. Notably, evidence shows that people regard advertisements with more text as more believable—as in “long copy sells”— even when they have no time to read it (Meyer-Hentschel 1984). Thus, by its mere presence, a disclaimer might encourage greater consumption of the product (e.g., taking a drug) that the disclaimer is intended to discourage.
In summary, attempts to change behavior using mandatory disclaimers are often ineffective and, in many cases, lead to effects that are opposite to those intended. When the government takes more responsibility, citizens take less. Most people do not like being told what to do and may rebel. They cannot justify devoting time to details that will not affect their decisions and struggle to understand disclaimers that do affect them. 3
A California court case involving a mandatory disclaimer ended abruptly on October 15, 2010, a day before one of us was scheduled to testify, when the judge granted a request for a directed verdict after the state had rested its case. The lawyers making the request pointed out to the judge that the state had not met its burden of justifying the mandatory disclaimer and that the survey experts for the State of California had misinterpreted the disclaimer in the research that they had done to support its use. The case had been going on for seven years (Michael L. Potts, D.D.S., and the American Academy of Implant Dentistry v. Brian Stiger, Director, California Department of Consumer Affairs, et al. 2010).
Legal Basis for Commercial Speech Restrictions
Congress shall make no law … abridging the freedom of speech.
Our reading of the First Amendment to the U.S. Constitution suggests that it establishes an unconditional right to free speech: the right to choose for oneself what to say and what not to say. When Benjamin Franklin (1731) wrote an editorial regarding his publication of a sea captain's advertisement containing a note that offended some of his readers, 4 he made no “commercial speech” distinction in his defense of free speech. The First Amendment apparently applied without restrictions until the late 1920s.
The captain's note stated, using colloquial language, that he would not provide passage on his ship to prostitutes or ministers of the Church of England under any circumstances.
Thierer (2011) argues that it is not possible to make a clear distinction between commercial and other speech. Indeed, the Supreme Court examined the difficulty of properly drawing such a distinction in Virginia State Board of Pharmacy et al. v. Virginia Citizens Consumer Council, Inc., et al. (1976). In their opinion in favor of consumers who challenged a statute that prevented pharmacists from advertising prices of prescription drugs, the justices stated (§ V), “We see no satisfactory distinction between the two kinds of speech. … As to the particular consumer's interest in the free flow of commercial information, that interest may be as keen, if not keener by far, than his interest in the day's most urgent political debate.”
Nevertheless, the Supreme Court has created a commercial speech distinction and has ruled that such speech has lesser First Amendment protection and therefore can be regulated. Our review of the legal basis for commercial speech regulations (see the supplementary essay “Have the Courts Protected Free Speech for Business People?” in the Web Appendix; http://www.marketingpower.com/jppm_webappendix) has led us to conclude that government and the courts justify regulations on the assumption that they protect consumers from making bad decisions. To our knowledge, there is no evidence to support that assumption, a conclusion that was also reached in a review of mandated disclosures (Ben-Shahar and Schneider 2011).
When the U.S. Supreme Court decided not to hear a mandatory disclaimer case, Justices Thomas and Ginsburg issued a dissenting opinion (Borgner, Richard A., et al. v. Florida Board of Dentistry et al. 2002). The dissenting justices stated (¶ 13), “If the disclaimer creates confusion, rather than eliminating it, the only possible constitutional justification for this speech regulation is defeated.” The justices said (¶ 13) that the case presented “an excellent opportunity to clarify some oft-recurring issues in the First Amendment treatment of commercial speech and to provide … guidance on the subject of state-mandated disclaimers,” including clarification of “the nature and the quality of the evidence a State must present to show that [a disclaimer] directly advances the governmental interest asserted.”
We suggest that to obtain proper evidence, it is necessary to conduct experiments to predict the effects of each and every restriction proposed. Situations change, so it is necessary to conduct further experiments over time to determine whether a net benefit still exists.
Prior Evidence on Government-Mandated Messages
We investigate the issue of mandatory messages by first determining whether they reduce confusion and then examining whether they have beneficial outcomes. To address these issues, we relied on empirical, especially experimental, evidence. For complex situations such as this, findings from nonexperimental studies are unreliable (Armstrong 2012; Armstrong and Patnaik 2009). 5 By examining experimental evidence on the effects of mandatory disclaimers, we treat the issue as a subject for scientific investigation rather than as a matter for voting or expert opinion. To the best of our knowledge, this is the first comprehensive review of the evidence on mandatory disclaimers.
This problem is not unique to advertising. It occurs in other fields, such as epidemiology, in which researchers and officials are often misled by analyses of nonexperimental data (Kabat 2008).
Our primary criterion for including a study was that it employed an experimental or a quasi-experimental design to compare the effects of using a disclaimer with those of not using one. We considered any fully disclosed study, regardless of whether it was published in an academic journal.
We conducted Google Scholar, ISI (Institute for Scientific Information), and JSTOR searches for articles or legal opinions that contained the terms “experiment” and “mandatory disclaimers,” “corrective advertising,” and related terms. We also examined articles that cited key papers, such as the review of corrective advertising by Wilkie, McNeill, and Mazis (1984). We also posted our working paper on the Internet for many months and sought comments.
Our most successful search efforts involved contacting legal scholars and leading researchers on the topic and checking references from key studies. To ensure that our summaries were accurate, we sent our paper to the authors of the studies. Their replies led to many corrections. 6 We also asked the authors whether we had overlooked evidence, and their responses helped us find relevant experiments.
Wright and Armstrong (2008) find that academic studies often improperly summarize findings from published research, partly because the authors fail to read the articles they cite.
Government-Mandated Messages Cause Confusion
Consumers often fail to understand government-mandated messages. For example, in an experiment on corrective advertising, 83 participants heard one of four versions of a Listerine mouthwash advertisement. Two of the four versions of the advertisement included an FTC mandated disclaimer. Of the responses from the 36 participants who recalled a disclaimer after prompting, 39% misperceived the disclaimer in ways that harmed their assessments of aspects of the brand that were not addressed by the disclaimer (Mazis and Adkinson 1976).
Lawyers for the FTC proposed two sets of three corrective advertising messages for the pain relief drugs Excedrin and Bufferin. To test understanding of the messages, 451 participants were given questionnaires on at least two of the proposed statements. The proposed statements were each followed by ten choices: one or two correct interpretations of the proposed statement, six or seven misinterpretations of the proposed statement, a “none of the above” response, and a “don't know” response. Only 24% of choices the participants made were correct interpretations of a proposed statement (Jacoby, Nelson, and Hoyer 1982). One reason for the result is that disclaimers typically use negative words, and statements with negative words are difficult to understand (Armstrong 2010, pp. 185–86).
Berlex Laboratories (part of Schering-Plough Corporation) was required to provide a disclaimer stating that it had no relationship with another company, Schering AG. The disclaimer read: “Schering AG, West Germany, is not connected with Schering-Plough Corporation or Schering Corporation, Kenilworth, New Jersey.” An advertisement with the disclaimer was compared with one with no disclaimer, as well as with one that claimed that the companies were related (i.e., a “claimer”). The 600 physician and pharmacist participants were given as much time as they wanted, and they responded to questions immediately after they reviewed the advertisements. The disclaimer reduced the incorrect responses from 58% to 46%. However, and surprisingly, the percentage of people who thought the companies were related was lower for the claimer than the disclaimer (Jacoby and Szybillo 1994).
Government-Mandated Messages Have Unintended Effects on Beliefs and Behavior
The FTC's policy requires that remedies correct consumers’ misperceptions but not harm their evaluations of firms. This does not seem to be the case in practice, however, as the following two studies show. When 58 participants viewed a corrective advertisement about one product of a firm, they reduced their ratings of that firm's unrelated products (Johar 1996). Similarly, in a series of five experiments, the 961 participants exposed to an advertisement that included a correction were less persuaded by subsequent advertising for a different product by the same firm and by an unrelated firm selling a similar product (Darke, Ashworth, and Ritchie 2008).
In a lab experiment, 64 participants read “original” and “acceptable” advertisements for Firestone tires, Listerine mouthwash, Freihofer's bread, and Crown petroleum. The original advertisements were those that had been judged as deceptive in FTC proceedings, and the acceptable advertisements were those that had been modified from the original using FTC guidelines by either eliminating or qualifying offending content. Despite the drafters’ intentions, the “acceptable” alternative advertisements had the same effects on participants’ beliefs as the original advertisements. This lack of effect is not surprising, considering that 30 raters regarded none of the product attributes of concern to the FTC as relevant to their purchase decisions (Glassman and Pieper 1980).
How should patients react if they are informed that their doctor has a conflict of interest in recommending a treatment? In two experiments involving 1704 participants in the role of the patient, those exposed to a required disclosure were less inclined to trust their doctor, to accept the recommended treatment, and to see the doctor in future. However, they worried that the doctor would believe that they were concerned about bias if they turned down the recommendation (Sah, Loewenstein, and Cain 2011).
In a field experiment, approximately 200 male high school students were exposed to warning signs stating “DANGER, Shallow Water, You Can Be Paralyzed, NO DIVING.” The study found that they were more likely to dive into the shallow end of the pool than a similar number of students who were not exposed to the sign (DeTurck and Goldhaber 1991).
In a laboratory experiment, 155 participants exposed to an advertisement (picture of a bottle or can of alcoholic beverage) with the U.S. Surgeon General's warning displayed underneath rated the benefits as greater and the risks as lower than those who were given the advertisement without the warning. In addition, male participants exposed to the warning reported greater drinking intentions than those who were not (Blood and Snyder 1993; Snyder and Blood 1992).
In another laboratory experiment, 360 participants were provided with brief descriptions of 12 made-for-television films. The participants exposed to warnings that a film contained violent content more often chose a violent film than those who were not exposed to warnings (Bushman and Stack 1996).
In a test of the effect of a warning from the Surgeon General about the relationship between fat consumption and heart disease, participants were given a choice of full-, reduced-, or no-fat cream cheese. The 120 participants exposed to labels that included information on fat content and the warning were more likely to want to taste the full-fat cheese than the other cheeses. This pattern was similar for the 120 participants not exposed to the information and warning, but the 120 participants exposed only to the information on fat content were more likely to want to taste the lower-fat, allegedly healthier, cheeses. When asked to choose one of the cheeses to taste, the participants were no more likely to taste the lower-fat cheeses when they were exposed to the Surgeon General warnings than when they were exposed only to the information about the cheeses (Bushman 1998).
Three laboratory experiments involving the consumption of regular or low-fat M&Ms found that when foods were labeled as low fat, consumers, especially overweight consumers, ate up to 50% more (Wansink and Chandon 2006). A possible explanation is that they felt less responsible for their own health.
An experiment on perceptions of a fictitious new energy supplement among 78 current dietary supplement users tested the effects of a warning (“Caution: Taking more than the recommended serving may result in side effects such as high blood pressure, heart attack, or stroke”) and a disclaimer (“This statement has not been evaluated by the FDA. This product is not intended to diagnose, treat, cure or prevent any disease”). Participants exposed to the warning considered the product less safe but more effective than those not exposed to the warning. Participants exposed to the mandatory disclaimer did not perceive the product any differently than those who were not exposed to the disclaimer. A second study, involving a diet supplement and 199 participants, led to the same finding: The warning was effective in changing perceptions in the intended direction, but the mandated disclaimer was not. Indeed, among participants who were dietary supplement users, exposure to the disclaimer improved perceptions of the product (Mason, Scammon, and Fang 2007).
To test the effect of a mandatory disclaimer, 1471 randomly selected U.S. residents were shown football jerseys during an interview in their homes. The participants were shown five jerseys either with or without a disclaimer and were then shown a National Football League authorized jersey for comparison. The disclaimer read, “Not authorized or sponsored by the N.F.L.” The disclaimer had no meaningful effect on confusion, quality perceptions, or purchase preferences (Jacoby and Raskopf 1986). The authors suggested that this was consistent with behavioral research on information processing and the use of negative words.
In two experiments, 146 participants were briefly shown claims about health and medical matters on a computer screen. Claims described as false were later incorrectly remembered as true. Repetition of the disclaimer inflated this false conclusion after three days. Furthermore, older adults were more prone to this “illusion-of-truth” memory problem (Skurnik et al. 2005).
Evidence shows that an extract of “saw palmetto” berries provides relief for problems caused by noncancerous prostate enlargement common among older men. When questioning a convenience sample of older men, Eggers and Fischhoff (2004) found that 40% of 15 men would make choices against their best interests when exposed to a disclaimer. This compares with 22% of 9 men who would make poor decisions in the absence of the disclaimer.
Variations of each of a test advertisement for a mock antihypertensive and a mock antiarthritic drug were shown to 676 participants in groups of 20—about half of whom had high blood pressure or arthritis—during a 17½ minute television show. The pairs of advertisements were shown at 1 and 12 minutes into the program. Participants exposed to commercials that included federally mandated disclosures of specific product risks were much less aware of and knowledgeable about the benefits of the drugs than participants exposed to the commercials without the mandated disclosures (Morris, Mazis, and Brinberg 1989).
Experimental Evidence Prepared for Court Case on Florida's Implant Dentistry Disclaimer
We conducted an experiment on the effects of a mandatory disclaimer that the State of Florida required dentists to use if they advertised credentials from the American Academy of Implant Dentistry (AAID). We refer to it as the Florida mandatory disclaimer, or FMD.
All dentists licensed by the Florida State Dental Board are permitted to perform implant dentistry even though few have received formal training in these procedures. The American Academy of Implant Dentistry program offers two credentials: associate fellow and fellow. Each requires substantial skill training and experience, as is described on the AAID website (http://aaid.com/credentialing/index.html).
Treatment, Participants, and Administration
We designed our experiment to measure the extent to which customers comprehend the disclaimers regarding AAID credentials and how these disclaimers affect their decisions. Given the large expenditure and the risk involved in implant dentistry, choosing a dentist to perform implant dentistry is a decision that typically involves serious deliberation. Thus, consumers and those advising them are likely to attend closely to advertisements and seek further information.
We pretested the materials with seven people. Minor modifications to the wording of some questions were made from this pretesting. We commissioned Gallup & Robinson to administer the experiment. Neither Gallup & Robinson nor CRG Global, the fieldwork firm Gallup & Robinson employed, was aware of the purpose of the study or the identity of the study's sponsor.
During November 2007, CRG Global interviewed 317 people face-to-face in malls in Orlando, Daytona Beach, and Fort Lauderdale. Potential participants were screened to ensure that they were over 18 years of age and were able to read the English language materials. The interviewers approached 1053 people in total; 599 declined to participate, and 137 did not meet the screening criteria. 7 CRG Global reported that there were no problems with the interviewing.
Confusion sometimes arises in court cases about the need for randomization in the experimental design, and the issue was raised in this case. Although it is important to ensure that participants are from the relevant population, the key issue is that the assignment to the experimental treatments be based on a probability design. On the chance that the effect of the disclaimer might have depended on local factors, we conducted the experiment in the area covered by the disclaimer—Florida.
For more than a century, in the social and medical sciences as well as in advertising, convenience samples have been standard practice in experiments. When a tangible item (e.g., an advertisement) must be shown to a respondent, face-to-face interviews have been required. Jacoby and Handlin (1991) find that marketing researchers used mall interviews in 95% of their face-to-face studies (the others are door-to-door [3%] and other central locations [2%]). None of the studies used probability designs to select the subject pool. Jacoby and Handlin also analyzed articles in academic journals that described “primary empirical research and used samples of people either individually or in groups.” Based on a sample of 446 articles from 34 academic journals, they find that 97% used convenience sampling and screening questions to select the participant pool.
The issue of the random selection of participants arose also in the previously mentioned Berlex case, in which a New Jersey company was required to provide a disclaimer that it was not affiliated with a West German company with a similar-sounding name. The defendants in that case insisted on conducting a replication study using randomly selected participants. The findings were nearly identical to those from the study that used a convenience sample of participants.
To simulate a high-involvement situation in a realistic way, the interviewers asked the participants to imagine that they had a friend in need of implant dentistry. CRG Global conducted the interviews in locations that provided privacy so that participants would not be distracted or feel rushed. The participants received two mock Yellow Pages advertisements, each on its own card. Each advertisement was for a single dentist (“Dr. Alan Reed” or “Dr. Barry Smith”) and was titled “IMPLANT DENTISTRY.” Both advertisements described the advertiser as a “general dentist” and carried the statement, “Implant dentistry is a technique by which artificial replacement teeth are fastened to metal posts surgically implanted in a patient's jaw bones.”
In addition to the preceding information, the advertisement for Dr. Smith included only his name in all treatments. In contrast, Dr. Reed's advertisement described him as a fellow of the AAID and a diplomate of the ABOI/ID (American Board of Oral Implantology/Implant Dentistry). The information on Dr. Reed's credentials was followed by (1) no disclaimer, (2) the FMD, or (3) a modified disclaimer. The “nodisclaimer” variation is illegal under Florida law. Therefore, we wrote the “modified disclaimer” variation with the objective of causing the smallest harm to dentists who advertise AAID credentials while meeting the expressed aims of the Florida legislature in requiring a disclaimer.
The purpose of an advertisement is to present the seller's strongest arguments. In the case of dentists advertising implant dentistry services, this would include the attainment of credentials in implant dentistry. Dr. Smith's advertisement listed no qualifications other than a DDS. Therefore, Dr. Smith appeared less qualified than Dr. Reed. Because Dr. Smith did not have formal implant dentistry qualifications, he was not obliged to include a disclaimer in his advertisement. Figure 1 displays the four advertisements.
Advertisements Used in the Experiment
After showing the advertisements to the participants, the interviewer asked them which dentist (Reed or Smith) they would recommend to their friend. In roughly half the interviews, the interviewer collected the advertisements before the participants were questioned about their understanding of them. Table 1 shows the numbers of participants for each treatment.
Number of Participants by Treatment
Interviewers next asked the participants which of the two dentists they believed had the better implant dentistry qualifications. This was an alternative way of asking which dentist would be the best one to perform implant dentistry on their friend. Finally, the participants answered demographic questions.
Effects of the FMD on Consumers’ Decisions
When the advertisement for the AAID-credentialed dentist did not include a disclaimer, 13% of participants reported that they would recommend the other dentist who had no apparent qualifications for implant dentistry. When the advertisement for the AAID-credentialed dentist included the FMD, 21% reported that they would recommend the dentist without implant dentistry credentials. Thus, the FMD led to 1.6 times as many inferior decisions (see Table 2).
Effects of the FMD on the Percentage of Participants Who Recommended the Less-Qualified Dentist (n = 317)
Among those who did not see a disclaimer, 12% of participants without and 13% of those with a college degree would recommend the less-qualified dentist. Among those who did see the FMD, 25% of the less-educated participants would recommend the less-qualified dentist, in contrast with 16% of the better educated who would do so. In other words, the disclaimer was especially harmful to those with less education. Similarly, the disclaimer was more harmful to women than it was to men.
Prior research has shown that for high-involvement products, advertisements should contain only material that is relevant to consumers’ decisions (Armstrong 2010). Given that the disclaimer provides information on organizational arrangements among the AAID, American Dental Association, and Florida Board of Dentistry only, it is difficult to understand why it would be relevant to potential customers or to anyone advising them. This is consistent with the observation of Justices Thomas and Ginsburg in their dissent on an earlier AAID case that “the mandated disclaimer is likely to foster more confusion” (Borgner, Richard A., et al. v. Florida Board of Dentistry et al. 2002, ¶ 12).
Effects of Involvement
We used two additional ways to encourage participants to process the advertisements carefully, as they would do in a high-involvement situation. These were allowing participants to retain the advertisements and retesting the participants.
Long Exposure: Participants who Retained the Advertisements
Half the participants were allowed to retain copies of the advertisements while answering the questions. Ironically, those who retained the advertisement that included the FMD were much more likely to be confused about the AAID's role and credibility. Substantially larger proportions of those who retained the advertisement agreed with three false statements regarding the AAID's credibility (see Table 3). In other words, the longer participants were exposed to the advertisements containing the FMD, the greater was their level of confusion. Note that the AAID is a bona fide credentialing organization that confers credentials only after extensive training and experience. The defendants in the court case did not challenge this.
Effect of Retaining the Advertisement Including the FMD on Perceptions About AAID Credibility
Retesting after Participants Reflected on their Understanding
To test whether additional time and examination would reduce confusion, we asked participants a second time to choose between the dentists after they had answered the questions pertaining to their understanding of the advertisements. This time, we asked them which dentist they believed was better qualified to perform implant dentistry.
Among the participants shown the FMD, 19% recommended the less-qualified dentist, compared with 12% of those not shown the disclaimer. These responses correspond closely to the prior results on recommendations of 21% and 13%, respectively. Among the participants who received advertisements with the FMD, 16% of those who returned the advertisements before answering the questions believed that the dentist with no apparent qualifications specific to implant dentistry was the better qualified, compared with 22% of those who retained the advertisements. In other words, providing time for reflection led to more confusion.
Effects of a Modified Disclaimer on Consumers’ Decisions and Confusion
We examined the extent to which a modified disclaimer might provide protection to consumers. We tested one possibility (see Figure 2) that we expected would lead to less confusion than the FMD. It included a negative word, a problem that we could not overcome while adhering to the state's aims in mandating the disclaimer.
Modified AAID Disclaimer
Of the participants given the advertisement with the modified disclaimer, 15% recommended the dentist with no qualifications. While this is less than the 21% for those who received the FMD, it is higher than the 13% for those given no disclaimer. Thus, although the modified disclaimer was less harmful to participants’ decision making than the FMD, it did not eliminate the harm.
Effects of Mandatory Disclaimer on Sellers
Confusion about aspects of the AAID's credibility was higher among participants who had received the FMD than among those who had not. Indeed, 48% of those exposed to the FMD agreed or strongly agreed that the AAID was “not a bona fide credentialing organization,” whereas 36% of those not exposed to the FMD believed this. Because the disclaimer unjustly damages the reputation of the AAID, the FMD also harms individual providers who have AAID credentials—which, in the long run, could cause further harm to consumers by reducing the motivation of dentists to improve their skills in implant dentistry.
Discussion of the Role of Evidence to Support Mandatory Disclaimers
Mandated disclaimers are not free. The costs are passed on to consumers as higher prices and higher taxes. Higher costs lead people, especially the poor, to consider inferior substitutes (e.g., standing on a chair rather than buying a ladder). It is reasonable to ask for evidence of benefits that are greater than these and other costs to support the imposition of a mandatory disclaimer. We were unable to find a single instance of a mandatory disclaimer for which experimental evidence of net benefit was produced. We therefore expect that properly applying that criterion to each proposed restriction of commercial speech would eliminate the use of mandatory disclaimers.
In practice, however, the “commercial speech” distinction is a tenuous one, and the process of providing experimental evidence and fighting for it in the courts is expensive. When regulators, legislators, interest groups, and competitors want to restrict speech, they are motivated to identify an economic interest on the part of a speaker. That would not be difficult to do. Those with an interest in restricting speech could then call for regulation of the commercial by constructing an argument that net welfare will increase if the speech is restricted.
Shortly before we submitted our final version of this article, we learned of a review of the evidence on the related policy of mandated disclosures (Ben-Shahar and Schneider 2011). The review covered not only advertising but other areas as well, including Miranda rights, informed consent, and institutional review boards. The issue of mandatory disclosures might seem less contentious than mandatory disclaimers because they involve simply providing more information to those who might find it useful. Indeed, mandated disclosures are widespread, and enormous sums of money are spent with the intention of making them useful. In their wide-ranging review of the evidence presented in court cases and in social science literature, however, Ben-Shahar and Schneider (2011) find that mandated disclosures seldom provide clear explanations. When consumers do read them—typically they do not—they become confused. In the rare instances when they are not confused, they are unlikely to remember the information, or if they do, they rarely use it properly. Ben-Shahar and Schneider were unable to find a single mandatory disclosure for which the benefits outweighed the costs.
When disclosures are shown not to work, regulators try to solve the problem with different (typically longer) disclosures. These efforts to improve disclosures, however, lead to greater harm. Ben-Shahar and Schneider (2011) explain that mandated disclosures fail because they are based on false assumptions about how people make decisions, and they require a chain of unlikely achievements by lawmakers, disclosers, and disclosees. 8 As in our study, the efforts of lawmakers and regulators to improve on the functioning of markets have proved to be “fatal conceits.” 9
Ben-Shahar and Schneider (2011) asked readers to imagine that they were a doctor whose duty was to inform a patient on the use of a drug with 26 side effects, including heartburn, stomach ulcers, hepatitis, inflammation of skin, itching, and life-threatening allergic reactions. They then asked which side effects they would describe and whether they would answer differently if they knew that the drug kills between 3000 and 10,000 people per year in the United States. Furthermore, they asked whether the readers’ answers would change if they knew the drug was aspirin.
As Friedrich Hayek (1988, p. 76) wrote in The Fatal Conceit, “The curious task of economics is to demonstrate to men how little they really know about what they imagine they can design.”
Experimental evidence is consistent with economic theory and prior research on consumer behavior in finding that mandated disclaimers disrupt the functioning of markets. Disclaimers confuse customers and cause them to be less vigilant when making decisions. Disclaimers restrict sellers’ ability to provide customers with important information about their products and lead them to follow rules set by officials with inferior knowledge of the market. Moreover, the existence of mandatory disclaimers as a policy option encourages lobbying of politicians and regulators by competitors and by interest groups. For additional discussion of the legal evidence and First Amendment issues, see the Web Appendix (http://www.marketingpower.com/jppm_webappendix).
Conclusions
Disclaimers can provide important information to consumers and have been widely used since the beginning of advertising. Our concern herein has been only with the special case of mandated disclaimers. We found that the laws that restrict speech identified as “commercial” with mandates rest on unrealistic economic assumptions about the motivations and behaviors of consumers, business managers, and government officials. Moreover, we found experimental evidence from behavioral research on persuasion that mandatory disclaimers are unlikely to influence consumers in the ways that drafters intend but rather are likely to influence them in unexpected and detrimental ways.
We then examined evidence from 18 experimental studies related specifically to mandatory disclaimers. In all cases, the mandatory disclaimers caused confusion among consumers. Mandated messages were ineffective or harmful in the 15 experiments that examined perceptions, attitudes, or decisions.
To date, mandatory disclaimers have been imposed at the discretion of officials in contravention of economic understanding, in violation of the findings of research on persuasion, and in the face of direct experimental evidence showing that they are detrimental. Mandatory disclaimers fail to meet the criterion suggested by Justices Thomas and Ginsburg on “the nature and the quality of the evidence a State must present to show that [a disclaimer] directly advances the governmental interest asserted” (Borgner, Richard A., et al. v. Florida Board of Dentistry et al. 2002, ¶ 13). We suggest an extension to the Thomas and Ginsburg criterion: A mandatory disclaimer should be considered only if experiments demonstrate that it will provide net long-term benefits without causing serious harm to any buyers or sellers. Such a test properly applied would likely end the use of mandatory disclaimers.
We used this extended criterion to reexamine the FMD that was the subject of Borgner, Richard A., et al. v. Florida Board of Dentistry et al. (2002). Our experiment shows that the disclaimer confused potential customers, led them to make poor decisions, and unfairly harmed sellers. The judge found our evidence compelling (Ducoin Francis J., D.D.S., et al. v. Dr. Ana M. Viamonte Ros, in her official capacity as the State Surgeon General, et al. 2009).
In practice, however, when costs and benefits are considered, free speech becomes conditional on the opinions of courts and regulators about whether sufficient evidence exists that a particular speech restriction would increase welfare. They may even decide that the increase of one group's welfare is more valuable than the consequent loss in another's. Free speech then ceases to be a right, as commonly understood and intended by the First Amendment, and becomes instead an uncertain privilege subject to the opinions of courts and government regulators.