1. Introduction

Satisfaction phenomena have been studied on their many aspects. Life satisfaction is an important theme in sociology:marital satisfaction in family sociology. Patient satisfaction study is done for hospital management. In the sphere of business management, job satisfaction is a major subject for personnel management. Consumer satisfaction, the theme of this dissertation, is a major study subject for marketing. Its literature has been accumulated from the seventies. The reason why satisfaction phenomena have been researched about their various aspects is that they are essential for understanding of human behavior. Consumer satisfaction is also an essential phenomenon for explanation, prediction and understandment of consumer behavior.

Consumer satisfaction has great meaning for the three major market actors; companies, consumers, and government. It is a key factor of marketing concept. It also has big influences to various consumer behaviors that is, consumer attitudes, repeat purchase behavior, consumer loyalty and consumer complaint behavior. Thus, it has great meanings for companies, especially, for marketers. It is very important phenomena for government who has the responsibility of consumer welfare enhancement and consumer protection. It's because consumer satisfaction is not only an essential indicator of consumer welfare but also a good evidence that consumers are protected well at the market.

Consumer satisfaction is very crucial phenomenon to consumers themselves, too. It is an evidence that they made wise purchase decisions and that they utilized scarce resources of themselves well. Also, consumer satisfaction is a very important indicator of consumer welfare as is noted above. The study of consumer satisfaction and the resultant explanation, prediction and understandment of it is a very crucial task of marketing scientist which helps firms, consumers, and government make good decisions.

2. Literature survey

Consumer satisfaction can be approached from two directions. The first direction is the study of consumer satisfaction as the independent variable. If we survey these studies, we can find that their subjects are the influences of consumer satisfaction upot consumer attitude , repeat purchase behavior, consumer loyalty and consumer complaint behavior.

The second paradigms of consumer satisfaction determinents are three; disconfirmation paradigm. equity paradigm and attribution paradigm. Disconfirmation paradigm deals with the influences of expectation, disconfirmation and perceived product performance upon consumer satisfaction. Adaptation level theory and comparison level theory are considered as the theoretical based of this paradigm.

The theoretical background of equity paradigm is equity theory of social psychology. The major contention is that consumers determine the equitableness of the exchange and firms. The more equitable the exchange are the more satisfied are the consumer.

The theorectical basis of attribution paradigm is attribution theory of social psychology. This paradigm asserts that consumer's purchase behavior can be regarded as an achievement behavior, and that when the purchase comes out to be a success, the internally consumers feel more satisfied than the externally attributing consumers.

This dissertation focused on the study of disconfirmation paradigm. All the above three paradigms can be contained in the research scope. But it is also a valuable task to focus upon a paradigm and to make a progress of it knowledge.

3. Research Hypotheses

Disconfirmation paradigm is the earliest researched and the most deeply researched of the three paradigms. The major variables of it is expectation, disconfirmation and product performance. This dissertation deals theoretically can empirically with three questions in disconfirmation prardigm which are not fully answered yet. The first of these questions is how many dimensions have expectation and disconfirmation respectively.

Most studies of diconformation paradigm consider expectation variable to be constructed as one dimension. But Miller maintains that expectation has four types; the ideal, the expected, the deserved and the minimum tolerable. Summers and Granbois asserts that there are two; the predictive expectation and the normative expectation.

Swan and Trawick discerns two types of expectation ; the predictive expectation and the desired expectation. Futhermore, Barbeau makes an assertion that expectation has two dimensions, the first of which is predictive dimension and the second normative dimension. He also maintains that the first acts as an adaptation level and the second as a comparison level.

But the desired expectation and the normative expectation have different characteristics. The two expectations told by Miller-the ideal and the minimum tolerable-are not expected by consumers in ordinary setting. Thus, this study considers the expectation variable to have three dimension; the predictive, the normative and the desired. The disconfirmation variable is also considered to have three dimensions in this dissertation. The hypothesis 1 is as follows.

Hypothesis 1:The satisfaction model which have three types of expectation and disconfirmation have greater explanatory and predictive power than the model which have one or two types.

The second question in this dissertation is the role of expectation in satisfaction process. Early studies of consumer satisfaction were based upon assimilation theory or contrast theory. The issue was which variable between expectation and disconfirmation affects consumer satisfaction. But Oliver maintains that expectation and disconfirmation are independent variables and these two variable influence satisfaction additively. Many researches follows Oliver's contention which became a popular view.

But if expectation is an adaptation level, it will not have direct influences upon satisfaction. It only acts as a criteria or a frame of reference.

A few empirical researches confirms the above arquments. For example Swan and Trawick found that satisfaction is determined solely by disconfimation and expectation does not affect satisfaction.

If disconfirmation is determined by expectation and perceived product performance, the former will have negative fluences upon disconfirmation while the latter will have positive influences.

Thus, expectation should have negative correlation with disconfirmation. A few researches exist which confirm these arquments. For example Bearde and Teel, Churchill Surprenant both found that expectation has negative correlation with disconfirmation. Churchill and Surprenant found that expectation have negative effect upon disconfirmation.

Thus, the hypothesis 2 is as follows.

Hypothesis 2-1:Expectation has not direct effect upon consumer satisfaction. All of three types of expectations act as a criterior or a frame of reference.

Hypothesis 2-2:Expectation has negative correlation with disconfirmation and negative effect upon disconfirmation. This contention applies to all of the three types of expectation ; the predictive, the normative and the desired.

The third issue of this dissertation is performance paradigm. Oliver and Desarbo regard performance as an independent paradigm. But this study regard it to be a variable in disconfirmation paradigm. Oliver insisted that it is axiomatic for product performance to have influence upon consumer satisfaction. But and axiomatic phenomenon can be a very important one. It should not be excluded from the model. The their hypothesis of this study is as follows.

Hypothesis 3:The model consisted of expectation, disconfirmation and product performance have more explanatory and predictive power than the model consisted of only two variables; expectation and disconfirmation.

4. Empirical Study

To test above-mentioned three hypotheses, and empirical study was performed. The product was food. The consumer satisfaction study in disconfirmation paradigm should measure consumers' expectation before they buy products and perceived product performance, disconfirmation and satisfaction after they purchase products. This longitudinal characteristics of disconfirmation paradigm makes its empirical study very difficult. Thus this study selects food as the object of research.

A middle-class Korean restaurant was selected. The customers of it were sampled. the reason of selection of middle-class restaurant is that because it has the least bias, its selection can enhance the reliability and validity of the research.

The questionaire was divided into two parts. The first half consists of introductory items and the items which measure predictive, normative and desired expectations.

The second half consists of items which measure perceived product performance, disconfirmation, satisfaction and demographic variable.

Statistical analysis uses LISREL technique. Hypothesis 1 was tested through confirmatory factor analysis and LISREL model comparison. The result of the confirmatory factor analysis reveals that the three expectation types have high correlation. The result of LISREL model comparison reveals that the squared multiple correlations of the model consisted of three expectation types are not higher than those of the models consisted of one or two expectation types. These results reject hypothesis 1.

An examination of the parameters of LISREL model shows that hypotheis 2-1 is accepted and the latter half of hypothesis 2-2 is accepted. In each model which is consisted of three expectation and disconfirmation types, two expectation and disconfirmation types and disconfirmation types and one expectation and disconfirmation type, all three types of expectation have not direct effect on satisfaction and they have negative effects upon disconfirmation. But expectation has not negative correlation with disconfirmation.

Hypothesis 3 was accepted as true. The fitness index of the model containing perceived product performance was higher than that of the model without it. This result confirms the assertion of this study that performance variable should included in disconfirmation paradigm.

5. conclusions

Expectation is a very important construct in consumer satisfaction research. It is essential to enhance it knowledge further. The studies which probe into the expectation's influences upon other consumer behaviors than consumer satisfaction should be done. The subject of the dimensions of expectation construct should also be dealed with by future studies.

Expectation should not be too high ; it does not influence satisfaction directly and has negative effect upon disconfirmation. But if its level is low, the possibility for consumers to purchase the product is low. The management of consumer expectation, thus, is an essential subject for marketing researchers.

Consumer satisfaction research has many tasks to be performed. The studies which contain all three paradigms in their research scope should be further investigated. So are the studies which deals with a paradigm deeply.

key words:predictive expectation, normative expectation, desired expectation, disconfirmation, perceived product performance consumer satisfaction disconfirmation paradigm.