Potential Moderators For Comparison Standards in Consumer Satisfaction Formation: Some Exploratory Findings

ABSTRACT - The extant literature suggest that consumers may use one or some of different comparison standards to evaluate actual product performance during consumer satisfaction formation. However, boundary conditions under which a particular standard operates have not been identified. This study explored such conditions by analyzing data collected over several product categories. Findings suggest that consumer involvement and product experience appear to influence the type of comparison standard used for evaluating product performance. Theoretical and managerial implications of these findings are discussed.

Citation:

Jong-Won Park and Jiho Choi (1998) ,"Potential Moderators For Comparison Standards in Consumer Satisfaction Formation: Some Exploratory Findings", in NA - Advances in Consumer Research Volume 25, eds. Joseph W. Alba & J. Wesley Hutchinson, Provo, UT : Association for Consumer Research, Pages: 124-131.

Advances in Consumer Research Volume 25, 1998      Pages 124-131

POTENTIAL MODERATORS FOR COMPARISON STANDARDS IN CONSUMER SATISFACTION FORMATION: SOME EXPLORATORY FINDINGS

Jong-Won Park, Korea University

Jiho Choi, Korea University

[We would like to thank Pulmuone Co. for their supporting the data collection and members of B.E.S.T. for their suggestions concerning interpretations of the results.]

ABSTRACT -

The extant literature suggest that consumers may use one or some of different comparison standards to evaluate actual product performance during consumer satisfaction formation. However, boundary conditions under which a particular standard operates have not been identified. This study explored such conditions by analyzing data collected over several product categories. Findings suggest that consumer involvement and product experience appear to influence the type of comparison standard used for evaluating product performance. Theoretical and managerial implications of these findings are discussed.

BACKGROUND

Consumer satisfaction has long been investigated by marketing and consumer researchers. In particular, a number of empirical studies have been published with regard to cognitive processes mediating consumer satisfaction (see Yi 1990 for a review). In these studies is typically employed a form of "standard-performance-disconfirmation" paradigm: (1) actual product performance is compared to an internal standard during satisfaction formation, and (2) the level of the internal standard, the level of actual product performance, and/or the magnitude of the difference between the two levels are determinants of a consumer satisfaction level. The literature suggest that the type of internal comparison standard is diverse. In some cases, for example, a brand expectation (the level at which the brand performance is expected to be) is such standad (e.g., Oliver 1980; Oliver and Linda 1981; Westbrook and Reilly 1983), and in other cases a product norm (the level at which a consumer feels the product performance should be: e.g., Boulding et al. 1993; Cadotte, Woodruff, and Jenkins 1987), ideal (the level at which the product performance is personally desired to be: e.g., Tse and Wilton 1988), or equity (the product performance level which is considered equitable to the price paid or effort invested: e.g., Oliver and DeSarbo 1988; Tse and Wilton 1988). Further, recent studies reported that more than one standards might be used simultaneously (e.g., Tse and Wilton 1988; Spreng, MacKenzie, and Olshavsky 1996). Then, it seems reasonable to suppose that the type of internal standards used might depend upon some contextual conditions. Identifying such conditions will increase our understanding of more precise cognitive mechanisms mediating consumer satisfaction and provide more meaningful managerial implications. Unfortunately, empirical investigations have rarely dealt with this important issue (cf., Oliver and Bearden 1983). The results reported here provide some insights into this matter.

Types of Comparison Standards

Perhaps, the work by Oliver (1980) is one of the pioneering studies formally looking into cognitive processes underlying consumer satisfaction. In that study, he proposed "expectation-disconfirmation" paradigm, in which the prepurchase "expectation" of a product served as a comparison standard in the satisfaction formation process. Following this work, a great deal of research effort has been devoted to test and extend it. In the meantime researchers also began to consider alternative comparison standards such as product norms, ideal, and equity. Following is a brief description of these alternative standards and of related research results. Since a much richer review is available elsewhere (Yi 1990), our description here will be brief.

Expectation. This standard is most frequently employed in satisfaction studies (e.g., Bearden and Teel 1983; Churchill and Surprenant 1982; Day 1984; Oliver 1980; Oliver and Linda 1981; Swan and Trawick 1981; Westbrook and Reilly 1983). Expectation is a prepurchase cognition about how the product performance will be and typically measured using a bipolar semantic differential scale such as "how good or bad did you expect the product performance would be."

Product Norm. Consumers may evaluate actual product performance against some sort of normative standard like "how the product performance should be." Such standard has been suggested as an alternative to the expectation standard by various researchers (e.g., Morris 1977; Sirgy 1984; Swan, Trawick, and Caroll 1982). The basis for forming the norm level can be varied. It might be based on the average performance level of products in the product category to which the focal product belongs (i.e., product norm). Alternatively it might be based on the performance level of the best brand in that product category (i.e., best-brand norm). Cadotte, Woodruff, and Jenkins (1987) empirically tested these possibilities. Their results suggest that the norm appeared to be based on the average performance level of the product category.

Equity. Consumers purchase a product with an expectation that they will receive at least as equitable value from the product as what they pay for it. This suggests that during satisfaction formation, consumers may compare input/output combinations in terms of fairness. Thus, in inequitable situations, consumers are likely to express their dissatisfaction (e.g., Fisk and Coney 1982; Mowen and Grove 1983; Oliver and DeSarbo 1988). Equity refers to such a normative level of product performance, given the cost they paid (e.g., price).

Ideal. Another alternative standard considered in the literature is ideal or desire. Westbrook and Reilly (1983) argue that consumers are likely to evaluate product performance based on how well they perceive the focal product fulfill their needs and wants. In this respect, the ideal standard represents the optimal product performance a consumer ideally would hope for (Tse and Wilton 1988) or what is personally desired from a product by a consumer (Spreng et al. 1996). Tse and Wilton (1988) empirically demonstrated that ideal was a viable comparison standard during satisfaction formation.

Comparison of Alternative Standards

Given several types of comparison standards identified, it is an important research issue to understand which type of standard operates better. Or, it might be a even more important issue to identify conditions under which a particular type of standard operates. Up to now only a few studies have empirically compared various types of standards (e.g., Cadotte et al. 1987; Tse and Wilton 1988). However, none of them explicitly considered boundary conditions under which a particular standard might be superior to others in explaining satisfaction formation processes.

For example, Cadotte et al. (1987) considered three alternative standards: brand expectation, product norm, and best-brand norm. Comparing three standards in terms of the power to explain the variance in satisfaction with restaurants , they found the product norm superior to the other standards in two out of three situations. On the other hand, Tse and Wilton (1988) tested the relative power of expectation, ideal, and equity in explaining satisfaction in the context of miniature record players. Their results seem to indicate that the brand expectation was better than the ideal or the equity.

Certainly, results from these studies indicate that different standards operate in different consumption situations (i.e., restaurants versus record players). However, it is hard to infer from the studies boundary conditions under which various standards are localized. For alternative standards were not simultaneously compared across a variety of settings (c.f., Cadotte et al. 1987). Furthermore, some methodological differences make it difficult to directly compare results between the studies. First, as noted above, the types of alternative standards considered were different across the studies. Second, causal models of satisfaction formation processes adopted in the studies were also different. For example, Tse and Wilton (1988) used a "full-path" model which allowed both direct and indirect paths from the standard to satisfaction and from the perceived performance to satisfaction. On the other hand, Cadotte et al.’s (1987) model constraint the paths such that the comparison standard and the product performance were to influence satisfaction only indirectly through the subjective disconfirmation. Third, while Cadotte et al. (1987) used different subjective disconfirmations corresponding to different standards in alternative causal models (i.e., disconfirmation from expectation, disconfirmation from product norm, disconfirmation from best-brand norm), Tse and Wilton used did only one type of subjective disconfirmation.

The studies to be reported were designed to overcome some of the limitations of previous research in speculating variables potentially moderating the validity of alternative standards. First, four comparison standards (expectation, product norm, ideal, and equity) were simultaneously tested in five different settings. Second, the same full-path causal model (differing only in the type of comparison standard) was employed to test alternative standards across all settings. The choice of a full-path model was based on previous studies suggesting that a comparison standard and perceived product performance would influence satisfaction levels not only indirectly through disconfirmations but also directly at least in some situations (e.g., Bearden and Teel 1983; Bolfing and Woodruff 1988; Bolton and Drew 1991; Churchill and Surprenant 1982; Oliver 1980; Oliver and DeSarbo 1988; Swan and Trawick 1981; Tse and Wilton 1988). Third, four types of subjective disconfirmations were measured corresponding to four different comparison standards. Fourth, all the constructs were measured by tw response scales and the reliability of each construct was assessed. Finally, all the measures were deliberately kept virtually identical across all settings.

METHOD

Overview of Data Collection

Five satisfaction studies. The data reported here were collected in Korea and through five consumer satisfaction surveys sponsored by Pulmuone, a national food company in Korea. The company is manufacturing and marketing a variety of general food items (e.g., tofu and noodles), health supplementary food items (e.g., calcium and aloes), and special food items (e.g., meal-replacement formula for diet). Most of them are positioned as high-quality/high-price products, and targeted mainly to those with high education, high income, and health concern. In the first two surveys, consumer satisfaction levels and their antecedents were measured with respect to the company’s tofu and noodle products. The third and forth surveys were about the calcium and diet meal products, respectively. The fifth survey dealt with dealer satisfaction. Store managers’ satisfaction with the company’s general food products were collected.

Product characteristics. Five different surveys were intended to provide various settings in which alternative standards were compared in terms of power to explain the variance in satisfaction. That is, different food items might represent different involvement and product experience situations. This speculation is explained below.

The typical meal in Korea consists of cooked rice with several side dishes. Tofu is frequently-used material for preparing a side dish, perhaps being used several times a week. Thus, the level of Korean consumers’ use experience with tofu is high. The Pulmuone’s tofu is made of beans grown without agricultural chemicals and its price is high, almost twice as high as the average price of the market (around $1.50 per unit). Thus, consumers’ involvement with the Pulmuone’s tofu is unlikely to be low. Noodles are used for a meal, usually as a replacement for the typical Korean meal of "rice and side dishes." Further, the Pulmuone’s noodle product is highly priced (around $2.50). Thus, it might be that consumers are at least somewhat involved with the product. However, noodles are prepared only occasionally, perhaps a few times a month. It might be that consumers’ use experience levels with noodles are not as high as their levels with tofu. The calcium and diet meal products are sold only by a package (3-month usage for the calcium and 4-week usage for the diet meal product) with prices of $110.00 and $120.00, respectively. Further, the people surveyed tended to be highly concerned with their health, and to have tried several other brands in the past. Therefore, their involvement and experience levels are presumably high. The fifth survey was with store managers who carried the Pulmuone’s general food items. Those managers seem minimally involved with the products because in most stores the tofu and noodle products account for only a small portion of the total sales. However, store managers’ transaction experience with the products must be very high since they carry the company’ products and other brands almost on a daily basis. Based on these characteristics, five surveys seem to provide different involvement and use experience settings for comparing alternative comparison standards during satisfaction formation: medium-involvement/high-experience (tofu), medium-involvement/medium-experience (noodles), high-involvement/high-experience (calcium & diet meals), and somewhat low-involvement/high-experience situations (dealer).

Subjects. Subjects for the surveys about tofu and noodles were housewives living in a metropolitan city in Korea. Sampled were only those who had purchased the focal products at least once during the last four weks before the time of the surveys. A quota sampling procedure was used for sampling based on the demographic profiles of the population such as age and residential area. The sample sizes were 600 and 260 housewives for tofu and noodle products, respectively. For calcium and diet meal products which are sold mainly by the company’s salesforces, customers’ lists are available. Thus, a probability sampling procedure was used for selecting subjects for the survey. In total, 300 subjects were sampled for each of the calcium and diet meal products. Finally, a list of supermarkets and stores which carried the company’s general food items was used for sampling dealer subjects. 300 store managers were selected through a quota sampling procedure.

Survey Procedure. Five surveys were administered by a professional survey organization. The data were collected through a face-to-face interview with each subject assisted by a structured questionnaire. All interviewers were female, well-trained, and received an orientation about the purpose of the surveys and the contents of the questionnaires. Surveys for tofu and noodle products were conducted first and took 14 days long. A month later, surveys for calcium and diet meal products were conducted. 12 days were taken for completion. Finally, the store managers were surveyed for 16 days.

FIGURE 1

CAUSAL MODEL OF SATISFACTION FORMATION PROCESS

Survey Instrument

Key Constructs. For the purpose of this research, a number of constructs were necessary to be measured. First, overall satisfaction levels with the focal products and actual product performance levels experienced by subjects were measured. Second, to comparatively test alternative comparison standards against one another, it was necessary to measure four types of comparison standards (expectation, product norm, ideal, and equity) simultaneously from each subject. Third, subjective disconfirmations corresponding to these standards were measured separately. In addition, several background variables were measured including consumer involvement levels and product experience. Two response scales were used to measure each construct (except product experience). They were of 9-point bipolar semantic differential scales and selected based on a review of previous studies and on several pretests during questionnaire construction. These measures are explained below.

Measures. First, overall satisfaction of the product was measured by two scales: ('1’-dissatisfied very much, '9’-satisfied very much) and ('1’-very bad purchase, '9’-very good purchase). After this, the perceived performance level of the product was measured. Subjects were asked about the product quality they actually experienced during consumption regardless of their satisfaction/dissatisfaction level. Two response scales were used: ('1’-very bad product, '9’-very good product) and ('1’-poor quality, '9’-excellent quality’).

Next, four alternative comparison standards and corresponding disconfirmation levels were measured. Subjects indicated the level of each comparison standard on two measurement scales. Then, they evaluated the product performance experienced against that standard level. Thus, the level of each comparison standard and that of corresponding disconfirmation were measured twice. These are explained in a more detail below.

Subjects were first asked about the level of product quality which they had expected prior to purchase would be (expectation). Two scales were provided: ('1’-very bad, '9’-very good) and ('1’-poor quality, '9’-excellent quality). Immediately after each response, subjects evaluated the experienced quality compared to that expectation level (expectation-disconfirmation) using the scale, ('1’- much worse, '9’- much better). After this, the average quality of the product category subjects assumed was measured (product norm) using two response scales: ('1’-very bad, '9’-very good) and ('1’-por quality, '9’-excellent quality). This was followed by evaluations of the experienced quality against that norm level (norm-disconfirmation). Next were measures of the product quality which consumers personally would desire to receive (ideal). Two response scales were provided: ('1’-average level, '9’-world best level) and ('1’-fair quality, '9’-superb quality). This was followed by evaluations of the experienced quality against that ideal level (ideal-disconfirmation). Finally, subjects reported the product quality which they assumed should be in light of the price they had paid (equity) on two scales: ('1’-very bad, '9’-very good) and ('1’-poor quality, '9’-excellent quality). Again, this was followed by evaluations of the experienced quality against that equity level (equity-disconfirmation).

Finally, two response scales were used in order to measure consumer involvement levels with the product: ('1’-not at all interested in the product, '9’-very much interested in the product) and ('1’ -not at all careful in choosing a brand, '9’-very much careful in choosing a brand). To measure product experience, the average frequency of purchase per week and the average amount of each purchase were asked by open-ended questions.

Causal Model and Validation Test

A full-path model was employed to test alternative standards. Figure 1 shows hypothetical paths among the constructs in the model. Four full-path causal models of the satisfaction process were developed. The models differed only in the type of standard employed. Then, the validity of the models was assessed using LISREL 8 (Joreskog and Sorbom 1993) for each product category. To compare the models’ validity, five conventional model-fit indicators were used: the chi-square statistics, the goodness-of-fit index (GFI), the adjusted goodness-of-fit index (AGFI), the root mean square residual (RMR), and normed fit index (NFI).

RESULT

Results to be reported here are organized as following. First, results from the causal modeling analyses about the tofu data are presented. Then, potential variables moderating the type of standard used will be speculated. After this, results about the other data (for the noodle, calcium and diet meal products as well as the dealer satisfaction) will be reported and integrated.

TABLE 1

FACTOR LOADINGS FOR FOUR-FACTOR SOLUTION

TABLE 2

MODEL-FIT INDICATORS FOR ALTERNATIVE-STANDARD MODELS IN TOFU DATA

Tofu Data

Reliability and validity of measures. As explained previously, all constructs included in the causal model of consumer satisfaction formation processes were measured using two scale items. The scores obtained from two items for each construct were averaged into a composite score for further analyses. First, reliability was assessed by Cronbach’s coefficient alpha for each construct. Analyses revealed that all the measures were highly reliable (all alpha values were higher than .86). Second, a factor analysis was performed on eight response scales measuring four comparison standards (two per each standard) in order to see if four standards considered were really distinct constructs. A principal component factor analysis with varimax rotation revealed that four factors emerged and they accounted for 90.1 percent of the total variance of the original items. This supports the validity of the measures.

Overall model tests. Four causal models of satisfaction formation processes were tested against the tofu data. Table 2 shows the values of various indicators conventionally used for comparing alternative causal models. As one can see, all indicators except chi-square values suggest that four models fit the data pretty well. However, chi-square statisticsshow that the only ideal-standard model had insignificant values. Further, the model had the lowest RMR as well as the highest GFI, AGFI, and NFI, representing very good fits. Therefore, causal modeling analyses supported the ideal as the best standard in the tofu case.

Subgroup analysis. Although it might be interesting to know that a particular standard was supported by the tofu data, it might be even more interesting to investigate the possibility that there are subgroups of subjects who might have employed a different standard to evaluate product performance. To explore such possibility, subjects were divided into four groups based on the levels of involvement and product experience. First, a composite involvement score was calculated by averaging responses on two scale items (Cronbach alpha=.85). Second, a product experience score was obtained by multiplying the average frequency of purchase per week by the average amount of each purchase. Using a median split, four subgroups of subjects were formed: low-experience/low involvement group, high-experience/low-involvement group, low-experience/high-involvement group, and high-experience/high-involvement group. Then, four alternative causal models were tested for each group. Summary results are shown in Table 3.

Some notable patterns emerged in the chi-square statistics. First, the ideal-standard model fit the data predominantly, representing the strong main effect for the tofu product’s characteristics (relatively a medium level of involvement and high level of experience). Second, and importantly, the standards supported other than the ideal varied depending on the involvement and experience levels. Specifically, in the high-experience/low-involvement group (C), the expectation-standard model was strongly supported, whereas in the high-experience/high-involvement group (D), two normative-standard models (norm and equity) were supported. By contrast, in the low-experience/high-involvement group (B), the data supported the ideal-standard model only. Finally, none of the models was strongly supported when both involvement and experience were low (A).

Overall, the data about the tofu product (which was presumed to be of high-experience/medium-involvement) were supportive of the ideal standard model. However, the subgroup analyses of the tofu data appeared to suggest that involvement and product experience might interactively influence the type of standard used. Briefly, expectation might become a dominant standard when the level of product experience is high but the involvement level is low. On the other hand, normative standards such as product norm and equity might operate well if the levels of both product experience and involvement are sufficiently high. To assess this possibility, the alternative causal models were further tested in the contexts of the noodle (medium-experience/medium-involvement), the calcium and diet meal (high-experience/high-involvement) and the dealer satisfaction (high-experience/low-involvement).

TABLE 3

MODEL-FIT INDICATORS FOR ALTERNATIVE-STANDARD MODELS IN TOFU SUBGROUPS

Noodle, Calcium, Diet Meal, and Dealer Satisfaction Data

Overall model tests. First of all, preliminary analyses showed that the measures were highly reliable and valid in all data sets. Next, four alternative causal models were tested against each data. All the model test procedures were identical to those in the tofu data analysis. Table 4 shows the values of model-fit indicators for alternative models across the data sets.

As one can see, chi-square statistics show that the model best supported differed in different data sets. First, the ideal-standard model was the best in the noodle data, same as in the tofu data. Second, both calcium and diet meal data supported the product norm, a normative standard. This is consistent with results from the high-involvement/high-experience subgroup of the tofu data. Finally, the expectation-standard model best fitted the dealer data. This is also consistent with the results form the low-involvement/high-experience subgroup of the tofu data.

In sum, different compaison standards were better supported across noodle, calcium, diet meal, and dealer data sets in a way consistent with our expectation based on the subgroup analyses of the tofu data. The results appeared to suggest that involvement and use experience interactively influenced comparison standards used for evaluating product performance during satisfaction formation. In interpreting these results, however, there is a different possibility that subjects’ use of different standards across the data sets might have been due to differences not in terms of involvement and use experience levels of the products, but in terms of some other product characteristics. To overcome this difficulty of interpretations at least to some extent, we conducted subgroup analyses of the noodle, calcium, and dealer data sets as we did earlier with the tofu data set.

Subgroup analysis. The procedure of subgroup analyses was virtually identical to that of the tofu data: (1) dividing subjects into subgroups of low-experience/low involvement, high-experience/low-involvement, low-experience/high-involvement, and high-experience/high-involvement and (2) testing four alternative causal models for each subgroup. It might be ideal to perform this analysis for each of the noodle, calcium, diet meal, and dealer data sets separately. Unfortunately, the total sample size of each data set was relatively small (n=300 or less) compared the size of the tofu data (n=560). Consequently, it was thought to be impractical to form four subgroups and to test causal models using LISREL for each subgroup because of its limited sample size (n=75 or less). Our strategy therefore was to conduct subgroup analyses for the combined data set across products. First, four subgroups were created for each of the noodle, calcium, and dealer data sets (the diet meal data was not included in this analysis because the use experience was not measured). Next, the subgroups of a given involvement and use experience condition of each data set were merged into one, resulting in four subgroups based on the combinations of involvement and experience levels for the combined data. Finally, four alternative causal models were tested for each of the four subgroups. If the results from these analyses were consistent with those for the tofu subgroups, then our earlier conclusions regarding interactive effects of involvement and experience could be made stronger. In fact, this was the case.

TABLE 4

MODEL-FIT INDICATORS FOR ALTERNATIVE-STANDARD MODELS IN NOODLE, CALCIUM, DIET MEAL, AND DEALER SATISFACTION DATA

Table 5 contains summary results of the subgroup analyses. First, in the high-experience/low-involvement group (C), the expectation-standard model was best supported, whereas in the high-experience/high-involvement group (D), two normative-standard models (norm and equity) were strongly supported. These results were consistent with those in the tofu subgroups. By contrast, in the low-experience/high-involvement group (B), the data predominantly supported the ideal-stand model. This was also consistent with the corresponding tofu subgroup. Finally, none of the models was supported (although the ideal model was most favored) when both involvement and experience were low (A). Again, this was also consistent with the corresponding tofu subgroup.

DISCUSSION

Results from this research suggested that consumer involvement and product experience might interactively influence the type of comparison standard used in the satisfaction formation process. The summary and implications of the results are now discussed.

The results suggest that in high-involvement/high-experience situations, normative standards such as product norm and equity are likely to operate. This was initially suggested by the results from the subgroup analysis of the tofu data, and in the analysis of the data for the calcium and diet meal (which were presumably of high involvement and high experience). It was further supported by the subgroup analyses of the combined data for the calcium, diet meal, and dealer satisfaction. Accrdingly, highly involved consumers with high experience might judge the product quality in reference to what they assume the product should provide. This finding seems consistent with Woodruff, Cadotte, and Jenkins (1983). They argued that norms were constrained by the performance consumers believe was possible as indicated by the performance of "known" brands. Consequently, having some experience with the product is a necessary condition to possess a norm standard. On the other hand, a normative standard is related to consumers’ emotional commitment in meeting their needs and wants by purchasing a product. The involvement concept seems to reflect such emotional commitment. A norm standard is then unlikely to be used by uninvolved consumers. Therefore, a normative standard is likely to operate well for highly involved consumers with ample product experience.

Both subgroup analysis of the tofu data and overall analysis of the dealer data appeared to suggest that the expectation standard might be a dominant standard in low-involvement/high-experience situations. This tended to be further supported by the subgroup analysis of the combined data. This finding appears consistent with some of previous findings from information processing research. Specifically, uninvolved consumers tend to simply use easily accessible information in memory to make a judgment (e.g., Park and Hastak 1994; Sanbonmatsu and Fazio 1990). Also, memory information like a prior brand evaluation tends to be more accessible when it is experience-based than when it is information-based (e.g., Berger and Mitchell 1989; Fazio and Zanna 1981). Since the expectation of a brand is in fact a previously-formed brand evaluation, it is likely to be easily accessible when the experience of the brand is accumulated. Consequently, relatively uninvolved consumers with high product experience are likely to simply retrieve and use the expectation standard to evaluate product performance.

Results also indicated that in high-involvement/low-experience situations the ideal-standard model stood out. This tended to be supported consistently by the subgroup analyses of the tofu and the combined data as well as by the overall analysis of the noodle data. This finding seems intuitively reasonable. Consumers in those situations would be highly concerned with meeting their needs and wants. However, they are unlikely to have a strong normative standard due to lack of product experience. Therefore, they simply evaluate the product performance against what they personally desire to receive from the product. On the other hand, no causal model was strongly supported in low-involvement/low-experience situations in the subgroup analysis of either the tofu or the combined data. One possibility is that satisfaction is predominantly determined by the actual product performance (c.f., Cronin and Taylor 1992). However, this clearly needs to be examined by future research.

One caveat to our conclusions should be emphasized. Our findings are exploratory in nature. None of the factors considered were not experimentally manipulated. In particular, high versus low levels of involvement and product experience variables were operationalized by a median split of measured scores and by different product categories. This certainly introduces a danger of confounding and thus weakens the validity of our interpretations about the results. While a replication of our findings is needed, future research involving experimental manipulations of the variables such as involvement and product experience is warranted to be of great value.

TABLE 5

MODEL-FIT INDICATORS FOR ALTERNATIVE-STANDARD MODELS IN SUBGROUPS OF COMBINED SAMPLES FROM NOODLE, CALCIUM, AND DEALER DATA SETS

REFERENCE

Bearden, William O. and Jesse E. Teel (1983), "Selected Determinants of Consumer Satisfaction and Complaint Reports," Journal of Marketing esearch, 20 (February), 21-8.

Boulding, William, Ajay Kalra, Richard Staelin, and Valarie A. Zeithaml (1993), "A Dynamic Process Model of Service Quality: From Expectations to Behavioral Intentions," Journal of Marketing Research, 30 (November), 7-27.

Bolfing, Claire P. and Robert B. Woodruff (1988), "Effects of Situational Involvement on Consumers’ Use of Standard in Satisfaction/Dissatisfaction Process," Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, 1, 16-24.

Bolton, Ruth N. and James H. Drew (1991), "A Multistage Model of Customers’ Assessments of Service Quality and Value," Journal of Consumer Research, 17(March), 375-84.

Cadotte, Ernest R., R. B. Woodruff, R. L. Jenkins (1987), "Expectation and Norms in Models of Consumer Satisfaction," Journal of Marketing Research, 24(August), 305-14.

Churchill, Gilbert A., Jr. and C. Surprenant (1982), "An Investigation into the Determinants of Customer Satisfaction," Journal of Marketing Research, 19(November), 491-504.

Cronin, J. Joseph, Jr. and Steven A. Taylor (1992), "Measuring Service Quality: A Reexamination and Extension," Journal of Marketing, 56(July), 55-68.

Day, Ralph L. (1984), "Modeling Choices Among Alternative Responses to Dissatisfaction," in Advances in Consumer Research, 11, Thomas C. Kinnear, ed. Ann Arbor, MI: Association for Consumer Research, 469-79

Fazio, Russell H. and Mark P. Zanna (1981), "Direct Experience and Attitude-Behavior Consistency," in Advances in Experimental Social Psychology, 14, ed. Leonard Berkowitz, New York: Academic Press, 161-198.

Fisk, Raymond P. and Kenneth A. Coney (1982), "Postchoice Evaluation: An Equity Theory Analysis of Consumer Satisfaction/Dissatisfaction with Service Choices," in Conceptual and Empirical Contributions to Consumer Satisfaction Complaining Behavior, eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University School of Business, 9-16.

Joreskog, Karl and Dag Sorbom (1993), LISREL 8: A Guide to the Program and Applications. Chicago: Scientific Software International.

Morris, Earl W. (1977), "A Normative Deficit Approach to Consumer Satisfaction," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction, H. Keith Hunt, ed. Cambridge, MA: Marketing Science Institute, 240-74.

Mowen, John C. and Stephen J. Grove (1983), "Search Behavior, Price Paid, and the 'Comparison Other’: An Equity Theory Analysis of Post Purchase Satisfaction," in International Fare in Consumer Satisfaction and Complaining Behavior, eds, Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University School of Business, 57-63.

Oliver, Richard L. (1980), "A Cognitive Model of the Antecedents and Consequences of Satisfaction Decisions," Journal of Marketing Research, 17(September), 46-9.

Oliver, Richard L. and William O. Bearden (1983), "The Role of Involvement in Satisfaction Processes," in Advances in Consumer Research, Richard P. Bagozzi and Alice M. Tybout, eds. Ann Arbor, MI: Association for Consumer Research, 250-255.

Oliver, Richard L. and G. Linda (1981), "Effects of Satisfaction and Its Antecedents on Consumer Preference and Intention," in Advances in Consumer Research, Kent B. Monroe, ed. Ann Arbor, MI: Association for Consumer Research, 88-93.

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

Park, Jong-won and Manoj Hastak (1994), "Memory-based Product Judgement: Effects of Involvement at Encoding and Retrieval," Journal of Consumer Research, 21(December), 534-547.

Sirgy, M. Joseph (1984), "A Social Cognition Model of CS/D: An Experiment," Psychology and Maketing, 1(Summer), 27-44.

Spreng, Richard A., Scott B. Mackenzie, and Richard W. Olshavsky (1996), A Reexamination of the Determinants of Consumer Satisfaction," Journal of Marketing, 60(March), 15-32.

Swan, John E. and I. Frederik Trawick (1981), "Disconfirmation of Expectations and Satisfaction with a Retail Service," Journal of Retailing, 57(Fall), 49-67.

Swan, John E., I. Frederick Trawick, and Maxwell G. Carroll (1981), "Satisfaction Related to Predictive, Desired Expectations: A Field Study," in New Findings on Consumer Satisfaction and Complaining Behavior, H Keith Hunt and Ralph L. Day, eds. Bloomington, IN: Indiana University, 15-22.

Tse, David K. and Peter C. Wilton (1988), "Models of Consumer Satisfaction: An Extension," Journal of Marketing Research, 25(May), 204-12.

Westbrook, Robert A. (1980), "Product/Consumption-Based Affective Responses and Postpurchase Processes," Journal of Marketing Research, 24(August), 258-70.

Westbrook, Robert A. and Michael D. Reilly (1983), "Value-Percept Disparity: An Alternative to the Disconfirmation of Expectations Theory of Consumer Satisfaction," in Advances in Consumer Research, Richard P. Bagozzi and Alice M. Tybout, eds. Ann Arbor, MI: Association for Consumer Research, 256-61.

Yi, Youjae (1990), "A Critical Review of Consumer Satisfaction," in Review of Marketing, Valarie A. Zeithaml, ed. Chicago: American Marketing Association, 68-123.

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