Threats and Promises in Advertising Appeals

Michael Menasco, University of California, Los Angeles
Penny Baron, University of Iowa
ABSTRACT - The research reported here attempted to determine the persuasive impact of desirable and undesirable consequences of physical and social advertising appeals. Research on these kinds of appeals is distinct from that on strictly physical fear arousal. Results of the present research appear to conform to predictions from studies in protection motivation theory (Rogers and Mewborn, 1976; Mewborn and Rogers, 1979), particularly in terms of mediational responses and persuasive impact. The influence of measured personality variables (self-esteem, fear of physical and social consequences) on persuasion was also assessed. Results generally did not follow predictions.
[ to cite ]:
Michael Menasco and Penny Baron (1982) ,"Threats and Promises in Advertising Appeals", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 221-227.

Advances in Consumer Research Volume 9, 1982      Pages 221-227

THREATS AND PROMISES IN ADVERTISING APPEALS

Michael Menasco, University of California, Los Angeles

Penny Baron, University of Iowa

ABSTRACT -

The research reported here attempted to determine the persuasive impact of desirable and undesirable consequences of physical and social advertising appeals. Research on these kinds of appeals is distinct from that on strictly physical fear arousal. Results of the present research appear to conform to predictions from studies in protection motivation theory (Rogers and Mewborn, 1976; Mewborn and Rogers, 1979), particularly in terms of mediational responses and persuasive impact. The influence of measured personality variables (self-esteem, fear of physical and social consequences) on persuasion was also assessed. Results generally did not follow predictions.

INTRODUCTION

Most research on the impact of threats in persuasive communication has focused on the relationship between the strength of a fear appeal and the degree of persuasion. Due to difficulties in calibrating the amount of fear induced by different threat manipulations this work has yielded apparently inconsistent results across studies (See Sternthal and Craig, 1974, for a review).

The more interesting questions from an applied perspective concern which kinds of fear appeals are more effective, what groups of people are more persuaded by fear appeals and how fear appeals stack up against appeals promising benefits. Current practice and much research on the general effects of rewards vs. punishments suggests that promises usually work better than threats but controlled experimental comparisons in commercial mass persuasion contexts are rare. The experiment reported here investigates how the persuasive impact of advertising appeals is moderated when the claimed consequences are promises or threats, when the consequences are physical or social in character and when the audience contains people with different levels of self-esteem and choices of fear.

PHYSICAL FEAR

With respect to kinds of fear appeals, most work has focused on the persuasive impact of physical threats. Early studies (Janis and Feshbach, 1953; Haefner 1956; Leventhal, 1966; Leventhal et al., 1965; Evans et al., 1970), while yielding inconclusive results on amount of fear, did consistently produce some persuasive impact. Applied work investigating the potential of using fear appeals in advertising (Ray and Wilkie, 1970; Stuteville, 1970; Wheatley and Oshikawa, 1970; Wheatley, 1971), while less encouraging, suggests a similar conclusion. There is also some evidence that fear appeals involving threats to physical well-being are more effective when they advocate a particular coping response designed to avoid the bad physical consequence than when they simply seek to induce fear of the bat consequence (Rogers, 1975; Rogers and Mewborn, 1976). Results from Rogers and Mewborn (1976) also indicate that fear operates on mediational responses rather than upon intentions. directly.

SOCIAL APPROVAL AND DISAPPROVAL

Other work suggests the kinds of fear appeal that may interact with personality variables in determining degree of persuasion. Results from Spielberger (1972), Denny (1966), and Spielberger and Smith (1966) suggest that people high in chronic anxiety react more extremely to social fear appeals but only when their self-esteem is threatened. Fear appeals involving strictly bad physical outcomes do not produce this differential response. Persuasability of a fear appeal may also be modified by self-esteem (Dabbs and Leventhal, 1960; Kormzweig, 1967; Leventhal and Trembly, 1968) and by chronic fear of social and physical consequences (Spielberger, 1972). Thus, it would appear that social disapproval may operate differently than physical threat, particularly when certain personality variables are included. There is little experimental evidence in support of this contention, although one study did compare the persuasibility of social approval and disapproval (Powell and Miller, 1967). Their experimental manipulations, however, appear to confound social and physical threats (donating blood vs. giving blood for payment).

Taken together this past work suggests that the kind of consequences, physical or social; its valence, positive or negative; and personality characteristics of the audience, level of self-esteem and chronic level of social and physical fear, may interact in determining the persuasive impact of a commercial message. The purpose of this research is to test this proposed interaction between main effects (social and physical consequences) and measured personality variables.

SELF ESTEEM, SOCIAL and PHYSICAL FEAR

Several theoretical bases for specific expectations about the form of this interaction exist. For example, according to Aronson and Linder's gain-loss theory (1965), gains and losses of positive outcomes have reward value in themselves, quite apart from the absolute level of reward represented by the outcomes. This suggests that high self-esteem people may be more persuaded by threats of negative social consequences, because such consequences are discrepant with own self-view and represent a potential loss to them. Low self-esteem persons may be more persuaded by promises of positive social consequences, since those outcomes are not consistent with their sense of self-worth and thus may represent potential reward.

Research by Rosenberg (1965) suggests an alternative set of predictions for the effects of self-esteem. Rosenberg found that low self-esteem people both thought poorly of themselves and expected others to think poorly of them as well. High self-esteem people thought well of themselves and expected others to hold similarly positive opinions of them. If so, low self-esteem people, because they expect others are likely to hold negative opinions of them, might be more persuaded by an appeal which threatened disapproval if an advocated action is not taken, and less persuaded by an appeal which promises approval, because the prospect of disapproval is more consistent with their expectations. Applying the argument to high esteem people suggests more persuasion when approval is promised for taking the advocated action than when disapproval is threatened for not taking it.

Differential effects of threats and promises of social consequences are likely to be stronger for people with high measured levels of social fear. For those with high physical fears, a stronger effect is expected for appeals threatening bad physical consequences for non-compliance. Appeals promising good physical outcomes for compliance are expected to be relatively more effective among those with low physical fears. Again, the rationale here is that those with high fear find threats more realistic while those with low fear find promises more believable and are thus likely to be more persuaded by them.

METHOD

Hypotheses

Given the discussion above, the following hypotheses are presented. Hypothesis two is based on expectations of behavior due to gains and losses (Aronson and Linder, 1965). Hypotheses two, three and four imply a three way interaction effect with measured personality variables. Thus, a main effects ANOVA, with the personality variables as covariates, is conducted first in order to assess any main effects due to covariates. Second, separate post hoc ANOVAs are conducted for each of the covariates employed as a between subjects factor

Note that hypotheses indicate impact should be on variables measuring persuasion. Work by Rogers and Mewborn (1976), however, would predict greater impact upon mediational responses. Additional dependent variables therefore are included as a partial test for possible mediational effects.

H1: Positive outcomes of proposed product use should produce greater persuasive impact than negative outcomes (This is represented by a main effect in the experimental design).

H2: Persons high in self-esteem should be more persuaded by threats of social disapproval, while persons low in self-esteem should be more persuaded by promises of social approval.

H3: Persons high in social fear should be more persuaded by threats of social outcomes, while those with low levels of social fear should be more persuaded by positive social outcomes.

H4: Persons high in physical fear should be more persuaded by threats of bad physical consequences, while persons low in physical fear should be more persuaded by positive consequences

A 2 x 2 x 2 factorial experiment involving 160 college students varied two products, deodorant vs. shampoo; desirability of the consequences of their use, promise of good outcomes for using the product vs. threat of bad outcomes for not using the product; and type of consequence, physical outcomes vs. social outcomes. Desirability of the consequences and type of consequence were between subject variables. Product was a within subject variable.

Recruitment of Research Participants

Students who had filled out a battery of psychological tests in connection with filling requirements for Introductory Psychology were recruited by phone. Students participated for a $1 participation payment and the chance to see and evaluate the experimental ads, which turned out to be an enjoyable task. Participants also received a free 1 sample of one of the advertised products of their choice, [Products were donated by the manufacturer but all advertising claims, while consistent with the benefits offered by the product advertised, were executed by the experimenters to meet the requirements of the experimental design. first. To cut final production costs, four final tapes were produced, each containing 4 ads appropriate to each condition in fixed order, ad situations 1-4 for the physical consequences conditions and ad situations 5-8 for the social consequence conditions.] although they were not told about this prospect until they finished the experimental task.

Independent Variables

Each participant was randomly assigned to see four 40 second commercials, two deodorant ads and two shampoo ads. The four ads seen by a participant represented one of the four treatment combinations, yielding a total of 16 different advertisements. The two ads seen by a participant for a given product each depicted a different usage situation, one experienced by a male and the other by a female. Half the participants saw usage situations involving physical consequences. Each usage situation was taped twice, once depicting positive consequences and once depicting negative consequences. Participants saw either all positive or all negative versions.

Below is a schematic of these possibilities:

TABLE

Because of the expense and time required to produce the ads, elaborate counterbalancing typical of treatments involving verbal or paper and pencil manipulations was prohibited. Instead, actors are confounded with situations x products and situations are confounded with products. Some informal protesting indicated that order effects were likely to be trivial, e.g., usage situations they saw

Advertisement Composition

As mentioned above, desirability of consequences varied whether the protagonist in the ad experienced a negative consequence for not using the advertised product (a threat) or a positive consequence for using it (a promise). Type of consequence varied whether the protagonist experienced a physical consequence (smooth, shiny, easy to brush hair vs. tangled, dull, hard to brush hair) or a social consequence (people approaching and hugging in new arrival at a party vs. ignoring and drawing away from him). Virtually the same verbal copy was used in both the negative and positive version of a given commercial. In the social consequence conditions, positive social responses were visually portrayed by having other people in the situation approach the protagonist, pat him on the back, hug him, smile at him and otherwise make accepting responses appropriate in the situation depicted. Negative social consequences were visually portrayed by having other people in the situation ignore or withdraw from the protagonist. Physical consequences were varied by portraying the protagonist's personal response to using the product as pleasurable or painful, e.g., smiling and nodding while stroking and combing smooth, shiny hair vs. frowning and whining while trying to comb tangled, dull hair. Verbal messages were designed to be appropriate across both positive and negative versions of an ad and across type of consequences from using a given product. The meanings which varied across conditions were conveyed by the visual portion of the ads. Illustrative excerpts from an ad for each product indicates the type of verbal copy used:

Deodorant--"Sometimes a deodorant lets you know it's on your side and sometimes it lets you know it isn't...with _____ deodorant you don't have to worry. It's always on your side..."

Shampoo--"Shampooing your hair can improve your life a little or make it a little worse. If you want a definite improvement, shampoo with _____ ..."

Dependent Variables

The principle dependent variables were measured with 10 point semantic differential type scales. Responses were coded from 0 to 9. The principal dependent variables measures for each product were: (1) preference for the advertised brand in comparison to other brands; (2) intentions to buy the brand; (3) how well the brand compares to other brands; (4) how believable the ads were; (5) how interested subjects were in more-information about the brand; (6) importance of product use; (7) likelihood of experiencing usage situations seen in the ads; (8) how interesting the ads were; and (9) how comfortable subjects felt while viewing the ads.

Several manipulation checks were obtained: (a) ratings of how the people in the commercials felt on a 10 point scale; (b) four recall measures of message content; (c) the subject's rating of the quality of results experienced by the people in the ads on a 5 point scale, from 1 equalling terrible to 5 equalling excellent and a similar rating of how the sponsor of the ad would evaluate the results experienced. These measures were obtained separately for the shampoo ads and the deodorant ads.

Subject Variables

Participants' chronic level of self-esteem was measured by a series of 17 semantic differential type scales (e.g., interesting-dull, likeable-unlikable, competent-incompetent). Levels of social and physical fear were measured by items taken from the Fear Survey Schedule (Rubin, et al., 1969) and three additional items, "being disliked" on the social fear component of the measure and "being hurt or injured," and "taking medicine" on the physical fear component. Responses to the scales were obtained six months prior to the experiment in an otherwise unrelated project.

Procedure

Participants, recruited by phone, were asked to take part in a research project designed to test the relative effectiveness of different kinds of appeals used in television advertising When they arrived at the testing room, they were informed of the procedures involved and signed standard informed consent agreements. Each participant was then randomly assigned to watch one of the 4 video tapes on a monitor in a viewing booth. Immediately after viewing the tape, the participants fillet out the questionnaire, were debriefed, paid their $1 and given a card entitling them to a free sample of either of the advertised products. There were 4 monitors in 4 separate booths allowing all four conditions to be run at each experimental session. Approximately equal numbers of males and females were run in each condition (the products used in the experiment were targeted to sell to both males and females in college age populations). Similar efforts were made to counter counterbalance for measured levels of self-esteem and social and physical fear. The experiment was run during a single week prior to the products advertised being widely available in the local market (Both were new products, making it unlikely that many subjects had prior experience with them).

RESULTS

A 2 x 2 x 2 ANOVA with repeated measures and three covariates was used by employing the BMD P2V analysis of variance program of the UCLA biomedical software package. This program follows a mixed effect design given by Myers (1972) in which the between factors are considered fixed and the within factor is random. All analyses of variance reported below included, in addition to the subject variables, the two between experimental terms (desirability and type of consequences) and the within (product) term. There are 160 subjects x 2 products = 320 observations in all repeated measures analyzed.

Products Effects

Products were found to be significantly different (p <.05) on seven of the nine dependent variables. In all cases but one, the largest mean values were associated with deodorant. These results may be due to deodorant being inherently more visible as a product class. Further, the deodorant ads may simply have been viewed as more interesting by subjects. Significant mean differences in variable 5 measuring interest in the ads show this to be the case (shampoo = 3.53, deodorant = 3.79, p <.02). Subsequent analyses from the BMD P2V program treats the between subject effects as averages over the product factors. Separate ANOVAs on each of the products did not produce results generally different from the repeated measures design.

Manipulation Checks

Of the 160 subjects, all correctly identified the brand names of the two test products. All correctly identified the manufacturer of the shampoo product and all but one correctly identified the deodorant manufacturer. Similarly, only four people incorrectly identified the product the protagonist in the ad was using (The brand advertised was the correct choice in the positive consequence conditions and "unidentified competitor's brand" was the correct choice in the negative consequences conditions). The results for "how pleased the people shown in the ads appeared to feel," also suggest that the manipulations were successful (See Table 1). Similarly, the mean ratings for quality of the results experienced by the protagonist and the sponsor's opinion of these results also indicate successful manipulation of consequence desirability (See Table 1).

TABLE 1

Between Subjects Results

The between subjects ANOVA revealed no significant main effects for any of the three covariates. The two primary dependent measures (intention and preference) also revealed no significant effects. Hypothesis 1 is not supported. Other dependent measures, however, did exhibit significance. These results are presented in Tables 2a through 2d. Each analysis is based upon one degree of freedom per treatment effect (ant interactions) and 153 degrees of freedom for error. For the between subjects analyses, the error term is defined as a nested term. Mean square error is estimated by measuring the variability among subjects given a level of a between factor interaction. Mean scores reported in Table 2 are the adjusted means due to the effects of covariates. The adjustments are small since the covariates exhibited no significant effects.

Although no significant results were found for the main dependent variables (preferences and intentions), the effects on the other dependent measures are consistent with results reported by Rogers and Mewborn (1976). That is, an appeal influences intentions via mediating variables. Measures on these variables represent cognitive (coping) responses that mediate the fear appeal. This schematic is represented in Figure 1. Using the variables comfort (9), believability (4), likelihood (7) 9 comparison (3) as surrogates, similar path analyses were conducted as those reported by Rogers and Mewborn (1976). The variable, believability, is a close surrogate for appraised severity in the sense that subjects were asked to respond to the believability of consequences depicted in the ads. The other variables (comfort, likelihood of experiencing consequences, and brand comparisons) are close replicates of those used by Rogers and Mewborn to measure fear arousal, expectancy, and efficacy, respectively.

TABLE 2

SIGNIFICANT RESULTS FOR BETWEEN SUBJECT EFFECTS

TABLE 3

SIGNIFICANT EFFECTS FOR SOCIAL FEAR (SF) AND PHYSICAL FEAR (PF)

FIGURE 1

PATH ANALYSES REPORTED BY ROGERS AND MEWBORN (1976)

Analyses were conducted by experimental cell and by product. Pearson correlations between variables (paths) shown in Figure 1 conformed closely to results given Rogers and Mewborn (1976), [Results reported here are purely exploratory in the sense that only simple path coefficients are reported. The authors have completed analyses using compound paths (Asher, 1979) on an additional set of data and have found a cognitive model for deodorant that differs from the mediation model reported by Rogers and Mewborn. Also, no recursive paths occur in the authors' model while one such path does exist in Figure 1.] with exception of reported comfort and believability (range -.12 to .35).

The correlations between believability and intentions and preferences were all significant (p < .10) in three cells (range .27 to .57) except for shampoo under the social-negative condition (r = .24 and .26 between believability and intentions and preference). Only one product correlation was significant under the remaining physical-positive condition, and that was for shampoo (believability and preference, r = .38).

Path correlations for believability and likelihood and for brand comparison (efficacy) and preference and intentions were significant over the two products (p < .07, range .29 to .73; and p < .03, range .34 to .75, respectively).

Other path coefficients were generally not significant between likelihood (expectancy) and brand comparison (efficacy) and between comfort and brand comparison with the exception of the social positive condition. Here, most inter-correlations between variables tended to be higher than in the other three conditions. These results are probably due to the generally positive nature of the messages presented in this condition (positive results of using the product within a group setting). Two of the three other conditions involve negative social and physical outcomes and conform closely to results on physical threat. Rogers and Mewborn studied the mediating effects due to cigarette smoking, driving safety and venereal disease.

Thus, it would appear that the positive or negative outcomes of social or individual (physical) appeals operate primarily on mediational responses and not directly upon preference or intention. Note that significant main effects occurred for believability, likelihood and, comfort; all significant variables included in path analyses.

Measured variables (self-esteem, social fear, and physical fear)

Since each of the measured variables showed no significance as covariates, a separate post hoc ANOVA was conducted for each variable. Variables were split at the median and subjects were assigned to high and low groups. These median splits resulted in another factor added to the design and with unequal cell sizes. The ANOVAs were again conducted on repeated measures (over the two products) in the event measured variables interacted with the product variable (repeated measure).

Self Esteem. Effects due to self-esteem were nil. Results do not support hypothesis 2. Products interacted with self-esteem on message interest (variable 5; F = 5. 6, p = .022), and self-esteem interacted with positive-negative consequences on importance of product use (variable 6; F = 4.05, p = .046). High self-esteem subjects rated deodorant higher on message interest than low self-esteem subjects. The opposite was true for shampoo. For product importance, high self-esteem subjects reacted more favorably to positive consequences, while low self-esteem subjects reacted more favorably to negative consequences. This result conforms to the expectations concerning self-esteem and valence of consequences, however, for only one variable (importance of product use).

Physical and Social Fear. Results on physical and social fear are a little more encouraging. These results are presented in Tables 3 and 4. Nearly the same dependent variables show significance for both PF and SF (Table 4).

Further, significance is found for interactions on the primary dependent measures of preference and intentions; thus, it would appear that the appeals did operate directly on intentions and preferences once these measured variables were considered. Results reported in Tables 3 and 4 indicate that physical fear interacts with physical-social consequences, while social fear is more likely to interact with positive-negative consequences. These results are a partial confirmation of hypothesis 4, but not hypothesis 3.

TABLE 4

MEANS FOR SIGNIFICANT INTERACTIONS INVOLVING MEASURED VARIABLES (PF AND SF)

Measured social fear interacted significantly with the desirability of the consequences portrayed on the one basic measure of persuasion (buying intention). A similar significant interaction also occurred for how well the advertised brand compared with other brands. Those with high measured social fear levels were more persuaded by the commercials which portrayed positive rather than negative consequences of using the advertised brand. Those whose measured level of social fear was low displayed opposite responses. They were more influenced by negative consequences. On two measures, product interest and product importance, there is a significant 3-way interaction (Tables 3 and 4),

Measured physical fear interacted similarly as SF but with physical-social consequences instead. Those with high measured levels of PF were persuaded by physical consequences rather than social consequences. The tendency was opposite for those with low levels of measured PF. They were more persuaded by social consequences. The direction of interactions was the same for all significant variables reported in Tables 3 and 4. A product and PF interaction also occurred for product importance, with high PF types rating shampoo much lower on importance of product use (Table 4). No significant three-way interactions involving PF were obtained on any measure.

DISCUSSION

Mediational Responses

On the principal measures, preference for the advertised brand and buying intention, no significant main effects were observed for either type or desirability of the consequences portrayed, nor was their interaction significant. Additional analyses on path correlations, however, confirmed the possibility that desirability of physical and social consequences may operate only indirectly on measures of persuasion through mediational responses. This is consistent with work in protection motivation theory (Rogers, 1975; Rogers and Mewborn, 1976; Mewborn and Rogers, 1979). Further research might address this question directly for appeals of this kind. That is, specific dependent measures should be constructed to test the hypothesis that certain mediational responses will influence preference and/or intentions for social and physical appeals. Causal path analysis can be utilized to determine the mediational effects upon persuasion. Traditional wisdom might predict a direct effect upon persuasion (intentions and preference). Evidence from this study, however, does not lead to that inference.

Subject Variables

For the specific personality characteristics measured in this study, the results are not entirely consistent with expectations. No three-way interactions on the principal dependent measures were obtained. With respect to self-esteem, it appears expectations of results were not confirmed. Threats and promises appear equally effective across the whole audience for both social and physical types of consequences.

The results for the two types of chronic fear were also unexpected. Differences in social fear interacted with desirability of consequences, positive or negative, but not with type of consequence. This result contradicts hypothesis 3. The form of the interaction, however, was consistent across dependent measures. Those high in social fear were more persuaded by positive consequences than by negative ones, i.e., promises worked better than threats, while people low in social fear seemed to be more persuaded by negative consequences. For physical fear only type of consequences, physical or social produced differential responses. In the physical consequences condition, both physical fear groups responded somewhat similar while in the social consequences condition, preference and buying intention ratings were definitely lower among the high physical fear people.

The findings, overall, suggest that segmentation of the potential audience by such personality characteristics may be worth considering when trying to achieve optimal persuasive impact across the total audience. It also suggests that, within the range of the independent variables included in the experiment, the best appeal may depend on these audience characteristics. Since this range is not likely to be greatly exceeded in actual commercial advertising situations, these results have applied implications. Choosing the best appeal may require assessment of the predominant personality characteristics of the intended audience. From an applied perspective, the appeal which produces the best response is not always intuitively obvious. For instance, greater differences in response among those measured high and low in physical fear occurred in the social consequences conditions in comparison to the physical consequence conditions.

Additional research should clarify the role such personality variables play in ads promoting promises or threats.

Evidence from this experiment does not conform to expectations. It does appear, however, that people may self-select themselves to certain persuasive conditions, i.e., high PF persons were more persuaded by physical outcomes, while high SF persons were more persuaded by promises (probably in a supportive context). Given that results from this study are contradictory to previous research (Spielberger, 1972; Powell and Miller, 1967; Leventhal and Trembly, 1968), perhaps future research should concentrate on the mediational effects upon persuasion as in the studies by Rogers and Mewborn (1976) and Mewborn and Rogers (1979). Their work is much more appealing in comparison to the traditional approach of directly measuring attitudinal and behavioral responses to fear appeals.

REFERENCES

Aronson, E., and Linder, D. (1965), "Gain and Loss of Esteem as Determinants of Interpersonal Attractiveness," Journal of Experimental Social Psychology, 1, pp. 156-172.

Asher, R. B. (1979), Causal Modeling, Beverly Hills, CA: Sage Publications.

Dabbs, J., and Leventhal, El. (1966), "Effects of Varying the Recommendations in a Fear-Arousing Communication," Journal of Personality and Social Psychology, 4, pp. 525-531.

Denny, J. P. (1966), "Effects of Anxiety and Intelligence on Concept Formation," Journal of Experimental Psychology, 72. pp. 596-602.

Evans, R., Rozelli, R., Laxater, T., Dembroski, T., and Allen, B. (1970), "Fear Arousal, Persuasion, and Actual vs. Implied Behavioral Change," Journal of Personality and Social Psychology, 16. pp. 220-227.

Haefner, D. (1956), "Some Effects of Guilt-Arousing and Fe r-Arousing Persuasive Communications on Opinion Change," American Psychologist, p. 359.

Janis, I., and Feshbach, S. (1953), "Effects of Fear-Arousing Communications," Journals of Abnormal and Social Psychology, 48, pp. 78-92.

Kormzweig, N. (1967), "Behavior Change as a Function of Fear Arousal and Personality," Unpublished doctoral dissertation, Yale University.

Leventhal, H., Singer, R., and Jones, S. (1965), "The Effects of Fear and Specificity of Recommendations upon Attitudes and Behavior," Journal of Personality and Social Psychology, 2, pp. 20-29.

Leventhal, H. and Trembly, G. (1968), "Negative Emotions and Persuasion," Journal of Personality, 36, pp. 154-168

Mewborn, C. R., and Rogers, R. W. (1979), "Effects of Threatening and Reassuring Components of Fear Appeals on Physiological and Verbal Measures of Emotion and Attitudes," Journal of Experimental Social Psychology, 15, pp. 242-253.

Myers, G. L. (1972), Fundamentals of Experimental Design, 29th ed., Boston: Allyn and Bacon, Inc.

Powell, F. and Miller, G. (1967), "Social Approval and Disapproval Cues in Anxiety-Arousing Communications," Speech Monographs, pp. 152-159.

Ray, M., and Wilkie, W. (1970), "Fear: The Potential of an Appeal Neglected by Marketing," Journal of Marketing, 34, pp. 93-114.

Rogers, R. W. (1975), "A Protection & Theory of Fe r Appeals and Attitude Change," Journal of Psychology, 919 pp. 93-114.

Rogers, R. W. and Mewborn, C. R. (1976), "Fear Appeals and Attitude Change: Effects of a Threat's Noxiousness, Probability of Occurrence and the Efficacy of Coping Responses," Journal of Personality and Social Psychology, 34, pp. 54-61.

Rubin, S. E., Lawlis, G. F., Tasto, D. L., and Namenek, T. (1969), "Factor Analysis of the 122-Item Fear Survey Schedule," Behavioral Research and Therapy, 7, pp. 381-386.

Spielberger, C. D. (1972), Anxiety: Current Trends in Theory and Research, Vol. 1, new York: Academic Press.

Spielberger, C. D. and Smith, L. H. (1966), "Anxiety (drive) Stress, and Serial-Position Effects in Serial-Verbal Learning," Journal of Experimental Psychology, 72, pp. 589-595.

Sternthal, B., and Craig, C. S. (1974), "Fear Appeals: Revisited and Revised," Journal of Consumer Research, 1, pp. 22-34.

Stuteville, J. R. (1970), "Psychic Defenses Against High Fear Appeals," Journal of Marketing, 34, pp. 39-45.

Wheatley, J. J., and Oshikawa, S. (1970), "The Relationship Between Anxiety and Positive and Negative Advertising Journal of Marketing Research, 7, pp. 85-89.

Wheatley, J. J. (1971), "Marketing and the Use of Fear-or-Anxiety Arousing Appeals," Journal of Marketing, 35, pp. 62-64.

----------------------------------------