Economic Constraints on Consumer Complaining Behavior

ABSTRACT - Consumer complaining behavior has rarely been subject to explanatory or predictive inquiry. In this paper, a simple model, drawing upon some basic tenets of consumer choice in economics, is applied to post-purchase behavior. Using only two independent variables, the model is able to explain a substantial portion of the variance in complaint ratios.

Citation:

Claes Fornell and Nicholas M. Didow (1980) ,"Economic Constraints on Consumer Complaining Behavior", in NA - Advances in Consumer Research Volume 07, eds. Jerry C. Olson, Ann Abor, MI : Association for Consumer Research, Pages: 318-323.

Advances in Consumer Research Volume 7, 1980     Pages 318-323

ECONOMIC CONSTRAINTS ON CONSUMER COMPLAINING BEHAVIOR

Claes Fornell, Northwestern University

Nicholas M. Didow, University of North Carolina

ABSTRACT -

Consumer complaining behavior has rarely been subject to explanatory or predictive inquiry. In this paper, a simple model, drawing upon some basic tenets of consumer choice in economics, is applied to post-purchase behavior. Using only two independent variables, the model is able to explain a substantial portion of the variance in complaint ratios.

INTRODUCTION

While most research on consumer behavior still centers around consumer purchase choice, recent years have witnessed a rapidly growing interest in the total consumption process, including post-purchase phenomena. Since consumer satisfaction/dissatisfaction (CS/D) is one measure of economic system performance and consumer complaints are indicators of low or declining performance, it is not surprising that the research on consumer post-purchase behavior has concentrated on these two variables.

As for most fields in the beginning of their development, basic definitional and measurement issues have yet to be resolved in CS/D research. These problems notwithstanding, some studies have gone beyond descriptive purposes. Hypotheses have been derived and tested from psychological concepts and models of assimilation-contrast (Anderson, 1973), comparison levels (LaTour and Peat, 1978), assimilation (Olshavsky and Miller, 1972), and cognitive structure (Olsen and Dover, 1976).

Studies on complaining behavior, on the other hand, have been almost completely dominated by a descriptive research orientation. Although methodologies and sampling procedures differ widely, two basic findings have emerged from this research: one is that a majority of dissatisfied consumers do not voice their complaints; another is that these who do complain exhibit distinct socio-economic and demographic characteristics (Robinson, 1978). However, it is probably fair to say that no significant advances have been made in terms of explaining or predicting consumer complaining behavior. Very few attempts have been made to go beyond descriptive empirical investigations. Among the pioneering efforts in this regard are studies by Granbois, Summers, and Frazier (1977) that found weak relationships in applying beliefs and expectation to explain complaining, and by Gr°nhaug (1976) that did little better than chance in predicting complaint incidents from demographics. Following a finding of Landon and Emery (1974), some researchers (Lawther et al., 1978; Kraft, 1977) have tested the hypothesis that attributions of responsibility play a central role in determining consumer complaint behavior, but they did not provide much evidence as to the explanatory power of this variable.

Although empirical studies of an explanatory or predictive orientation have been very scarce to date, several conceptual models of the consumer dissatisfaction/complaining process have been proposed (Day and Landon, 1977; Landon, 1977; Fornell, 1976). For example, Landon's model is an attempt to integrate previous research on complaint behavior. It postulates that dissatisfaction, importance, benefit from complaining, and personality are the basic determinants of complaining behavior. Fornell's model of the consumer post-purchase evaluation specifies a functional relationship between perceived performance and prior expectations that is affected by both consumer and market variables.

To the extent that the empirical research and the conceptual models have theoretical underpinnings, these are typically found in psychology and social psychology. Economic theory and structural variables have not found much application. Some exceptions are LaTour and Peat's (1978) test of some propositions in Hirschman's (1970) theory of quality decline and, to some degree, the model presented by Fornell, in which structural variables such as competition and availability of substitute products are integrated in a theory of product performance evaluation.

In order to gain an understanding of consumer response to dissatisfaction, there is a need for a theoretical framework that allows empirical determination of the conditions under which consumers are likely to voice their concerns and the conditions under which complaints are likely to be suppressed. Consequently, it is necessary to go beyond descriptive studies. It is also necessary to study non-complainers and their reasons for not complaining. According to a recent review of the literature in the field (Robinson, 1978), very few studies have focused on the non-complainer. The limited findings are again of a descriptive nature and correspond well to the results of complainant studies (Best and Andreasen, 1977; Warland, Herrmann, and Willits, 1975).

In sum, then, for the advancement of knowledge of consumer post-purchase behavior, it seems necessary to (1) move beyond description and toward explanation and prediction, and to (2) examine both complainers and non-complainers in the same study. It also seems desirable to (3) expand the rather narrow disciplinary scope of the field and test formulations from disciplines other than psychology and social psychology. This paper is an initial attempt in this direction. By deriving and testing two basic propositions from Hirschman's (1970) theory and by incorporating variables that take both complainers and non-complainers into consideration, this study uses two structural economic variables in an attempt to explain the variation in complaint levels across different products and services.

THEORY--DERIVATION OF HYPOTHESES

While research in consumer complaining behavior has not drawn much upon economic theory, economists have not demonstrated a great deal of interest in studying complaint behavior or the role and impact of verbal consumer protest on market behavior. The work of Hirschman (1970) is an interesting exception. In reference to his application of some fundamental concepts in economics to the analysis of consumer response to quality decline, it seems that economics has much to offer the study of consumer complaining behavior.

Economic theory of consumer choice seeks to describe how the consumer chooses among goods and services. Choice is viewed as a function of preference and availability of alternatives (which is subject to certain constraints). It is assumed that the consumer has well-defined preferences and that he/she is rational. Making the transition to the consumer post-purchase process, it is not difficult to cast complaining behavior in terms of choice, preference, and availability of alternatives. Ignoring the possibility that consumers may complain for the sake of complaining, let us assume that dissatisfaction (with a particular purchase episode) is a necessary, but not sufficient, condition for a complaint to be voiced. Thus, if complaining does not automatically follow dissatisfaction, there must be more than one possible consumer reaction to dissatisfaction. For example, besides complaining, the consumer may indicate his displeasure with the offending firm by abandoning it in favor of a competing firm. Given at least two alternative ways of reacting to unacceptable seller and/or product performance, the study of consumer complaining behavior is a study of consumer choice.

One difference between choice in the traditional economic model and complaining behavior is that the former pertains to pre-purchase events; whereas the latter concerns post-purchase events. The objects of choice are also different. Traditionally, these objects consist of goods and services. In Hirschman's theory the objects are exit (brand or patronage shifting, purchase termination, product class transfers) and voice (complaints directed to the offending firm or to a third party such as governmental agency, industry association, mass media, etc.). It should also be noted that, as in the traditional case, the individual is not forced to take overt action. At any given point in time, the consumer may choose to do nothing.

Given that the dissatisfied consumer faces a choice, what does economic theory postulate? Under what conditions will dissatisfied consumers use the market to improve their situation? Under what conditions will they voice their discontent in the form of complaints? This study will examine two economic determinants of the voice action. Assuming that the pre-purchase choice theory is relevant for post-purchase situations, voice would be a function of (1) preference and (2) the availability of alternatives. Let us discuss the availability of alternatives first.

Availability of Alternatives

If the exit option was blocked or unavailable, complaining would be the only way the dissatisfied consumer could (overtly) react. Economic theory predicts that complaints would be more frequently registered in markets characterized by monopoly than in markets resembling the notion of perfect competition.

At the other extreme, when a large number of close substitutes are available, the exit alternative becomes not only possible, but also a more probable course of action for the dissatisfied consumer. By switching brand or changing patronage, the dissatisfied consumer sets in motion forces of the market which can lead to either a recovery for the firm that has declined in relative performance, or to the elimination of that firm from the market. The proliferation of products and increased competition tend to increase cross-elasticities of demand and thereby increase the probability that the dissatisfied consumer takes some form of market action.

Consequently, economic theory suggests that market structure is a powerful factor in determining the level of consumer complaints. In markets with many competing firms with identical or similar offerings, dissatisfied consumers are less prone to voice complaints than they are in markets where substitutes are limited. The derived hypothesis is then:

H1:  The number of voiced complaints relative to the number of dissatisfied consumers is negatively related to the number of competing organizations offering identical or similar goods and services in the relevant trading area.

Not only is it hypothesized that the number of available alternatives is negatively associated with complaint actions, but, as implied by the theory, that the availability of alternatives is a very powerful structural constraint on consumer voice and thus should explain a substantial part of the variation in complaint levels.

Preference

Besides the availability of alternatives, economic theory suggests another important constraint on consumer voice. Consumer preference for the voice option is a function of the expected value of the outcome from complaining minus the costs involved. If the cost of complaining (in terms of money, time, effort) is greater than zero, and if the probability of satisfactory outcome is less than unity, the rational consumer will refrain from complaining unless the expected reward exceeds the costs.

As the consumer spreads his/her income and time over an increasing number of purchases, each individual product and purchase episode represents a declining proportion of both income and time. It would seem that the cost of devoting the time and effort to complain about any one unsatisfactory event is likely to exceed the expected benefits in many situations. Thus, the rational consumer may not complain about nonrecurrent negative experiences in any one of his daily or weekly interactions with a particular seller, simply because it is not likely to be worth the effort. For this reason, economic theory of rational choice postulates that the frequency of buyer-seller interaction affects the preference for voice. The more frequent the interaction (as in the case of grocery shopping), the less likely it is that the consumer will be able to "afford" to invest the time and effort required to register a complaint about any one of these interactions. A second hypothesis of this study can thus be formulated:

H2:  The number of voiced complaints relative to the number of dissatisfied consumers is negatively related to the frequency of the buyer-seller interaction.

Although other factors may impact upon consumer complaining behavior, it is hypothesized that availability of alternatives and frequency of buyer-seller interaction account for a substantial part of the variation in complaint levels. Thus, a simple general model can be written:

C_ij/D_ij = f(AA_i, FI_ij)   (1)

THE STUDY

Data

The data analyzed in this study were obtained from telephone interviews with 254 randomly selected households in Evanston, Illinois, during the fall of 1978. Respondents were the head-of-household who acted as a key informant on behalf of the entire household. For a total of 32 product/services covered, the household spokesperson was asked a series of questions relating to dissatisfaction and complaining behavior. [For a detailed descriptive analysis of consumer problems relating to these 32 products and services, see Didow (1978).] First, it was determined whether or not someone in the household had purchased the particular product/service within a specific time frame. For frequently purchased items, the time frame was set to the past three months. For other items, the time frame was the past year.

If the time frame recency requirement was set, the respondent was asked several questions about the most recent purchase of the goods and services. For example, the respondent indicated general satisfaction or dissatisfaction with the most recent grocery purchase. If dissatisfied, the respondent was asked in an open-ended format, what was done, if anything, to resolve the problem.

The measurement methodology followed the aided recall approach as suggested and applied on a national basis to a consumer post-purchase context by Best and Andreasen (1977). However, the categories of goods and services were broadened in this study to include household interactions with suppliers of public utilities, which were not analyzed by Best and Andreasen, but have been investigated in other studies (e.g., Day and Bodur, 1977).

From the initial measures on the 32 product/service categories, 13 were selected for analysis. The remainder were discarded because of few reported dissatisfactions. Although the data base refers to observations on 254 individuals, the data to be analyzed can conveniently be summarized in a 13 x 3 matrix (see Table 1).

The 13 product/service categories listed in Table 1 that were found to be subject to the most prevalent consumer dissatisfaction correspond well to the findings of studies based on national probability samples. For example, a TARP (1976) investigation identified automobiles, television/radio, appliances, mail service, clothing, telephone/telegram, food, and household items as the most common sources of consumer problems. Similarly, the only other major study in the area (Best and Andreasen, 1977) reported that automobiles, grocery items, clothing, and furniture were among the most frequently mentioned products that involved consumer dissatisfaction.

TABLE 1

THE DATA

Measures

Consumer voice.  Column A of Table 1 lists the Voice Index for each of 13 products. This index reflects the extent to which consumer dissatisfaction is verbally communicated to the retail seller organization. The focus of the variable is the purchase episode--the interaction between buyer and seller. Consequently, the source or target of dissatisfaction can be both product performance and seller behavior. Following the model in (1), Voice Index (VI) for any seller/ product i across N households is defined as:

EQUATION    (2)

Examples of consumer behaviors counted in the numerator of the Voice Index were complaining to the sales staff or management of the seller organization. Excluded were actions such as engaging in negative word-of-mouth about the seller, changing patronage, and doing nothing. Since complaining to a third party is almost always preceded by complaints directly to the source organization (Diamond et al., 1976; Hannigan, 1977), such behaviors were also excluded.

Even though the sample was locally restricted in the present study, the estimated Voice Index conforms fairly well to estimates at the national level. The index for grocery items in this study amounted to 16; in Best and Andreasen's survey it was 16.5. Other ratios were: auto repair 46 vs. 48.4, furniture 57 vs. 48. Only two major differences between the two studies were found. The Voice Index for appliance repair in the present study was much higher (78) than in the previous study (51.4), and the estimate for clothing was somewhat lower (11 vs. 26.0).

Availability of alternatives.  The availability of alternatives (AA) was operationalized as the number of retail outlets in the trading area for each product. Trading area is defined as the geographic region in which the consumer may reasonably expect to find goods and services at competitive prices (Stern and El-Ansary, 1977). The direct measurement of this area for each product category requires extensive data on each individual consumer's shopping habits. Such data were not available in this study. Instead, the extent of trading area was measured indirectly for each product category by counting the number of outlets listed in the yellow pages of the 1978 Evanston telephone directory. Given scrambled merchandising, exclusive and selective distribution, this is a crude measure. On the other hand, it is quick, inexpensive and free from interviewer biases. It should reflect management judgment as to whether or not an individual resides in an area from which the respective retail outlet draws its customers. For example, 92 autonomous outlets for clothing and apparel were identified (many of them outside Evanston), but only a single source for telephone, water, and natural gas services.

Frequency of buyer-seller interaction.  The frequency of buyer-seller interaction was operationalized as purchase frequency where applicable (e.g., auto repair, clothing, etc.). For public utilities it was necessary to distinguish consumption from buyer-seller interaction. Although the water service is used on a daily basis, actual interaction with the supplier is less frequent. Consequently, for each utility, the respondent was asked directly to indicate the frequency with which the household interacted with the supplier organization. For example, for telephone service the interaction involved such matters as service on the phone, installment of a new phone, repairs, etc., but not ordinary operator assistance.

An annual Frequency of Interaction (FI) rate was calculated across the sample of households for each product/service. As can be seen from Column C of Table 1, the typical household in the study purchased grocery products 69.20 times a year, while it interacted with the supplier of water service .16 times a year.

Analysis

Testing model.  Given the general model (1), multiple regression is a convenient method both for the simultaneous test of the two hypotheses, and for assessing the joint explanatory power of the proposed independent variables. The regression equation is thus:

VI_i = a - b₁AA_i - b₂FI_i + e_i   (3)

where a is a constant and e the error term.

Estimation.  Because the unit of analysis is "purchase episode" (relating to a product/service category), and not "survey respondent," the resulting data matrix is small. As a consequence, Ordinary Least Squares estimation might produce unstable estimates. Therefore, a modified procedure using Tukey's (1958) jackknife pseudovalues was adopted. The jackknife statistic for each diagnostic statistic and coefficient is the mean of the pseudovalue estimators for that parameter, That is,

EQUATION   (4)

where:

J_L = jackknife statistic L; and

f_Li = pseudovalue for statistic L derived on iteration i.

f_Li is the pseudovalue estimate of statistic L derived from the set composed of all available cases except case d, where d goes from 1 to k. Both regression and correlation coefficients were estimated by using the jackknife procedure. By holding out one observation for each coefficient computation, a total of k estimates were obtained for each coefficient. That is, 13 pseudo-values (f_Li) were calculated, with a different observation omitted each time. The final jackknife coefficients (J_L) were then obtained by averaging the corresponding pseudovalues.

Jackknife estimation eliminates biases which are inversely proportioned to sample size (Mantel, 1967). It has also been shown that total biases of the jackknife estimate are less than the biases obtained from conventional estimation techniques (Gray and Schucany, 1972). Moreover, jackknife estimation is particularly efficient for small samples and for measures that divert from normality (Ireland and Uselton, 1975). If an unbiased estimator exists, it will be found by the jackknife procedure (Crask and Perreault, 1977).

Results.  Table 2 presents the jackknife estimators of the correlation and regression coefficients. All the regression coefficients are statistically different from zero at a significance level of .10 (using only the 13 cases as a base and not the total number of observations on respondents). The negative direction of both regression coefficients is consistent with the hypotheses; that is, the results suggest that the greater the availability of substitutes and the more frequent the buyer-seller interaction, the less likely it is that dissatisfied consumers complain.

TABLE 2

CORRELATION MATRIX AND MULTIPLE REGRESSION ANALYSIS (JACKKNIFED ESTIMATORS)

The overall F-ratio of 11.344 of the regression model is highly significant, but it is the explanatory power of the model that is noteworthy. As much as 70% of the variance in the Voice Index is accounted for by the two independent variables. The results support the contention from theory that the availability of alternatives is a very powerful "suppressor" of consumer voice. As can be seen from the correlation matrix this variable alone accounts for more than 54% (-.734²) of the variance in the criterion variable.

A distinct advantage of the jackknife estimation procedure, as it relates to small samples, is its provision for assessment of coefficient stability. Table 3 summarizes the diagnostic statistics concerning the distribution of pseudovalues. The low variance for most values is evidence of the reliability one can place in the results.

DISCUSSION

Sample restrictions and limitations of measurement methodology notwithstanding, the results of this study strongly suggest that a simple model, derived from some fundamental tenets of economic theory, may explain a good portion of consumer complaining behavior. The findings also suggest that structural variables deserve more attention in consumer research. To the extent that the market structure allows for consumer mobility, it is (indirectly) implied that dissatisfied consumers use the market to improve their position, thereby punishing faltering firms by withdrawing patronage.

The findings have interesting implications at both the micro and the macro levels. At the micro level, it appears that the availability of substitutes providing for consumer mobility requires effective marketing mix management to insure customer satisfaction. By simply counting the number of competitors in the yellow pages and estimating the frequency of purchase (or interaction), it seems that the retailer would be able to get a reasonable picture of revenue losses due to customer dissatisfaction. As the level of competition increases, the dissatisfied customer is more prone to exit than to voice a complaint. As a result, the firm might not be aware of the existence of specific customer dissatisfactions or their causes.

TABLE 3

JACKKNIFED ESTIMATORS OF COEFFICIENTS AND DIAGNOSTIC STATISTICS

In the absence of competition, the firm becomes less vulnerable to consumer dissatisfaction reaction and perhaps less sensitive to consumer concerns, because complaints will then be the dominant vehicle for consumer protest and the threat of direct revenue losses is minimal. On the other hand, in the monopoly situation where the exit alternative is either non-existent (e.g., water service) or severely circumscribed (e.g., telephone service), the importance of voice increases from a communication point of view.

Thus, a case can be made for the encouragement of voice for both the monopolist and for the firm in a competitive environment. For the monopolist, it would be equivalent to strengthening the democratic process by having consumers articulate their views and influence corporate decision making; for the competitive organization, it would be a means of inducing change before suffering revenue losses. As indicated by the study findings, it would be more difficult for the firm in a competitive environment to develop the voice mechanism. For company survival and profitability, it would also be more crucial.

The normative macro implications are perhaps subject to more controversy. One reason for this is that two major forces of societal and economic development--the political democracy model and the capitalistic economic model--appear to take opposite sides. According to normative economics, it is in the best interest of consumers to communicate their desires and dissatisfactions via market forces. The competitive firm is presumed to respond to consumer withdrawal and patronage shifting by improving its offerings. Voice does not have a place in normative economic theory. The political model, which otherwise is in close harmony with the economic model, puts the emphasis on voice. Articulate, active, and vocal citizens are held as essential for a healthy democracy.

Given its importance, and the previous neglect by cons u met research, the present interest in consumer post-purchase behavior is likely to gain increasing momentum. An understanding of how people react to unsatisfactory products and seller practices is of vital import for individual business firms, nonprofit organizations, and for public policy. The implications range from individual complaint-handling procedures to the formulation and enforcement of antitrust policy. Perhaps the attempt of this study in applying basic economic theory to consumer complaining behavior will stimulate further inquiry in this direction.

REFERENCES

Anderson, R. E. (1973), "Consumer Dissatisfaction: The Effect of Disconfirmed Expectancy on Perceived Product Performance," Journal of Marketing Research, 10, 38-44.

Best, A. and Andreasen, A. R. (1977), "Consumer Response to Unsatisfactory Purchases: A Survey of Perceiving Defects, Voicing Complaints, and Obtaining Redress," Law and Society, 11, 701-742.

Crask, M. and Perreault, W. D., Jr. (1977), "Validation of Discriminant Analysis in Marketing Research," Journal of Marketing Research, Vol. XIV, 60-68.

Day, R. E. and Bodur, M. (1978), "Consumer Response to Dissatisfaction with Services and Intangibles," in H. K. Hunt (ed.), Advances in Consumer Research, Vol. V, 263-272.

Day, R. L. and Landon, E. L., Jr. (1977), "Toward a Theory of Consumer Complaining Behavior," in A. G. Woodside, J. N. Sheth, and P. E. Bennett (eds.), Consumer and Industrial Buying Behavior, (New York: North-Holland), 425-437.

Diamond, S. L., Ward, S., and Faber, R. (1976), "Consumer Problems and Consumerism: Analysis of Calls To A Consumer Hot Line," Journal of Marketing, pp. 58-62.

Didow, N. M. (1978), "Dissatisfaction in the Grand Old Dowager--Analysis of the Consumer Affairs Task Force Telephone Survey," Consumer Affairs Task Force, City of Evanston, Illinois.

Fornell, C. (1976), Consumer Input for MarketingDecisions: A Study of Corporate Departments for Consumer Affairs, New York: Praeger.

Granbois, D., Summer, J. O., and Frazier, G. (1977), "Correlates of Consumer Expectations and Complaining Behavior,'' paper presented at the Second Annual Conference on Consumer Satisfaction/Dissatisfaction and Complaining Behavior, University of Indiana, Bloomington, Indiana, April 20-22.

Gray, H. L. and Schucany, W. R. (1972), The Generalized Jackknife Statistic, New York: Marcell Dekker, Inc.

Gr°nhaug, K. (1977), "Exploring Complaining Behavior: A Model and Some Empirical Results," in W. D. Perreault, Jr. (ed.), Advances in Consumer Research, Vol. IV, 159-163.

Hannigan, J. A. (1977), "The Newspaper Ombudsman and Consumer Complaints: An Empirical Assessment," Law & Society, Vol. 11, 679-699.

Hirschman, A. O. (1970), Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States, Cambridge, Mass.: Harvard University Press.

Ireland, M. E. and Uselton, G. C. (1975), "A Distribution-Free Statistic for Management Scientists," in W. W. Menke and C. H. Whitehurst (eds.), Proceedings, Eleventh Annual Meeting of the Southeastern Chapter, TIMS, 25-26.

Kraft, F. B. (1977) "Consumer Perceptions of Business Fraud: A Key Variable in Complaint and Repatronage Behavior,'' paper presented at the Southwestern Marketing Meetings, New Orleans, March 23-26.

Landon, E. L., Jr. (1977), "A Model of Consumer Complaint Behavior," paper presented at the Second Annual Conference on Consumer Satisfaction/Dissatisfaction and Complaining Behavior, Indiana University, Bloomington, Indiana, April 20-22.

Landon, E. L., Jr. and Emery, D. R., Jr. (1974), "Causal Attributions of Consumer Dissatisfaction as a Predictor of Consumer Complaint Action: A Preliminary Report," paper presented at the Association for Consumer Research Meetings, Chicago, Illinois, November 9.

LaTour, S. and Peat, N. C. (1978), "Effects of Prior Experience and Quality of the Best Alternative on Consumer Satisfaction, Complaints, and Purchase Termination Responses to Quality Decline," working paper series #5, Center for Consumer Research, College of Business Administration, University of Florida.

Lawther, K., Krishnan, S., and Valle, V. (1978), "The Consumer Complaint Process: Directions for Theoretical Development," paper presented at the Third Annual Conference on Consumer Satisfaction/Dissatisfaction and Complaining Behavior, Chicago, Illinois, October 5-7.

Mantel, N. (1967), "Assumption-Free Estimators Using U-Statistics and a Relationship to the Jackknife Method," Biometrics, 23, 567-571.

Olsen, J. C. and Dover, P. (1976), "Effects of Expectation Creation and Disconfirmation on Belief Elements of Cognitive Structures," in B. B. Anderson (ed.), Advances in Consumer Research, Vol. III, 168-175.

Olshavsky, R. W. and Miller, J. A. (1972), "Consumer Expectation, Product performance and Perceived Product Quality," Journal of Marketing Research, 9, 17-21.

Robinson, L. M. (1978), "Consumer Complaint Behavior: Review With Implications for Further Research," paper presented at the Third Annual Conference on Consumer Satisfaction/Dissatisfaction and Complaining Behavior, Chicago, Illinois, October 5-7.

Stern, L. W. and E1-Ansary, A. I. (1977), Marketing Channels, Englewood Cliffs, N. J.: Prentice-Hall, Inc.

TARP (Technical Assistance Research Programs) (1976), "A National Survey of the Complaint-Handling Procedures Used by Consumers," report prepared for the Department of Health, Education, and Welfare, Washington, D. C., Office of Consumer Affairs.

Tukey, J. W. (1958), "Bias and Confidence in Not Quite Large Samples," (abstract) Annuals of Mathematical Statistics, 29, 614.

Warland, R. H., Herrmann, R. O., and Willits, J. (1975), "Dissatisfied Consumers: Who Gets Upset and Who Takes Action?" Journal of Consumer Affairs, 9, 148-163.

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