Discussion of Psychological Contributions

ABSTRACT - The three research papers on deviant behavior, structure of human values, and measurement of social power have one thing in common: They are all psychological contributions to the consumer behavior area. After evaluating each paper, several suggestions are made about the future directions in each area with respect to theory development, methodological issues, and testing propositions in the consumer behavior context.


Jagdish N. Sheth (1979) ,"Discussion of Psychological Contributions", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 355-358.

Advances in Consumer Research Volume 6, 1979      Pages 355-358


Jagdish N. Sheth, University of Illinois


The three research papers on deviant behavior, structure of human values, and measurement of social power have one thing in common: They are all psychological contributions to the consumer behavior area. After evaluating each paper, several suggestions are made about the future directions in each area with respect to theory development, methodological issues, and testing propositions in the consumer behavior context.


At a first glance, the three papers seem to have very little in common except that each is a psychological contribution to the area of consumer behavior. Even this may be argued as only partially true because the same issues are equally researched in sociology and to some extent in consumption economics. However, a more careful look at the three research papers reveals at least four distinct common attributes.

First, all of them are anti-attitudinal papers and explicitly deviate from the more common multiattribute attitude research. This is remarkable in light of the extreme popularity of multiattribute models in consumer behavior, and may indicate that the tide of attitude research may be receding in consumer behavior.

Second, all of them are direct borrowings from psychological and sociological disciplines. However, unlike previous blind borrowings from the behavioral and social sciences, each paper has attempted either to critically examine what is theorized or measured in psychology or add richness of thinking and measurement to the original contributions. For example, Mills and Bonoma strongly dispute the labeling theory or deterrence research as explanations for the deviant consumer behavior, and they offer a new approach in terms of relative balance of power between the buyer and the seller in the market place. Similarly, Gutman and Vinson propose and test that Rokeach's Value Survey consists of many values which are not mutually exclusive but cluster together either in a complimentary or in an inverse manner. Finally, Swasy attempts to develop a psychometric scale with which to measure the six-factor power theory proposed by French and Raven.

Third, all the three papers deal with more fundamental and general constructs reminiscent of the golden days of personality and motivation research. Thus, Gutman and Vinson rely on fundamental human values as a relevant construct in consumer behavior and both Swasy and Mills and Bonoma rely on social power and balance of power as the relevant constructs in consumer behavior.

Fourth, all of them are pilot studies on a convenient sample of college students, and utilize a survey research approach. Of course, the sample sizes and specific instruments vary from one study to another.

We will first evaluate each research paper in terms of its strengths and weaknesses and then offer some suggestions for future research and theory in each area.


In light of the recent zeal and almost missionary approach toward protecting the consumer against the malpractices of the market place, it is refreshing to note that someone has attempted to research and provide a theoretical foundation on shoplifting, vandalism and deception practiced by consumers against the marketer.


The biggest strength of the paper is the choice of the problem area. Deviant consumer behavior especially among the young adults has reached crisis proportions as a part of the overall syndrome of juvenile delinquency. Unfortunately, sociologists and psychologists have ignored the consumer behavior aspects in favor of deviant social behavior creating the need for us to research the area.

A second area of strength in the paper is the proposed context-specific balance of power approach as a way to understand and predict deviant consumer behavior. The classification of retail stores having images of unilateral, bilateral, and the group welfare power structures is a new approach to retail store image research.

A third strong point about the paper is its attempt to actually measure and test specific propositions derived from the theory in the retailing context. As I have commented elsewhere (Sheth, 1974), it is almost mandatory that authors of new theories also develop test instruments and hopefully carry out some empirical-experimental research on their theoretical perspectives.


Unfortunately, this research paper has more weaknesses than strengths. I will summarize the weaknesses in terms of three categories: conceptual, methodological, and analytical weaknesses.

Conceptually, it is very difficult to agree that all deviant consumer behavior is a function of the perceived balance of power between the store and the consumer. It is much more logical to suggest that deviant consumer behavior is a function of the interaction of three distinct factors: Opportunity to shoplift, vandalize or deceive the store, personal and socioeconomic characteristics of the consumer, and the perceived balance of power. Without such a complex three-way interaction model, it is not possible to explain why (a) some consumers never shoplift from any power base store, (b) some weak power stores such as the "mama and papa" stores get vandalized and held up, and (c) stores with comparable sizes, sales and power structure have different rates of shoplifting, vandalism and deception by consumers.

In addition, there are two other conceptual weaknesses. First, why should there be only three types of power systems? There is no rationale about the typology in the paper, and I presume that it is fully detailed in the Bonoma paper referred to in the study. Second, several subhypotheses are highly collinear and, therefore, unnecessary to be tested. For example, larger stores have greater sales volumes and spend more on promotional campaigns. Given this multicollinearity among size, sales and promotion, it is unnecessary to have three separate tests of hypotheses.

The methodological weaknesses center around the sampling and the data collection procedures. It is unfortunate that despite the national statistics about juveniles being the main segment of deviant consumers and its use to justify the power theory proposed by the authors, they chose to sample university students who are considerably older and more mature. Similarly, the authors chose to use the scenario method which is not context specific enough. There are too many other specific factors besides the general stereotypes which attract deviant behavior to a specific retail store. In fact, this is somewhat incongruent in light of the heavy emphasis the authors give to making the power theory more context specific.

The analytical weaknesses can be specified with respect to each of the two pilot studies. The first pilot study involved a chi-squared analysis of the relationship between size of the department store and level of shoplifting. Since shoplifting incidence was measured as a percentage of sales, and store size is more or less a collinear variable of sales, the rejection of the null hypothesis is more due to contamination of the dependent-independent variables than a true relationship between level of shoplifting and store size. It is also unfortunate that the amount of covariance contained between the two variables is not reported in the study.

In the second pilot study, the results of scenario manipulations are highly tentative. Since most of the mean values range between 11.0 and 15.0 on a scale of 5 to 25, it is obvious that the statistical significance is achieved by sample size rather than by amount of variance explained. In other words, rejection of the null hypothesis is not sufficient to accept that the experimental hypothesis is proved or successfully tested. We need to compare the results not against null hypotheses but against a priori hypotheses (Overall and Klett, 1974).

In summary, however, Mills and Bonoma must be applauded for pointing at a very relevant but neglected problem area of consumer behavior, and for making early efforts at both theorizing and testing the phenomenon of deviant consumer behavior.


This paper is an attempt to critically examine the structure of values as proposed by Rokeach. While the authors seem to accept the Rokeach theory of values and its relevance to consumer behavior, they are reluctant to accept the rank ordering of 36 terminal and instrumental values proposed in the Value Survey as a method of measuring values. They believe that many of the values are highly collinear and, therefore, cluster together and possess fewer underlying dimensions. In order to understand this clustering and reduced dimensionality of values in the Value Survey, they utilize the multidimensional scaling (MDS) approach.


The primary strength of this paper is to point out weaknesses of the Rokeach's Value Survey instrument which had been utilized in the past without a careful methodological look and in a straight-forward manner in consumer behavior studies. The authors very correctly point out that the instrument is very difficult to administer and obtain error-free responses because of similarities of many values and the rank ordering procedure recommended by Rokeach.

It is indeed refreshing to note that researchers in consumer behavior have the confidence and maturity to question theories and measurement instruments proposed by psychologists rather than to blindly borrow them as was so typical of the era of personality research in marketing.


The choice of MDS as a method to understand clustering and dimensionality of values puzzles me. It is simply a very complicated procedure requiring numerous compromises in data collection and data analysis to warrant its usage for clustering purposes (Sheth, 1976). Most of my negative comments stem more from the use of MDS as a method to recover underlying dimensionality of Rokeach's values.

First, the problem of rank ordering of values is compounded in the MDS approach since it requires ranking of pairs of values. Thus, to obtain distances or similarities among 36 values in the Rokeach scale would require a total of 666 stimuli to be ranked which is admittedly a more difficult task than ranking 36 values. In fact, MDS users have often resorted to fractional factorial designs to reduce the comparison task. Even in this study, the authors limited themselves to only 14 of the 36 values in order to make the task of making judgments manageable. Furthermore, they were forced to use a distance rating scale for each pairwise combination rather than the ordinal rank ordering procedure because of still very large number (91) of pairs of values. The use of MDS, therefore, seems to be inconsistent with their rightful criticism of rank order requirement in the Rokeach scale.

Second, the three-dimensional MDS map cannot be generalized to the full set of values since it is derived from only a subset of 14 values. Its dimensionality is likely to increase as more values are added.

Third, it is not always possible to attribute the dimensions as due to perceived similarity of stimuli. The dimensions recovered in the MDS are often reflections of sample heterogeneity and dimensionality of a single stimulus. Both of these seem to be present in this study as indicated by large ranges of subject saliences for each of the three dimensions, and by mixed loadings of a value on two or more dimensions (Anderson and Sheth 1976).

Fourth, the three-dimensional INDSCAL solution is not at all meaningful despite the brave efforts by authors to find a structure in the data. For example, the first two dimensions do not have a single value loaded above 0.467 which is clearly a poor fit between values and the underlying dimensions. The third dimension has only one value loaded high enough to justify interpretation of that dimension in terms of clustering of values. As Redinger and Sheth (1977) have pointed out, this is to be expected when the data collection in MDS requires too many pairwise comparisons and the distance scale used is restricted to a seven point scale as was the case in this study.

Finally, except for the fourth behavior (getting high), all other behaviors have very comparable average rating as perceived instruments to attaining or blocking all of the 14 values. In other words, there is no real discrimination in terms of certain clusters of values being attained or blocked by certain behaviors as one should expect if values are useful in social or consumer behavior.

Rather than using MDS, the authors would have been considerably better off in deriving value structures if they had relied upon factor analysis or cluster analysis of the distance matrix between values. In fact, structure of values could be also revealed by simple or partial correlations among values in terms of their desirability or saliency to the individual.


The Swasy paper on measurement of social power is an excellent example of psychometric scaling and scale development. This paper has far more strengths than weaknesses and, therefore, it is a pleasure to discuss it.


The most important strength of this paper is the careful and rigorous procedures followed by Swasy in operationalizing, measuring and validating the Social Power scale. There is a very succinct and clear description of French and Raven typology of social power, an excellent critique of past efforts at operationalizing the construct, and a superb description of the procedures followed in deriving the social power scale. The author clearly has a very balanced background in both social psychology and testing and measurement which is often rare in psychology.

Swasy is also a very articulate writer. His paper reads with complete clarity and he is equally good in describing both theoretical and methodological aspects of his study.


As I mentioned before, there are only minor weaknesses in this paper. Most of them relate to author's assertations about the goodness of fit between his theory and the data.

First, the amount of total variance retained by all factors seldom exceeds 53 percent. Similarly, the amount of variance summarized in the general factor is seldom above 35 percent. Those are simply very low percentages. It indicates that the item overlap is not as strong as one would expect in scale development. My suspicion is that this is due to heterogeneity of his sample despite the fact that they are all undergraduate accounting majors at UCLA (Sheth and Armstrong, 1969). Alternatively, it suggests that there is a lot of response error in the data among his subjects which is highly ad hoc and specific to each respondent.

A second weakness is in the validation stage. The results are not as clear-cut as Swasy would make us believe with respect to specific situations reflecting each type of social power base. For example, we should expect a true bimodal distribution of all situations on a particular social power base with mean values dichotomized at or near the extremes of the scale range between 3 and 15. The actual distributions of the mean values across all six social power bases are much narrower, and represent only one third of the possible maximum difference.

Furthermore, a particular situation is often having a large mean value across two or more social power bases. For example, this is true of situation 12 which reflects both reward and information bases of social power. Similarly, situation 10 reflects both expert and referent bases of social power. This might indicate that the six factor typology of social power is not as mutually exclusive as we are led to believe. It would be interesting to test this by examining the factor score correlations among the six uni-factors.

Finally, it is puzzling to note that the best fitting situations for each power base do not achieve comparable mean values. For example, while the mean values of the two situations depicting the expert base of social power is above 11, it is only 9.75 and 6.07 for the legitimate base.

Overall, this paper is excellent. My only concern is that it has very little to do with consumer behavior, although it can be easily applied to study the process of persuasion and communication in marketing.


While it is relatively easy to discuss and critique other people's research efforts, it is much harder to offer constructive suggestions as to where to go from here. There is no question that some time in the future, we will look back at each of the three studies as pioneering efforts in offering new directions in consumer behavior. I will attempt to offer some suggestions as to where the future research is likely to, or ought to emerge in each of the three areas.

Value Research

The biggest payoff in value research is likely to be in the development of consumption specific values, identifying their structure and creating a standardized psychometrically adequate scale. I will call this as the need to develop a Consumption Values Inventory (CVI). I think the real payoff in terms of explanation, prediction, dissemination and managerial use is likely to be somewhere between very fundamental values proposed by Rokeach and the multiattributes emphasized in attitude research. Furthermore, it should be specific, if not unique, to consumption behavior (Sheth, 1974). It is interesting to note that Daniel Yankolevich (1964) demonstrated the usefulness of consumption values for segmentation purposes at least a decade ago. Similarly, it is much more important and probably socially more relevant to examine consumer choice behavior at product class level rather than at brand choice level in linking consumption values to specific buying behavior. Due to the extreme market variety, predominantly artificial brand differences, and low degree of consumer involvement, it is probably neither possible nor relevant to build value based research at the level of brand choice behavior.

Third, we need to develop a comprehensive and large scale research on a representative national sample of the population to fully comprehend the missing linkages between a set of values and a set of product choice behaviors. For example, it is not very difficult to think of using the BLS typology of consumer expenditures and link them with consumption values. This will generate research findings which will be much more meaningful to both the public policy makers and the private marketing practitioners.

Finally, there is a considerable degree of opportunity to learn more about cross-cultural differences based on consumption values. This may explain more why the same product flourishes in one country and fizzles out in another. In many ways, Dichter (1962) presented some interesting cross-cultural anecdotes based on value research.

Deviant Consumer Behavior

The problem area of deviant consumer behavior needs several new research directions to be more substantive than issue-oriented research.

First, we need a much clearer definition of deviant consumer behavior which will offer a good typology of deviant consumer behavior by time, place and nature of deviancy. It is not enough to limit it to shoplifting, vandalism and deception in the retail store setting. This will be a very difficult task if the research on consumer protection and consumerism is any guide as the other side of the coin.

Second, deviant consumer behavior is likely to be much more fruitful if the sociology of consumption tradition is brought to bear in theorizing about it rather than relying upon the psychological processes. This will inevitably force thinking and research at a more aggregate level in terms of groups or classes of consumers, and types of opportunity settings for deviant consumer behavior to manifest itself. It is my belief that any microlevel modeling at the individual consumer level will not be fully complete since there are too many time, place and person specific situational influences which intervene between a consumer's propensity for deviant behavior and its actual manifestation.

Finally, we need far more complicated theory than what has been proposed so far. As I mentioned earlier, deviant consumer behavior is more likely to be an interactive process. I have identified three factors: opportunity, consumer values possibly indicated by demographics, and perceived legitimacy of behavior in a particular setting. There can be more factors but the crucial point is that deviant consumer behavior is an interactive outcome of several factors.

Measuring Social Power

It is much more difficult to imagine the directions of future research related to social power. The concept has been already heavily utilized in terms of buyer-seller interaction process, opinion leadership and interpersonal communication.

I think the best way to bring social power theory more closely to consumer behavior is to drop the adjective "social" and focus more sharply on the role of power in a transaction or exchange between the buyer and the marketer. It would be most interesting to translate the traditional four elements of marketing mix (product, place, price and promotion) into the six bases of power (reward, coercion, legitimate, expert, referent, and information). In other words, is there a dominance of one or more power bases in each element of the marketing mix? Similarly, does the consumer have any power base, and if so, how does he acquire them?

A second area of research is to understand consumer perception of various power bases under different types of competitive structures in the market place. For example, a monopoly or even an oligopoly structure may be perceived as coercive power, regulated monopoly or perfect competition as legitimate power, service industry as expert or information power, and monopolistic competition as reward power. This may be most useful in assessing corporate image and the role of business institutions in society.

A third area of future research is to correlate consumer characteristics with preferences for specific power bases. For example, do higher socioeconomic people prefer information or expert base and lower socioeconomic people prefer reward or referent base of power? Do children prefer reward or coercive power bases more than adults, for example?

Finally, it will be most interesting to link specific products and services as means to exercising certain power bases. For example, food is often used for a reward base of power, appliances for a referent base, technical books for an expert base, mass media for an information base, and dangerous objects such as guns for the coercive base of power.

I hope these suggestions about the future directions of research in each of the three areas is useful to the authors and others in consumer behavior.


Gary A. Anderson and Jagdish N. Sheth, "A Comparative Analysis of Multidimensional Scaling of Portfolio Investment," Faculty Working Paper No. 344, October 13, 1976.

Ernest Dichter, "The World Customer," Harvard Business Review, 40 (July - August, 1962), 116-120.

John E. Overall and C. James Klett, Applied Multivariate Analysis, (New York, McGraw Hill, 1972), Chapter 17.

Robert Redinger and Jagdish N. Sheth, "Multidimensional Scaling of Binary Data for Homogeneous Groups," Faculty Working Paper No. 411, June 29, 1977.

Jagdish N. Sheth, "Problems and Potential of Multidimensional Scaling & Conjoint Analysis," Paper presented at AMA Educators Conference, Memphis, August 8-11, 1976.

Jagdish N. Sheth, "The Next Decade of Buyer Behavior Theory and Research," in J. N. Sheth (ed.) Models of Buyer Behavior, (New York, Harper & Row, 1974), 391-406.

Jagdish N. Sheth and Scott Armstrong, "Factor Analysis of Marketing Data: A Critical Evaluation," in P. R. McDonald (ed.) Marketing Involvement in Society and the Economy, (American Marketing Association, 1969), 137-144.

Daniel Yankelovich, "New Criteria for Market Segmentation,'' Harvard Business Review, 42 (1964), 83-90.



Jagdish N. Sheth, University of Illinois


NA - Advances in Consumer Research Volume 06 | 1979

Share Proceeding

Featured papers

See More


L7. The Joy of Shopping: Reconciling Mixed Effects of Positive Emotions on Shopping Behavior

Kelley Gullo, Duke University, USA
Duncan Simester, Massachusetts Institute of Technology, USA
Gavan Fitzsimons, Duke University, USA

Read More


K7. Responses to Online Behavioral Advertising Disclosures: Effects of Disclosure Source Trustworthiness and Message Type on Advertising Outcomes

Iris van Ooijen, University of Twente

Read More


Exiting Etsy? When Collaboration Among Market Co-Creators Come Undone

daiane scaraboto, Pontificia Universidad Católica de Chile
Eileen Fischer, York University, Canada

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.