Presumed influence of endorser and fear appeal in DTC prescription drug advertising: Are they overpowering consumers’ judgments?

Abstract

As consumers are increasingly exposed to direct-to-consumer prescription drug advertising, critics are concerned about frequent uses of emotional appeals and influential endorsers that may mislead consumers’ judgments in favor of the advertised drug. To investigate this issue, we developed a set of hypotheses based on the marketing and consumer behavior literature and tested independent and interactive effects of fear appeal and endorser types (physician, peer, and no endorser) on consumers’ cognitive and attitudinal responses to direct-to-consumer ads. Results from an experiment suggest that the endorser and fear appeal can influence consumers’ evaluation of the ad message, but do not necessarily overpower their judgments regarding the drug. Findings on the interactive effects suggest that direct-to-consumer advertisers must give a careful consideration to the right match of an endorser type and an emotional appeal to avoid any negative unintended effects. Implications for future research and practices for direct-to-consumer advertising are discussed.

Keywords

Pharmaceuticals marketing communication direct-to-consumer advertising endorser types fear appeal

Introduction

Direct-to-consumer advertising of prescription drugs (DTCA) refers to ‘any promotional effort by a pharmaceutical firm to present prescription drug information to the general public through the lay media.’¹ Advertising and other promotional activities for prescription drugs had been primarily targeted at physicians. However, the FDA permitted DTCA on print media in 1985 and further relaxed its regulations for broadcast media in 1997. Since then, DTCA has become one of the fastest growing ad categories, reaching US$4 billion in 2011.² Consumers are exposed to as many as 16 hours of DTCA on television alone each year.³ Given DTCA’s considerable return on investment (US$4.20 in drug sales for every US$1 spent on DTCA⁴), the pharmaceutical industry’s investment on DTCA is expected to grow continuously.

The prominence of DTCA on consumer media, however, has raised concerns about its impact on consumers’ healthcare decision-making. Particularly, DTCA has been criticized for its skewed emphasis on the drug’s benefits over risks, interference with the physician–patient relationship, and undue influence on consumers’ requests for prescription drugs. Furthermore, critics argue that DTCA’s frequent employment of emotional appeals impedes consumers’ ability to make an appropriate treatment-seeking judgment and misleads them into believing that the drug is better than it actually is.⁵ In fact, several content analysis studies revealed that the majority of DTCA on television (95%) and magazines (67%) relied on emotional appeals, especially fear appeal, and that there was a patterned association between certain emotional appeals and health conditions (e.g. fear appeal for hypertension and high cholesterol; humor appeal for arthritis and asthma).^3,6

In addition, one of growing trends in DTCA is featuring various types of endorsers such as physicians and typical consumers. The significant persuasive influence of endorsers is documented extensively in the literature. Critics of DTCA argue that endorsers shift consumers’ attention away from important drug information and lead to misinterpretation of drug endorsement. Particularly, critics are concerned that endorsers used in DTCA do not always possess real expertise or experience to make valid claims about the drug.⁷ More problematically, the endorsers are sometimes actors who play a physician or a lay consumer. One of the most controversial cases was Pfizer’s DTCA for its anti-cholesterol drug Lipitor, which featured Dr. Robert Jarvik. Although he was introduced as a distinguished doctor who invented the artificial heart in the ad, it was later revealed that he had never been licensed as a medical doctor and could not legally prescribe a drug. The claim regarding the invention of the artificial heart was also refuted by former colleagues.

Despite the criticisms and debates regarding persuasive tactics used in DTCA, however, a paucity of research has examined their effects on consumers’ cognitive and attitudinal responses to DTCA. In this study, we seek to address this void in the current literature and test independent and interactive effects of endorser types and fear appeal in DTCA via an experiment. Specifically, we examine the relative persuasive influence of the physician and peer as endorsers and also compare their influences against when no endorser is used in the ad. In addition, we assess consumers’ engagement with the ad, perceived message quality, and attitudes toward the ad and drug brand as a result of fear appeal (vs. no-fear appeal). Finally, we propose and test two alternative hypotheses regarding interactive effects of endorser types and fear appeal. The findings of our study address some of the aforementioned issues raised by DTCA critics and provide important implications for DTCA practitioners and policy-makers.

Literature review

Endorser types

Physicians can exert significant persuasive influence as endorsers of DTCA via perceived expertise. Physicians’ assertions on prescription drugs are considered valid because they possess specialized knowledge about the product as a result of extensive experience and formal training. Their product-relevant expertise can help consumers solve a specific problem (e.g. health condition), thereby increasing perceived value and utility of information in DTCA. Researchers hold that the expert endorser’s persuasive influence is especially powerful because the expert’s opinions, once adopted, are integrated into consumers’ own belief system and sustained even if the endorser is forgotten later.⁸

Peers, normally portrayed by typical consumers, draw their appeal from similarity with consumers. Similarity may be based on observable physical characteristics (e.g. age, gender, and ethnicity) or internal traits (e.g. beliefs, interests, and attitudes).⁹ Peer endorsers enhance consumers’ judgmental confidence about the product because they are believed to evaluate the product from the same point of view as consumers. Perceived relevance and acceptance of the message increases as the endorser is viewed as more prototypical and representative of the consumer’s membership group.¹⁰

A substantial amount of research has shown significant influence of expert (vs. non-expert) endorsers in various persuasion contexts. Yet, empirical evidence regarding persuasive influence of similar (vs. dissimilar) endorsers is limited in the current literature and the relative effects of expert endorsers versus similar endorsers remain largely unexplored. However, studies investigating endorser-by-product interactions suggest that the effectiveness of an endorser type may be determined by the type of product advertised. In comparison of an expert, a celebrity, and a typical consumer as endorsers, Friedman and Friedman¹¹ contended that an expert is the most suitable endorser for products that are associated with high financial, performance, or physical risks, whereas a celebrity is highly influential for promoting products associated with social or psychological risks. They considered a typical consumer to be an appropriate endorser for products with little inherent risk. In Feick and Higie’s study,¹² consumers relied on a dissimilar expert’s opinions for products that are typically evaluated with objectively measurable standards and vary little in individual tastes and preferences (low preference heterogeneity; e.g. auto mechanics). For products that are evaluated mostly based on individual preferences (high preference heterogeneity; e.g. night clubs), consumers valued an inexperienced, but similar endorser’s opinions. More recently, Stafford et al.¹³ and Lord and Putrevu¹⁴ showed that utilitarian (vs. hedonic) products and products appealing to consumers’ informational (vs. transformational, sensory, or social) motives were promoted effectively by expert endorsers as they conveyed benefits and values of the product most convincingly.

The match-up hypothesis explains these endorser-by-product interactions in terms of congruence between endorsers’ and products’ attributes and images. The endorser whose attributes and images are congruent with the advertised product’s attributes and images communicates product benefits quickly and influences product evaluation positively.¹⁵ The endorser’s product-congruent attributes carry informational value and help consumers encode and recall brand attributes readily. Petty and Cacioppo¹⁶ argued that product-relevant endorser attributes serve as a persuasive argument and strengthen the quality of the message about the product.

Drawing on the above literature, we propose that the physician will exert greater influence as an endorser than the peer in engaging consumers with the DTCA message, enhancing perceived quality of the message, and forming favorable perceptions and attitudes toward the ad, drug, and health issue featured in the ad. Prescription drugs are utilitarian (vs. hedonic), low (vs. high) preference heterogeneity products and associated with high perceived physical risks. Prescription drugs are also more likely to appeal to consumers’ informational (vs. transformational) motives. Consumers will thus consider the physician as a more appropriate referent for the evaluation of a prescription drug than the peer and accept the physician’s message as valid assertions. From the match-up perspective, the physician is intrinsically matched to the product category. This natural match-up will increase the effectiveness of the physician as an endorser, resulting in greater attitude change than prompted by the peer endorser. Considering the criticism regarding undue influence of endorsers used in DTCA, we also assess the endorser effects of the physician and peer compared with when no endorser is used. Based on previous research findings on positive effects of endorser expertise and similarity, we predict that the physician and peer will have greater effects than when no endorser is used. Specifically, we propose the following hypotheses. H1

Involvement with the ad message (H1a), perceived message quality (H1b), attitudes toward the ad (H1c), perceived drug efficacy (H1d), drug inquiry intentions (H1e), and perceived severity of the health issue (H1f) will be greater when the physician (vs. peer or no endorser) is used as the endorser in the ad.

Involvement with the ad message (H2a), perceived message quality (H2b), attitudes toward the ad (H2c), perceived drug efficacy (H2d), drug inquiry intentions (H2e), and perceived severity of the health issue (H2f) will be greater when the peer (vs. no endorser) is used as the endorser in the ad.

Fear appeal

The effects of fear appeal on message processing

Fear appeal has been a popular message strategy in advertising. Its popularity stems from the notion that the presentation of information alone is not sufficient to motivate behavior change, but some form of emotional arousal is necessary. Fear, in particular, is identified as the most ‘compelling persuader’ in all emotions¹⁷ and fear-arousing ads have been shown to enhance interest and recall of the ad and brand, desire to seek further information, and adoption of the recommended behavior.¹⁸ In comparison of multiple message strategies, studies find that the ads portraying fear are remembered better and recalled more frequently than the ads showing warm, upbeat, other positive feelings, or no emotional content.¹⁹ Similarly, individuals exposed to different types of message appeals rate fear appeal to be more effective than rational appeal or humor appeal.²⁰

According to the drive response model²¹ and the arousal theory,²² fear appeal heightens drive and tension resulting in greater vigilance to the ad message. When fear is not elicited or elicited at a low level, consumers are little motivated to process the message as they deem the recommended change in the message inconsequential and unnecessary. A sufficiently strong fear appeal can introduce interest and engage consumers in an active search of solution to the problem presented in the ad. In a similar vein, the limited capacity model of motivated mediated message processing²³ posits that an emotional content in the media message automatically activates two fundamental motivational systems (appetitive and aversive systems) that underlie basic human emotions. These motivational systems direct energy and cognitive resources so to address particular kinds of adaptive problems. The appetitive system is activated by a positive emotional message and encourages exploration of the new environment and information intake. The aversive system is triggered by a negative emotional message and produces self-protective responses (e.g. fight, flight, or freeze reactions).²⁴

Many researchers consider the valence of emotion as a significant determinant of consumers’ distinct kinds of message processing. They suggest that a negative feeling state engenders more analytical, systematic processing with increased attention to details and closer scrutiny of message quality,²⁵ whereas positive affect induces more simplified, heuristic processing and greater reliance on mental shortcuts and peripheral cues of the ad.²⁶ Based on the functional approach of emotion, this view holds that the current feeling state serves as information about the emotion-arousing environment.²⁷ Negative affect signals a threat to personal well-being and motivates consumers to engage in careful causal analysis of the environment. Positive affect, in contrast, indicates that the environment is safe and current understanding of the environment is sufficient to make a judgment. Consumers in a positive affective state thus seek to reduce the complexity of information in order to simplify their judgment and maintain the current affective state.

Appraisal theories of emotion extend the valence-based view and emphasize qualitatively different contents of discrete emotions. The appraisal theories focus on cognitive antecedents of emotional experience and posit that each discrete emotion arises from the unique process by which consumers appraise personal significance of an emotion-evoking event along several cognitive dimensions.²⁸ Fear, for example, is associated with pleasantness, certainty, and control dimensions of appraisal. It is aroused when personally important values and goals are threatened (unpleasantness) due to a situational factor (situational control) and the consumer is uncertain about what will happen and feels vulnerable to the threat (uncertainty).²⁴ These cognitive appraisals account for unique adaptive responses in terms of thoughts (how bad things could get), action tendencies (running away), and motivational goals (get to a safe place) that occur as a result of fear arousal.²⁹ Increased elaboration of the message, in particular, is attributed to the uncertainty appraisal of fear. Tiedens and Linton²¹ found that consumers engaged in systematic processing of the message that elicited fear. When the message elicited other discrete emotions characterized by certainty appraisal (e.g. disgust, happiness, contentment), they observed greater influence of heuristic cues such as source credibility and stereotypes.

The effects of fear appeal on attitude change.

Although fear appeal is generally thought of as a catalyst for attitude change, empirical results and theoretical explanations for the effects of fear appeal on attitude change have not been consistent in the literature. Particularly, there is a lack of consensus on the relationship between intensity of fear felt and amount of attitude change. Early research tended to adopt the view of inverted-U relationship, in which a moderate level of fear exerts the greatest persuasive influence.³⁰ However, this view was later criticized for its elusive conceptualization and ad hoc delineation of mixed findings. Instead of theorizing the persuasiveness of fear appeal as a function of levels of fear arousal, more recent research focuses on the elements of fear appeal that produce either facilitating or inhibiting effects on attitude change.

The protection motivation theory (PMT)³¹ and the extended parallel processing model (EPPM)³² are both grounded on the premise that fear appeal can lead to desirable coping responses or dysfunctional responses depending on individuals’ threat appraisals and coping appraisals. Threat appraisals involve assessment of severity and vulnerability of the threat, whereas coping appraisals involve individuals’ belief in the effectiveness of the recommended action for averting the threat (response-efficacy) and confidence in their ability to perform the recommended action (self-efficacy). To induce desirable attitudes and behaviors, it is necessary that fear be elicited first and then followed by an assurance message that instills the above efficacy beliefs.³³ When fear is not elicited, individuals lack interest and motivation to pay attention to the message. When fear is elicited, the message that enhances the efficacy beliefs leads to desirable adaptive responses, such as changes in attitudes and behaviors as suggested in the message (danger control). When a fear appeal lacks the assurance message, it results in maladaptive responses, such as message derogation and avoidance, denial of the threat, and distortion of the ad’s meaning (fear control).

In summary, the preceding review of literature suggests that the negative valence and cognitive antecedents associated with fear appeal promote a careful analysis of the message and enhance consumers’ engagement of message processing. The above literature also suggests that fear appeal produces desirable attitude change if it is accompanied by the assurance message containing efficacy beliefs. In DTCA, the assurance message is naturally incorporated since the prescription drug is presented as an effective means to alleviate the health condition.^34,35 Therefore, we hypothesize that DTCA using fear appeal will engender positive effects on variables related to processing of the ad message and attitudes toward the ad, drug, and health issue, compared to DTCA not using fear appeal. The following is the specific hypothesis proposed. H3

Involvement with the ad message (H3a), perceived message quality (H3b), attitudes toward the ad (H3c), perceived drug efficacy (H3d), drug inquiry intentions (H3e), and perceived severity of the health issue (H3f) will be greater when fear appeal is used (vs. not used) in the ad.

Endorser types and fear appeal

The effectiveness of an ad is usually determined by multiple elements of the ad that are at work simultaneously. Particularly, endorser factors and message factors can have both significant independent and interactive effects on consumers’ responses to the ad. For example, many researchers have found that a strong (vs. weak) message argument in the ad reduces (vs. increases) the persuasive influence of an endorser because it focuses consumers’ attention on the central message feature rather than a peripheral endorser factor.¹⁶ In Jones et al.’s study,³⁶ message framing had interactive effects with endorser credibility such that a positively framed message from a credible endorser (a medical doctor) yielded greater message elaboration and more favorable attitude and behavior changes toward exercise than a negatively framed message from a credible or non-credible endorser (a high school science student). They concluded that the message effectiveness is enhanced when a credible endorser emphasizes benefits of regular exercise rather than risks of irregular exercise. Miller et al.³⁷ found that the message using controlling language negatively affected perceptions of the endorser’s expertise, trustworthiness, and sociability, whereas the message using concrete language positively influenced these endorser perceptions. These studies as well as other similar studies³⁸ imply that messages that are perceived to threaten one’s well-being or freedom lead to psychological reactance and endorser derogation. Receivers of a threatening message attempt to reduce reactance pressure by devaluating the credibility of the endorser. According to this line of research, an expert endorser such as the physician will likely have greater effects than the peer endorser or no endorser if fear appeal is not used (vs. used) in the ad.

Opposing views and evidence have been presented by other researchers, however. Hewgill and Miller³⁹ conducted a series of experiments to test interactive effects of fear appeal and endorser credibility. They consistently found that a strong fear-arousing message produced greater attitude change than a mild fear-arousing message when the message was attributed to a highly credible endorser. They found no difference in the persuasive effects of the strong versus weak fear-arousing messages that were attributed to a low credible endorser. They argued that while a strong fear-arousing message induces cognitive imbalance in message receivers, it is psychologically difficult for message receivers to resist the message argument if the endorser is expert and trustworthy. Message recipients thus adopt the attitudes advocated in the message as a means to restore cognitive imbalance. If the endorser has low credibility, the message argument is discredited, resulting in little persuasive influence regardless of the level of fear arousal. From the cognitive response theory perspective,⁴⁰ a fear-arousing message presented by a highly credible endorser inhibits counter-argumentation and facilitates the pursuit of adaptive action via the danger control process. Contrary to the studies reviewed earlier, this body of research suggests that an expert endorser such as the physician yields greater effects than the peer endorser or no endorser if fear appeal is used (vs. not used) in the ad.

These conflicting views on the relationships between endorser types and fear appeal indicate that there is insufficient empirical evidence and a lack of established theories in the current literature to develop unidirectional hypotheses. We, therefore, propose two competing hypotheses based on the above literature and predict different moderating effects of fear appeal on endorser types. Specifically, one hypothesis predicts greater effects of the physician endorser (vs. peer or no endorser) when fear appeal is not used in the ad (H4), whereas the other hypothesis predicts greater effects of the physician endorser (vs. peer or no endorser) when fear appeal is used in the ad (H5). We put forth the following hypotheses. H4

The physician will induce greater effects on involvement with the ad message (H4a), perceived message quality (H4b), attitudes toward the ad (H4c), perceived drug efficacy (H4d), drug inquiry intentions (H4e), and perceived severity of the health issue (H4f) when fear appeal is not used (vs. used) in the ad.

The physician will induce greater effects on involvement with the ad message (H5a), perceived message quality (H5b), attitudes toward the ad (H5c), perceived drug efficacy (H5d), drug inquiry intentions (H5e), and perceived severity of the health issue (H5f) when fear appeal is used (vs. not used) in the ad.

Methods

Design and procedure

The study employed a 3 (endorser types: physician/peer/no-source) × 2 (fear appeal: fear/no-fear) between-subject experimental design. Undergraduates (N = 263) with diverse majors at a large southern state university (47 different majors including business, liberal arts, and communications) participated in the study in exchange for extra course credit. The majority were female (63%) and Caucasian (57%) followed by Hispanic (13%) and Asian (13%). The average age was 19 (SD = 1.88). College students were chosen as the study sample due to the relevance of the health issue examined in the study (sleep disorder) and accessibility of the sample. Results need to be interpreted with caution, however, since they are not a representative sample of general consumers and they are relatively light users of prescription medicines.

Participants completed the study in a computer laboratory using individual laptop computers. They first rated their level of involvement with sleep disorders. They then viewed one of six randomly selected stimulus ads at their own pace and completed a questionnaire measuring dependent variables, covariates, and manipulation checks.

Stimulus ads

Six versions of a stimulus ad were created for a fictitious prescription drug brand that treats sleep disorders. The topic of sleep disorders was chosen because national surveys report that college students are among the most sleep-deprived segments of population and highly vulnerable to sleep disorders.⁴¹ A general lack of sleep and irregular sleeping behaviors among college students have been linked to depression, heart diseases, and injuries resulting from automobile accidents. Consistent with national data, undergraduates in our pretests rated sleep disorders significantly higher on illness severity, personal vulnerability, and impact on their lives than 27 other health conditions (e.g. sexually transmitted infections, skin cancer, and flu). In addition, sleep-aid medication is one of the most heavily advertised prescription drug categories to consumers (e.g. Lunesta and Ambien CR were among the top ten brands with the highest advertising expenditures from 2007 to 2011).⁴²

Except for the no-endorser condition, the stimulus ad contained an endorser’s (physician or peer) picture and a copy incorporating the fear appeal manipulation. The ad for the no-endorser condition contained the copy only. Since the study sample was to be drawn from a female-dominant population, only female endorsers were considered for the study. While participants’ sex was statistically controlled in hypothesis testing, a preliminary analysis showed that it did not have a significant effect on dependent variables. A fictitious physician who was described as ‘the director of the National Association of College Health Centers’ and having ‘been caring for the health of college students for over 20 years’ and a layperson who appeared to be a college student were selected based on a series of pretests, which indicated that the physician and peer produced distinct perceptions of endorser expertise and similarity, respectively (e.g. perceived expertise of the physician, M = 5.97/7, SD = 1.39; perceived similarity of the peer, M = 5.36/7, SD = 1.49).

Guided by the EPPM-based research,⁴³ fear appeal was manipulated by incorporating the threat and coping appraisals. The threat appraisal was stimulated by describing severe life-threatening symptoms (e.g. choking and gasping for air while sleeping) and salient consequences of the illness to college students (e.g. depression, reduction of sexual desire, weight gain, car accidents and injuries). For the no-fear appeal, mild symptoms (e.g. clumsiness, difficulty paying attention, and irritability) and less salient consequences such as common chronic illnesses (e.g. diabetes, heart diseases, and obesity) were described. Overall, negative aspects of the illness were emphasized in the fear appeal, whereas the illness was described in a neutral, factual tone in the no-fear appeal. For both appeals, the advertised drug was presented as an effective means to treat sleep disorders (coping appraisal), while side effects of the drug and the prompt to ask a doctor about the drug were included as typically found in DTCA.

Measures

For dependent variables, message involvement, perceived message quality, attitudes toward the ad, perceived efficacy of the drug, drug inquiry intentions, and perceived severity of sleep disorders were measured. Most measures were adapted from previous research, which are cited in the following. When the measure consisted of multiple items, internal consistency was assessed using Cronbach’s alpha. The alpha value of each measure obtained in this study is reported below.

Message involvement was measured with seven questions on a 7-point scale (1 = not at all, 7 = very much) regarding the extent to which participants were involved in processing the written message in the ad (e.g. how interested/engaged/involved were you in the written message in the ad?; α = 0.94).⁴⁴ Semantic differential scales were used to measure perceived message quality, attitudes toward the ad, and perceived drug efficacy. Perceived message quality was rated with five items (e.g., illogical/logical; not compelling/compelling; α = 0.91),⁴⁵ attitudes toward the ad with three items (bad/good; unfavorable/favorable; negative/positive; α = 0.90),⁴⁶ and perceived drug efficacy with two items (unreliable/reliable; ineffective/effective; α = 0.90). Drug inquiry intentions were measured with six statements on a 7-point scale (1 = very unlikely, 7 = very likely),⁴⁷ assessing intentions to talk to health professionals (e.g. physicians, nurses, pharmacists), friends, and family about the advertised drug and to seek additional information from mediated sources (e.g. the Internet, magazines), if they thought they had a sleep disorder (α = 0.91).

The manipulation of fear appeal was checked by asking participants to evaluate the ad in terms of fearfulness, anxiousness, and nervousness on a 7-point scale (1 = not at all, 7 = very much; α = 0.91).⁴⁸ The endorser was evaluated for perceived expertise (e.g. experienced, knowledgeable; α = 0.95), trustworthiness (e.g. sincere, honest; α = 0.95), and similarity (e.g. has similar/dissimilar values as me, quite a bit like me/not at all like me; α = 0.90) using 15 semantic differential items.⁴⁹ Trustworthiness was measured additionally because it is often conceptualized as a critical dimension of endorser credibility and thus could explain potential effects of endorsers that were not hypothesized. The endorser’s perceived congruence with the product was assessed with four semantic differential items (e.g. compatible/consistent with the product; α = 0.97) adapted from previous research.⁵⁰

In addition, several covariates that could potentially affect dependent variables were measured. Involvement with sleep disorders was measured using five semantic differential items (e.g. unimportant/important; of no concern/of concern to me; means nothing to me/means a lot to me; α = 0.94).⁵¹ Participants’ demographic characteristics (e.g. sex, age, and ethnicity), history of sleep disorders, and prescription drug usage for a sleep disorder were also measured. Participants were asked to indicate their ethnicity on a multiple choice question (e.g. Caucasian/White, Black, Asian, Hispanic, etc.). Each category was dummy-coded before it was entered in the hypothesis testing model as a covariate. A set of four dichotomous questions (yes/no) were asked to assess the history of sleep disorders and prescription drug usage for a sleep disorder (e.g. Have you been diagnosed with a sleep disorder?; Have you ever taken a prescription medication for a sleep disorder?)

Results

Manipulation checks

An ANOVA test indicated that participants felt greater intensity of fear when fear appeal was used in the ad (M = 3.82, SD = 0.89). By contrast, they showed little emotional reaction when the no-fear appeal was used in the ad (M = 1.42, SD = 0.76), F(1, 261) = 7.31, p < 0.01. ANOVA tests and subsequent pairwise comparisons revealed that endorser types had significant effects on all endorser attributes except perceived trustworthiness (Table 1). The physician (M = 4.70, SD = 1.35) was rated significantly higher than the peer (M = 3.75, SD = 1.23) on perceived expertise, F(1, 177) = 24.44, p < 0.001. The physician (M = 4.92, SD = 1.38) was also perceived to be significantly more congruent with the product than the peer (M = 4.32, SD = 1.41), F(1, 177) = 8.47, p < 0.001. By contrast, the peer (M = 4.79, SD = 1.58) was perceived to be significantly more similar to participants themselves than the physician (M = 3.80, SD = 1.56), F(1, 177) = 7.00, p < 0.01. The two endorsers, however, had similar perceptions of trustworthiness to each other (M = 4.57, SD = 1.39 for physician; M = 4.52,, SD = 1.09 for peer), F(1, 177) = 0.07, p > 0.78).

Table 1.

Perceived endorser attributes

	Physician	Peer	F (df)
Perceived expertise	4.70 (1.35)	3.75 (1.23)	24.44*** (1, 177)
Perceived trustworthiness	4.57 (1.39)	4.52 (1.09)	0.07 (1, 177)
Perceived similarity	3.80 (1.56)	4.79 (1.58)	7.00** (1, 177)
Perceived congruence	4.92 (1.38)	4.32 (1.41)	8.47** (1, 177)

Means and standard deviations (in parentheses) are reported in the physician and peer columns.

p < 0.01, ***p < 0.001.

Hypothesis testing

Prior to hypothesis testing, significant covariates were identified and entered in the hypothesis testing models (Table 2). Hypotheses were tested through ANOVA or ANCOVA, where appropriate, with endorser types, fear appeal, and interaction term as independent variables. Post hoc pairwise comparisons were conducted for any significant main effects of endorser types and interaction effects found in the ANOVA/ANCOVA tests.

Table 2.

Results of hypotheses tests

	Endorser types (E)				Fear appeal (F)			E × F
	Physician	Peer	No endorser	F(df)	Fear	No-Fear	F(df)	F(df)
Message involvement¹	4.38_a (1.25)	3.99_a (1.42)	4.20_a (1.50)	2.07 (2, 255)	4.25_x (1.43)	4.10_x (1.36)	.64 (1, 255)	2.91* (2, 255)
Perceived message quality^1,2	4.51_a (1.27)	4.53_a (1.28)	4.47_a (1.33)	.91 (2, 254)	4.61_x (1.25)	4.39_x (1.32)	1.87 (1, 254)	3.31* (2, 254)
Attitudes toward the ad¹	4.48_a (1.19)	4.56_a (1.16)	4.23_b (1.30)	4.73** (2, 255)	4.41_x (1.16)	4.45_x (1.26)	1.39 (1, 255)	1.19 (2, 255)
Perceived drug efficacy	4.44_a (1.27)	4.32_b (1.15)	4.07_c (1.36)	2.97* (2, 256)	4.33_x (1.20)	4.23_x (1.31)	.07 (1, 256)	.70 (2, 256)
Drug inquiry intentions	3.98_a (1.59)	4.09_a (1.36)	3.62_a (1.58)	2.47 (2, 255)	4.03_x (1.53)	3.76_x (1.57)	1.52 (1, 255)	1.17 (2, 255)
Perceived severity of sleep disorders¹	4.26_a (1.32)	4.04_a (1.22)	3.87_a (1.47)	1.24 (2, 256)	4.26_x (1.25)	3.86_y (1.40)	5.91* (1, 256)	1.43 (2, 256)

Involvement with sleep disorders was a significant covariate.

Participants’ age was a significant covariate.

^{a–c, x–y}Means and standard deviations (in parentheses) are reported in the columns. Means sharing the same subscript are not statistically different from one another.

p < 0.05, **p < 0.01.

The results showed that endorser types had significant main effects on ad attitudes and perceived drug efficacy. For ad attitudes, the physician (M = 4.48, SD = 1.19) and peer (M = 4.56, SD = 1.16) had a similar effect to each other, while both had a significantly greater effect than the no-endorser condition (M = 4.23, SD = 1.30), F(2, 255) = 4.73, p < 0.01. For perceived drug efficacy, the physician produced the greatest effect (M = 4.44, SD = 1.27), followed by the peer (M = 4.32, SD = 1.15) and then by the no-endorser condition (M = 4.07, SD = 1.36), F(2, 256) = 2.97, p < 0.05. The main effects on message involvement, perceived message quality, drug inquiry intentions, and perceived severity of sleep disorders were not statistically significant, although the physician and peer generally produced similar means to each other, while both endorser types had greater means than the no endorser condition. H1c was supported for the physician versus no-endorser effect, but not for the physician versus peer effect. H1d, H2c, and H2d were supported. The rest of H1 and H2 were not supported.

As for fear appeal, a significant main effect was found for perceived severity of sleep disorders, F(1, 256) = 5.91, p < 0.02. The fear appeal (M = 4.26, SD = 1.25) induced greater perceptions of severity of sleep disorders than the no-fear appeal (M = 3.86, SD = 1.40). The main effects on the other dependent variables were not statistically significant, although the fear appeal yielded greater means than the no-fear appeal for the most part. H3e was supported, but the rest of H3 was not.

In addition, significant endorser types X fear appeal interaction effects were found for message involvement and perceived message quality (Figures 1 and 2). For message involvement, the physician (M = 4.49, SD = 1.21) produced a significantly greater effect than the peer (M = 4.01, SD = 1.29) and no-endorser condition (M = 3.88, SD = 1.54) when the no-fear appeal was used. When the fear appeal was used, the no-endorser condition (M = 4.53, SD = 1.42) produced a greater effect than the physician (M = 4.26, SD = 1.30) and peer (M = 3.97, SD = 1.55). For perceived message quality, the physician (M = 4.65, SD = 1.34) engendered a greater effect than the peer (M = 4.32, SD = 1.33) and no-endorser condition (M = 4.30, SD = 1.28) when the no-fear appeal was used. When the fear appeal was used, the pattern was reversed. The peer (M = 4.75, SD = 1.16) and no-endorser condition (M = 4.64, SD = 1.39) engendered a greater effect than the physician (M = 4.38, SD = 1.21). These patterns of interaction effects were generally consistent with the prediction of H4 rather than H5. Specifically, H4a was supported for the physician versus no-endorser effect, but not for the physician versus peer effect. H4b was supported. H5 and the rest of H4 were not supported.

Figure 1.

Message involvement.

Figure 2.

Perceived message quality.

Discussions

Pharmaceutical companies’ active marketing of prescription drugs directly to consumers has raised concerns about its impact on consumers’ healthcare decision-making. Particularly, the use of emotional appeals and influential endorsers in DTCA has been criticized for persuading rather than informing consumers about the drug’s effectiveness and impeding consumers’ careful processing of the drug’s facts. However, the current literature provides little empirical evidence on the actual effects these executional elements of DTCA have on consumers’ message processing and attitude change toward the advertised drug. To fill this void, we developed a set of hypotheses regarding independent and interactive effects of endorser types and fear appeal and tested them via an experiment.

As predicted, our results indicated that the physician endorser produced significantly greater perceptions of drug efficacy than the peer endorser. Attitudes toward the ad were also more favorable when an endorser (either physician or peer) was used than when no endorser was used. Significantly greater effects of the physician or endorsers in general were limited to these dependent variables, however. Although the physician was indeed perceived to be significantly more expert and better matched to the advertised product, the physician’s endorser effects on consumers’ involvement with the ad message, perceptions of message quality, intentions to inquire about the drug, and perceptions of severity of sleep disorders were not considerably greater than the peer’s endorser effects or the effects produced when no endorser was used.

A few inferences can be drawn from the above results. First, the comparable effects of the two endorsers may be explained by their similar perceptions of trustworthiness. Researchers have pointed out the important role of consumer trust in the effectiveness of DTCA.⁴⁷ Although the physician’s perceived trustworthiness was not manipulated in this study, it may have been enhanced as a result of high perceptions of expertise of the physician. Endorser expertise and trustworthiness are highly correlated with each other and identified together as critical dimensions of endorser credibility in the literature.⁵² The peer’s increased perceptions of trustworthiness were not expected. However, some researchers contend that peers can be viewed as trustworthy if they are believed to be unbiased and share a common interest with consumers.⁹ Although the stimulus ad did not state that the peer endorser had a sleep disorder, participants of our study may have assumed it due to common DTCA practices. As they viewed the peer endorser to share similar demographic and illness-related traits as themselves, they evaluated the endorser’s trustworthiness and ad messages favorably. Second, the physician and peer were rated positively for their attributes. The physician was perceived to be expert and trustworthy. The peer was viewed as similar to the participants themselves and also trustworthy. However, these favorable perceptions of the endorsers did not necessarily transfer to participants’ evaluation of the ad message or have overpowering influence on attitudes toward the drug and health issue. In fact, their overall persuasive effects were not substantially greater than when the information about the drug and health condition was provided simply without an endorser. This finding tends to corroborate previous research that showed limited persuasive influence of endorsers and other peripheral cues of the ads when the ad contains a personally relevant topic.¹⁶ Considering that sleep disorders are a personally relevant issue for many college students, our undergraduate participants possibly found the informational content in the ad to be more valuable and were less affected by endorsers. Third, the lack of statistically significant effects of endorser types on some dependent variables may be due to the small sample size of the study. The mean differences among the three endorser conditions could have reached statistical significant with a larger sample.

Regarding fear appeal, the significant main effect was found for perceived severity of sleep disorders but not for the other dependent variables, although the mean differences between fear and no-fear appeals were in the expected direction. These results may be due to a weak manipulation of fear appeal. Although the manipulation check proved to be successful, the manipulation of fear appeal may not have been strong enough to elicit distinct reactions to the ad. In line with our discussion above, it is also possible that the personal relevance of the health condition in the ad led to greater attention to the message, while reducing the influence of an executional style of the ad such as fear appeal. A follow-up study should investigate this issue by improving the fear appeal manipulation and ruling out methodological problems.

Despite the limited main effects of fear appeal, we found noteworthy interaction effects between fear appeal and endorser types on message involvement and perceived message quality. Particularly, the interaction effect found for perceived message quality supported the view that a non-threatening message communicated by an expert endorser is more effective than a threatening message by an expert endorser. As reviewed earlier, researchers maintain that this pattern of interaction effects is due to psychological reactance to fear appeal, which results in endorser derogation. A thorough theoretical framework needs to be developed to account for the potential mediation role of psychological reactance in the interactive effects of fear appeal and endorser types. Meanwhile, it may be inferred that psychological reactance was induced in our study because the assurance message that instills efficacy beliefs was not strong enough after fear arousal. Cope and Richardson⁵³ reports that a fear-arousing message leads to positive speaker evaluation when it incorporates a reassuring recommendation, compared to when it does not. Another possible explanation is the congruence of message appeal and endorser type. Expert endorsers such as physicians may be viewed more appropriate for the ad focusing on the facts than the ad arousing an emotion. The reduced interest and engagement of the message and lower perceptions of message quality found in our study may have resulted from the perceived mismatch of the physician endorser and fear appeal. This reasoning is consistent with source-message (in)congruity effects shown in Artz and Tybout’s study.⁵⁴ The expert source exerted greater persuasive influence when the message claim contained quantitative information. When the message claim contained non-quantitative information, the non-expert source had greater effects. It suggests that consumers have certain expectations about the type of message delivered by expert versus non-expert sources (source bias). The violation of these expectations can lead to less favorable evaluation of the source and message.

As with most research, our study had a few limitations. Perhaps the most notable is the use of a convenient student sample. Undergraduate students are a unique, relatively homogeneous group. Their responses may not be representative of those of general US consumers. The use of a female-dominant sample and female endorsers further reduces the generalizability of the study findings. As discussed earlier, the small sample size also could have affected the results of the study. A future study should verify our findings by employing a larger sample from a more diverse background. Another issue to be explored in future research concerns fear appeal. Some researchers suggest that a single emotional appeal may evoke multiple discrete emotions and produce unintended effects.⁵⁵ Although we measured fear-related emotions as a manipulation check of fear appeal, other discrete emotions (e.g. anger and disgust) could have been evoked inadvertently and affected consumers’ psychological reactance and other responses to fear appeal. A future study that delineates the dynamic relationships of multiple discrete emotions aroused by an ad will help extend the understanding of fear appeal’s complex effects.

Despite these limitations, our study contributes to the current literature by addressing the ongoing debates about the use of fear appeal and endorsers in DTCA. Our study suggests that these executional elements of the ad can have some influence on consumers’ cognitive and attitudinal responses to DTCA and the drug, but their influence is not necessarily as substantial as to overpower consumers’ attitudes and judgments about the drug. Especially when the ad discusses a personally relevant health condition, consumers are likely to evaluate the ad and drug based on the value of information rather than peripheral elements of the ad. Our study also provides important implications for DTCA practitioners that emotional appeals and endorser types should be selected based on a careful consideration because they may produce unintended, boomerang effects if they are mismatched or consumers’ psychological reactance is not effectively alleviated.

Footnotes

Hyojin Kim is an assistant professor in the Department of Advertising of College of Journalism and Communication at University of Florida. She received her PhD in Advertising at the University of Texas at Austin. Her research interests include social marketing, pharmaceutical advertising, and message source effects.

Chunsik Lee is an assistant professor at University of North Florida and teaches advertising and public relations. He earned his PhD from University of Florida in Gainesville and MA from University of Minnesota in Minneapolis. His research focuses on the effectiveness of different formats of online advertising.

References

Kessler

Pines

. The federal regulation of prescription drug advertising and promotion. J Am Med Assoc 1990; 264(18): 2409–2415.

Media Intelligence

TNS

. Ad$pender [Database], New York: TNS Media Intelligence, 2012.

Frosch

Krueger

Hornick

. Creating demand for prescription drugs: A content analysis of television direct-to-consumer advertising. Ann Fam Med 2007; 5(January/February): 6–13.

Kaiser Family Foundation. Views on prescription drugs and the pharmaceutical industryhttp://www.kff.org/spotlight/rxdrugs/upload/Rx_Drugs.pdf (accessed 20 March 2008).

Wolfe

. Direct-to-consumer advertising: Education or emotional promotion? N Engl J Med 2002; 346(July): 524–526.

Sokol

Wackowski

Lewis

. Marketing pharmaceutical drugs to women in magazines: A content analysis. Am J Health Behav 2010; 34(4): 402–411.

Federal Trade Commission. In the matter of request for comments on consumer-directed promotionhttp://www.ftc.gov/be/v040002 text.pdf (accessed 20 February 2012).

Kelman

. Processes of opinion change. Publ Opin Quart 1961; 25(1): 57–78.

Simons

Berkowitz

Moyer

. Similarity, credibility, and attitude change: A review and a theory. Psychol Bull 1970; 73(1): 1–16.

10.

Briñol

Petty

. Source factors in persuasion: A self-validation approach. Eur J Soc Psychol 2009; 20: 49–96.

11.

Friedman

. Endorser effectiveness by product type. J Advert Res 1979; 19(September/October): 63–71.

12.

Feick

Higie

. The effects of preference heterogeneity and source characteristics on ad processing and judgments about endorsers. J Advert 1992; 21(Summer): 9–24.

13.

Stafford

Day

. A contingency approach: The effects of spokesperson type and service type on service advertising perceptions. J Advert 2002; 31(Summer): 17–35.

14.

Lord

Putrevu

. Informational and transformational responses to celebrity endorsements. J Curr Issue Res Advert 2009; 31(Spring): 1–13.

15.

Kahle

Homer

. Physical attractiveness of the celebrity endorser: A social adaptation perspective. J Consum Res 1985; 11(March): 954–961.

16.

Petty

Cacioppo

. Communication and persuasion: Central and peripheral routes to attitude change, New York: Springer-Verlag, 1986.

17.

Tomkins

Affect theory. In: Scherer

Ekman

(eds). Approaches to emotion, NJ: Erlbaum, 1984; 163–195.

18.

Flora

Maibach

. Cognitive responses to AIDS information: The effects of issue involvement and message appeal. Commun Res 1990; 17(6): 759–774.

19.

Lee

Davie

. Audience recall of AIDS PSAs among U.S. and international college students. Journ Mass Commun Quart 1997; 74(1): 7–22.

20.

Santa

Cochran

. Does the impact of anti-drinking and driving public service announcements differ based on message type and viewer characteristics? J Drug Educ 2008; 38(2): 109–129.

21.

Ray

Wilkie

. Fear: The potential of an appeal neglected by marketing. J Market 1970; 34(January): 54–62.

22.

Henthorne

LaTour

Nataraajan

. Fear appeals in print advertising: An analysis of arousal and ad response. J Advert 1993; 22(2): 59–69.

23.

Lang

. Using the limited capacity model of motivated mediated message processing to design effective cancer communication messages. J Commun 2006; 56: s57–s80.

24.

Norris

Bailey

Bolls

. Effects of emotional tone and visual complexity on processing health information in prescription drug advertising. Health Commun 2012; 27: 42–48.

25.

Tiedens

Linton

. Judgment under emotional certainty and uncertainty: The effects of specific emotions on information processing. J Person Soc Psychol 2001; 81(6): 973–988.

26.

Griskevicius

Goldstein

Mortensen

. Fear and loving in Las Vegas: Evolution, emotion, and persuasion. J Market Res 2009; 46(June): 384–395.

27.

Schwarz

Clore

. Mood, misattribution, and judgments of well-being: Informative and directive functions of affective states. J Person Soc Psychol 1983; 45: 512–523.

28.

Lazarus

. How emotions influence performance in competitive sports. Sport Psychol 2000; 14(3): 229–252.

29.

Roseman

Wiest

Swartz

. Phenomenology, behaviors, and goals differentiate discrete emotions. J Person Soc Psychol 1994; 67(2): 206–221.

30.

Janis

Feshbach

. Effects of fear-arousing communications. J Abnormal Soc Psychol 1953; 48(1): 78–92.

31.

Rogers

. A protection motivation theory of fear appeals and attitude change. J Psychol 1975; 91: 93–114.

32.

Witte

. Putting the fear back into fear appeals: The extended parallel process model. Commun Monogr 1992; 59(4): 329–349.

33.

Tanner

. Read this or die: A cognitive approach to an appeal to emotions. Int J Advert 2006; 25(3): 414–416.

34.

Kavadas

Katsanis

LeBel

. The effects of risk disclosure and ad involvement on consumers in DTC advertising. J Consum Market 2007; 24(3): 171–179.

35.

Pharmaceutical

Executive

. The secret of my success. 2007; 27(7): 82–83.

36.

Jones

Sinclair

Courneya

. The effects of source credibility and message framing on exercise intentions, behaviors, and attitudes: An integration of the elaboration likelihood model and prospect theory. J Appl Soc Psychol 2003; 33(1): 179–196.

37.

Miller

Lane

Deatrick

. Psychological reactance and promotional health messages: The effects of controlling language, lexical concreteness, and the restoration of freedom. Human Commun Res 2007; 33(2): 219–240.

38.

Quick

Bates

. The use of gain- or loss-frame messages and efficacy appeals to dissuade excessive alcohol consumption among college students: A test of psychological reactance theory. J Health Commun 2010; 15: 603–628.

39.

Hewgill

Miller

. Source credibility and response to fear-arousing communications. Speech Monogr 1965; 32(June): 95–101.

40.

Greenwald

Cognitive learning, cognitive response to persuasion, and attitude change. In: Greenwald

Brock

Ostrom

(eds). Psychological foundations of attitudes, New York: Academic Press, 1968; 147–170.

41.

Center for Disease Control and Prevention. National Health and Nutrition Examination Survey: 2007-2008. http://www.cdc.gov/nchs/nhanes/nhanes2007-2008/SLQ_E.htm (2009. accessed 14 March 2010).

42.

Vargas

. Under pressure. Pharmaceut Execut 2008; 28(May): 87–94.

43.

de Hoog

Stroebe

de Wit

JBF

. The impact of vulnerability to and severity of a health risk on processing and acceptance of fear-arousing communications: A meta-analysis. Rev Gen Psychol 2007; 11(Autumn): 258–285.

44.

Laczniak

Muehling

. The relationship between experimental manipulations and tests of theory in an advertising message involvement context. J Advert 1993; 22(Autumn): 59–74.

45.

Zhang

. Responses to humorous advertising: The moderating effect of need for cognition. J Advert 1996; 25(Spring): 15–32.

46.

MacKenzie

Lutz

. An empirical examination of the structural antecedents of attitude toward the ad in an advertising pretesting context. J Market 1989; 53(April): 48–65.

47.

Huh

DeLorme

Reid

. Factors affecting trust in on-line prescription drug information and impact of trust on behavior following exposure to DTC advertising. J Health Commun 2005; 10(December): 711–731.

48.

Lerner

Keltner

. Fear, anger, and risk. J Person Soc Psychol 2001; 81(January): 146–159.

49.

Ohanian

. Construction and validation of a scale to measure celebrity endorsers' perceived expertise, trustworthiness, and attractiveness. J Advert 1990; 19(Autumn): 39–52.

50.

Rifon

Choi

Trimble

. Congruence effects in sponsorship: The mediating role of sponsor credibility and consumer attributions of sponsor motive. J Advert 2004; 33(Spring): 29–42.

51.

Mittal

. A comparative analysis of four scales of consumer involvement. Psychol Market 1995; 12(July): 663–682.

52.

Hovland

Janis

Kelly

. Communication and persuasion: Psychological studies of opinion change, New Haven, CT: Yale University Press, 1953.

53.

Cope

Richardson

. The effects of reassuring recommendations in a fear-arousing speech. Speech Monogr 1972; 39(2): 148–150.

54.

Artz

Tybout

. The moderating impact of quantitative information on the relationship between source credibility and persuasion: A persuasion knowledge model interpretation. Market Lett 1999; 10(January): 51–62.

55.

Leshner

Bolls

Thomas

. Scare ’em or disgust ’em: The effects of graphic health promotion messages. Health Commun 2009; 24: 447–458.