The impact of warnings, disclaimers, and product experience on consumers' perceptions of dietary supplements.

Two experiments explored the impact of disclosures (warning and disclaimer) and product experience (heavy, light, and nonusers) on consumers' perceptions of dietary supplements. In general, the government-mandated disclaimer did not impact consumers' beliefs about either the efficacy or the safety of dietary supplements; for most consumers, the disclaimer is unlikely to achieve its intended results. However, when product experience was examined, analyses show that heavy product users responded differently to the disclosures than did light product users and nonusers. Results point to the importance of understanding the influence of individual differences on the interpretation of product disclosures by specific consumer segments.

Information interventions have long been an important policy tool for enhancing consumers' ability to navigate effectively in the marketplace. The availability of relevant, accurate information is thought to assist consumers in making meaningful decisions about product and brand use. Frequently important product information is mandated for inclusion on product labels and in advertising. The Dietary Supplement Health and Education Act (DSHEA 1994) established new standards for information disclosure on dietary supplement labels. These standards were adopted in part to stimulate the availability of health-related products and to encourage the flow of information in the marketplace.

DSHEA's standards exemplify policymakers' attempts to balance the rights of sellers to make accurate and truthful claims about their products and the rights of consumers to have access to complete and nondeceptive information. The standards made fairly subtle distinctions among different types of product claims (health or disease claims, nutrition content claims, and structure-function claims) and introduced the requirement for a disclaimer previously unfamiliar to consumers. The wording of the disclaimer, mandated when a structure-function claim is made, is prescribed by DSHEA and states:

This statement has not been evaluated by the FDA. This product is
not intended to diagnose, treat, cure or prevent any disease.
This mandated disclaimer is intended to inform consumers that the regulatory review environment for structure-function claims for dietary supplements is different than what consumers are accustomed to for foods or drugs and for other types of claims. DSHEA also distinguished dietary supplements from drugs and exempted dietary supplements from Food and Drug Administration (FDA) premarket approval or safety testing requirements (Mason 1998). Without such requirements, specific hazards associated with use may not be known until extensive consumer use demonstrates adverse effects. Some specific products and/or brands do carry warnings associated with known hazards, but the presence of such warnings is not consistent because of difficulties in identifying risks and enforcing specific warning. The ongoing debate surrounding the energy supplement known as ephedra (65 FR 30677 June 4, 1997; 65 FR 17474-17477 April 3, 2000; and 68 FR 10417 March 5, 2003) illustrates these problems. Although the product has been reported to have contributed to numerous adverse (and severe) reactions, specific warnings about the supplement have been difficult to enforce. Consumers may be aware that risks exist for some products based on their own experience or word-of-mouth, but they may be confused by the presence/absence of warnings on various brands and products. The new requirements in DSHEA posed the challenge of finding effective ways to convey rather subtle messages to consumers (Mason and Scammon 2000).

Since 1994, additional rules have been adopted and new proposals made to further refine the disclosures required when health or disease claims are made in ads (Federal Trade Commission 1994) and on labels (FDA 2003). In response to issues raised in Pearson v. Shalala (1999), the FDA has attempted to make the standards for scientific evidence required to support health claims more flexible (FDA 2004; International Food Information Council 2005). In contrast to the evolution of standards for health claims and disclosures to be used with qualified health claims, little regulatory or research attention has been paid to structure-function claims and their government-mandated disclaimer.

In the current regulatory environment, finer distinctions between types of claims are being made and the ways in which claims can be qualified are being expanded, in part, to support sellers' rights to provide product information (Pearson v. Shalala 1999) and, in part, to provide more useful information to consumers (FDA 2003). However, evidence that the qualifications and clarifications allowed (or mandated) are likely not to work as intended (France and Bone 2005; Murphy 2005) raises questions about the factors that contribute to the diverse impacts such messages may have. Dietary supplements represent a unique context for expanding knowledge about the impacts of warnings and disclosures. Specifically, the information environment is unique because warnings of product risks may or may not be available (such disclosures are only required if postmarket consumer experience reveals risks) and the government-mandated disclaimer deals with the absence of FDA involvement in review of claims rather than with product-specific risks. Further, consumers who use dietary supplements are likely to do so for health reasons and thus may expect supplements to contribute positively to their health while not posing any adverse health consequences. Correcting or modifying these initial beliefs presents a particular challenge. Thus, research on consumers' understanding of disclosures in this unique environment could be helpful to policymakers as they attempt to find effective ways to ensure that consumers understand product information.

The purpose of this study was to explore consumers' understanding of product disclosures (1) in a changed regulatory environment. We are particularly interested in consumers' understanding of the disclaimer required for dietary supplements and of warnings appearing in the unique information environment. The structure-function claims and disclaimers are less specific and less familiar to consumers than health claims and warnings, which are more common for foods and drugs. Warnings appear more sporadically, with some relating to an entire product category and others referring to a specific brand. Of particular interest is whether and how responses to warnings and disclaimers differ across consumer segments. We focus on one individual difference variable, product user status, and its impact on consumers' interpretations of disclosure messages. We examine the meanings that heavy, light, and nonusers of dietary supplements take away from structure-function claims and disclosures and also examine the ways in which these messages impact consumers' perceptions of product efficacy, safety, and overall product evaluations.

DSHEA AND PRODUCT DISCLOSURES

DSHEA established structure-function claims as a new type of claim that describes the role of a nutrient or a dietary ingredient in the normal structure or function of the body. These claims may characterize the process by which the product acts to maintain structure or function and may describe the general well-being derived from consumption of the product. Examples of such claims include "calcium builds strong bones," "fiber maintains bowel regularity," or "antioxidants maintain cell integrity." Structure-function claims are less specific than disease claims and convey an overall qualitative sense of the product's benefits rather than fact-based support for the claim. These claims do not require FDA premarket approval, but sellers are responsible for ensuring their accuracy and truthfulness. Structure-function claims thus represent a different type of claim than what consumers are accustomed to with foods and health products.
DSHEA mandates that whenever a structure-function claim is made for a dietary supplement, a product disclosure in the form of a disclaimer must appear. The disclaimer is triggered by the seller making an affirmative structure-function claim and is thus designed to be interpreted in conjunction with the claim. The mandated disclaimer is unique from other disclosures because it is intended to stimulate consumers' consideration of the more global and complex product and information environment rather than to improve their knowledge about a single product attribute (e.g., fat content, pricing terms) or their understanding of expected performance based upon a summary of multiple product attributes (e.g., nutrition content based on fat, sodium, and others). Unlike a qualified health claim that is intended to help consumers understand the strength of support for product claims and unlike a warning that communicates a direct product effect, the mandated disclaimer makes a more abstract reference to the regulatory environment rather than to the product. Specifically, the mandated disclaimer provides general information about the lack of FDA review of products and their claims and is intended to put consumers on notice that they should be cautious in interpreting sellers' claims.

Dietary supplement labels may also be required to carry a more familiar warning message if a specific health concern related to product use is known (in general or for specific consumer segments) or when specific usage precautions are needed. Product warnings are a strong, explicit type of risk information intended to make consumers aware of a specific safety or health hazard related to use of the product. Generally, warnings do not contradict the purpose for which consumers seek the product; they merely provide additional relevant information. Consumers are likely to be familiar with product warnings since they appear on a wide variety of health and consumer products (e.g., pharmaceuticals, alcohol, and tobacco products). However, the warnings on dietary supplements may have a different impact on consumers because of the nature of the product category. Unlike tobacco and alcohol products (the context in which much warning research has been conducted), dietary supplements are not generally taken for hedonic purposes but rather for health reasons. With tobacco and alcohol products (and in the case of intentional misuse of drug products), warnings may be understood but ignored for the sake of pleasure and gratification. Because dietary supplements are promoted and consumed for health purposes, claims and warnings are likely to be understood/viewed in much the same way as those for pharmaceuticals. However, unlike drugs, which may carry warnings of multiple risks identified in premarket testing, dietary supplements may only carry warnings if consumer use has identified a specific risk. It is unclear under these conditions as to what meaning consumers will take from a warning. The risk may be viewed as more important (because of its presence) or as less important (because the overall product category is presumed to be safe).

Given the unique nature of the disclaimers and warnings, dietary supplements offer an interesting environment for examining the impact of disclosures including the ways in which consumers use such information to form product evaluations. We present the results of two experiments examining consumers' reactions to disclosures in the context of structure-function claims for dietary supplements. Our goal is to help marketers and policymakers gain a better perspective on how consumers understand qualifications to or disclaimers about product claims, especially when messages introduce unfamiliar information.

HYPOTHESES

Our hypotheses are framed in the context of the regulatory environment for dietary supplements and the "direct impact" anticipated by many public policy information interventions (see France and Bone 2005). When health risks are well known, warnings are required on dietary supplements. These warnings are very specific and intended to impact perceptions of product safety. The disclaimer is required when structure-function claims are made and refers to the lack of FDA review of the product and its claimed benefits. This disclaimer is intended to suggest to consumers that they should generally be cautious in interpreting the product claims, thus it is intended to impact perceptions of product efficacy. However, prior research on the impact of both health warnings and product disclaimers suggests that these intended impacts may not occur due to both message (disclosure variables) and audience (person variables) factors (Rogers, Lamson, and Rousseau 2000). We develop our hypotheses with both influences in mind.

The Impact of Product Disclosures

Warnings

The literature on the impact of product warnings is quite mixed (see Rogers, Lamson, and Rousseau 2000; Stewart, Folkes, and Martin 2001; Stewart and Martin 1994, for reviews); but research suggests that impact may vary significantly depending on message characteristics. Warnings communicating product risks can be specific or general, familiar or unfamiliar, and presented using more or less vivid language and images. Although the severity and nature of the hazard may impact individuals' perceptions of risk, consumers are more likely to notice warnings and remember precautions they can take to avoid the risk when warnings are particularly vivid (Slovic et al. 2002). The more specific the risk about which consumers are warned, the more likely they are to appreciate the actual risk (Stewart and Martin 1994). In research on the disclosure of risks related to prescription drugs, Morris, Mazis, and Brinberg (1989) found that the disclosure of specific risks led to greater risk information awareness than did disclosure of general risks. Similarly, in research on warnings for over-the-counter (OTC) antacids, Houston and Rothschild (1980) found that specific warnings led to greater recall and greater knowledge than did general warnings. Specific vivid warnings facilitate processing of the message and provide greater ease of encoding, which can aid interpretation of the risk message, impacting specific risk beliefs and more global attitudes (Bettman, Payne and Staelin 1986; Rucker and Petty 2006). Based on the findings that warnings do impact consumers' perceptions, we propose the following:

H1: Consumers exposed to the warning will display less favorable product safety evaluations than those exposed to the disclaimer or no-risk disclosure (control).
Less research on the impact of other types of disclosures on consumers has been reported, but it is likely that similar factors influence their impact. Disclosures are typically required to facilitate accurate global judgments or inferences. As an example, nutrition facts panels are required on food products' labels so that consumers can have information by which they can judge health claims about the product. Research on the impact of disclosures has examined their impact on both global judgments (e.g., healthfulness) and specific beliefs (e.g., sodium content) (see, e.g., Andrews, Burton, and Netemeyer 2000; Andrews, Netemeyer, and Burton 1998; Kelley, Gaidis, and Reingen 1989). However, the mandated disclaimer for dietary supplements presents a different situation. This disclaimer, which is required when a structure-function claim is made, includes a general statement not about risks associated with a particular product but rather about the general nature of the regulatory environment and the role of the FDA. The disclaimer thus includes an unfamiliar general message about an unfamiliar situation. The disclaimer presents an interesting case in which to examine consumers' responses relative to that of the more studied product warning. A recent study (France and Bone 2005) evaluated the impact of the disclaimer on subjects' beliefs about the issues specifically mentioned in the disclaimer. The study found that the disclaimer had no discernable impact on subjects' beliefs about the FDA's involvement in evaluating product claims or about the effectiveness of supplements in preventing, treating, or curing the focal disease. Overall, the authors concluded that the disclaimer did not have its intended effects in communicating to consumers' specific information about the regulatory environment for dietary supplements. Beyond the beliefs about the explicit information in the disclaimer, a further question is whether the disclaimer has an impact on more global judgments about the product. Johar and Simmons (2000) reported that disclaimers are less likely to impact inferences about global product or brand characteristics, especially when the information would require modification of previously held beliefs. Only vivid, explicit disclosures were able to impact consumer inferences (and then only under conditions of high motivation and high opportunity). In light of this prior research on the impacts of specific and general disclosures, we specifically test the impact of the disclaimer with respect to its intended purpose of raising cautions as to the efficacy of structure-function claims and propose the following:

H2: Consumers exposed to the disclaimer will display less favorable product efficacy evaluations than those exposed to the warning or no-risk disclosure (control).
The Impact of Risk Disclosures in Light of Product Usage

Another consistent finding in the warning literature is that the impact of warnings depends on audience characteristics (Rogers, Lamson, and Rousseau 2000; Stewart and Martin 1994). Research suggests that personal characteristics (i.e., age, personality), cognitive variables (i.e., attitudes, beliefs), and behavioral variables (i.e., experience with the product or the risk) can impact consumers' responses to warning messages. We focus on findings that link cognitive and usage variables to response to disclosures. Communications research suggests that there is a general tendency for consumers to interpret messages in light of their current beliefs and behaviors. This confirmatory bias enables consumers to maintain consistency between their beliefs and behaviors. In their review of warning research, Stewart and Martin (1994) suggested that different reactions to warnings can be explained in part by differences in the relevance of the message to consumers and consumers' predisposition toward the message. Product usage is likely to influence such predispositions. One's experience using a product is likely to contribute to differences in beliefs about product benefits. In turn, these product-related beliefs are likely to impact how consumers interpret warnings and other product disclosures.

Familiarity with a product, product category, or particular risk, especially gained through usage, may influence how consumers respond to disclosure messages. At the product category level, it is likely that heavy product users will have more positive prior beliefs about products than light product users, which impacts their response to disclosures and their perceptions of product-related risks. When investigating the impact of alcohol warnings, Patterson, Hunnicutt, and Stutts (1992) found that product usage levels impacted survey respondents' perceptions of product risks. Using an experimental design, MacKinnon and Lapin (1998) found that user status (drinking/nondrinking) influenced students' perceptions of product risks, and warnings had little effect on product perceptions. Andrews, Netemeyer, and Durvasula (1991) found that students who were frequent alcohol users reported greater disbelief in alcohol warnings than occasional users and nonusers. Research further suggests that if no risks have been experienced during use, consumers may become overconfident, leading to lower perceptions of risk (Rogers, Lamson, and Rousseau 2000). Such studies suggest that positive prior beliefs and previous benign experience with a product category may contribute to warnings or disclosures having less impact on consumers.

This, however, may not be the case with dietary supplements. In a product category where industry trust varies significantly (France and Bone 2005), consumers may be especially concerned with product risks. Because of their extensive experience with the product category, heavy product users may be more aware of the potential for health risks related to product use. This greater knowledge of risk potential within the product category should cause heavy product users to consider warnings more carefully than light product users. Light product users are likely to be less familiar with products overall and less aware that unknown risks may result from use. Warnings thus may impact the safety perceptions of heavy and light product users differently.

With regards to efficacy perceptions, other information sensitivities may be at work. Heavy product users are likely to have strong initial product-related beliefs resulting from varying reasons for and extent of experience that influence their reactions to product information (Andrews, Netemeyer, and Durvasula 1991; Slovic 1987). When individuals have strong initial views, these views may be resistant to change, and the strongly held beliefs may influence the way in which subsequent information is approached and interpreted. Slovic (1987) argues that "... new evidence appears reliable and informative if it is consistent with one's initial beliefs; contrary evidence tends to be dismissed as unreliable, erroneous, or unrepresentative" (281). In the context specifically of the disclaimer required with dietary supplements, France and Bone (2005) examined the impact of consumers' pre-existing beliefs and behaviors on their beliefs about the disclaimer message and product efficacy and concluded that prior beliefs act as "filters" through which the disclaimer information is interpreted. If the disclaimer is to have its intended effect of reducing expectations regarding promised benefits, we expect that this will be more likely with light product users. Heavy product users may disregard the disclaimer due to inconsistency with beliefs. Light product users are likely to be less committed to the benefits promised by supplements; thus, their efficacy evaluations may be more impacted by disclaimers. Thus, we propose the following general impacts of disclosures and interactions between type of disclosure and product usage.

H3: Consumers exposed to the warning will display less favorable product safety evaluations than those exposed to the disclaimer. The effect is greater for heavy product users than for light product users.

H4: Consumers exposed to the disclaimer will display less favorable product efficacy evaluations than those exposed to the warning. The effect is greater for light product users than for heavy product users.

STUDY 1: HEAVY AND LIGHT PRODUCT USERS' REACTIONS TO DISCLOSURES

Stimuli Design

To test the hypotheses, we designed an experiment in which disclosures associated with a fictitious energy supplement were manipulated. Exploratory research suggested that this product was relevant with consumers across age groups. A variety of supplements were available to consumers seeking energy, but no option dominated the market. To introduce the product to subjects, a Web page highlighted an illustration of an herbal plant; stated "Introducing Komar, a new, all-natural energy herb"; featured a structure-function claim, stating either "promotes physical and mental vitality" or "supports energy, stamina, and vitality"; and included the appropriate disclosure treatment. The structure-function claims used were actual claims that appeared on seller Web sites. These claims met existing FDA policy guidelines for dietary supplement labeling classification of a structure-function claim (FDA 2002). A pretest with seventy students examined participants' general understanding of the meaning of the claims. Subjects' verbatim responses to open-ended questions reflected the FDA definition for structure-function claims. Individuals viewing the claims referred generally to low energy, feeling tired or run down, or similar comments reflecting a temporary state of lowered vitality.

The disclosure messages used for the experimental manipulations were (1) the mandated disclaimer that must appear on any dietary supplement, making a structure-function claim and (2) a modified version of the warning proposed by the FDA for ephedra products (see 65 FR 30677 June 4, 1997; 65 FR 17474-17477 April 3, 2000; and 68 FR 10417 March 5, 2003, for debate on proposed wording). The actual disclosures were "This statement has not been evaluated by the FDA. This product is not intended to diagnose, treat, cure or prevent any disease" (disclaimer) and "Caution: Taking more than the recommended serving may result in side effects such as high blood pressure, heart attack, or stroke" (warning). The warning message specifically mentioned personal control (appropriate dosage) and side effects. The control condition involved exposure to the benefit claim but no disclosure, a realistic condition for consumers who view claims on the front of the label but not the more extensive information on the back panel.

Study Design and Procedure

This experiment used a 3 (disclosure: warning vs. disclaimer vs. control) x 2 (usage level: heavy product user vs. light product user) between-subjects factorial design. Product usage was assessed from subjects' survey responses regarding frequency and extent of current supplement use. The focus in this study was broadly on product category usage. In the context of a hypothetical product, it is likely that consumers' general comfort level with products in a given category influences their assessments, particularly when there is only limited product information available. A median split was used to divide respondents into two groups: light product users (take three or fewer supplements daily) and heavy product users (take four or more supplements daily).

Persons entering a nutrition retailer were approached, screened for product use, and asked to participate in a study about a new dietary supplement. After obtaining informed consent, participants were randomly assigned to a treatment group and exposed to the product Web site with the appropriate disclosure. Following exposure, they were asked to complete a survey that included questions about product evaluations and other attitudinal and demographic information.

Dependent Measures

Five semantic differential scaled items were used to assess product perceptions. Factor analysis (with varimax rotation and principle component extraction) revealed two underlying dimensions to product evaluations--efficacy and safety. Based on the analyses, composite constructs were created. The efficacy variable consisted of three 6-point items (ineffective/ effective, unhelpful/helpful, and not useful/useful), with factor loadings of .86, .84, and .86. The safety construct consisted of two 6-point items (unsafe/safe and harmful/harmless), with factor loadings of .89 and .93. In both constructs, higher ratings refer to more favorable evaluations (more effective and safer, respectively). These two factors explained 83.5% of the variance between items. Cronbach alphas were .89 for the efficacy construct and .86 for the safety index.
Participants

The sample consisted of seventy-eight consumers who were current users of dietary supplements. Subjects were primarily females (72%), between 26-35 and 36-45 years old (29% and 26%, respectively), and had some post-high school education but had not graduated from college (47%). They represented a cross-section of annual household income levels, ranging from less than $15,000 (12%) to more than $75,000 (12%). Their supplement use varied in terms of length, intensity, and purpose. Approximately 21% of subjects reported supplement usage of two years or less, while 17% reported usage of more than ten years. While most took a multivitamin for general health, several subjects consumed supplements for treatment of health conditions. Approximately 65% of subjects used a supplement for alleviating a sporadic illness (i.e., a cold, sore throat) and 34% for an ongoing chronic illness. Regarding health issues, 82% of subjects had visited a medical provider in the past year, 47% were living with a serious or chronic illness, 19% had no health insurance coverage, and 26% had no prescription drug benefit coverage.

Analysis and Results

Two structure--function claims were used to increase the likelihood that any effects occurring were due to the claim type rather than a specific message. Prior to testing for our hypotheses, independent t-tests on the dependent variables revealed no significant differences between the claims; thus, these messages were combined for subsequent analyses.

To assess the impact of disclosures and test our hypotheses, we ran two analyses of variances (ANOVAs), with disclosure (warning vs. disclaimer vs. none) and product usage (heavy product user vs. light product user) as the independent variables. Product safety and efficacy perceptions were the dependent variables for the respective ANOVAs. The ANOVAs revealed several significant effects on safety and efficacy perceptions, which are shown in Table 1 and discussed below.
H1 suggests that the warning message will impact perceptions of product safety, while H2 suggests that the disclaimer will impact perceptions of product efficacy. A significant main effect of disclosures on safety perceptions was found ([M.sub.warning] = 3.20, [M.sub.disclaimer] : 4.34, [M.sub.control] : 4.00; F(2, 70) = 6.44, p = .003, [[eta].sup.2] =.16). Post hoc Bonferoni tests revealed that safety perceptions were significantly lower when subjects were exposed to the warning compared to the disclaimer (p = .001) and no-disclosure control (p = .03). As expected, given the strength of the warning compared to the disclaimer, subjects viewing the warning perceived significantly lower product safety, thus supporting H1. It is also important to note that no difference was found between the disclaimer and the no-disclosure control (p = .79). It appears that the disclaimer did not have a significantly different impact on perceptions of safety than did a message without any disclosure information.

With regards to the impact of the disclosure messages on perceptions of efficacy, a significant main effect for disclosure was found ([M.sub.warning] = 3.51, [M.sub.disclaimer] = 2.90, [M.sub.control] = 2.84; F(2, 70) = 3.50, p = .04). Post hoc Bonferoni tests indicated a significant difference in efficacy perceptions between exposure to the warning and the control (p = .04). When subjects saw a warning, their efficacy ratings were significantly higher than when viewing no disclosure. No difference was found in efficacy perceptions for subjects exposed to the disclaimer compared to the control; thus, H2 is not supported. It is particularly noteworthy that the mandated disclaimer did not impact either efficacy perceptions (as intended) or safety perceptions (as might be expected, given the general nature of the disclaimer) any differently than the control message.

H3 and H4 propose interaction effects between type of disclosure and usage level, that is, the two types of disclosures will have different effects for light and heavy product users. H3 focuses on safety perceptions and suggests that for heavy product users more than for light product users, a warning will lead to lower safety perceptions. Our findings reveal a significant interaction between product usage and disclosure; F(2, 70) = 10.59, p = .003, [[eta].sup.2] = .16) (Table 1; Figure 1). Heavy product users viewing a warning reported significantly lower safety evaluations than light product users across conditions and heavy product users viewing the disclaimer or control. Thus, H3 is supported. H4 focuses on efficacy perceptions and suggests that for light product users more than for heavy product users, a disclaimer will lead to lower efficacy perceptions. Contrary to our hypothesis, no significant differences in efficacy evaluations were found between heavy and light product users. Thus, H4 is not supported. The significant interaction between type of disclosure and product usage appears to be driven by differences in the perceptions of heavy product users.