On Assessing the Emotionality of Advertising Via Izard's Differential Emotions Scale

Chris T. Allen, University of Cincinnati
Karen A. Machleit, University of Cincinnati
Susan S. Marine, University of Cincinnati
ABSTRACT - Recently, several researchers have called for additional work on measurement methods that might capture the multidimensional richness of ad-evoked emotion. This paper involves an empirical investigation of two instruments that derive from Izard's (1977) differential emotions theory. Izard's influential theory postulates ten discrete emotion types. Data are reported from two studies which suggest his measures can be adapted to a variety of contexts while maintaining their desirable psychometric properties. The data also provide some insights about the nature of emotions commonly associated with advertising.
[ to cite ]:
Chris T. Allen, Karen A. Machleit, and Susan S. Marine (1988) ,"On Assessing the Emotionality of Advertising Via Izard's Differential Emotions Scale", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 226-231.

Advances in Consumer Research Volume 15, 1988      Pages 226-231


Chris T. Allen, University of Cincinnati

Karen A. Machleit, University of Cincinnati

Susan S. Marine, University of Cincinnati


Recently, several researchers have called for additional work on measurement methods that might capture the multidimensional richness of ad-evoked emotion. This paper involves an empirical investigation of two instruments that derive from Izard's (1977) differential emotions theory. Izard's influential theory postulates ten discrete emotion types. Data are reported from two studies which suggest his measures can be adapted to a variety of contexts while maintaining their desirable psychometric properties. The data also provide some insights about the nature of emotions commonly associated with advertising.


A variety of forces have converged in recent years to yield surging interest in affective and/or emotional topics among consumer and especially advertising researchers (e.g., Aaker, Stayman and Hagerty 1986; Batra and Ray 1986; Havlena and Holbrook 1986; Holbrook and Westwood 1987). A major focal point of this interest has involved the attitude-toward-the-ad (Aad) construct (e.g., Lutz 1985; MacKenzie, Lutz and Belch 1986). Aad researchers have produced empirical support for the inference that Aad is an important mediator of advertising's impact on brand attitude. However, as this stream has matured, some concern has arisen regarding the inherently inhibiting nature of Aad. That is, this construct leads researchers to consider ad-evoked affect as a unidimensional, bipolar variable which involves only simple evaluations concerning how good or likeable an ad is; thus the concern that Aad researchers "have thereby missed the multidimensional richness of the emotional responses that presumably mediate advertising effectiveness" (Holbrook and O'Shaughnessy 1984, p. 48).

Recently, Holbrook and O'Shaughnessy's concern about Aad's constraining nature has been recognized and reaffirmed by others (cf. Aaker et al. 1986; Batra and Ray 1986; Zeitlin and Westwood 1986). With this recognition has come a call for more research about the diversity of emotions that may be generated by exposure to advertising. If emotional response to advertising is a rich, multifarious phenomenon, can it be captured in data? What types of emotional responses are frequently provoked during ad exposure, and can measurement systems be adapted or developed to reliably gauge them? The present project presents data germane to these questions.

Psychologists have studied emotional structure for decades: their work furnishes a tempting starting point for those who are interested in exploring the richness of ad-evoked emotion. Indeed, psychologists have developed a number of comprehensive emotional taxonomies (e.g., Smith and Ellsworth 1985) and correspondent measurement systems which may be adaptable to the advertising domain (cf. Batra and Ray 1986). While Plutchik's (1980) emotional typology has been a focus of empirical work (Havlena and Holbrook 1986; Holbrook and Westwood 1987; Zeitlin and Westwood 1986), it is apparent that "much work remains to be done in constructing, comparing and validating competing typologies of emotional content" (Holbrook and O'Shaughnessy 1984, p. 59). The focus of this project is one of Plutchik's major theoretical competitors CIzard's (1977) differential emotions theory.

The taxonomies advanced by theorists like Izard and Plutchik were obviously not conceived as systems for categorizing reactions to ads; it is by no means obvious that they will prove useful in helping us measure or understand advertising response. Indeed, we assume that- examining these typologies for their potential applicability in an advertising context is an ambitious borrowing endeavor. This assumption guided our approach to the data collection reported herein, and led us to try to challenge the applicability of the Izard framework for the advertising domain. The issue of whether advertisements do provoke Izard's (1977) ten fundamental emotions is treated as an empirical question.

Two studies were conducted using Izard's differential emotions scale. Before turning to these data a brief discussion of Izard's system is provided, and concerns are elaborated about the adaptability of such a system for the advertising context.


Several typologies (e.g., Izard 1977; Plutchik 1980) have been proposed which seek to delineate finite sets of primary emotions. These theorists contend that all emotional experience can then be defined in terms of various mixtures or blends of the primary emotions. Izard's (1977) theory postulates ten primary emotions: interest, joy, surprise, sadness, anger, disgust, contempt, fear, shame and guilt. Izard (1977, p. 64) defines emotion as an intra-individual process characterized by specific neurophysiological activity and distinctive facial expression. He also proposes that each emotion type is associated with a unique, consciously-experienced, feeling state. It is this experiential component of emotion that is examined herein.

There are a number of reasons why Izard's system merits the attention of consumer researchers. First, Izard's typology is motivating new research in the emotions literature (e.g., McHugo, Smith and Lanzetta 1982; Mosher and White 1981), and his measurement procedures have been refined through considerable validation work (e.g., Fuenzalida et al. 1981; Izard 1977; Kotsch, Gerbing and Schwartz 1982). Also, Izard's taxonomy includes two added emotions (shame and guilt) not appearing in the Plutchik system advocated by Holbrook and his colleagues. Explicit consideration of these additional emotion types could prove germane for communication research involving guilt arousal (e.g., Bozinoff and Ghingold 1983). Finally, Izard's framework supplied the structure for Batra and Ray's (1986) recent review and synthesis of the emotion-typologies literature.

To assess emotion Izard and others have developed the Differential Emotions Scale (DES) (see Izard 1977, p. 126). The DES "is a standardized instrument that reliably divides the individual's description of emotion experience into validated, discrete categories of emotion" (Izard 1977, p. 124). The DES was formulated to gauge the emotional state of individuals at that specific point in time when they are responding to the instrument. A minor variant (the DES II) also allows assessment of emotions experienced over extended time periods. DES II allows the researcher "to determine how often one experiences each of the fundamental emotions" (Izard 1977, p. 125) in a specific context or over a specified time period.

The DES and DES II are both formulated around a thirty-item adjective checklist, with three adjectives for each of the primary emotions. Another variant has also been developed which expands each of the adjectives into phrases that describe the subjective feelings associated with each primary emotion. This version is referred to as the DES III (Kotsch et al. 1982). Both the DES II and III are used herein.


It can not be assumed that the desirable psychometric properties of Izard's instruments will automatically extend to an advertising application (cf. Peter and Churchill 1986). However, borrowing concerns in this instance extend beyond a call for reaffirming psychometric properties in the new context. Consumer researchers also need to consider the more fundamental question of whether the focal phenomena that psychologists seek to understand really occur as part of day-to-day ad processing. That is, do advertisements really provoke the kinds of experiences that Izard, Plutchik and others think of as emotion?

Contemporary theorists like Izard and Plutchik portray emotion as an adaptive, motivating mechanism. As summarized by Smith and Ellsworth (1985):

...emotions represent adaptive responses to the demands of the environment. Interpretation of the perceived situation in terms of prevailing goals creates an emotional experience that allows the organism to respond adaptively. This view implies that cognitive appraisals will lead to an emotional response primarily when they are perceived as having adaptive significance for the organism's well-being (p. 836).

Given this view of emotion as a process which facilitates individual adaptation to a demanding environment, it seems reasonable to question whether advertisements actually provoke emotion. Indeed, the scientific meaning psychologists ascribe to the emotion construct may have little relevance to the phenomenology of ad processing.

Consumer researchers have a notable reputation as ambitious borrowers (cf. Olson 1982; Sheth 1982), and scholars like Sheth (1982) have contended that "there is a clear surplus of borrowed constructs and a critical shortage of self-generated constructs in consumer behavior" (p. 14). Such a charge certainly merits reflection in the context of ad-evoked emotion. It is conceivable that the theories of Izard and others take for granted a level of attention to and involvement with environmental stimuli that simply does not characterize most ad processing. Thus, while commercials clearly do have the capacity to alter one's subjective feelings, these "altered feelings" may be simpler, less differentiated, and less intense than the focal phenomena of differential emotions theory.

There are theoretical alternatives to the Izard-type taxonomy which do conceptualize feelings as simpler, less differentiated forms of response: these are noteworthy because they facilitate translating the borrowing dilemma noted above into an empirical competition. Recent social cognition research has established that low-intensity affective states can impact cognition and behavior (e.g., Allen and Madden 1985; Isen 1984). Furthermore, empirical evidence indicates that this simple feelings construct is best conceived as two unipolar and largely independent dimensions which correspond to pleasant and unpleasant affect (cf. Abelson et al. 1982; Isen 1984). Such low intensity affects are likely to be a common part of day-to-day ad processing (e.g., Lutz 1985; MacKenzie, Lutz and Belch 1986).

If this simplistic feelings construct more effectively represents the subjective experience of ad processing, then Izard's ten-dimensional system should prove unwieldy in gauging advertising response. Alternatively stated, if the simpler conceptualization is more correspondent with common, ad-evoked feelings, one might expect the structure of the Izard instrument to collapse into a more parsimonious, two-dimensional system.

Interestingly, recent empirical work indicates that Plutchik's eight primary emotions collapse into two dimensions of response when consumption experiences (Havlena and Holbrook 1986) or advertising reactions (Holbrook and Westwood 1987) are the research focus. Our project supplies data concerning whether the complex structures of Izard's DESII and DESIII maintain their integrity in the ad-response context. First, however, a pilot study which examined more basic concerns about the psychometric properties of Izard's DESII and DESIII is reported.


One of our fundamental concerns about the DESII involved its adjective checklist format and whether grouping items by emotion type would influence reliabilities of each scale and/or the structure of the overall instrument. When these adjective items are grouped (rather than mixed randomly) in a questionnaire, it can be obvious to subjects that they are responding to related sets of items. This could yield a substantial response bias if a subject tried to be consistent within sets of items, rather than responding independently to each item. Such bias could inflate coefficient alpha (cf. Peter and Churchill 1986), and if it were substantial enough, the observed dimensionality of the overall instrument could be affected. Study I was conducted to assess how sensitive the reliabilities and ten-dimensional structure of Izard's instruments are to this response bias.

Beyond mixing the adjectives, another way that one might evoke more careful responses would be to enrich the individual items so that each is more thought provoking. Indeed, Izard's DESIII is primarily an enriched version of DESII, since in the DESIII each adjective is expanded into a phrase which offers more depth in describing the target emotion (cf. Kotsch et al. 1982). Because DESIII may be viewed as an enrichment of the DESII, the expectation was that DESIII would prove less susceptible to response bias contamination.

To examine these concerns, two questionnaires were developed that contained the DESII, then a page of "filler" items, followed by the DESIII. In version one of this instrument, the DESII adjectives were grouped by emotion type, while DESIII's phrases were mixed randomly; in version two, the adjectives were randomly mixed and the phrases were grouped. The context of the study involved the emotional character of college life. A total of 208 undergraduates from introductory business courses completed one of the two versions of the questionnaire.

Coefficient alpha's were calculated for each of the emotions in the four formulations of the DES: most fell in a range from .75 to .90. As expected, they demonstrated that the reliability of each scale is impacted by item grouping; however, this was the case for both the adjectives (DESII) and the phrases (DESIII). In a number of instances, mixing the items lowered the reliabilities from an acceptable range (.78 to .85) to more marginal levels (.63 to .68).

These data were also used to test the theoretical, ten-dimensional structure of the instruments. First, confirmatory factor analyses (via LISREL VI) were examined for the four DES variations. While no variant yielded an acceptable fit for the ten-dimensional model, the goodness of fit indices ranged from .75 to .77, indicating little discrepancy among the four versions in their actual dimensionality. Exploratory factor analyses were also run to offer further insight concerning actual dimensionality: results are summarized in Table 1. Factor loadings of .6 and above were used to interpret the derived factors. Seven to nine factors proved interpretable for each of the DES variants.

The most complex structure was the nine factor solution for the mixed adjectives. Notably, while the guilt/sadness scales collapsed to yield a single factor, this result was probably due to the fact that two of the guilt items (guilty and blameworthy) appeared before and after one of the sadness items (downhearted) in the checklist Without this quirk in the item ordering, it is likely that the mixed adjectives would have yielded the ten-dimensional structure. Furthermore, the collapsing of scales observed for the grouped versions of DESII and III also involved juxtaposed scales. These data suggest that a response bias can contaminate the observed dimensionality of the Izard instruments. The best way to address the problem would seem to involve a careful and complete mixing of the individual items.



These results identify an interesting trade-off inherent in the use of the Izard scales. If one wishes to maximize scale reliabilities, the best approach seems to entail grouping items by emotion type. However, such groupings appears to capitalize on a bias which can then contaminate the observed dimensionality of the instrument. Because of this trade-off two versions of the Izard instrument were again used in Study II. Study II utilizes the DESII with adjectives mixed and DESIII with phrases grouped.


Study II was conducted to examine psychometric properties of the DESII and III in the ad-response context. In addition, it furnishes evidence about the types of emotional experience that individuals commonly associate with ad exposure, and provides a test of the viability of Izard's ten-dimensional framework.


To help subjects get in touch with the kinds of subjective experiences that advertisements may evoke, a videotape containing twenty television ads was created. The first eight ads on the tape were 1983 Clio award winners: these were selected because they seemed emotionally provocative. The remaining twelve ads were selected in pairs from six different types of programming: 1) prime time, 2) day time, 3) late night, 4) weekend sports, 5) syndicated reruns, and 6) a movie. The objective in selecting these was to create a group of ads like one might encounter in day-to-day television viewing.

The study was introduced to subjects as an assessment of how ads make people feel. They were asked to try to remember the kinds of feelings they experience during normal commercial viewing. They were told that a videotape of twenty ads had been prepared simply to help them recollect the types of feelings that ads can provoke. Participants were shown the complete set of twenty ads, and then were asked to describe three other commercials which they remembered as being strong "feeling producers". Next, subjects reported how frequently they experience each of Izard's ten emotions during normal ad viewing via the DESII. Following a dummy task, subjects reported ad-evoked emotions on the DESIII.

The above procedure was conducted in six undergraduate and MBA classes, yielding 180 completed instruments. While the videotape was used simply to help participants recall the kinds of feelings ads produce in them during normal television viewing, it seemed an interesting empirical question to see if exposure to the tape actually influenced their responses. Thus, two additional classes were run as a control group; in these the above procedure was followed, except no videotape was shown to remind subjects of the kinds of feelings -that ads may produce. A total of 37 students were processed in this tape-absent condition.


Reliability coefficients for each of the ten emotions ranged from .72 to .89 in the tape-present groups, and from .57 to .87 in the tape-absent condition. Alpha's in the tape-absent condition were a bit erratic, but for the main body of subjects, the reliabilities were very sound.



Table 2 displays mean scores and standard deviations for each emotion type. Table 2 thus offers information concerning the emotions individuals are most likely to associate with advertising. The pleasant emotions-joy and interest--are the one's most associated with advertising for this group of subjects. Surprise is next in order of frequency, followed by disgust. The remainder of the negative emotions show little activation, and the low means for fear, shame and guilt indicate that these emotions may be evoked very infrequently.

Interestingly, the presence of the ad videotape had little impact: none of the means between the tape-present and tape-absent groups differed by as much as one scale point. For the DESII, the only pairs of means that are significantly different at the .05 level are those for joy and interest. Differences between means were even less pronounced on the DESIII. One might conclude from these data that both groups of subjects were providing, as requested, their sincere recollections of the kinds of emotions they experience as part of normal ad processing.

The final issue to be considered involves dimensionalityCis Izard's ten-dimensional structure supported in these data? This issue was examined via confirmatory factor analysis (using LISREL VI) for both the DESII and III. Findings for the two versions of the scale were nearly identical, so only the data for DESII are considered hereafter. Responses from subjects run in the tape-absent group were not included. Results for tests of six different models are summarized in Table 3.



It is conceivable that advertisements do not provoke the rich, discrete emotions postulated by Izard's theory. As previously discussed, there are theoretical alternatives which offer a less complex "feeling construct" involving only pleasant and unpleasant affects (cf. Abelson et al. 1982; Allen and Madden 1985; Isen 1984). If this simpler conceptualization is more consistent with the phenomenology of ad response, one might expect the DESII to collapse into plain pleasant and unpleasant dimensions. This issue was examined by specifying a simple structure for the DESII around the positive versus the negative emotions, with surprise handled in two ways. For the two factor models (M1 and M4) in Table 3, surprise was treated as a possible amplifier of both positive and negative emotion (cf. Tomkins 1980): it was specified to load with both emotional categories. In the three factor models (M2 and M5), surprise was treated as an indicator for arousal, which is often handled as a response dimension separate from valence (cf. Smith and Ellsworth 1985). Confirmatory factor analysis was then used to test whether these simplistic models fit the ad-response data better than Izard's complex structure.

The first three models tested the two, three and ten factor solutions. As shown in Table 3, none of the goodness of fit indices (GFI's) surpassed the .85 convention for an acceptable fit. However, the GFI for M3 does approach the cut-off, while those for M1 and M2 do not. Here there is no indication that parsimony yields an improved fit.

To further explore the dimensionality issue, a second set of models were tested: in these, the emotion types fear, sadness, guilt and shame were excluded for two reasons. First, the means in Table 2 suggest that these four negative emotions are not commonly provoked by advertisements. We thus reasoned that participants' responses on these types might be contributing little more than error variance. Second, the three remaining negative emotions (i.e., anger, disgust and contempt) are considered by Izard (1977) to be a theoretically-related set, which he labels the "hostility triad." Thus, tests of models M4, M5 and M6 entail the hostility triad along with joy, interest and surprise. Excluding four of the negative emotions did improve model fits. However, an acceptable fit is observed in Table 3 for just the most complex model (i.e., M6).

These results indicate that Izard's instruments do capture categories of emotional responses which are meaningful to subjects in the advertising context. They appear to reaffirm Izard's claim that the DES "reliably divides the individual's description of emotion experience into validated, discrete categories of emotion" (Izard 1977, p. 124).


These studies demonstrate instruments for assessing emotion which appear to have considerable versatility. The DESII and III separate individuals' reflections on their emotional experiences into the primary types derived from Izard's theory. They possess sound psychometric properties in diverse applications, including the ad-response context. In addition, the complex structure of the instruments seems compatible with advertising response data. Thus, for those inclined to move beyond the unidimensionality of the Aad construct, the DESII and III appear to be sound instruments for studying the multidimensional richness of ad-evoked emotion. More empirical work would then seem appropriate concerning what emotion types or combinations of types are most commonly experienced during ad processing. When used in conjunction with measures of advertising effectiveness, research with the DESII and III might then begin to offer insights about emotional experiences which enhance advertising's impact (cf. Holbrook and O'Shaughnessy 1984).

It is important to emphasize that while Izard's measures do allow the researcher to detect diverse and sophisticated emotional experience, the crucial question regarding which of these emotions are commonly evoked by advertisements is far from answered. Borrowing theories like those of Izard or Plutchik should prove valuable in consumer research, but as an alternate research strategy there is also much to be said for generating our own "affective" constructs. This alternate strategy is nicely illustrated by the work of Aaker and his colleagues (Aaker and Bruzzone 1985; Aaker et al. 1986) who identify "irritation" and "warmth" as two feeling states commonly provoked by advertisements. Aaker et al.'s conceptualizations are driven by the phenomenology of ad response, rather than by basic psychological theory. Merging this emphasis on ad-evoked phenomena with the insight provided by basic theory may help pinpoint emotional constructs of special significance to consumer behavior.

Related to the above issues concerning the frequency of ad-evoked emotion are questions about the intensity of such experience. One might suspect substantial differences in intensity levels of ad-evoked emotion compared to other emotions experienced in day-to-day life. However, what emotions are experienced, and their intensity in an advertising context, are empirical matters which could be examined with Izard's (1977) original Differential Emotions Scale. Whereas the DESII was formulated to tap frequency of experience, the DES was created to gauge intensity. The intensity issue might be examined by exposing subjects to ads presumed to be emotionally provocative, and then measuring their emotional states (via the DES) immediately after exposure to each ad. This sort of study could prove helpful in attacking the difficult question of whether sophisticated and distinctive emotions really are common in day-to-day ad processing.

Data generated in Study II suggest that the emotion types joy, interest, surprise and disgust characterize the subjective experiences individuals most commonly associate with advertising. Obviously, it is just this sort of finding that needs to be examined with additional studies before any strong inferences can be offered regarding the nature of ad-evoked emotion. We would thus advise that future studies should continue to work with all ten dimensions of the Izard instruments. However, it is noteworthy that other recent investigations of advertising emotionality have also found positive/pleasant emotional types as dominant response dimensions (Batra and Ray 1986; Zeitlin and Westwood 1986).

It is possible that the greatest discrepancy between the phenomenology of advertising response and the focal phenomena of psychological theoreticians comes within the category of negative emotions. That is, feeling states generated by fear or guilt appeals in advertisements may have little in common with the subjective experience which accompanies Izard's emotion types of the same name. Moreover, if as Isen (1984) argues, negative emotions are more cognitively complex and harder to evoke than positive affects, it may be that ads designed-to produce fear or guilt simply do not "work" as often or as reliably as those designed to yield joy or interest. Indeed, the continuing debate over the virtues of irritating consumers with ads (cf. Aaker and Bruzzone 1985; Zeitlin and Westwood 1986) serves to highlight the special need for advertising research in the area of negative emotions. Because Izard's taxonomy is dominated by negative emotions, it seems to offer a useful starting point for further work.


Aaker, David A. and Donald E. Bruzzone (1985), "Causes of Irritation in Advertising," Journal of Marketing, 49 (Spring), 47-57.

Aaker, David A., Douglas M. Stayman and Michael R. Hagerty (1986), "Warmth in Advertising: Measurement, Impact, and Sequence Effects," Journal of Consumer Research, 12 (March), 365-381.

Abelson, Robert P., Donald R. Kinder, Mark D. Peters and Susan T. Fiske (1982), "Affective and Semantic Components in Political Person Perception," Journal of Personality and Social Psychology, 42 (April), 619-630.

Allen, Chris T. and Thomas J. Madden (1985), "A Closer Look at Classical Conditioning," Journal of Consumer Research, 12 (December), 301-315.

Batra, Rajeev and Michael L. Ray (1986), "Affective Responses Mediating Acceptance of Advertising," Journal of Consumer Research, 13 (September), 234-249.

Bozinoff, Lorne and Morry Ghingold (1983), "Evaluating Guilt Arousing Marketing Communications," Journal of Business Research, 11 (June), 243-255.

Fuenzalida, C., R. N. Emde, B. J. Pannabecker and C. Stenberg (1981), "Validation of the Differential Emotions Scale in 613 Mothers," Motivation and Emotion, 5 (March), 37-45.

Havlena, William J. and Morris B. Holbrook (1986), "The Varieties of Consumption Experience: Comparing Two Typologies of Emotion in Consumer Behavior," Journal of Consumer Research, 13 (December), 394-404.

Holbrook, Morris B. and John O'Shaughnessy (1984), "The Role of Emotion in Advertising," Psychology and Marketing, 1 (Summer), 45-64.

Holbrook, Morris B. and Richard A. Westwood (1987), "The Role of Emotion in Advertising Revisited: Testing A Typology of Emotional Responses," in Cognitive and Affective Responses to Advertising, Pat Cafferata and Alice M. Tybout, eds., Lexington, MA: D. C. Heath, forthcoming.

Isen, Alice M. (1984), "Toward Understanding the Role of Affect in Cognition," in Handbook of Social Cognition: Vol. III, Robert S. Wyer, Jr. and Thomas K. Srull, eds., Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 179-236.

Izard, Carroll E. (1977), Human Emotions, New York, NY: Plenum Press.

Kotsch, William E., David W. Gerbing and Lynne E. Schwartz (1982), "The Construct Validity of the Differential Emotions Scale as Adapted for Children and Adolescents," in Measuring Emotions in Infants and Children, Carroll E. Izard, ed., New York, NY: Cambridge University Press, 251-278.

Lutz, Richard J. (1985), "Affective and Cognitive Antecedents of Attitude Toward the Ad: A Conceptual Framework," in Psychological Processes and Advertising Effects: Theory, Research and Application, Linda F. Alwitt and Andrew A. Mitchell, eds., Hillsdale, IN: Lawrence Erlbaum Associates, Publishers, 45-63.

MacKenzie, Scott B., Richard J. Lutz and George E. Belch (1986), "The Role of Attitude Toward the Ad as a Mediator of Advertising Effectiveness: A Test of Competing Explanations," Journal of Marketing Research, 23 (May), 135-143.

McHugo, Gregory J., Craig A. Smith and John T. Lanzetta (1982), "The Structure of Self-Reports of Emotional Responses to Film Segments," Motivation and Emotion, 6 (December), 365-385.

Mosher, Donald L. and Barbara B. White (1981), "On Differentiating Shame and Shyness," Motivation and Emotion, 5 (March), 61-74.

Olson, Jerry C. (1982), "Toward a Science of Consumer Behavior," in Advances in Consumer Research: Vol IX, Andrew A. Mitchell, e i., Ann Arbor, MI: Association for Consumer Research, v-x.

Peter, J. Paul and Gilbert A. Churchill, Jr. (1986), "Relationships Among Research Design Choices and Psychometric Properties of Rating Scales: A Meta-Analysis," Journal of Marketing Research, 23 (February), 1-10.

Plutchik, Robert (1980), Emotion: A Psychoevolutionary Synthesis, New York, NY: Harper and Row.

Sheth, Jagdish N. (1982), "Consumer Behavior Surpluses and Shortages," in Advances in Consumer Research: Vol. IX, Andrew A. Mitchell, ed., Ann Arbor, MI: Association for Consumer Research, 13-16.

Smith, Craig A. and Phoebe C. Ellsworth (1985), "Patterns of Cognitive Appraisal in Emotion," Journal of Personality and Social Psychology, 48 (April), 813-838.

Tomkins, Sylvan S. (1980), "Affect as Amplification: Some Modifications in Theory," in Emotion: Theory, Research and Experience: Vol 1, Robert Plutchik and Henry Kellerman, eds., New York, NY: Academic Press, 141-164.

Zeitlin, David M. and Richard A. Westwood (1986), "Measuring Emotional Response," Journal of Advertising Research, 26 (October/November), 34-44.