The Power of Affective Reports in Predicting Satisfaction Judgments

Laurette Dube-Rioux, University of Montreal
[ to cite ]:
Laurette Dube-Rioux (1990) ,"The Power of Affective Reports in Predicting Satisfaction Judgments", in NA - Advances in Consumer Research Volume 17, eds. Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, Provo, UT : Association for Consumer Research, Pages: 571-576.

Advances in Consumer Research Volume 17, 1990      Pages 571-576

THE POWER OF AFFECTIVE REPORTS IN PREDICTING SATISFACTION JUDGMENTS

Laurette Dube-Rioux, University of Montreal

This paper reports a field study that empirically investigated the relative ability of affective reports and cognitive evaluations to predict consumer satisfaction judgments. The results of this research revealed that consumers' affective reports were highly predictive of the level of satisfaction. This effect was independent of, and more powerful than cognitive evaluations. The predictive value of affective reports and cognitive evaluations varied as a function of the emotional intensity of satisfaction judgments.

THE ROLE OF AFFECTIVE REPORTS IN PREDICTING SATISFACTION JUDGEMENTS

Research in consumer behavior has traditionally viewed satisfaction as an affective response to the comparison of actual consumption experiences with some internal cognitive standards. In most studies, consumers are asked to rate the performance of a product on a series of salient attributes and to express how this performance measures up to their expectations. Satisfaction reflects the outcome of a linear combination of discrepancies between actual performances and past expectations, and is directly related to the extent to which expectations are met (Oliver, 1980; Oliver and DeSarbo, 1988; Tse and Wilton, 1988; Woodruff, Cadotte and Jenkins, 1983). Typically, consumers' expectations reflect belief probabilities of attribute occurrence, corresponding to pre-purchase predictions of what the product performance will be (Bearden and Teel, 1983; Oliver, 1980; Westbrook, 1980).

The unique status of probabilistic expectations has recently been challenged by the inclusion of additional cognitive standards in models of consumer satisfaction. Best brand, ideal product performance, and equity-based comparison between costs and anticipated rewards have been empirically investigated (Cadotte, Woodruff and Jenkins, 1987; Tse and Wilton, 1988; Woodruff et al, 1983). Furthermore, the continuous and monotonic relationship between the magnitude of disconfirmation of expectations (or of any other standards), and the emotional intensity of satisfaction judgments has also been questioned. Woodruff and his collaborators (1983; 1987) have suggested that consumers' perceptual limitations create a "zone of indifference" surrounding cognitive expectations, so that positive or negative disconfirmation results only when a perceived performance falls outside this zone. The existence of these differentiated zones, i.e., zone of "indifference" and zones of "positive or negative emotional reactions" has yet to be demonstrated empirically. Nonetheless, recognizing a discontinuous function between expectancy disconfirmation and emotional intensity of satisfaction judgments suggests the possibility that there may be no uniform type of affective responses or cognitive processes involved in post-purchase responses to consumer goods and services.

Although these recent conceptual developments have helped to improve the predictive ability of consumer satisfaction models, their scope is still exclusively cognitive. Affective responses to consumption have thus far been seen only as the outcome of extensive combinatorial processes. No research yet has investigated the feelings consumers experience as they consume products as a unique source of information on the resulting satisfaction. (For an exception, see Westbrook, 1987). Because affective reports often reflect motivation more directly than cognitive evaluations, they may surpass the latter in predicting satisfaction judgments and post-purchase behavior. Moreover, if satisfaction judgmental in zones of differentiated emotional intensity, the relative predictive ability of affective reports and cognitive.evaluations may vary as a function of the location of a specific consumption experience.

We addressed these research issues in a field study. The objectives were to compare the relative power of affective reports and cognitive evaluations in predicting satisfaction judgments, and to investigate the relative efficiency of these affective and cognitive predictors as a function of the emotional intensity of satisfaction judgments.

The role of affect in satisfaction Judgments

Research findings from the areas of psychology and marketing have demonstrated that affective states influence many aspects of consumer behavior ( for a review see Gardner, 1985). For instance, it has been shown that manipulating affective states influences the way consumers process information ( for a review see Isen, 1987), and the level of satisfaction they express for products commonly used (Isen, Shalker, Clark, and Karp, 1978). We therefore hypothesized that affective states experienced during the consumption of goods or services would also directly influence post-purchase responses such as satisfaction and the likelihood of repeat purchase. Indeed, affective reports differ qualitatively from cognitive evaluations and may be more directly related to behavioral predictions. The role of affective reports in determining consumer satisfaction may become essential for products in which the experiential aspect of consumption is important (e.g., services and highly symbolic goods) -- these products hardly being reducible to a list of discriminable characteristics on which adequate cognitive evaluations can easily be made.

Some empirical evidence in a related domain supports the prediction that consumers' reports of their affective states may be superior to cognitive evaluations of the same product in predicting satisfaction. Drawing on the results of two national surveys, Abelson, Kinder, Peters, and Fiske (1982) demonstrated that affect patterns very powerfully predict political preferences. They found that the ability of affective reports to predict political preferences was independent of, and even surpassed, respondents' cognitive evaluations of a candidate. Thus, although feelings reflect one's cognitive appraisal to a certain extent, the information provided by affective and cognitive measures seems not to be redundant. In cognitive evaluations, respondents provide information about others, focusing on external stimuli, whereas they deliver information about the self in affective reports. Since affective reports concern the internal state of respondents, they may provide much more complex and realistic information than cognitive evaluations (Abelson et al,1982). Furthermore, it seems that affective responses are less subject to consistency pressure than are cognitive evaluations. Cognitive evaluations seem to be "semantically filtered" to a greater extent than are affective responses (Abelson et al, 1982). That is, the overall conception respondents attempt to convey is weighted heavily in the ratings of individual items of cognitive evaluations. As a result, cognitive evaluations are more likely to be under the organizing influence of pre-existing expectation i than are affective responses, which are a more "naive" and direct reflection of experiences. Therefore, we further hypothesized that the correlation between positive and negative components of cognitive evaluations would be higher than those of affective reports.

The relative role of affective reports and cognitive evaluations in satisfaction judgments

How different can satisfaction judgments be as a function of their emotional intensity? Are processes typically associated with neutral satisfaction judgments different from those associated with very positive or very negative affective responses to consumption experiences ? Extending the cognitive categorization approach (for a review see Cohen and Basu, 1987), we suggest that with repeated exposures to a product, consumers learn to divide their experiences into neutral or emotional categories to form satisfaction judgments. The structure of these two affective categories -neutral and emotional -- and the processes they involve differ. Fazio et al (1986) have shown that "affective" evaluation may indeed range in nature from a very "hot" affect ( when the attitude object is associated with a strong emotional response), to a "colder" and more cognitively-based judgment of affect. We suggest that emotional categories include consumption experiences that have generated conscious positive (cases of satisfaction), or negative (cases of dissatisfaction) affective responses, with attentional capacity being devoted to satisfaction judgments. Consumption experiences included in the neutral category are these having generated a "scripted behavior", reflecting simply the absence of any significant purchase or usage related problems. Such an affective response reflects the absence of dissatisfaction, and occurs when a consumption experience meets expectations. Gardner (1985) suggests that neutral affective responses rarely interrupt ongoing behaviors, simply coloring attentional processes and evaluations. Furthermore, Mandler (1982) suggests that a person may in fact be unable to report his or her true feelings for neutral affective responses further than a valuation of familiarity. Thus, we hypothesized that for neutral affective responses, cognitive evaluations would be superior to affective reports as predictors of satisfaction judgments. In addition, since both affective reports and cognitive evaluations result from a top-down processing of expectations, they should be more highly correlated in neutral than in emotional judgements. On the contrary, since the raise of emotional responses results from the bottom-up processing of affective reactions to some part of a consumption experience, we hypothesized that within this category, affective reports should be the best predictor of satisfaction.

METHOD

Subjects

Subjects were 52 customers (29 males and 23 females) recruited from three casual restaurants. The average age of subjects was 35. The three restaurants were mid-priced restaurants offering a diversified menu and a non-formal service style in a relaxed atmosphere.

Procedure

Subjects were intercepted at the end of their meal and were asked to fill out a questionnaire on consumer satisfaction. The questionnaire had to be completed before leaving the restaurant. Within three days of the actual dinner, subjects were asked to report the emotional intensity of their satisfaction with the dinner as a whole, and more specifically with the service. They were also asked about the frequency of neutral and emotional experiences they had with the service in casual restaurants at dinner time. These data were collected as part of a more extensive interview. Thirty six of the 52 original subjects completed the second part of the study.

Measures

The questionnaire included three multi-item scales. These scales measured satisfaction, affective reports and cognitive evaluations. Except for one item in the satisfaction scale, all measures specifically related to service.

Satisfaction measures: The ordinal scale of satisfaction judgment included five items. Subjects were asked to express how satisfied they were with the dining experience as a whole, and with the service in particular (1 = not at all satisfied, to 7 = completely satisfied). They were also asked the extent to which the service interaction met their needs and wants at this time (1 = extremely poor, to 7 = extremely well), the nature of their feelings about the service (Westbrook's delighted-terrible scale, Westbrook 1980), and how happy they would be to be served by the same person on another visit (1 = very happy, to 7 = very unhappy).

TABLE 1

RELIABILITY OF SCALES OF SATISFACTION, AFFECTIVE REPORTS, AND COGNITIVE EVALUATIONS

Affective reports: The scale of affective reports included five positive ("warm feelings", "enthusiastic", "being valued", "surprised", "interested"), and five negative ("irritated", "annoyed", "unpleased", "bored", "indifferent") items. Subjects were asked how strongly these feelings described their reaction to the service they just had. The scale ranged from 1 to 7 points (1 = not at all, 7 = very strongly). The adjectives were selected from those most frequently mentioned in a pre-test in which comparable subjects were asked to freely elicit how they would describe their own feelings toward servers in casual restaurants, in dinners that either left them indifferent or with very strong feelings (good or bad). The adjectives were listed in random order on the questionnaire.

Cognitive evaluations: Subjects were asked to indicate the degree to which a list of adjectives described their perception of the server who just waited on them. The list of descriptors included six positive items ("friendly", "cooperative", "knows what to do", "knows what to say", "attentive", "in control"), and five negative ("hesitant", "looks stressed", "demanding", "unpleasant", "distant"). The scale ranged from 1 to 7 points, being anchored by "not at all" and "very strongly" respectively. These items were selected on the basis of the same pre-test used for affective reports. The items were listed in random order on the questionnaire.

Affective categorization of consumption experiences: Subjects were asked to categorize the service they receive on the target dinner in one of three categories: (I) "the service interaction was o.k., and did not leave me with any unusual or particular feelings" (Neutral); (2) "the service interaction was positive, and left me with very good feelings" (Emotional-positive); (3) "the service interaction was negative and left me with very bad feelings" (Emotional-negative). This retrospective measure is a conservative assessment of consumers' actual emotional responses. Twelve interactions were categorized as neutral and 24 were emotional positive. No service interactions left any respondent with very bad feelings. Subjects were also asked the following question: "If you could think of 100 dinners you had in casual restaurants, in how many of these would the service interaction have been...? The description of neutral and emotional categories presented above followed.

RESULTS

Reliability of the scales

Table 1 presents the reliability scores for the scales of satisfaction, affective reports, and cognitive evaluations as well as for their positive and negative subscales. All series of measures present a good reliability. The negative items were reversed and different summative scores were computed for further analysis: overall satisfaction, overall affective reports, overall cognitive evaluations, as well as positive and negative affective reports, and positive and negative cognitive evaluations.

Predictive ability of affect reports in satisfaction judgments

Recall that we hypothesized that affective reports would significantly account for satisfaction judgments, over and above their effect through cognitive evaluations. The critical dependent variable to test this hypothesis was the satisfaction judgment expressed by subjects at the end of the meal. A factor analysis with oblique rotation was conducted on the multi-item scale of satisfaction judgment. It revealed a unidimensional structure. Only one factor had an eigenvalue bigger than one (3.83) and accounted for 64% of the variance. The analysis of satisfaction judgments was conducted on the summative score of these five items.

Multiple regression analysis was conducted to test the predictive ability of affective reports and cognitive evaluations. In conformity with our hypothesis, both affective and cognitive assessments significantly accounted for the variability in satisfaction judgments, affective reports being slightly superior to cognitive evaluations (Standardized b's: affect = .48, ;2 < .001; cognition = .39, p < .005). A regression model was also run to further investigate the specific contribution of positive and negative items within the affective and cognitive scales. Both positive and negative summative scores of affective reports accounted for a significant amount of the variance, whereas only positive items of the cognitive scale were significant (see table 2).

TABLE 2

REGRESSION COEFFICIENTS OF AFFECTIVE REPORTS AND COGNITIVE EVALUATIONS ON SATISFACTION RATINGS

Relative predictive ability of affective reports and cognitive evaluations within affective categories

Neutral consumption experiences constituted the bulk of our subjects' past experiences with service in casual restaurants (mean frequency = 60%). On average, 30% of the visits generated strong positive feelings, and only 10% left very bad feelings. Turning to the relative ability of affective reports and cognitive evaluations to predict satisfaction judgments within each category, recall that we hypothesized that cognitive evaluations would be the best predictors of satisfaction within the neutral category, being outperformed by affective reports in emotional categories. The results confirmed this hypothesis. The overall cognitive evaluation was the unique significant predictor of satisfaction in neutral judgments (Standardized b = .604, g < .060), whereas affective reports played the leading role in the case of satisfaction and dissatisfaction (Standardized b = .504, p < .024).

Correlational patterns

Interesting insights on possible differences in structures and processing strategies among affective categories are provided by the correlations presented in table 3. Because cognitive judgments are typically more "semantically filtered" than affective reactions, and are therefore more susceptible to the organizing influence of existing expectations, we predicted that positive and negative items of the cognitive evaluation scale would be more highly correlated than these of the affective reports. The expected pattern of correlations between positive and negative items (averaged across affective categories) was marginally significant (Cognitive evaluations: r= .71; affective reports: r = .51; Z = 1.67, p < .10).

We also observe that neutral and emotional categories do not present the same pattern of correlations between positive and negative items within the affective and the cognitive scales. For instance, as emotion arises in satisfaction judgments, positive and negative items of cognitive evaluations become more sharply correlated, whereas positive and negative items of the affective report scale become more independent ( respectively r = .75 and r =.42; Z = 1.70, p < .10). Satisfaction judgments in the neutral category do not show such a pattern (respectively r = .60 and r = .66; Z = .13, p < .40). These findings parallel Abelson et al's (1982) observation that with strong emotional commitment, not only subjects show a significant halo effect that tints all cognitive evaluations, but also, and somewhat paradoxically, they become more aware of both good and bad feelings they experience.

TABLE 3

CORRELATIONS BETWEEN SUMMATIVE SCORES OF AFFECTIVE REPORTS, COGNITIVE EVALUATIONS AND OVERALL SATISFACTION FOR NEUTRAL AND EMOTIONAL SATISFACTON JUDGEMENTS

DISCUSSION

In summarizing the major results of this research, we have found firstly that consumers' affective responses were highly predictive of their level of satisfaction, adding significant variance explanation over and above the part due to cognitive evaluations. Secondly, we have provided some empirical evidence for the existence of affective categories that organize satisfaction judgments on the basis of their emotional intensity.

Perhaps the most psychologically provocative result of this research is the different behavior of affective responses and cognitive evaluations in neutral judgments, which we termed absence of dissatisfaction, compared to consumption experiences that generated emotional responses, which we suggested to be real satisfaction and dissatisfaction judgments. Recall that satisfaction is more directly related to affective reports in emotional categories, and to cognitive evaluations (the measure most often used in traditional studies of consumer satisfaction) only in the neutral category. In addition, when consumer satisfaction judgments resulted from very good or very bad feelings, positive and negative items of the cognitive evaluation scale became more highly correlated, whereas the magnitude of the relation between good and bad items of affective reports decreased. We take this as an indication of increased attentional capacity being devoted to emotional responses in these instances that are not assimilated into preexisting expectations.

Can these results be explained on methodological grounds? This research presents limitations. First, it is a correlational study and observed differences may partly be due to subjects' self assignment to affective categories. Second, the constraints imposed by the small sample size also have to be acknowledged. The very low sample size in the neutral category in particular resulted in low statistical power in testing some hypotheses. In addition, the study will have to be replicated to include consumption experiences that generated negative affective states.

Are these results bound to the specific domain r of this research? We studied restaurants, and have focused on consumer satisfaction with the service. But we think that the two types of satisfaction judgments -- neutral, more top-down and theory driven -- versus emotional -- more bottom -up and data-driven -- can be observed in all types of consumption experiences. The likelihood of activation of one or the other types of processes will vary with the nature of the product, its importance, and situational differences.

This research shows that affective reports differ qualitatively from cognitive evaluations and may be more directly related to behavioral predictions. Using consumers' affective reports should further our understanding of the processes involved in satisfaction judgments, and our ability t to predict their outcome. Investigating the conditions and mechanisms mediating the effect of affective responses on satisfaction judgments should significantly contribute to a more accurate forecasting of consumer satisfaction and repeat purchase behavior.

REFERENCES

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, 619-630.

Bearden, William D. and Jesse E. Teel (1983), "Selected Determinants of Consumer Satisfaction and Complaints Reports", Journal of Marketing Research, 20 (November), 21-28.

Cadotte, Ernest R., Robert B. Woodruff, and Roger L. Jenkins (1987), "Expectations and Norms in Models of Consumer Satisfaction", Journal of Marketing Research, 24 (August), 305-314.

Cohen, Joel B. and Kunal Basu (1987), "Alternative Models of Categorization: Toward a Contingent Processing Framework", Journal of Consumer Research, 13 (March), 455472.

Fazio, Russel H. (1986), "How do attitudes guide behaviors?", in The Handbook of Motivation and Cognition: Foundation of Social Behavior, Richard M. Sorrentino and E. Tory Higgins (eds.), NY: Guilford Press, 204-243.

Gardner, Meryl P. (1985), "Mood States and Consumer Behavior: A Critical Review", Journal of Consumer Research, 12, 281-299.

Isen, Alice M. (1987), "Positive Affect, Cognitive Processes, and Social Behavior", in Advances in Experimental Social Psychology, Leonard Berkowitz (ed.), 20, 203-253.

Isen, Alice, M., T. Shalker, M. Clark, and L. Karp (1978), "Affect, Accessibility of Material in Memory and Behavior.: a Cognitive Loop?", Journal of Personality and Social Psychology, 36, 1-12.

Mandler, George (1982), "The structure of Value: Accounting for Taste", in Affect and Cognition: The 7th Annual Carnegee Symposium, Margaret S. Clarke and Susan T. Fiske (eds.), Hillsdale: V. J. Erlbaum, 3-36.

Miller, John A. (1977), "Studying Satisfaction, Modifying Models, Eliciting Expectations, Posing Problems and Making Meaningful Measurements", in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction, Keith H. Hunt (ed.), Cambridge, MA: Marketing Science Institute, 72-91.

Oliver, Richard L. (1980), "A Cognitive Model of the Antecedents and Consequences of Satisfaction Judgments", Journal of Marketing Research, 17 (November), 204-212.

Oliver, Richard L. and Wayne S. DeSarbo (1988), "Response Determinants in Satisfaction Judgments", Journal of Consumer Research, 14, 495 -507.

Tse, David K. and Peter C. Wilton (1988), "Models of Consumer Satisfaction: An Extension", Journal of Marketing Research, 25: May, 204212.

Westbrook, Robert A. (1980), "A Rating Scale for Measuring Product / Service Satisfaction", Journal of Marketing, 44 (fall), 68-72.

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

Woodruff, Robert B., Ernest R. Cadotte, and Roger L. Jenkins (1983), "Modeling Consumer Satisfaction Processes Using Experience-Based Norms", Journal of Marketing Research, 20 (August), 296-304.

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