An Investigation of the Attribute Basis of Emotion and Related Affects in Consumption: Suggestions For a Stage-Specific Satisfaction Framework

Richard L. Oliver, Vanderbilt University
ABSTRACT - Emotions having the potential to be evoked in the consumption experience were related to attribute-specific and overall judgments of satisfaction/ dissatisfaction in an effort to more fully understand the role of emotion in consumption. Results of a multidimensional scaling analysis of the attribute dimensionality of automobile purchases and a property-fitting of emotions and satisfaction measures revealed two attribute dimensions and meaningful interpretations for the position of satisfaction/ dissatisfaction, enjoyment, interest, surprise, and general negative affect on this space. The attribute and emotion solution suggests a consumption-stage interpretation of the evocation of emotions in usage. Propositions consistent with this framework are offered for future work in the area.
[ to cite ]:
Richard L. Oliver (1992) ,"An Investigation of the Attribute Basis of Emotion and Related Affects in Consumption: Suggestions For a Stage-Specific Satisfaction Framework", in NA - Advances in Consumer Research Volume 19, eds. John F. Sherry, Jr. and Brian Sternthal, Provo, UT : Association for Consumer Research, Pages: 237-244.

Advances in Consumer Research Volume 19, 1992      Pages 237-244

AN INVESTIGATION OF THE ATTRIBUTE BASIS OF EMOTION AND RELATED AFFECTS IN CONSUMPTION: SUGGESTIONS FOR A STAGE-SPECIFIC SATISFACTION FRAMEWORK

Richard L. Oliver, Vanderbilt University

[The author thanks Wayne S. DeSarbo, Graduate School of Business, University of Michigan, for his assistance with the scaling analysis used here, and the Dean's Fund for Faculty Research of the Owen Graduate School of Management of Vanderbilt University for providing partial support for this project.]

ABSTRACT -

Emotions having the potential to be evoked in the consumption experience were related to attribute-specific and overall judgments of satisfaction/ dissatisfaction in an effort to more fully understand the role of emotion in consumption. Results of a multidimensional scaling analysis of the attribute dimensionality of automobile purchases and a property-fitting of emotions and satisfaction measures revealed two attribute dimensions and meaningful interpretations for the position of satisfaction/ dissatisfaction, enjoyment, interest, surprise, and general negative affect on this space. The attribute and emotion solution suggests a consumption-stage interpretation of the evocation of emotions in usage. Propositions consistent with this framework are offered for future work in the area.

Work on the role of emotion in consumption is beginning to intrigue consumer researchers. Prompted by calls for greater emphasis on non-cognitive elements of purchasing and consumption phenomena (e.g., Holbrook and Hirschman 1982), researchers have, of late, begun to investigate this research domain (Havlena and Holbrook 1986; Holbrook 1986; Holbrook et al. 1984; Oliver 1989, 1990; Westbrook 1987; Westbrook and Oliver 1991). The present paper continues this tradition in the following manner. First, it examines the dimensionality of attribute reactions (satisfactions in the present case) in consumption. Second, it relates reported emotional experiences in consumption to the derived dimensions, illustrating the capacity for attributes to evoke emotion. And third, overall satisfaction and dissatisfaction are positioned within the emotion and attribute configuration.

Background

Prior to recent years, few writers had considered responses other than satisfaction/dissatisfaction to the consumption experience. In fact, few had speculated as to whether satisfaction was an emotion or not. One exception is Hunt (1977) who referred to satisfaction as "an evaluation of an emotion." Two streams of research are available to provide insight to this issue. The first, represented by the work of Leavitt (1977), Maddox (1981), and Swan and Combs (1976), relates to the dimensionality of satisfaction attributes. The second, more recently, investigates the emotional correlates of consumption and includes the works of Holbrook (Havlena and Holbrook 1986; Holbrook et al. 1984) and Westbrook (1987; Westbrook and Oliver 1991) with respect to overall satisfaction with products. Interestingly, the two streams of work do not overlap in that the "attribute basis" of the earlier works is not reflected in the emotional basis of the more recent writings.

In the following discussion, these perspectives are reviewed to properly frame the present study. First, literature on the attribute basis of satisfaction and dissatisfaction is discussed in an effort to explain how others have viewed attribute categorizations. Second, works on emotion in consumption are reviewed in an effort to position satisfaction relative to other emotions and, to some extent, respond to Hunt's (1977) query about whether satisfaction is an emotion or the evaluation of emotional responses in purchasing. Finally, the present effort links attribute satisfaction, overall satisfaction, and emotion in a manner suggestive of an emerging framework. At present, only Oliver's (1989) theoretical framework containing both satisfaction and emotion is available to researchers. This conceptualization does not address the role of product attributes, however.

Prior Works on the Attribute Basis of Satisfaction/Dissatisfaction

Three studies are known which have attempted to propose or test the role of attributes in product satisfaction judgments. In an investigation of responses to newly acquired apparel, Swan and Combs (1976) argued that attributes can be classified into "satisfiers" and "dissatisfiers" based on whether the product features were expressive or instrumental. Instrumental apparel features were defined as physical product attributes such as "wash and wear," while expressive features were non-product characteristics such as a friend's admiration. Using a method wherein subjects recalled satisfying and dissatisfying clothing usage situations and the reasons for their (dis)satisfaction, the authors found that more expressive than instrumental attributes were mentioned in satisfying scenarios, while instrumental attributes tended to be mentioned more frequently in dissatisfying scenarios.

Maddox (1981) replicated this approach on a larger selection of products and found that, if one restricts analysis to personal care items (including clothing), the separate factor interpretation is supported. However, if other products such as durables are considered, the results tend to appear in a direction opposite to that proposed by Swan and Combs (1976). Maddox suggested that the instrumental/expressive breakdown is flawed because the results appear to be product-specific.

In an alternative approach, Leavitt (1977) used the Herzberg (1966; Herzberg, Mausner, and Snyderman 1959) distinction between intrinsic and extrinsic work factors to classify a variety of marketing mix descriptions (product, price, advertising, and place) into intrinsic (product) and extrinsic (price, advertising, and place) categories. When correlated with separate measures of satisfaction and dissatisfaction, Leavitt found virtually identical correlations (neglecting signs) across categories and concluded that the intrinsic/extrinsic hypothesis was not supported.

Thus, current knowledge about the attribute basis for satisfaction/ dissatisfaction is lacking. Neither the instrumental/expressive nor the intrinsic/extrinsic breakdown proposed in prior works has proven fruitful. Because the former has been unsuccessfully replicated and because the latter showed little promise as a classification scheme for consumer products, these perspectives were not pursued here. Rather, an inductive attempt was made to shed light on the attribute satisfaction phenomenon in consumption.

Studies of the Emotional Content of Product Experiences

To date, four published studies have examined the potential emotional basis underlying experiences with products. The first, by Holbrook et al. (1984) examined the Mehrabian and Russell (1974) pleasure-arousal-dominance (PAD) emotional typology in the context of video games. Their results showed that mastery of the game affected pleasure and dominance perceptions, but not arousal. In a comparison of the PAD framework with eight discrete emotions proposed by Plutchik (1980), Havlena and Holbrook (1986) found that the Mehrabian and Russell PAD typology resulted in more attractive psychometric properties (dimensionality, variance explained, etc.) over some 35 hypothetical consumption experiences. However, neither overall nor attribute-specific satisfaction was investigated in either study.

Westbrook (1987), correlated responses to Izard's (1972, 1977) Differential Emotions Scale, consisting of ten discrete emotions (to be discussed), with various satisfaction and dissatisfaction measures regarding automobiles and cable TV and posited that consumers form two summary affect states, one based on the positive affects in consumption and the other on the negative affects. His results showed that the affects of joy and interest loaded on a factor separate from that of anger, disgust, and contempt, and that these factors were relatively uncorrelated. Moreover, both were related to satisfaction in the expected direction although Westbrook concluded that satisfaction was not fully represented by these emotional dimensions.

In the latest paper, Westbrook and Oliver (1991) replicated Izard's (1977) typology on a new sample of automobile ownership, to study further the dimensionality of consumption. Their results modified the earlier Westbrook (1987) findings somewhat in that they found a broader negative dimension which they labeled "hostility," a positive dimension combined with surprise or "pleasant surprise," and an interest dimension. These dimensions explained approximately 40-45% of the variance in various satisfaction measures.

Thus, these prior works suggest that emotion may play a distinct role in postexposure processes including satisfaction/dissatisfaction. However, the evidence to date is not able to suggest a correspondence between specific states of satisfaction/dissatisfaction, the emotions evoked in consumption, and the attribute basis evoking these emotions. In short, the emotion literature speaks to one link in the three-part attribute-emotion-satisfaction scheme, while prior literature on the categorization of attributes reflects a second -- that between attributes and satisfaction. Missing, however, is an integrated analysis of the attribute basis for emotion and of the relation of satisfaction to these variables.

Such an analysis would perform three functions in efforts to understand how emotions operate within the product consumption experience. First, it would suggest the tendency for attribute experiences to evoke specific types of emotions. Second, it would position satisfaction and, perhaps separately, dissatisfaction within the attribute profile, thus suggesting the tendency for attributes to promote either satisfaction or dissatisfaction. Finally, by positioning satisfaction and dissatisfaction within the emotion-on-attribute space, the correspondence of satisfaction/dissatisfaction to specific emotions can be observed. This would permit an emotional interpretation of some common satisfaction and dissatisfaction measures.

Thus, the goals of this paper are to: (a) spatially describe an attribute configuration of attribute satisfactions, (b) fit an emotional typology on this space to suggest the correspondence between attributes and their potential to evoke emotions, and (c) study satisfaction/dissatisfaction as a property of emotional experience and attribute experience, particularly with regard to the juxtaposition of satisfaction vis-a-vis the emotions present in consumption experiences.

METHOD

To address these objectives, a methodological approach and research vehicle were needed with the flexibility to accommodate adequate numbers of attributes and emotions within the same design and analysis. Precedent for this problem has been established in the works of Westbrook (1987; Westbrook and Oliver 1991) and Havlena and Holbrook (1986). The Westbrook studies used Izard's (1977) Differential Emotions Scale containing ten fundamental emotions in automobile ownership and consumption. This approach is combined with the Havlena and Holbrook scaling solution to respond to the study objectives.

Sample

The sample consisted of 125 owners of recently purchased automobiles contacted in shopping centers of a major northeastern city. Details of the sampling procedure appear in Westbrook and Oliver (1991).

Measures

Attribute Satisfactions. To achieve a reasonable list of attributes, Consumer Reports and focus groups of graduate and adult evening school students were consulted to determine salient automobile features. Consideration of the overlap and redundancy in the frequency of attributes mentioned across sources produced a list of 19 features. The attributes used were acceleration, body integrity (no squeaks, rattles), fuel economy, handling, image, interior features, luggage space, noise, power, price, quality, repair frequency, ride, roominess, safety, service, sound system, styling, and uniqueness.

Because automobile "consumption" is an ongoing process and not a one-time purchase/usage experience, allowance was made for the possibility that the respondents would have observed favorable and unfavorable experiences with the same attribute. For example, although initial fuel economy may have been poor, later improvements may have occurred through servicing and normal "break-in." Thus, consumers could have been both satisfied and dissatisfied with the same attribute. To account for this, subjects were asked to state the degree to which each attribute provided satisfying and dissatisfying experiences on six-point scales ranging from none to a great amount. The attribute satisfaction score used here was the net of the satisfying and dissatisfying scores.

Emotions. Because of its validated properties and the precedent set by Westbrook (1987), Izard's DES II was used in the present study. Specific scale items represent the ten fundamental emotions of interest, enjoyment, surprise, sadness, anger, disgust, contempt, fear, shame, and guilt. This 30-item scale, consisting of three items per emotion, appears in two forms. It can be used as a "state" measure whereby subjects indicate their intensity of emotion and as a "trait" measure whereby subjects indicate the frequency of experiencing the emotion. Whereas Westbrook and Oliver (1991) used the frequency scales, the intensity scales are used here because the present focus studies attribute experience which Westbrook and Oliver did not address. Since some very important attribute experiences may occur only once (e.g., an intensely negative service experience) and still impact satisfaction greatly, it was felt that intensity would be more meaningful in the attribute context. Moreover, the original DES, as well as Westbrook's (1987) study used intensity measures and greater scale validation is available for this version. Izard (1977) reports original test-retest reliabilities for the ten scales ranging from .68 to .87.

Satisfaction. Because separate measures of satisfaction and dissatisfaction were desired, an alternative to the conventional bipolar scale was needed. Thus, a twelve-item Likert scale expanded from Oliver (1980; Oliver and Westbrook 1982) was adapted to the present study. This scale consists of a set of statements reflecting agreement/disagreement with phrases such as "happy with product," "right decision," etc. To provide a measure of satisfaction only, all items were worded in the positive. A separate negative scale was constructed by repeating the twelve items in the negative (e.g., unhappy with product, wrong decision). Because these are agreement scales, it was not expected that the subjects would simply give complementary answers to the two parallel scales. For example, a subject could conceivably indicate strong agreement with "happy with product," but indicate neutrality as opposed to strong disagreement with unhappiness.

Alpha reliability estimates for the ten emotion intensity scales range from .70 to .94 and are similar to those for the frequency scales in Westbrook and Oliver (1991). Reliabilities for the satisfaction and dissatisfaction scales were very high (.98 and .97 respectively), indicating strong agreement over items in these scales.

Analysis

The two-way matrix of attribute ratings (125 subjects by 19 attributes) was row-standardized to equate subjects' variances and means so that idiosyncratic anchor points were removed and submitted to MDPREF (Carroll 1972), a vector MDS model. Here, subjects are represented as vectors emanating from the origin of the space, and attributes are represented as points. The projection or scalar products of these attribute points onto subject vectors renders a graphical summary of how each subject's level of satisfaction/dissatisfaction derives with respect to each attribute. Since MDPREF produces orthogonal dimensions, one can examine a scree plot of the variance accounted by dimension to select the most parsimonious interpretation

Once the joint space of subject vectors and attribute points is estimated, one can use simple regression procedures to property-fit (cf. Green and Rao 1972) the ten emotions from the DES individually. This would indicate the nature of the relationship between the ten scales and the derived stimulus space. Similarly, one can perform the same property-fitting application with the satisfaction and dissatisfaction scales. The net result, therefore, is a graphical summary of the internal components of attribute satisfaction and how these derived components are related to the various emotions and overall satisfaction/dissatisfaction.

FINDINGS

The MDPREF analysis of the attribute satisfaction scores revealed two orthogonal dimensions explaining 42% and 15% of the variance respectively. Based upon interpretation and scree tests, the remaining dimensions were found to offer no new insight to the analysis. The 57% of variance explained by the first two dimensions is quite good considering the large number of subjects. The Figure presents the two derived dimensions where both the lengths of the attribute points and subject vectors were normalized for convenience. For ease of illustration, the subject and subsequent emotion vectors are not shown. Rather, the end points (termini) are represented as hollow dots for subjects and crosses for the emotions and satisfaction variables.

The horizontal dimension (Dimension 1) consists of satisfaction with instrumental performance features (i.e., those which provide ongoing dynamic performance, satisfaction, and pleasure such as ride, acceleration, stereo quality, and styling), while the vertical dimension (Dimension 2) consists of satisfaction with less variable "constants" (e.g., one-time events or factors considered in the initial purchase decision such as price, safety, quality, and size). Note that these dimensions do not fit either the notion of "satisfiers" or "dissatisfiers," or of instrumental/expressive or intrinsic/extrinsic attribute classifications as described earlier. The data reveal, further, that all attributes are essentially satisfying in this sample in that they are positively situated on the dimensions with the exception of fuel economy. Apparently, satisfaction with fuel economy is at odds with satisfaction with the other attributes on Dimension 1. This is easily explained, however, as satisfaction with features such as acceleration and power on this dimension can only come at the expense of fuel economy.

FIGURE

TWO-DIMENSIONAL JOINT SPACE PLOT

The outer circle of subject vector termini represents the subjects' respective positions vis-a-vis their net satisfaction with specific attributes. Generally, the sample appears quite diverse with regard to individuals' satisfaction and dissatisfaction with the attribute dimensions of the car. For example, subjects scoring positively on Dimension 1 apparently derive satisfaction from the performance characteristics of the car at the expense of fuel economy while those scoring negatively on this dimension derive satisfaction from high fuel economy at the expense of performance.

Correlations between the derived attribute dimensions and the emotion intensity ratings are shown in the Table. The ten emotions are shown at the vector termini on the outer normalized radius circle. This mapping reveals four distinct sets of emotional profiles, namely interest, enjoyment, surprise, and general negative affect. Interestingly, the attributes tend to group on two of these, as the following discussion describes.

Enjoyment appears to represent more of the attributes than any other dimension profile although, interestingly, only satisfaction with (the absence of) noise and squeaks appear directly related to this dimension. Generally, enjoyment correlates more highly with the attributes on Dimension 1 (dynamic performance) than on Dimension 2.

TABLE

CORRELATIONS BETWEEN THE ATTRIBUTE DIMENSIONS AND THE CONSUMPTION RESPONSES

Interest also appears to be evoked primarily by the Dimension 1 variables and displays a modestly negative correlation with Dimension 2. Apparently, the subjects in this study derived interest from the ongoing performance characteristics of their cars, perhaps because these are capable of maintaining interest. Surprise, in contrast, is captured only by Dimension_2 where satisfaction levels of the attributes are associated with a lack of surprise. Thus, unsurprising levels of price, repairs, service, and roominess, for example, tend to characterize this perspective. Finally, all of the negative affects including fear, sadness, disgust, contempt, and anger group together. Somewhat more representative of Dimension 2, these affects do not appear 180 degrees apart from enjoyment, a purer form of positive affect. Anger, in particular, is oriented more closely with Dimension 2.

When satisfaction and dissatisfaction were fit in this same space, satisfaction is shown in the Figure to share more in common with enjoyment than with interest. It also appears that satisfaction is more representative of the attribute satisfactions than is enjoyment. Alternatively, dissatisfaction is more moderately positioned than are the negative affects, correlating almost equally with Dimensions 1 and 2.

DISCUSSION

The discussion proceeds in three sections. First, the attribute satisfaction space as it relates to the solution found here is discussed. Second, the relation between these attributes and the ten discrete emotions are elaborated. Third, satisfaction is discussed, both with regard to its relation to the attributes and to the emotions. At each stage in this process, research issues are proposed to stimulate further work in the area.

Attribute Dimensions

The data suggest two groups of attributes, those which relate to satisfaction with the continuing performance of the product such as acceleration, ride, fuel economy, etc. and those which are one-time (e.g., price), infrequently accessed (e.g., service), or unchanging characteristics (e.g., safety, quality). This dynamic/static description of attribute satisfaction was evident in this study due to the nature of the product. Unlike one-time or short-life products (e.g., foods, entertainment, disposables), automobiles have the capacity to display both "purchase phase" and "usage phase" qualities which have the capacity to evoke a wider range of emotions. Conceivably, there exist other dimensions of performance based on the temporality or emergence of different performance stages.

For example, an initial stage of acquisition attributes (e.g., price, terms) is proposed which includes the known attributes of the product (e.g., size). This is followed by certain "unknowns" which manifest themselves very shortly (e.g., initial defects). The next stage (if appropriate to the product) is a longer period of "steady-state" performance whereby the variable performance dimensions are observed to change and subjective attributes (e.g., image) are reinterpreted. Thus, the following proposition is offered:

Proposition 1. Satisfaction from consumption and, hence, satisfaction with attribute performance may be viewed in stages, each of which has the potential for evoking diverse emotions and further cognition. The first is the acquisition and known characteristics stage where various properties of the product are known with certainty (e.g., size, color). This may be followed by the initial discovery of unknowns as in the number of factory defects and service experiences with the dealer. Finally, there may exist longer term consumption experiences where basically known elements of the product are monitored for change.

Note that the framework suggested by the present data does not correspond to separate-factor theories as previously discussed. Rather than "satisfiers" or "dissatisfiers," consumption-stage attributes capable of producing either satisfaction or dissatisfaction appear evident. Nor do the expressive/instrumental categories, operationalized as product and non-product benefits, correspond to the present findings. It appears, instead, that product features and external features exist together on both dimensions (e.g., power and image on Dimension 1, roominess and service on Dimension 2). This does not suggest that the prior conceptualizations are incorrect, as a different methodology was used here. Rather, it suggests that alternative interpretations are plausible.

Emotions

The location of the emotions on the attribute satisfaction space can be viewed as consistent with the attribute interpretation above. Whereas Westbrook and Oliver (1991) found dimensions of hostility, pleasant surprise, and interest, an alternative analysis of intensity ratings from the same data set shows that the emotions of enjoyment, interest, surprise, and negative affect appear to operate as responses to underlying attribute dimensions. Thus, these findings differ from those of the Westbrook and Oliver frequency data in that enjoyment and surprise may initially be separate responses in consumption that become additive over time (as consumption experiences become more frequent -- i.e., accumulate) to produce the pleasant surprise dimension.

Of greater interest is the correspondence of the emotions to the attribute space. Specifically, enjoyment was fairly moderately situated among the attributes although it was more highly related to Dimension 1. Interest, however, was clearly aligned with the first dimension, representing dynamic performance and suggesting that continued interest in a product relates to satisfactions with ongoing performance. In contrast, surprise appears to relate solely to Dimension 2 where less (or no) surprise is related to satisfaction with the one-time or static aspects of the product. Finally, general negative affect appears to reflect the negative dimension of enjoyment although these extremes are not 180 degrees apart. Rather, the group of negative affects is approximately 155 degrees from enjoyment. Anger, more so than the other negative emotions, is a Dimension 2 affect, suggesting that dissatisfaction on the Dimension 2 attributes is capable of provoking anger. Alternatively, dissatisfaction on the Dimension 1 attributes is somewhat more likely to evoke the more general negative affects.

The implications of these findings are rather interesting for it appears that the nature of the attributes or, more accurately, the relevance of an attribute to a consumption stage, may evoke certain emotions characteristic of that stage. For example, the data suggest the possibility that dissatisfaction with known attributes may provoke anger while dissatisfaction with unknown but subsequently observed attributes may produce surprise or anger and surprise. Alternatively, dissatisfaction with the more dynamic attributes possibly may generate less specific forms of negative affect. Based on these observations, the following proposition is offered:

Proposition 2. There may exist consumption-stage-specific affects which are evoked by the attributes unique to or observed in these stages. Initial stage consumption, consisting of observation of previously known and soon to be observed attributes, produces, in the dissatisfying case, short duration but fairly intense emotions such as surprise and anger. Attribute satisfaction in this initial stage may produce similar types of positive emotions of the "surprisingly good" variety such as delight and excitement. Alternatively, later consumption stages may produce emotions capable of being sustained along with the dynamic nature of the later stage attributes. Such emotions in the satisfaction case appear to be interest and enjoyment. Parallel emotions in the dissatisfaction case may be described in general terms such as sadness or resignation.

Satisfaction/Dissatisfaction

Two issues are relevant to the role of satisfaction/ dissatisfaction in the present framework. The first concerns the relation between attribute satisfaction and overall satisfaction while the second concerns the underlying emotional basis for perceptions of satisfaction. The data suggest that overall satisfaction includes a fairly global summary judgment of attribute satisfaction. This perspective is suggested by the very central position of the satisfaction vectors among the attribute satisfactions. Thus, it appears that satisfaction is a summary attribute phenomenon, at least in part.

Finally, the results suggest that satisfaction shares commonalities with, but differs from enjoyment (pleasure). This is consistent with Oliver's (1989) position that pleasure is but one of perhaps five satisfaction prototypes. Interest and surprise, other somewhat distinct affects found here, are not well represented on the satisfaction vector. This suggests the possibility that other consumption emotions, and perhaps cognition, coexist with satisfaction/dissatisfaction and have separate, but equally important effects on later purchase behavior. These observations suggest the following:

Proposition 3. Satisfaction and dissatisfaction reflect the general affective tone of the consumption experience, appearing to equally represent static and dynamic purchase elements. However, other purchase emotions (e.g., surprise) may coexist with satisfaction so that a fuller range of postpurchase psychological responses is needed to properly describe the cognitive and affective reactions to consumption.

Limitations

Given the early stages of this research stream, limitations encountered here exist which will serve to alert future researchers to potential pitfalls. One such pitfall is the restriction imposed by a limited sample of respondents and a single product. Automobiles are known to be a high involvement product category which has the capacity to evoke greater levels of both affect and cognition. Low involvement products would not be expected to elicit high levels of processing and may result in the "contentment mode" of satisfaction as proposed by Oliver (1989). As a result, the findings here may not replicate on frequently purchased, everyday products.

A second limitation is the necessity for a greater sampling of emotions. The ten "fundamental" emotions from Izard (1977) were selected because of the precedent set by Westbrook (1987). One problem with this scale, observed in retrospect, is that, although the subjects were largely pleased with their cars, seven of the ten affects in the Izard scale are negative in character. Thus, other typologies, consisting of other sets of emotions need to be studied, perhaps in conjunction with those of Izard. Plutchik's (1980) framework as used by Havlena and Holbrook (1986) is one such example.

Another limitation concerns the methodology and analysis, specifically with regard to the use of the property-fitting procedures in MDPREF. For example, the vector model representation could reflect a misspecification error if an ideal point model were the true underlying structural form. Thus, the suggested research propositions are both data-dependent and specific to the methodology and analysis used here and are not theory-driven. Until more work is done in this area, however, early representations such as that found here will hopefully serve to stimulate future efforts.

CONCLUSION

Drawing on the precedent set by other works in the satisfaction and emotion areas, an effort was made to provide early insights into the role played by emotion in the consumption experience with particular emphasis on attribute and overall satisfaction. The data suggested attribute dimensions differing on a usage stage basis and a mapping of emotion suggestive of early stage and later stage experiences. Satisfaction/dissatisfaction was found to mimic a general affect, more representative of enjoyment than of interest or surprise. Propositions are offered, suggesting a framework for the future study of emotions deriving from consumption and its stages.

REFERENCES

Carroll, J. Douglas (1972), "Individual Differences and Multidimensional Scaling," in Multidimensional Scaling: Theory and Applications in the Behavioral Sciences, Vol. I: Theory, Roger N. Shepard, A. Kimball Romney, and Sara Beth Nerlove, eds. New York: Seminar Press, 105-155.

Green, Paul E. and Vithala R. Rao (1972), Applied Multidimensional Scaling: A Comparison of Approaches and Algorithms, New York: Holt, Rinehart and Winston.

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.

Herzberg, Frederick (1966), Work and the Nature of Man, Cleveland: World Publishing.

Herzberg, Frederick, Bernard Mausner, and Barbara B. Snyderman (1959), The Motivation to Work, 2nd Ed., New York: John Wiley & Sons.

Holbrook, Morris B. (1986), "Emotion in the Consumption Experience: Toward a New Model of the Human Consumer," in The Role of Affect in Consumer Behavior: Emerging Theories and Applications, Robert A. Peterson, Wayne D. Hoyer, and William R. Wilson, eds., Lexington, MA: D. C. Heath, 17-52.

Holbrook, Morris B., Robert W. Chestnut, Terence A. Oliva, and Eric A. Greenleaf (1984), "Play as a Consumption Experience: The Roles of Emotion, Performance, and Personality in the Enjoyment of Games," Journal of Consumer Research, 11 (September), 728-739.

Holbrook, Morris B. and Elizabeth C. Hirschman (1982), "The Experiential Aspects of Consumption: Consumer Fantasies, Feelings, and Fun," Journal of Consumer Research, 9 (September), 132-140.

Hunt, H. Keith (1977), "CS/D -- Overview and Future Research Directions," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction, H. Keith Hunt, ed., Cambridge, MA: Marketing Science Institute, 455-488.

Izard, Carroll E. (1972), Patterns of Emotions, New York: Academic Press.

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

Leavitt, Clark (1977), "Consumer Satisfaction and Dissatisfaction: Bi-Polar or Independent?," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction, H. Keith Hunt, ed., Cambridge, MA: Marketing Science Institute, 132-149

Maddox, R. Neil (1981), "Two-Factor Theory and Consumer Satisfaction: Replication and Extension," Journal of Consumer Research, 8 (June), 97-102.

Mehrabian, Albert and James A. Russell (1974), An Approach to Environmental Psychology, Cambridge, MA: The MIT Press.

Oliver, Richard L. (1980), "A Cognitive Model of the Antecedents and Consequences of Satisfaction Decisions," Journal of Marketing Research, 17 (November), 460-469.

Oliver, Richard L. (1989), "Processing of the Satisfaction Response in Consumption: A Suggested Framework and Research Propositions," Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, 2, 1-16.

Oliver, Richard L. (1990), "Cognitive, Affective, and Attribute Bases of Usage/ Postpurchase Response." Paper presented at the Annual Conference of the Association for Consumer Research, October 1990.

Oliver, Richard L. and Robert A. Westbrook (1982), "The Factor Structure of Satisfaction and Related Postpurchase Measures," in New Findings on Consumer Satisfaction and Complaining, Ralph L. Day and H. Keith Hunt, eds., Bloomington, IN: Indiana University School of Business, 11-14.

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

Swan, John E. and Linda J. Combs (1976), "Product Performance and Consumer Satisfaction: A New Concept," Journal of Marketing, 40 (April), 25-33.

Westbrook, Robert A. (1987), "Product/Consumption-Based Affective Responses and Postpurchase Processes," Journal of Marketing Research, 24 (August), 258-270.

Westbrook, Robert A. and Richard L. Oliver (1991), "The Dimensionality of Consumption Emotion Patterns and Consumer Satisfaction," Journal of Consumer Research, 18 (June), 84-91.

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