Deviant Consumer Behavior: a Different View

Michael K. Mills, University of Pittsburgh
Thomas V. Bonoma, University of Pittsburgh
ABSTRACT - Previous research in the area of deviant consumer behavior in retail contexts has been theoretically and methodologically exploratory, with correspondingly little ability to get at the real problem--the motivation behind such behavior. A new theoretical perspective is offered which utilizes three prototypical power systems in a conflict based approach to the problem. Pilot tests of the theory's applicability to explaining theft problems are presented which show encouraging support for the theory.
[ to cite ]:
Michael K. Mills and Thomas V. Bonoma (1979) ,"Deviant Consumer Behavior: a Different View", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 347-352.

Advances in Consumer Research Volume 6, 1979      Pages 347-352


Michael K. Mills, University of Pittsburgh

Thomas V. Bonoma, University of Pittsburgh


Previous research in the area of deviant consumer behavior in retail contexts has been theoretically and methodologically exploratory, with correspondingly little ability to get at the real problem--the motivation behind such behavior. A new theoretical perspective is offered which utilizes three prototypical power systems in a conflict based approach to the problem. Pilot tests of the theory's applicability to explaining theft problems are presented which show encouraging support for the theory.


Deviant consumer behavior is defined here as behavior in a retail store that society considers inappropriate or in conflict with a previously accepted societal norm. Some examples are shoplifting, price altering, destroying or damaging merchandise, marring in-store fixtures or restrooms, and consumer fraud. Such behavior is rampant and is on the increase. Indeed, one observer argues that the Gross National Dishonesty Quotient is rising steadily (Moneysworth, July 24, 1972).

Deviant consumer behavior involves a substantial cost to merchants. Stolen merchandise alone costs retailers between sixty and one hundred million dollars annually (Wright, 1972). FBI figures show that the cost of recovered merchandise from arrested shoplifters was approximately 22 million dollars in 1975 (Uniform Crime Reports, 1975); the bulk, of course, goes unmeasured. The total estimated cost to retailers in shoplifting-related expenses totaled three billion dollars in 1959 (U.S. News and World Report, March 2, 1970). A conservative current estimate would place this close to twenty billion dollars annually.

In these inflationary times, stock "shortages" are a major marketing problem. Yet, there has been relatively little research done on deviant consumer behavior or the motivation behind such behavior (Cameron, 1964; Cohen and Stark, 1974; Geurts et. al, 1975; Pedrini, 1972). Additionally, these studies and other work in the area suffer from the lack of an actionable theoretical base.


Essentially all previous studies have approached the problem through labeling theory (Becker, 1960; Erick-son, 1962), or have focused attention on the social psychological variables suggested by the deterrence literature (Andenaes, 1966; Gibbs, 1966). However, vast disagreement exists even among practitioners of each of these theoretical perspectives as to their applicability, or usefulness in explaining, controlling, or predicting deviant phenomena (Meade, 1974; Tittle, 1975). Further, neither labeling nor deterrence theory enjoys unequivocal or even firm support in the literature and, as some suggest, much of that which is cited in regard to the issue is of dubious value (Tittle and Logan, 1973).

Thus, we suggest that the theoretical foundation upon which many current views of deviant consumer behavior are based is weak. Further, the competing theoretical orientations available allow for no integrated approach to the problem. Because of the increasing urgency of deviant consumer behavior to retailers and to marketers generally, a new theoretical perspective seems called for.


This study utilizes a new theoretical perspective to explain deviant consumer behavior in retail stores, as well as the motivations behind such behavior. Basically it is our belief, supported by previous retail image research (Mills, 1976), that retail stores transmit social communications to their customers through their physical and marketing image. We believe consumers in "interaction" with stores react to such signals and modify their behavior accordingly. Thus, the deviant behaviors consumers engage in vis-a-vis retail stores are also social communications-generated in response to messages contained in store "image" manipulation.

This approach to the problem extends both retail image studies and a power-context interaction theory of Bonoma (Bonoma, 1976). In this view, consumers' perceptions of stores' relative power play an important role in understanding the motivations underlying both nondeviant and deviant consumer behavior. Research on such power perceptions may then suggest how power can be manipulated in social interaction (Bacharach and Lawler, 1976). Shoplifting, vandalism and other acts may thereby be removed from a nonactionable category of personal "character disorders" and put into the category of social communications, making them available for managerial control through increased attention to store image.

This general approach, novel in its application, lies on a solid theoretical and practical base. The concepts of conflict, cooperation, trust, power and social influence form a theoretical foundation that has generated much interest and has enjoyed far-ranging application in the social disciplines (Boulding, 1962; Deutsch 1973; Gamson, 1968). Bonoma (1976), however, has recently extended much of the power literature in arguing for a context-specific definition of power. He contends that a somewhat finer-grained examination of social episodes in which the constructs of power, conflict, cooperation and trust may be implicated shows clearly that these episodes fall not into one grand conceptual category, but into three. These are the unilateral, bilateral, and group welfare systems.


We claim that the three systems outlined by Bonoma have their equivalents in retail store settings with corresponding implications for the analysis of deviant consumer behavior (Mills, 1976). For example, a large department store chain might make it explicitly clear to the consumer that the chain is strong and the consumer is weak by its size, by its strict adherence to a no-sale policy, by its tight credit policies, or by its appeal to one very small (e.g., upper-middle class) segment of the retailing market. Another chain, contrarily, might encourage more equal comparisons between the consumer and itself by vigorous branching, by appeal to a wide segment of the target market, and numerous sales promotions. Even the provision or lack of parking may affect the power image transmitted to the customer. It is our claim that when stores are broken up in this fashion and put into one of the three theoretical categories, clear and consistent differences will be found in deviant consumer behavior occurring in various types of stores. The point is that store image generates impressions of store power in consumers; they may feel weaker, about equal, or a "part of" the selling organization, and will steal and vandalize more or less accordingly.

Evidence from previous empirical work would tend to support our contention that the power context of a store influences the degree of customers' deviant acts. In several studies (Kraut, 1976; Robin, 1963), highly ranked reasons given for shoplifting included the fact that the person wanted to get back at the store, had political motives, or felt shoplifting was acceptable.

Power is implicated further by a second factor which shows that most shoplifting and other deviant consumer behavior is done by juveniles (Cohen and Stark, 1974; Kraut, 1976; Robin, 1963) -- the "weaklings" of adult society. Further, rebellion, alienation, and powerlessness are important reasons given by apprehended juveniles for their behavior (Pedrini, 1972). Involvement of the young age group is demonstrated by the fact that 35 percent of larceny-theft crimes cleared in the nation's cities are solved by arrest of persons under 18 years of age (Uniform Crime Reports, 1975).


The relationships specified above, the findings of previous research in the area and the researchers' best conjectures suggest the following theoretical assumptions and consequent hypotheses:

H1: The power context of a retail store as perceived by the customer and management can be identified and classified into one of our three types. Each type will differentially influence the incidence of deviant consumer behavior occurring in that store. Theoretical Justification: Retail stores can be classified as having more or less relative expendable influence resources vis-a-vis single consumers. Grossly speaking, those having the most power toward the unilateral ideal-point; those with the least, toward the group welfare.

H2: Perceived power differences between stores will vary as a consequence of various store attributes. Specifically,

H2a: Larger stores will be perceived as having a greater amount of power vis-a-vis the customers than smaller stores. Theoretical Justification: Size is an obvious indicator of resource base.

H2b: Stores with a greater number of branch operations will be perceived as having more power vis-a-vis the consumers than will stores having a fewer number of branches. Theoretical Justification: Same as for H2a, with the possible exception that vigorous branching may be perceived as in customers' best interests and hence, give impressions of power equalization.

H2c: Stores catering to the elite or upper middle class shopper will be perceived as having greater power vis-a-vis the customer than will stores catering to the lower or lower middle class. Theoretical Justification: Stores capable of dealing with "powerful" consumers will be seen as more powerful than those dealing with "weak" consumers.

H2d: Stores with greater sales volume will be perceived as more powerful vis-a-vis the customer than will stores of lesser volume. Theoretical Justification: Dollar volume is an obvious indicator of resource base.

H2e: The greater the reliance on promotional campaigns or advertising by stores, the more will be the perceived power of the stores vis-a-vis the customer. Theoretical Justification: Promotional expenditures, like size and sales, indicate many disposable resources.

H3: The greater the customer's perceptions of power differences between store and customer, the greater will be the incidence of deviant consumer behavior as a counterpower move by consumers. Specifically,

H3a: The greatest perceived power differences and the greatest incidence of deviant consumer behavior should occur in the unilateral retail power system. Theoretical Justification: Implied by the theory and assumptions.

H3b: The least amount of perceived power differences and hence, the least incidence of deviant consumer behavior will occur in the bilateral retail power system. Theoretical Justification: Implied by theory and assumptions.

H3c: The amount of perceived power and the incidence of deviant consumer behavior of the mixed (bargaining) power system will fall somewhere in-between the unilateral and bilateral power systems. Theoretical Justification: Implied by the theory and assumptions.

H4: The retail power image of a given store will be manipulable by changes in the above attributes. Thus,

H4a: Rates of deviant consumer behavior will be manipulable by conscious setting of store image. Theoretical Justification: Necessary consequent of theory validation.


As a partial test of the soundness of the hypotheses an archival-based preliminary pilot test was undertaken. Archival data consisting of stock shortage figures for retail department stores, gathered and tabulated by the National Retail Merchants Association and published in their annual volume, Financial and Operating Results, was examined and analyzed for the fifteen-year period 1961-1975. Stock shortages here serve as a proxy variable for shoplifting, since NRMA officials estimate that over eighty percent of stock shortages represent theft. Several analyses were made, with the following results:

A chi-square analysis of levels of shoplifting to department store size was made; the six groupings listed in the raw data were combined into three groupings for analysis, and the criteria used to assign cell frequencies was based on low-to-high percentage values of shoplifting to sales for each year (this was done to assure adequate representation in each cell and to meet intuitive criteria). The null hypothesis that store size and level of shoplifting are independent was rejected at the p < .005 level. Store size and level of shoplifting are related, with larger stores having a reliably greater incidence of shoplifting occurrences.

A chi-square was also done for the six original groupings of department store sizes represented in the Financial and Operating Results to level of shoplifting. The null hypothesis that store size and level of shoplifting are independent was rejected at the .005 level. Again, larger stores appear to have a greater incidence of shoplifting.

While only a first step, this pilot test revealed very significant and encouraging results as to the soundness of one of the hypothesized relationships--that of the relationship of store size (one contributing aspect of perceived store power) to levels of shoplifting.


In order to get a "second opinion" on the validity of the theory which was not limited by the vagaries of an archival data base, we also conducted an experimental study at the University of Pittsburgh with college-student subjects.

The Instrument

We constructed six separate scenarios designed to operationalize one or another facet of the unilateral, bargaining, or bilateral power system within retailing establishments. The six scenarios were constructed in such a fashion that certain additional data, deemed useful commentary on the theory, could be collected besides a simple overall judgment of whether the theory could be generally confirmed by the test. Basically, the scenarios were as follows:

Scenario A described a large "prestige" department store catering to customers in the upper-middle income ranges, whose pricing structure was moderately high, which advertised frequently, had its own credit cards, a downtown location, and somewhat difficult access. Scenario A typified a department store in the unilateral power context.

Scenario B depicted a department store as large as Store A, but located in a suburban environment, catering to wide income segments, offering significant sales through promotional media, giving free parking, and so on. Store B was deliberately made to he a mixed description not characteristic of any of our three pure power-contexts.

Scenario C was an automobile dealership, whose primary characteristic was the negotiable nature of the pricing structure. It was representative of a bargaining power-context in a nondepartment store setting.

Store D represented a retail grocery cooperative in a university district. The store is owned by its members, who also provide the labor to run the store. The store aims at the lower and lower middle income classes, is not-for-profit in structure, and uses only word-of-mouth advertising. Store D was intended to characterize a group welfare, power-context store in a nondepartment setting.

Store E was a medium sized department store with a few branches, aiming at a full range of the income market, which sometimes negotiates for merchandise, which has liberal credit policies, which provides free parking, which does not advertise frequently, and depends on word-of-mouth about its good bargains to bring in customers. Store E was designed to illustrate a bargaining power-context within a department store setting.

Finally, store F was characterized as a small recreational clothing cooperative. Store F was designed to characterize a bilateral power-context department store setting.

Choice of these six scenarios gave us a number of interesting comparisons to perform including

(1) An overall comparison between the unilateral power-context scenario (Scenario A), and the bargaining (Scenarios C and E) and bilateral (Scenarios D and F) contexts,

(2) A within department store only comparison,

(3) An across retailing establishment comparison using say, a department store (Scenario A) for the unilateral power system, an auto dealer (Scenario C) for the bargaining system, and a vacationer's buying cooperative (Scenario D) for the bilateral system, and

(4) What we termed a "Big Chicken" test of a possible alternative account for our archival results. A colleague suggested that store size was possibly the only mediator of higher or lower levels of deviant consumer behavior. That is, he suggested the hypothesis that the bigger you are, the more likely you are to be the target of deviant consumer behavior. We not very kindly named this notion the "Big Chicken" theory, and set about to test it in this pilot study by constructing a scenario B which displayed a large department store with characteristics of each of the unilateral, the bargaining, and the bilateral systems, so that we could check his speculations against the pure types of systems depicted by the other scenarios.

Subjects and Method

The subjects were 66 MBA and undergraduate business students at the University of Pittsburgh. The sample was split about equally between graduate and undergraduate students, and about equally between male and female subjects. Subjects volunteered for the study with no course credit and no other incentive for completion of the survey.

Subjects were approached on an in-class basis, and asked to take a copy of the 'Retail Store Survey" home with them during the month of January, 1978. They were asked to complete the survey and to return it to a central University location by 1 February, 1978. Eighty surveys were handed out in this manner; 66 were returned as promised. Each subject received a booklet in which the six scenarios described above were placed in randomized order. This was done to minimize the effects of sequence on their responses.

Dependent Measures

Each scenario was followed by a list of fifteen questions, broken up into three types. The manipulation check section asked five questions about the type of store depicted in the scenario, or about S's relationship between the store and himself.

The remaining questions asked subjects whether they had ever destroyed or damaged merchandise in this type of store, whether merchandise was ever returned fraudulently to the store, whether shoplifting occurred in this kind of store, whether this kind of store would be vandalized, and whether fraudulent complaint would be made in this type of store. The questions were asked in two different ways. One set of questions asked whether the subject ever engaged in these actions. The other set of questions asked how likely subjects thought it would be that "someone" would engage in the behaviors so described. The "self" questions contained only a "yes," "no," or "doesn't apply" response listing. The second, "someone" set of questions, contained a five point Likert scale from very unlikely to very likely.

We included these "self" and "other" questions because we regarded it as highly unlikely, without a randomized research design, that subjects would actually report they had shoplifted, vandalized, or engaged in other fraudulent behaviors in a department store. Yet, it was our belief that subjects would be likely to endorse that "someone" might engage in such behaviors, and we were willing to accept this data as a preliminary commentary on the theory.

The "self" questions were combined to form a summary index which could range from 0 (all "no" or it "doesn't apply") to 5 (all "yes"). The "someone" questions were also summed to form a composite index varying from 0 (all "very unlikely") to 5 (all "very likely").

The three summary indices - one for the manipulation check, and one each for the "someone" and "self" questions - were subjected to a one-way repeated measures analysis of variance. Planned comparisons, consistent with the description given above, were performed on means generated for each of the scenarios.


Tables 1 and 2 respectively, present the results for the manipulation check and the "someone" results. The "self" results, as expected, yielded generally insignificant results (though in the direction predicted by the theory), and to save space are not reported here. Turning to Table 1, an assessment can be made of whether our somewhat casually constructed scenarios did indeed operationalize the three power contexts of unilateral, bargaining, and bilateral reactions. The F-test computed on the summary of the five manipulation check questions detailed tn Table 1 yielded a highly significant F-ratio of 40.16, reliable at the .0001 level.

The first check that was made on the individual scenario means comprising this F-ratio was an overall comparison. The results indicated that We were indeed successful in operationalizing the unilateral, bargaining, and bilateral systems. As Table 1 shows, the unilateral scenario generated the highest value on our summary index, the bargaining and intermediate value, and the bilateral system the lowest value. All differences (by two-tailed t-tests) were significant at the .05 level.

When the manipulation check was completed within department stores only, and across retailing contexts, the results again suggested that we for the most part successfully operationalized the three power systems. Even though some of the individual planned comparisons were nonsignificant, all means were in the predicted direction.





Finally, to check whether an equally large but not uni-laterally-disposed department store generated higher or lower (our prediction) perceived power than a strictly unilateral department store, we compared the mean for scenario A against that with scenario B. As Table 1 shows, the perceived power score for the mixed type department store was indeed lower than that for the strictly unilateral department store. However, the results were non-significant.

The results on whether subjects would endorse that "someone" would destroy merchandise, fraudulently return it, shoplift, or vandalize, or fraudulently complain in one or another of the store scenarios were considered to be the main ones of this study. Again, an overall comparison, a within department stores only, an across retailers comparison, and a "Big Chicken" comparison were made on the scenario means.

Over all the scenarios broken down into unilateral, bargaining, and bilateral power types, the means were in the theoretically predicted direction. Subjects reported that "someone" would more likely destroy merchandise, fraudulently return it, shoplift, vandalize, or fraudulently complain in the unilateral power system. They endorsed such behavior less extremely in the bargaining system, and least in the bilateral system. However, all means were not significant in planned comparisons. Only the unilateral/bilateral and bargaining/bilateral comparisons were significant. Nonetheless, the results must be taken as encouraging for the theory.

Within department stores only, a similar set of results was obtained. That is, the only reliable planned comparisons were between the unilateral and bilateral systems, and the bargaining and bilateral systems. In each case, the bilateral system elicited a lower rate of endorsement for these deviant consumer acts than did the other system. The unilateral-bargaining comparison was nonsignificant, although the bargaining scenario did generate a slightly higher level of endorsement of deviant consumer acts, a result which is contrary to the power-context theory.

Across retailers, the means were again all in the theoretically predicted order of unilateral > bargaining > bilateral. Again, only two of the three possible comparisons were significant. This time, the bargaining-bilateral comparison was not reliable, although the other two comparisons were.

Finally, the test of the "Big Chicken" theory, in which a strictly unilateral power-context store was pitted against one equally large but mixed in its characteristics, yielded a theoretically disconfirming result. That is, both stores were equally large, but the one in a suburban location, with free parking, with a more "bargaining oriented" base generated higher levels of endorsement for the deviant consumer act cited than the strict unilateral power-context department store. This result lends some credence to the argument that size alone may be an important factor in whether or not subjects would behave deviantly within retail contexts.


The results of our second pilot study, conducted with business students responding to scenarios operationalizing the three power systems, lend some strong support to the power-context theory. When combined with the results of the first pilot test, a strong case can be made that the power-context theory appears to offer a rational explanation for deviant consumer behavior, and is worthy of more controlled and complete tests under more realistic circumstances.


While the conceptual approach we have developed has yet to be more than tested in an exploratory way, it has potentially far-reaching implications on theoretical and practical grounds.

Theoretical Implications

Given support for our hypothesized relationships, much theoretical integration occurs. Support for the hypothesized relationships would offer a viable alternative to the labeling versus deterrence chaos that now exists in studying consumer deviance. The point is not that existing theories of deviance must be abandoned. Rather, they may need to be modified to include principles of "context translation" which explain how the theoretical constructs may be modified across various power systems.

Practical Implications

There are also practical benefits to be gained here, with a number of parties who stand to benefit from further research. One such party is the nation's retailers. If the power difference in a store-customer relationship affects deviant consumer behavior, this suggests enormous challenges to retailers in the development of a favorable retail image. Among other things, it would certainly suggest a "lowering of posture" or more efforts at conscious image building, or similar attempts to play down perceived store-customer power differences. Such modifications should result in decreased incidence of deviant consumer behavior resulting in increased profits for retailers.

Though precise specification of managerial image-control implications will have to wait for future research, we can pose some illustrative research questions: What are the effects of an "information hotline" in the store in reducing perceived power disparities, and hence, deviant behavior? Should a single branch-store be put in the downtown area, where it is accessible (good) but parking is impossible (bad), or should it go in the relatively inaccessible but less dense suburbs? What are the effects of a "cash-only" or "house credit only" policy on shoplifting? Do armed guards encourage shoplifting, or do they retard it?


If our view of deviant consumer behavior is supported by more data, then several advances may be toted up for the area in general:

- An integrated, interactive theory of deviant consumer behavior will be available for further development and refinement.

- The "onus" of deviant consumer acts like shoplifting will be shifted from the consumer's narrow shoulders to the joint interaction of the consumer and the retailer.

- A set of actionable and directly controllable variables, namely the marketing mix, will be identified as the major mechanism by which deviant consumer behavior can be reduced by the retailer.

- A great deal of money and other resources may be able to be partially reallocated from security and theft-control mechanisms toward more effectively meeting consumer needs at a more competitive price.


J. Andenaes, The General Preventative Effects of Punishment, University of Pennsylvania Law Review, 114, (1966) 949-983.

S. B. Bacharach and E. J. Lawler, "The Perception of Power," Social Forces, 55, (1976), 123-134.

H. Becker, Outsiders: Studies in the Sociology of Deviance, (New York: Free Press, 1960).

T. V. Bonoma, "Conflict, Cooperation and Trust in Three Power Systems," Behavioral Science, 21, (1976), 499-514.

K. E. Boulding, Toward a Theory of Peace. In R. Fisher (Ed.), International Conflict and Behavioral Science, (New York: Basic, 1964).

M. O. Cameron, The Booster and the Snitch, (New York: Free Press, 1964).

L. Cohen and R. Stark, "Labeling Theory and the Five-Finger Discount," Journal of Criminal Justice, 2, (1974) 23-36.

M. Deutsch, The Resolution of Conflict. (New Haven: Yale, 1973).

K. Erickson, "Notes on the Sociology of Deviance," Social Problems, 9, (1962), 307-314.

W. A. Gamson, The Strategy of Social Protest. (Homewood, IL: Dorsey, 1975).

M.D. Geurts, J. Reinmuth, and P. R. Andrus, "Researching Shoplifting and Other Deviant Consumer Behavior, Using the Randomized Response Research Design," Journal of Retailing, 50, (1975), 43-48.

J.P. Gibbs, "Sanctions," Social Problems, 15, (1966), 14, 147-159.

R. E. Kraut, "Deterrent and Definitional Influences on Shoplifting," Social Problems, 23, (1976), 358-368.

D. Matza, Becoming Delinquent. (Englewood Cliffs: Prentice Hall, 1969).

A. C. Meade, "The Labeling Approach to Delinquency: State of the Theory as a Function of Method," Social Forces, 53, (1974), 83-91.

M. K. Mills, "The Effects of Perceived Conflict, Power, and Influence on Retail Store Image: An Exploratory Study in Three Power Systems," unpublished manuscript, Graduate School of Business, University of Pittsburgh, 1976.

Moneysworth, July 24, 1973, 3.

T. Parsons, On the Concept of Influence. In T. Parsons (Ed.), Sociological Theory and Modern Society, (New York: Free Press, 1967).

D. T. Pedrini, "Psychology and the Problems of Shoplifting.'' St. Louis, MO: Paper presented at the Midwest Section of the American Business Association Conference, April, 1972.

"Rising Wave of Shoplifting," U.S. News and World Report, (March 2, 1970), 56-58.

G. Robin, "Patterns of Department Store Shoplifting," Crime and Delinquency, 9, (1963), 163-173.

C. R. Tittle, "Deterrents or Labeling?," Social Forces, 53, (1975), 399-410.

C. R. Tittle, and C. H. Logan, "Sanctions and Deviance: Evidence and Remaining Questions," Law and Society Review, 7, (1973), 372-392.

Uniform Crime Reports (Washington: Federal Bureau of Investigation, 1975).

R. A. Wright, "Nation's Retail Merchants Mobilize Security Systems to Combat Fast-Growing Shoplifting Trend," New York Times, 121, (1), 51.

N.B. Zabriskie, "Fraud by Consumers," Journal of Retailing, 48, (1972), 22-27.