Document not found! Please try again

Research in Marketing - Science Direct

15 downloads 1352 Views 909KB Size Report
J. of Research in Marketing 13 (1996) 319-329. The relationships among arousal potential, arousal and stimulus evaluation, and the moderating role of need for ...
International Journal of

ELSEVIER

Intern. J. of Research in Marketing 13 (1996) 319-329

Research in Marketing

The relationships among arousal potential, arousal and stimulus evaluation, and the moderating role of need for stimulation Jan-Benedict E.M. Steenkamp

a,b,* Hans Baumgartner c, Elise van

der Wulp b

a Catholic Unicersity ofLeucen, Naamsestraat 69, 3000 Leucen, Belgium b Wageningen Unicersity, Hollandseweg 1. 6706 KN Wageningen, The Netherlands c The Man' Jean and Frank P. Smeal College ~fBusiness Administration, The Pennsyh'ania State Uniz'ersity, Unicersitv Park. PA 16802, USA

Received 1 August 1994:accepted 4 April 1996

Abstract We develop a series of hypotheses delineating the impact of arousal potential on arousal and of arousal on stimulus evaluation, and we introduce the concept of need for stimulation (NST) as a key moderating variable which takes into account other sources of stimulation and individual differences in preferred level of stimulation. The hypotheses are investigated in an exploratory study involving a series of fear-appeal ads. We find that there is a monotonically increasing relationship between the arousal potential of a stimulus and the arousal it induces in the consumer, and that the relation between arousal evoked by the stimulus and the consumer's evaluation of the stimulus takes the form of an inverted U. Support is also obtained for the moderating role of NST in this process. The effect of arousal potential on arousal is stronger for individuals with higher NSTs, and the level of arousal at which stimulus evaluation reaches a maximum is higher for people with higher NSTs.

1. Introduction It is a common observation that people frequently do not act to minimize arousal, as has traditionally been assumed in drive-reduction types of theories, but are rather actively looking for stimulation from the environment or through internal means (Mc-

" Correspondingauthor. Fax: +32-16-326732; e-mail: [email protected] order of the first two authors' names is arbitrary;both contributedequally to this study. An earlier version of this paper won the Houghton-MifflinBest Paper Award at the 1994 EMAC Conference.

Guire, 1976). In the consumer context, examples of behaviors which entail the potential for stimulation and arousal include watching exciting movies, attending to emotionally-charged stimuli such as feararousing ads, gambling for high stakes, participating in risky activities such as skydiving, switching brands for the sake of variety, consuming intriguing products such as mystery books or crossword puzzles, and engaging in intrinsically-motivated information search out of curiosity. A considerable body of psychological literature suggests that certain properties of stimuli have the capacity to induce arousal in people and that the arousal evoked by a stimulus may be an important

0167-8116/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved. PII S0167-81 16(96)00013-4

320

J.-B.E.M. Steenkamp et al. / Intern. J. of Research in Marketing 13 (1996) 319-329

determinant of the evaluation of the stimulus (e.g., Berlyne, 1960, 1971; Deci, 1975; Fiske and Maddi, 1961; Zuckerman, 1979, 1994). The importance of the concept of arousal and of arousal theories for understanding various aspects of consumer behavior has also been acknowledged in consumer research (e.g., Bettman, 1979; Holbrook and Hirschman, 1982; Raju, 1980; Steenkamp and Baumgartner, 1992). However, with the exception of a number of studies that investigate arousal as a dimension underlying emotional reactions - for example, studies using the pleasure-arousal-dominance (PAD) paradigm of Mehrabian and Russell (1974) (e.g., Donovan and Rossiter, 1982; Holbrook and Gardner, 1993; Olney et al., 1991) - the arousal construct has not received much empirical attention in consumer research. In a recent review paper, Bagozzi, 1991 (p. 153) states that arousal is a pervasive factor in everyday consumer behavior, and he calls for more empirical research studying the role of arousal in the consumer context. The purpose of this paper is to develop and test four interrelated hypotheses concerning the role of arousal in evaluations of marketing stimuli. ~ We discuss how arousal-generating stimulus properties are linked to arousal and how arousal influences stimulus evaluation, and we introduce the concept of need for stimulation as a key moderating variable which takes into account other sources of stimulation and individual differences in preferred level of stimulation. The hypotheses are tested in an exploratory study dealing with consumers' responses to feararousing ads.

2. Theoretical background In what follows, we will first define the terms arousal potential, arousal, and stimulus evaluation, and then propose two hypotheses about the relationships between arousal potential and arousal, and

No claim is made that arousal is the only variable affecting consumers' evaluations of stimuli (Bettman, 1979). However, the present paper concentrates on the role of arousal since it has received relatively little attention in consumer research.

between arousal and stimulus evaluation. Subsequently, we will introduce need for stimulation as a person-related characteristic which is posited to moderate the relations between these three constructs.

2.1. Arousal potential, arousal, and stimulus evaluation Arousal potential is defined as the extent to which a stimulus is capable of raising arousal (Berlyne, 1978). It refers to the "psychological strength" or intensity of a stimulus (Berlyne, 1967), and "represents something like the overall power to excite the nervous system" (Berlyne, 1973, p. 14). Its meaning is similar to the notion of stimulus impact in the framework of Fiske and Maddi (1961) and the concept of information rate proposed by Mehrabian and Russell (1974). Three types of stimulus properties are posited to contribute to arousal potential (Berlyne, 1971): psychophysical properties, which depend on the physical and chemical characteristics of the stimulus (loudness, color, temperature, etc.); ecological properties, which refer to variables that are associated with specific requirements for health and survival (e.g., stimulus factors related to sexual appetite or fear); and collative properties (e.g., novelty, incongruity, complexity), which, for their motivational effects to occur, depend on comparison, or collation, of various stimulus elements with each other or with previous experiences. Arousal is defined as a motivational state pertaining to the level of alertness or activation of an individual, ranging on a continuum from extreme drowsiness to extreme wakefulness (Berlyne, 1966; Mehrabian and Russell, 1974). Arousal is in many ways similar to the earlier concept of drive and reflects the intensity aspect of psychological functioning. The final construct in the framework, stimulus evaluation, is defined as a general evaluative reaction vis-h-vis the stimulus (Breckler, 1984). This evaluative reaction may take the form of cognitive appraisals (e.g., judgments of the utility or value of the stimulus), affective responses (e.g., emotional reactions elicited by the stimulus), or behavioral tendencies (e.g., approach-avoidance behaviors directed at the stimulus). In the present context we are

J.-B.E.M. Steenkamp et al. / Intern. J. of Research in Marketing 13 (1996) 319-329

mostly concerned with verbal-report measures of the degree of favorability toward some stimulus. Fig. 1 combines the constructs discussed above within an integrated framework. It summarizes the hypothesized relations between arousal potential, arousal, and stimulus evaluation, and depicts the moderating role of need for stimulation. Below, we will develop the rationale for these hypotheses. 2.2. Relationships between arousal potential, arousal, and stimulus evaluation Fiske and Maddi (1961) and Mehrabian and Russell (1974) postulated a monotonically increasing, and possibly linear, relation between arousal potential and arousal. The underlying logic is that the greater the intensity of the stimulus in terms of collative, ecological, or psychophysical properties, the more aroused a person exposed to this stimulus is likely to be. Consistent with this theorizing, experiments have shown that arousal is higher for surprising stimulus patterns, for stimuli with higher luminous intensity, for blurred and novel stimuli, and for more complex, irregular, and incongruous figures (Mehrabian and Russell, 1974). Berlyne's position concerning the relation between arousal potential and arousal is not completely clear (Berlyne, 1960, 1963, 1971). On the one hand, he posited a U-shaped relation between the two constructs, suggesting that arousal would be mini-

321

mized when the arousal potential of a stimulus was at some optimal value. On the other hand, he conceded that in many cases the relationship might be monotonically increasing by stating that "the greater the arousal potential in a stimulus situation the more aroused an organism is likely to be" (Berlyne, 1971, p. 70). Further, he arrived at a U-shaped relation by assuming that, under conditions in which the arousal potential of an environment has been low for a long time, a state of boredom ensues which, according to him, produces high arousal. Berlyne's state of boredom refers to extreme situations, such as conditions of sensory deprivation for an extended period of time, and it is expected that these situations rarely, if ever, occur in a consumer behavior context. For some people, today's television environment consisting of bland quiz shows and banal soap operas, interspersed with dull ads at high frequencies, may be an example of sensory, or at least mental, deprivation. However, even this television environment does not come close to Berlyne's extreme state of boredom which gives rise to his U-shaped hypothesis. It still provides a much higher arousal potential (admittedly in terms of psychophysical properties, but they constitute an important component of arousal potential) than sensory deprivation experiments in which (virtually) no external stimulation is present (Zuckerman, 1979). Thus, in a consumer behavior context only the right-hand or increasing portion of Berlyne's U-

H3

H4

Moderating effects of need for stimulation

I Relationships between AP,

Arousal potential

Stimulus evaluation

Arousal

AR and SE

AP

H1

H2

Fig. 1. Theoretical framework.

Note: The figure is an illustrative and stylized representation of the relationships among the various constructs. For example, a linear relationship between AP and AR is depicted for HI; other types of monotonic relations are also possible.

322

J.-B.E.M. Steenkamp et al. / Intern. J. of Researeh in Marketing 13 (1996) 319-329

shaped curve appears relevant. Hence, the theories of Fiske and Maddi, Mehrabian and Russell, and Berlyne converge in suggesting that a monotonic relation between arousal potential and arousal is most plausible. Consistent with this expectation, it has been found that musical tempo (a psychophysical stimulus property) and uniqueness of an ad (presumably reflecting novelty-related factors) are positively related to arousal (Holbrook and Gardner, 1993; Olney et al., 1991). We therefore hypothesize that (cf. Fig. 1): H I : There is a monotonically increasing relationship between the arousal potential of a stimulus and the arousal induced by that stimulus. Arousal, in turn, is hypothesized to influence stimulus evaluation. There is general agreement among researchers that extremely high levels of arousal evoked by a stimulus reduce a person's evaluation of that stimulus (Berlyne, 1960; Fiske and Maddi, 1961; Hebb, 1955; Mehrabian and Russell, 1974). In general, humans are not very well equipped to cope satisfactorily with highly arousing stimuli, which may be explained by the general adaptation syndrome stress reaction (Mehrabian and Russell, 1974). Very high arousal elicits strong bodily reactions (e.g., production of adrenalin, gastrointestinal changes) which may become threats to the person's health (causing headaches, insomnia, etc.) and may lead to feelings of fatigue and exhaustion. There is less agreement on the effect of low levels of arousal on stimulus evaluation. Some theorists (Fiske and Maddi, 1961; Hebb, 1955; Mehrabian and Russell, 1974) have hypothesized that the evaluation of a stimulus is lower at low levels of arousal. There is a large body of evidence supporting this position, based on studies with both animals and humans, indicating a greater preference for moderate degrees of arousal associated with more novel and varied stimuli than for uniform, familiar, and boring stimuli that induce low levels of arousal (see Mehrabian and Russell, 1974; Deci, 1975; and Berlyne, 1978 for reviews). Combining the notions that both high and low levels of arousal lead to lower stimulus evaluation, several researchers have posited that the evaluation of a stimulus is an inverted U-shaped function

of arousal (Fiske and Maddi, 1961; Hebb, 1955; Mehrabian and Russell, 1974). In contrast, Berlyne (1960, 1963), in his early work, adhered to an arousal-reduction type of theory by arguing that, just as a person strives to reduce tension, a reduction in arousal is always preferable. This implies that low levels of arousal are associated with positive stimulus evaluation so that the relation between the two constructs is negative. In his later work, Berlyne (1967, 1971) moved toward the inverted U-shaped relation as he recognized that many experiments with humans and animals - involving such diverse stimuli as shocks of varying intensity, poems that differ in complexity, and variations in tone sequences - showed that increases as well as decreases in arousal can be positively evaluated. As shown in Fig. 1, we therefore hypothesize that: I-I2: The relation between arousal induced by a stimulus and the evaluation of that stimulus takes the form of an inverted U. 2.3. Moderating effects o f need f o r stimulation

The relations between arousal potential, arousal, and stimulus evaluation suggested in H1 and H2 are affected by contextual factors and person-related characteristics (Fiske and Maddi, 1961; Berlyne, 1966, 1967; Zuckerman, 1988). A key moderator is what we will call a person's need f o r stimulation, defined as the discrepancy between an individual's optimum stimulation level and his or her current stimulation level at the time of exposure to the stimulus. 2 Every person is assumed to have a uniquely determined, homeostatic level of stimulation called optimum stimulation level (OSL) at which he or she feels most comfortable (Zuckerman, 1994). This ideal level of stimulation pertains to the amount

2 In the context of this paper it is important to distinguish between the activating effects of a particular stimulus of interest (e.g., a TV program, an advertisement, etc.) and the activating effects of all other environmentaland internal sources. Following Berlyne (1978), we will use the terms arousal potential and arousal from the stimulus to refer to the influence of a particular stimulus of interest and the terms optimal and current stimulation to denote all sources of activation.

J.-B.E.M. Steenkamp et a l . / lntern. J. of Research in Marketing 13 (1996) 319-329

of stimulation a person prefers in general, from all possible internal and external sources across all possible situations and over time (Maddi, 1961; Zuckerman, 1979). A person's actual exposure to stimulation may deviate from the optimal level, and the term current level of stimulation is used to refer to the amount of stimulation from all sources that a person experiences at a particular point in time. We refer to the difference between the optimum and current stimulation levels as a person's need for stimulation (NST). The higher a person's optimum stimulation level relative to his or her current level of stimulation, the greater (more positive) NST. Conversely, the lower a person's optimum stimulation level relative to his or her current level of stimulation, the smaller (more negative) NST. Thus, NST ranges on a continuum from negative values, in which case the current level of stimulation exceeds the person's OSL, to positive values, in which case the current level of stimulation falls short of the person's OSL. When NST is positive, the person has a need for an i n c r e a s e in stimulation, whereas when it is negative, he or she has a need for a d e c r e a s e in stimulation. For simplicity, we will use the term need for stimulation, noting that higher NST may refer to less negative values (the perso n has a lesser need for a decrease in stimulation) or more positive values (the person has a greater need for an increase in stimulation). Prior empirical research, if it has taken into account differences between individuals at all, has generally not considered deviations from the preferred level of stimulation and has instead used measures of OSL (see Steenkamp and Baumgartner, 1992 for a review), although some authors have acknowledged that it is the difference between optimum and current stimulation which results in attempts to reduce or augment stimulation (e.g., Steenkamp and Baumgartner, 1992; see also Deci, 1975, Hebb, 1955). An exception is the work of Wahlers and Etzel (1985), who related OSL scores as well as scores reflecting the discrepancy between current and optimum levels of stimulation to evaluations of different dimensions of vacations and found that the discrepancy scores yielded somewhat better results than the OSL scores. What are the moderating effects of NST on the relationship between arousal potential and arousal?

323

In general, researchers have not been very clear on this point. However, analogous to research on perceptual sensitization, where the impact of deprivation-relevant stimuli has been found to be higher for deprived people, we posit that people who are higher on the NST continuum should be more sensitive to stimuli with arousal potential than people who are lower on the NST continuum. Our expectation is consistent with the "'personality-emotional sensitivity" hypothesis of Mehrabian and Russell (1974) which states that particular personality dimensions may be characteristically associated with certain heightened emotional sensitivities, and that stimuli that elicit certain emotions have the strongest effect on persons who possess a particular emotional sensitivity (e.g., anxious people tend to be more afraid of a fear-eliciting stimulus than non-anxious people). Indirect empirical evidence for the moderating role of NST is provided by studies of Smith et al. (1986), Smith et al. (1989) and Mehrabian and Russell (1974) who found that arousal potential has a greater effect on arousal for people who have higher OSLs, and that this effect is especially strong at high levels of arousal potential (see also Zuckerman, 1988). Thus, we hypothesize that (cf. Fig. 1): H3: The effect of arousal potential on arousal is stronger the higher the individual is on the need for stimulation continuum. NST is also hypothesized to moderate the relationship between arousal and stimulus evaluation. In general, research suggests that people prefer stimuli that induce intermediate levels of arousal. However, a person's NST should influence at what level of arousal stimulus evaluation reaches a maximum. Assume there are two persons, A and B, with person A having a higher position on the NST continuum than person B. Then, the most preferred level of arousal in a stimulus should be higher for A than for B (Deci, 1975; Leuba, 1955). In other words, the optimal influx of arousal (Berlyne, 1960) will be positively associated with a person's score on the NST continuum. For people who are relatively high (low) on the NST continuum, a stimulus should be evaluated most favorably when it induces relatively high (low) levels of arousal. Consistent with this notion,

324

J.-B.E.M. Steenkamp et al. / Intern. J. of Research in Marketing 13 (1996) 319-329

humans as well as animals have been found to be less inclined to welcome arousing stimuli when they are highly stimulated (Berlyne, 1967). For example, subjects in an experimental treatment designed to raise arousal chose complex visual patterns less frequently for further inspection than subjects in a control group. Thus, we hypothesize (see Fig. 1 for a graphical representation):

1-14: The level of arousal at which stimulus evaluation reaches a maximum is higher for individuals who are higher on the need for stimulation continuum. As is evident from the wording of H3 and H4, these hypotheses are derived for NST as a continuous variable. Thus, the rationale for these hypotheses is independent of the sign (positive or negative) of NST. H3 and H4 are equally valid for situations in which all consumers in a sample are overstimulated or understimulated, and the hypotheses hold when some people are overstimulated and some are understimulated. For example, even if all consumers were to be overstimulated in a particular situation (i.e., all consumers have a negative value for NST), people who are less overstimulated (have a less negative value for NST) should be more sensitive to increases in arousal potential, and the level of arousal at which stimulus evaluation reaches a maximum should be higher, compared to people who are more overstimulated (have a more negative value for NST).

3. Method A study was designed to test the four hypotheses developed in the previous section. Advertisements were used as the stimulus of interest. Specifically, subjects watched ads varying in the strength of the fear appeal employed. The framework developed in the present paper seems eminently applicable to the context of fear appeals as it is generally acknowledged that fear appeals are arousing (Strong and Dubas, 1992). In terms of Berlyne's classification (Berlyne, 1967), the arousal potential of fear-appeal ads is primarily due to their ecological properties. Three variables were included as covariates in the

analyses, viz. the severity of the threat associated with the issue addressed in the ad, the perceived likelihood that the negative consequences of the threat will occur, and gender. The former two are key salient beliefs formed in cognitive processing of the threat (Rogers, 1983), while the importance of gender in reactions to fear-appeal messages is indicated in the stress literature (Miller and Kirsch, 1987).

3.1. Subjects and procedure Subjects were 89 undergraduates (22 males and 67 females) from a university in the Netherlands. Optimum stimulation level was measured 8 weeks before the actual experiment was conducted. In the main experiment, participants were run in a single group, with three experimenters available to answer questions and provide assistance. Subjects first completed a questionnaire designed to measure their current level of stimulation. Next, subjects were shown a series of seven fear-appeal ads which were presented in random order in terms of degree of arousal potential involved. The seven ads dealt with four important societal issues (smoking during pregnancy, smoking in general, AIDS, and child abuse). None of the subjects had seen any of the ads before. Immediately after seeing each ad, subjects responded to measures of arousal potential, arousal, and ad evaluation. Subjects also indicated whether they had seen the ad before. After everybody had provided these ratings for a given ad, the next ad was shown. Care was taken to proceed in an unhurried manner, and subjects were given ample time between ads. Subjects were also allowed to take a short break after the fourth ad. After all ads had been shown, subjects rated the probability of negative outcomes and the severity of unfavorable consequences associated with each of the issues addressed in the various fear appeals. At the conclusion of the study, subjects were asked to comment upon the purpose of the study. An analysis of their responses indicated that subjects showed no understanding of the true purpose behind our research. This lack of awareness and our focus on nonmonotonic and interaction effects (except for H I ) helps to rule out the likelihood of demand effects (Holbrook and Gardner, 1993). In

J.-B.E.M. Steenkamp et al. / Intern. J. of Research in Marketing 13 (1996) 319-329

total, the experiment lasted approximately one and a half hours. 3.2. Measures

Subjects assessed the arousal potential of each ad on the following four 7-point bipolar scales: not f e a r s o m e / f e a r s o m e , not anxiety p r o v o k i n g / a n x i e t y provoking, not w o r r i s o m e / w o r r i s o m e , and not frightening/frightening. 3 Subjects' arousal due to watching the ads was measured using the general arousal scale of Mehrabian and Russell (1974). This scale consists of six 7-point bipolar items pertaining to subjectively experienced or f e l t arousal (cf. Cohen and Areni, 1991). Ad evaluations were assessed on three 7-point bipolar scales: u n f a v o r a b l e / f a v o r a ble, b a d / g o o d , and n e g a t i v e / p o s i t i v e . Some of the scales were reverse coded to control for yea-saying effects. Confirmatory factor analysis on the measures for arousal potential, arousal, and ad evaluation supported the construct validity of the three factors. A three-factor solution yielded an adequate fit: X2(62) = 424.93, CFI = 0.93, GFI = 0.90. Convergent validity was evidenced by statistically significant factor loadings (the minimum t-value was 9.99) that were also high in magnitude (the average loading was 0.79) (Anderson and Gerbing, 1988). Correlations among the three factors were all significantly lower than unity ( p < 0.001), which supports the discriminant validity of the measures (Anderson and Gerbing, 1988). Coefficient o~ for the constructs ranged from 0.87 for arousal to 0.91 for arousal potential. Multiple-item measures were constructed by averaging the ratings on the individual items. Subjects' optimum stimulation level was measured using a 7-item version of the Change Seeker Index (Steenkamp and Baumgartner, 1995), which achieved a coefficient o~ of 0.81. Current stimulation level was assessed by Z u c k e r m a n ' s 15-item State

3 These items measure subjects' perceptions of potentially arousal-inducing properties of an ad. Berlyne (1978, p. 130) advocated this approach in the case of complex stimuli such as ads.

325

Sensation Seeking scale (coefficient ot of 0.84) (Zuckerman, 1979). Responses to both scales were measured on 5-point scales. NST was computed as the difference between the average score of a person on the Change Seeker Index and his or her average score on the State Sensation Seeking scale. The reliability of the difference score (Lord, 1963) was 0.79.4 Higher scores on NST indicate a greater need for stimulation (i.e., a greater need for increases in stimulation or a lesser need for decreases in stimulation). For each of the four issues addressed in the fear-appeal ads, subjects provided their personal assessment of the probability of occurrence of the threat to them as well as the severity of the unfavorable consequences associated with the threat, using 7-point Likert scales ranging from totally disagree to totally agree.

4. Tests of hypotheses Hypotheses 1 through 4 were tested using regression analysis across subjects and commercials. 5 As argued above, gender (coded as l for males and 0 for females), severity of unfavorable consequences, and probability of occurrence of the threat were added as covariates. Since the last variable was never signifi-

4 The use of difference scores has recently been criticized by Peter et al. (1993). However, for several reasons difference scores are appropriate in the present context. Theory indicates that it is the difference between optimum and current stimulation level, rather than either of the two constructs by themselves, which affects consumer reactions. Moreover, frequently occurring problems with difference scores are high correlations between its components, low reliabilities, and restrictions in the variance of the difference variable. None of these problems was apparent in the present study. The correlation between the Change Seeker Index and the State Sensation Seeking scale was only 0.16 (p > 0.10), the reliability of 0.79 is adequate for basic research (Nunnally, 1978), and the variance in NST scores was actually greater than the variances of either component. 5 Predictor variables involved in quadratic effects and/or interactions were mean-centered to reduce multicollinearity. Since the hypotheses are directional, all statistical tests will be one-tailed (Ferguson, 1981).

J.-B.E.M. Steenkamp et al. / Intern. J. of Research in Marketing 13 (1996) 319-329

326

cant in the present study, it was dropped from further consideration. 6 4.1. Arousal

H1 predicts a monotonically increasing relation between arousal potential and arousal, while H3 states that this relation is moderated by NST such that the effect of arousal potential on arousal increases with NST. These hypotheses were tested simultaneously by regressing arousal on NST, the linear (AP) and quadratic (AP 2) terms of arousal potential, and their interactions with NST, plus the covariates gender (SEX) and perceived severity of unfavorable consequences (SEV). The following regression estimates were obtained (t-values in parentheses): A R - - - 4 . 1 0 + 0 . 4 2 A P + 0.04AP 2 (15.44)

(2.99)

0.07 NST ( - 1.10)

+ 0.06 NST *AP + 0.04 NST *AP 2 (1.66)

-

0.21 S E X ( - 2.85)

(2.21)

0.01 SEV ( - 0.29)

R 2 = 0.343, p < 0.001. The coefficients of both the linear and quadratic effects of arousal potential are positive and significant ( p < 0.001 and p < 0.01, respectively). The positive coefficient for AP 2 indicates that the effect of arousal potential on arousal accelerates at higher levels of arousal potential, which is consistent with the theory of Berlyne (1960), if we ignore his situation of extreme boredom (see above). Thus, H1 was supported.

6 The robustness of the results was assessed in several ways. We standardized each item of the CSI and SSS-state scales to account for potential systematic response scale differences between CSI and SSS-state. NST was then computed using these standardized scores. All analyses were also performed using six dummy variables to control for differences among the seven ads used in the study. Finally, the potential influence of responses to previous ads on subsequent ads (carryover effect) was examined in a series of regression analyses which also included main effects and interactions of a lagged term of the predictor variables (arousal potential, arousal). In all cases, the results remained substantively the same, and led to the same conclusions.

Consistent with H3, arousal potential and NST interacted. Significant interactions of NST were found with both the linear ( p < 0.05) and quadratic ( p < 0.05) terms of arousal potential. This indicates that the effect of arousal potential on arousal is greater the higher the NST of the individual. The significant NST*AP 2 interaction indicates that the effect is strongest at high levels of arousal potential. Arousal was also higher for females than for males ( p < 0.001), supporting similar findings in the stress literature (Miller and Kirsch, 1987). 4.2. Stimulus evaluation

The necessary prerequisite for an inverted Ushaped relationship between arousal and ad evaluation (AAo) as predicted by H2 is that the regression coefficient for AR 2 be negative and significant. Support for the moderating effect of NST hypothesized in H4 (i.e., that the level of arousal at which stimulus evaluation reaches a maximum is higher for individuals with higher needs for stimulation) requires that the regression coefficient for the interaction between NST and AR 2 be positive and significant. The following regression results were obtained (t-values in parentheses): AAD=4.76+0.41AR(6.47)

-

0.08 AR 2 ( - 1,65)

0.21 NST ( - 2.05)

0.01 N S T * A R + 0 . 1 8 N S T * A R 2 (-0.13)

(2.92)

+ 0.38 S E X (2.65)

0.28 SEV ( - 4.75)

R 2 = 0.122, p < 0.001. As evidenced by the significant negative coefficient for the squared term of arousal ( p < 0.05), the relation between arousal and stimulus evaluation can be modeled by an inverted U. This supports H2. Consistent with H4, this relation is moderated by NST as shown by the significant positive coefficient for the interaction between NST and AR 2 ( p < 0.01). The higher a person's NST, the higher the level of arousal at which stimulus evaluation reaches a maximum. The effects of the covariates severity of unfavorable consequences and gender were also significant. When the consequences were perceived to be more

!

J.-B.E.M. Steenkamp et al. / Intern. J. of Research in Marketing 13 (1996) 319-329

severe, ad evaluations were lower ( p < 0.001). Furthermore, males liked the ads more than females ( p < 0.01).

5. Discussion This paper proposed a set of hypotheses about the relationships between arousal potential, arousal, and stimulus evaluation, and the moderating role of NST in this process. The hypotheses were in part based on, and in part extended, extant arousal theories. We found support for the role of arousal and for the moderating role of NST as a person-related characteristic. Specifically, we found that the relation between arousal potential and arousal was monotonically increasing and that this relation was moderated by NST. In addition, we found that the relation between arousal and stimulus evaluation was curvilinear and that it too was moderated by NST. Our findings imply that arousal is not inherently aversive, motivating people to reduce it, but rather can increase the evaluation of marketing stimuli. As always, several limitations should be kept in mind when interpreting the results. First, the research was conducted with a convenience sample of undergraduate students. Although this should not compromise the internal validity of the studies, caution should be exercised in generalizing the results to other subject populations. In support of external validity, however, we used real ads that have actually been aired in various countries, rather than artificially manipulated stimuli typically employed in this type of research. Second, this study only considered the role of arousal as a consequence of arousal potential and as an antecedent to stimulus evaluation. Other variables such as the immediate "goals" of the individual interacting with a stimulus, cognitive processes, and a person's level of involvement will also influence how that stimulus is perceived, and they may also affect arousal and stimulus evaluation. The relatively modest R2's obtained in the regression analyses, especially for stimulus evaluation, suggest that other constructs also serve as important explanatory variables. A third possible limitation is that we did not measure need for stimulation directly (a validated

327

direct measure of NST is presently not available), but instead used the difference between CSI and SSS-state as an indirect measure. Reviewers were concerned that since we used different scales to assess a person's current and optimum stimulation levels, it may not be appropriate to compute NST as the difference between the two average scale scores. In defense of our operationalization of NST, the empirical support for our hypotheses concerning NST attests to the construct validity of our scale (Bagozzi, 1980), particularly when one considers the wellknown difficulties of confirming interaction predictions in nonexperimental research (McClelland and Judd, 1993). Moreover, standardization of the item scores before computing NST to account for potential scale differences between CSI and SSS-state did not lead to different conclusions. To further address this concern, we conducted the following scale validation study. We administered the Change Seeker Index (CSI), the State Sensation Seeking (SSS) scale, and a newly developed direct measure of need for stimulation (NSTDIR) to an independent sample of 56 undergraduate students. The nine items included in the NSTDIR scale were sleepy (reversed), energetic, tired (reversed), intense, vigorous, aroused, stimulated, excited, and bored (reversed). Subjects were asked to indicate for each item how they felt at this moment compared to their most desired or preferred state, using a 9-point scale with endpoints of "much more than desired" (9) and "much less than desired" (1) and amidpoint of "about the same as desired". Note that higher numbers on NSTDIR indicate overstimulation, and thus a negative correlation between NSTDIR and NST is expected. The indirect measure of need for stimulation (NST) was computed as before. The reliabilities of CSI, SSS, NST, and NSTDIR (after appropriate reversals) were 0.84, 0.88, 0.80, and 0.85, respectively. The correlation between NSTDIR and NST was - 0 . 4 4 ( p < 0.001). The disattenuated correlation was - 0 . 5 3 which, according to Zuckerman (1994), indicates a strong relation in the context of personality constructs. Thus, the validation study provides additional support for the validity of our operationalization of NST as the difference between average scores on CSI and SSS. Despite these encouraging findings, we acknowledge that more research is needed on the construct validity of our operationalization of NST,

328

J.-B.E.M. Steenkamp et al. / Intern. J. of Research in Marketing 13 (1996) 319-329

and it seems worthwhile to continue working on the development and validation of a scale that directly measures a person's NST. Future research can take several other directions. One important issue concerns the incorporation of arousal concepts into extant theories of consumer behavior. Since the mid-1970s consumer behavior research has been dominated by the information processing paradigm, which concentrates on how consumer responses to marketing stimuli can be explained through intervening cognitive processes. The information processing approach has been criticized as being overly rational, according too little attention to emotions and motivations, and in response some recent work has focused on the role of emotions in consumer behavior (Cohen and Areni, 1991). In that context, arousal has been studied as a major dimension underlying emotional experiences. However, with few exceptions (e.g., Kroeber-Riel, 1979; Sanbonmatsu and Kardes, 1988) consumer researchers have not regarded arousal as an important construct in its own right, and the potential of integrating the concept of arousal into models of consumer information processing has not been realized. Such an integration would not only broaden the scope of cognitive models, but would also help to address the issue of why arousal increases consumers' evaluations of marketing stimuli, at least up to a point. Consistent with the work of Berlyne (1960, 1963, 1978) and Fiske and Maddi (1961), the arousal framework tested in this paper does not consider the possibility that experienced arousal may vary in hedonic tone (cf. Watson and Tellegen, 1985). Future research could investigate whether there are situations in which it is useful to take into account the hedonic tone of arousal. The arousal framework presented in this paper also has implications for other areas such as innovativeness, brand switching, advertising wearout, and advertising alternation schemes. For example, a new brand usually has a higher arousal potential than a brand which the consumer has tried before, due to its novelty and other collative properties. Depending on the consumer's NST at that point of time and level of arousal induced by the new brand, the evaluation of the new brand may be higher than that of familiar brands, and the consumer may decide to try the new

brand. It would also be interesting to track the arousal potential of ads over time. It may be expected that ads regain part of their arousal potential after they have not been shown for a while, after which they may be used again. However, the rate of decline in arousal potential should be faster the second time. The potential for innovative work in this area is high, and we hope that the present study will prove useful to researchers working on the role of arousal in consumer behavior.

References Anderson, J.C. and D.W. Gerbing, 1988. Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin 103, 411-423. Bagozzi, R.P., 1980. Causal models in marketing. New York, NY: John Wiley. Bagozzi. R.P., 1991. The role of psychophysiology in consumer research. In: Th.S. Robertson and H.H. Kassarjian (eds.), Handbook of consumer behavior, 124-161. Englewood Cliffs, NJ: Prentice Hall. Bettman, J.R., 1979. An information processing theory of.consumer choice. Reading, MA: Addison-Wesley. Berlyne, D.E., 1960. Conflict, arousal, and curiosity. New York, NY: McGraw-Hill. Berlyne, D.E., 1963. Motivational problems raised by exploratory and epistemic behavior. In: S. Koch (ed.), Psychology: A study of a science, Vol. 5, 284-364. New York, NY: McGraw-Hill. Berlyne, D.E., 1966. Curiosity and exploration. Science 153, 25-33. Berlyne, D.E., 1967. Arousal and reinforcement. In: D. Levine (ed.), Nebraska symposium on motivation, 1-110. Lincoln, NE: University of Nebraska Press. Berlyne, D.E., 1971. Aesthetics and psychobiology. Englewood Cliffs, NJ: Prentice Hall. Berlyne, D.E., 1973. The vicissitudes of aplopathematic and thelematoscopic pneumatology (or the hydrography of hedonism). In: D.E. Berlyne and K.B. Madsen (eds.), Pleasure, reward, preference, 1-33. New York, NY: Academic Press. Berlyne, D.E., 1978. Curiosity and learning. Motivation and Emotion 2, 97-175. Breckler, S.J., 1984. Empirical validation of affect, behavior, and cognition as distinct components of attitude. Journal of Personality and Social Psychology 47, 1191-1205. Cohen, J.B. and Ch.S. Areni, 1991. Affect and consumer behavior. In: Th.S. Robertson and H.H. Kassarjian (eds.), Handbook of consumer behavior, 188-240. Englewood Cliffs, NJ: Prentice Hall. Deci, E.L., 1975. Intrinsic motivation. New York, NY: Plenum. Donovan, R. and J.R. Rossiter, 1982. Store atmosphere: An

J.-B.E.M. Steenkamp et al. / Intern. J. of Research in Marketing 13 (1996) 319-329 environmental psychology approach. Journal of Retailing 58, 34-57. Ferguson, G.A., 1981. Statistical analysis in psychology and education, 5th ed. New York, NY: McGraw Hill. Fiske, D.W. and S.R. Maddi, 1961. A conceptual framework. In: D.W. Fiske and S.R. Maddi (eds.), Functions of varied experience, 11-56. Homewood, IL: Dorsey Press. Hebb, D.O., 1955. Drives and the C.N.S. (central nervous system). Psychological Review 62, 243-254. Holbrook, M.B. and M.P. Gardner, 1993. An approach to investigating the emotional determinants of consumption durations: Why do people consume what they consume for as long as they consume it? Journal of Consumer Psychology 2, 123-142. Holbrook, M.B. and E.C. Hirschman, 1982. The experiential aspects of consumption: Consumer fantasies, feelings, and fun. Journal of Consumer Research 9, 132-140. Kroeber-Riel, W., 1979. Activation research: Psychobiological approaches in consumer research. Journal of Consumer Research 5, 240-250. Leuba, C., 1955. Toward some integration of learning theories: The concept of optimal stimulation. Psychological Reports 1, 27-33. Lord, F.M., 1963. Elementary models for measuring change. In: C.W. Harris (ed.), Problems in measuring change, 21-38. Madison, WI: The University of Wisconsin Press. Maddi, S.R., 1961. Exploratory behavior and variety seeking in man. In: D.W. Fiske and S.R. Maddi (eds.), Functions of varied experience, 253-277. Homewood, IL: Dorsey Press. McClelland, G.H. and C.M. Judd, 1993. Statistical difficulties of detecting interactions and moderator effects. Psychological Bulletin 114, 376-390. McGuire, W.J.. 1976. Some internal psychological factors influencing consumer choice. Journal of Consumer Research 2, 302-319. Mehrabian, A. and J.A. Russell, 1974. An approach to environmental psychology. Cambridge, MA: MIT Press. Miller, S.M. and N. Kirsch, 1987. Sex differences in coping with stress. In: R.C. Barnett, L. Biener and G.K. Baruch (eds.), Gender and stress, 278-307. New York, NY: Free Press. Nunnally, J.C., 1978. Psychometric theory. 2nd ed. New York, NY: McGraw-Hill. Olney, Th.J., M.B. Holbrook and R. Batra, 1991. Consumer responses to advertising: The effects of ad content, emotions, and attitude toward the ad on viewing time. Journal of Consumer Research 17, 440-453. Peter, J.P., G.A. Churchill, Jr. and T.J. Brown, 1993. Caution in the use of difference scores in consumer research. Journal of Consumer Research 19, 655-662.

329

Raju, P.S., 1980. Optimum stimulation level: Its relationship to personality, demographics, and exploratory behavior. Journal of Consumer Research 7, 272-282. Rogers, R.W., 1983. Cognitive and physiological processes in fear appeals and attitude change: A revised theory of protection motivation. In: J. Cacioppo and R. Petty, Social psychophysiology, 153-176. New York, NY: Guilford Press. Sanbonmatsu, D.M. and F.R. Kardes, 1988. The effects of physiological arousal on information processing and persuasion. Journal of Consumer Research 15, 379-385. Smith, B.D., W.M. Perlstein, R.A. Davidson and K. Michael. 1986. Sensation seeking: Differential effects of relevant, novel stimulation on electrodermal activity. Personality and Individual Differences 7,445-452. Smith, B.D., R.A. Davidson, D.L. Smith, H. Goldstein and W.M. Perlstein, 1989. Sensation seeking and arousal: Effects of strong stimulation on electrodermal activation and memory task performance. Personality and Individual Differences 10, 671-679. Steenkamp, J-B.E.M. and H. Baumgartner, 1992. The role of optimum stimulation level in exploratory consumer behavior. Journal of Consumer Research 19. 434-448. Steenkamp, J-B.E.M. and H. Baumgarmer, 1995. Development and cross-cultural validation of a short form of CS1 as a measure of optimum stimulation level. International Journal of Research in Marketing 12, 97-104. Strong, J.T. and K.M. Dubas, 1992. The processing of marketing threat stimuli: A comprehensive framework. In: J.N. Sheth (ed:), Research for marketing, Vol. I1, 221-263. Greenwich, CT: JAI Press. Wahlers, R.G. and M.J. Etzel, 1985. A consumer response to incongruity between optimal stimulation and lifestyle satisfaction. In: E.C. Hirschman and M.B. Holbrook, Advances in consumer research, Vol. 12. 97-101. Provo. UT: Association for Consumer Research. Watson, D. and A. Tellegen, [985. Toward a consensual structure of mood. Psychological Bulletin 98, 219-235. Zuckerman, M., 1979. Sensation seeking: Beyond the optimal level of arousal. Hillsdale, NJ: Lawrence Erlbaum. Zuckerman, M., 1988. Behavior and biology: Research on sensations and reactions to the media. In: L. Donohew, H.E. Sypher and E.T. Higgins (eds.), Communication, social cognition, and affect, 173-194. Hillsdale. NJ: Lawrence Erlbaum. Zuckerman, M., 1994. Behavioral expressions and biosocial bases of sensation seeking. Cambridge: Cambridge University Press.