Extensions of the Basic Social Class Model Employed in Consumer Research

ABSTRACT - Social class research in consumer behavior has a long history, but the contributions have indeed been minimal. This is partly because the model that has dominated thinking and research is both inconsistent with major trends in American society and incompatible with the socialization perspective that represents a major orientation in studying consumer behavior. This paper proposes extensions to the standard procedure for operationalizing social class and presents data that support the rationale for these extensions.

Citation:

Terence A. Shimp and J. Thomas Yokum (1981) ,"Extensions of the Basic Social Class Model Employed in Consumer Research", in NA - Advances in Consumer Research Volume 08, eds. Kent B. Monroe, Ann Abor, MI : Association for Consumer Research, Pages: 702-707.

Advances in Consumer Research Volume 8, 1981      Pages 702-707

EXTENSIONS OF THE BASIC SOCIAL CLASS MODEL EMPLOYED IN CONSUMER RESEARCH

Terence A. Shimp, University of South Carolina

J. Thomas Yokum, University of South Carolina

ABSTRACT -

Social class research in consumer behavior has a long history, but the contributions have indeed been minimal. This is partly because the model that has dominated thinking and research is both inconsistent with major trends in American society and incompatible with the socialization perspective that represents a major orientation in studying consumer behavior. This paper proposes extensions to the standard procedure for operationalizing social class and presents data that support the rationale for these extensions.

INTRODUCTION

An historical examination of a scientific discipline entails an exhaustive review of that discipline's literature. Analysis of contemporary journal articles and conference proceedings provides insight into those scientific units (i.e., theories, concepts, and methods) that are the subject of current interest, while a review of textbooks over a period of years reveals those enduring scientific units that have transcended the discipline's occasional excursions into faddish areas.

Such an analysis of the consumer behavior literature would likely lead to a major conclusion that many of the cherished theories and concepts which have remained virtually intact for years in textbooks have done so in the absence of any significant advances in insight or explanatory ability. The social class concept is a notable example. Consumer behavior texts invariably include a chapter on social stratification, and the material coverage is, as invariably, highly similar. This literature presumably reflects current knowledge and thinking, yet the scientific value and practical usefulness of social class for understanding and predicting consumption phenomena is dubious.

This paper presents a critical analysis of the manner in which social class has been treated in consumer behavior and offers suggestions for improvement. These suggestions are predicated on the belief that the conceptual model of social class prevalent in consumer behavior is both inconsistent with major trends in American society and incompatible with the socialization perspective that represents a dominant orientation in studying consumer behavior.

SOCIAL CLASS IN CONSUMER BEHAVIOR

Consumer behavior interest in social class has waxed and waned over the past two decades. Martineau's (1958) seminal article provided the momentum for other solid works which appeared in the 1960s. Most notable were the writings of Coleman (1960), Carman (1965), and Levy (1966). Where Levy demonstrated that significant differences do exist among social classes with regard to various general behaviors, Coleman cautioned that the social class concept has been misunderstood and over-simplified. Carman, though critical of the overly simplistic manner in which researchers had applied the social class concept, presented a strong case supporting its potential usefulness.

The reflection and conceptualization personified by these works was followed by a period of empirical activity. Rich and Jain (1968) examined the relationship between social class and shopping behavior and concluded that social class has dubious usefulness in understanding this behavior.

Their skeptical perspective was countered by Wasson (1969), who argued, with a modicum of supportive evidence, that social class is superior to income as a predictor of consumer behavior. The social class vs. income debate was initiated, and a series of studies appeared (Hisrich and Peters 1974; Mathews and Slocum 1969; Myers, Stanton, and Haug 1971; Myers and Mount 1973; Slocum and Mathews 1970).

Empirical activity has been sporadic since then, exceptions are the store patronage research by Foxall (1975) and Prasad (1975) and Jain's (1975) novel application of conjoint analysis for delineating individuals' implicit social class concepts.

Interest in social class has not vanished, however. This is reflected in Nicosia and Meyer's (1976) writing on the sociology of consumption and in the provocative piece by Zaltman and Wallendorf (1977). These latter writers argue convincingly that a sociological perspective is needed to supplement the prevailing psychological orientation if a comprehensive understanding of consumer behavior is to be achieved. Their sentiments are echoed by Sheth (1977) who claims that demographic and SES variables, including social class, should not be discarded prematurely; instead, more sophisticated measurement and application are needed. Carman (1978) arrives at a similar conclusion while asserting that social class research has ignored his earlier criticisms (Carman 1965).

CRITICAL ANALYSIS

A variety of reasons account for why social class has provided minimal explanation of consumption phenomena. In fact, there are nearly as many reasons as there are empirical studies (cf. Dominquez and Page 1978; Zaltman and Wallendorf 1979). This critical analysis will provide a useful framework for suggesting improvements in the application of social class to consumer behavior. First, however, it will be useful to examine the exact nature of social class and also explore the reason why social class should provide a useful explanation of consumption phenomena.

The Nature of Social Class

A fundamental difficulty in working with the social class concept is arriving at a clear understanding of its meaning. From a strict sociological perspective, social class is just one dimension of a more general social stratification construct. Class is a power-based concept, including both economic and political elements. But stratification also includes a prestige dimension, or what sociologists term "status."

Carman (1965) has argued that a third dimension, cultural class, is one of primary relevance to consumer behavior. Cultural classes are self-perpetuating sub-cultures which differ with respect to the value placed on education, solidarity of the family, religious involvement, media exposure, recreational activities, etc. Cultural class can be operationalized with real variables (e.g., values, attitudes, behaviors) or with such proxy variables as occupation and education (Carman 1965).

Viewed from the perspective that a particular social class is a group of individuals who share a common culture and manifest similar life styles, the concept of social class should be useful for understanding and predicting various consumption phenomena: Since human behavior is determined in large part by the particular elements of culture which are learned and transmitted from generation-to-generation, different social classes should exhibit differences in values, motives, and other precursors to consumption behavior.

Why has it "Failed?"

Applications of social class in consumer behavior have not been extensive. The only sustained research has involved the social class vs. income controversy, and this research cannot be regarded as definitive since non-equivalent tests were employed, including different products, diverse data collection procedures, and different operationalizations of social class. The failure of social class rests more on the prevailing orientation that has guided the research.

The basic research approach has been to: administer a standard social class index (e.g., Warner's or Hollingshead's); use the household head's (typically husband's) score as the basis for determining the family's social class; and to statistically test whether different social classes differ with regard to product-, store-, or brand-choice behavior.

Three major assumptions are implicit in this "basic model." First, it is assumed that social class is capable, by itself, of explaining choice behavior, oftentimes without regard to whether or not the consumption behavior in question is subject to cultural influence. A second assumption is the tacit belief that a husband's social class by itself is the sole determinant of a household's class position. An additional assumption is that current consumption behavior can be explained by present social class, without regard to one's class standing during that period of primary socialization influence. These assumptions are now discussed in detail as the "choice behavior," "husband-only," and "present social class" fallacies.

The Choice Behavior Fallacy.  The only substantive conclusion that can be drawn from empirical applications of social class is that neither social class nor income is an adequate explanation of consumer choice behavior. It is simplistic to expect a single variable to account for behavior that is caused by multiple determinants. Kassarjian (1971) has made this point cogently in his review of personality research, and it is equally applicable here.

A second aspect of the choice behavior fallacy stems from the fact that although social class is an indicator of life style, different products can be used to fulfill the same life style, and the same product can be consumed by different people, in different ways, for different reasons (cf., Kernan 1977). It is thus unreasonable to expect social class to explain and predict choice behavior. Its proper role should probably be to account for the values and motives that underlie this behavior rather than attempting to explain the behavior per se, especially when that behavior may not be subject to cultural influence (e.g., brand choice in a low-unit-value product category).

The Husband-Only Fallacy.  Measurement of social class based exclusively on the class position of the household head, who is typically the husband in conventional American families, assumes that the husband's social class is the sole determinant of a household's class standing. In years past this assumption may have been tenable (in paternalistic societies the assumption still holds), but in contemporary America it is not. Indeed, nearly a third of all U.S. households include a wife who is working outside the home at least part-time (Pralle 1980). In addition to augmenting the family's purchasing power, this employment status, together with educational achievements, must have some effect on the family's class standing.

The social class categorization of households based exclusively on the husband's class indicators would be less unacceptable if there were a high degree of similarity (homogamy) in husbands' and wives' backgrounds and educations. Research performed by Blau and Duncan (1967) indeed reflects this. "For men 45 to 54 years of age, the correlation between husband's and wife's education is .580, and for men 55 to 64 years old it is no less than .632" (p. 190). These correlations, though apparently invalidating the above argument, actually suggest two reasons why a husband-only measure of social class is inadequate: (1) although husbands and wives share common backgrounds and educations in the majority of families-of-procreation, a large number of pairings are not homogamous; (2) assuming the lower correlation for the younger age cohort reflects a trend (Bleu and Duncan did not suggest this), then American families are becoming even less homogamous.

The Present Social Class Fallacy.  Operationalizing social class based on the husband's present class position assumes implicitly that every household that is classified into the sane social class will exhibit the same consumption behavior because they share this one commonality. The assumption would not be so limiting if American society were free of vertical mobility, but this is not the case. The father-son correlation for occupational status is of the order of .4 (Blau and Duncan 1967). Evidently, a very considerable amount of status modification does occur.

The consumer socialization literature (Moschis and Churchill 1978; Ward 1974; Ward, Wackman, and Wartella 1977) affords an explanation of why this extensive status modification invalidates the assumption that current choice behavior is determined by one's present social class position alone. Basic consumption skills are acquired through a lifetime of information receipt and value inculcation. One's present values, beliefs, and attitudes are partially attributed to recent acculturation but, in addition, are in large part the result of the childhood enculturation process.

Since behavior is partly a function of deep-seated values, beliefs, etc., it follows that consumption behavior is largely determined by past social position in addition to present position. In short, people arrive at the same class position in different ways; just because they occupy the same class position at a particular point in time (the time at which a cross-sectional study is performed) is no reason why their consumption behavior should be the same.

This point takes on added significance when the nature of vertical mobility is examined. The greatest mobility occurs among the "sons" of skilled and semi-skilled workers, i.e., children from upper-lower and lower-middle class families are moving both up and down the class hierarchy. Since the mass market for most businesses (and researchers also) consists of the upper-lower, lower-middle, and upper-middle classes, the respondents in a social class study are the ones who are most likely to occupy a particular social class position that is different from the one in which they received their initial consumption socialization.

EXTENDED OPERATIONALIZATIONS OF SOCIAL CLASS

Revision and extension of the basic model that has dominated social class research in consumer behavior is long overdue. Previous extension efforts have achieved some success. Coleman's (1960) consumer privilege concept and Peter's (1970) relative occupational class income concept have both enhanced the ability of social class to predict select consumption phenomena. Neither extension goes far enough, however,

A number of extensions are possible. However, the present effort is not intended to provide an exhaustive treatment. The aim, instead, is merely to illustrate two extended concepts and to suggest the consumption conditions when each may augment the explanatory ability of the standard operationalization of social class. The logic of discovery rather than that of justification provides the philosophy of science rationale for these efforts (Hunt 1976).

A Husband-and-Wife Combined Measure of Social Class

The identification of a household's social class based exclusively on the husband's occupation, education, or other social class indicators has been criticized as a fundamental weakness of social class research in consumer behavior (Zaltman and Wallendorf 1979). Sociologists (e.g., DeJong et al. 1971) have also called for a re-examination of the use of husband-only indicators as the sole determinants of a family's social position. However, how husband's and wife's individual social class indicators should be combined to form an overall family social class score is an open issue.

One method would be to simply average husband's and wife's separate social classes (Haug 1973). This method, though computationally convenient, would be wrong conceptually. The United States is a "leveling-up" society: the marriage of an individual from an upper social class with one from a lower class would not produce a middle-class couple. An alternative would be to assign the higher social class of the husband or wife to the family as a whole (Haug 1973; Haug and Sussman 1971). Conceptual problems also plague this approach. In fact, it would yield results different from the husband-only approach only in those situations where a wife's social class exceeds her husband's.

Both methods ignore the joint influence of husband and wife in consumption-oriented decision making. They would structure a process wherein all consumption decisions would be predicted using the same set of spouse weightings. What is needed is a weighting scheme that assigns weights based on each spouse's relative influence in a specific decision process.

We propose a procedure that places weights on husbands' and wives' social classes proportionate to their relative influence in the consumption situation at issue. The procedure is based on two premises. First, it is assumed that much consumption behavior involves joint decision making. The foundation for this premise is solid (e.g., Davis 1976). A second premise is that the particular social class concept and corresponding operationalization used in consumer research should be situation-specific. To make an analogy, just as a single measure of IQ is inadequate across all cultures, a single social class measure is inadequate across all consumption situations.

For example, if the objective is to predict differences in product use for a product which involves a husband-dominant decision and whose use is not related to deep-seated enculturated values (say the purchase of a lawnmower), then the standard, husband-only concept may serve quite well. If, however, the objective is to account for the frequency of family vacation behavior, which is known to involve joint decision making (Davis and Rigaux 1974) and which likely is influenced greatly by childhood experience, then it is doubtful that the standard concept will account for much variance in vacation frequency.

Implementation of the proposed husband-and-wife-combined measure of social class necessitates two types of measurements. One would entail measurement of the relative influence of husband and wife in the consumption activity at issue; the other measure would involve indicators of husband's and wife's social class. Each is examined separately, and then we turn to the issue of indexing them into an overall family social class score.

Measurement of relative influence of husband and wife would require precise specification of the consumption situation, since the roles of husbands and wives are variable by decision process stage. Once this is done, it becomes a relatively simple matter of assigning influence to each partner. A common procedure would be to have husband and wife rate themselves on rating scales ranging from "husband decided" to "wife decided" (Davis 1976). Another method would be to have respondents divide a number of points between husband and wife, using a constant sum scale, to indicate each partner's influence (Belay and Overholser 1975).

The separate social classes of husbands and wives could be measured using traditional indices: Warner's Index of Status Characteristics, Hollingshead's Index of Social Position, and Carman's Index of Cultural Classes. Although the validity of these has been challenged in terms of both the choice and weighting of indicators (Dominquez and Page 1978; Haug 1977; Haug and Sussman 1971; Jackson and Curtis 1968), we will assume for simplification sake that the indices are valid so that we can turn to the more fundamental concern of the paper.

The procedure for combining the separate social classes of husbands and wives is illustrated below. The discussion assumes that separate measures of social class have been obtained for both husband and wife, using for illustration Hollingshead's two-factor index. This procedure places each respondent into one of seven occupational and educational categories. The occupational score is weighted by seven, the educational score by four, and the indexed score is then assigned to one of five social classes (cf., Bawling, Coney, and Best 1980). Using this procedure, the derived family social class would be a weighted sum of each spouse's social class:

EQUATION   (1)

where F_SC is the family's overall social class; w_i is the weight for each spouse, derived from the measure of relative role for the decision process et issue; and SC_i is each spouse's social class, ranging from class 1 (the highest) to class 5 (the lowest), following Hollingshead's convention,

A simple numerical example will illustrate the procedure. Consider a decision making situation that varies from household-to-household in terms of relative influence of husband and wife. Assume for illustration that the decision involves the choice of vacation site. Data for four hypothetical families are presented.

TABLE

These illustrations show how social class assignment for a family is weighted in the direction of the spouse who has a greater decision influence. Where the husband-only measure would have categorized Family 1 as class 2, the husband-and-wife-combined measure yields an indexed score that is closer to the wife's class position, since she plays the dominant role in the particular decision. The combined procedure has the opposite affect for Family 3, where the weighting emphasizes the husband's dominant influence. The illustration for Family 4 reflects a situation where the family would be assigned to the wife's social class, because she has sole influence for this particular decision.

Family 2 reflects the only case where the husband-only index provides an equivalent result to our combined procedure. This occurs because the husband is the sole decision maker. Such a situation is the only occasion where it is theoretically appropriate and empirically prudent to designate a family's social class based exclusively on the husband's class position.

A Conjoint, Past-and-Present Measure of Social Class

The focus now turns to the issue of devising a social class index that extends beyond a family-of-procreation's present social class by incorporating the social classes of the families-of-orientation in which husband and wife received their early consumption socialization. In other words, what we are proposing is a social class index which combines "past" and "present" social classes. The considerable amount of intergenerational mobility or status modification provides the theoretical justification for such an index. It is unrealistic to expect to explain a household's current consumption behavior when this behavior in many instances is due to habits, choice rules, and other choice-determining factors that were acquired when family members occupied different (typically lower) class positions during childhood.

The rationale for combining past and present social classes is conceptually sound, but procedures for accomplishing this are not so obvious. For present purposes, we will simply conceptualize what the determining factors might be. Again, the premise is invoked that the amount of weight or influence that is assigned to the to-be-indexed components, past and present social classes, must vary by consumption situation, since there is also variability in the determinants of the particular behavior that the social class measure is attempting to explain. Vacationing behavior illustrates this point. While the decision to take a summer vacation may, for a particular family, be influenced greatly by what husband and wife were accustomed to doing as children, the decision concerning the particular vacation destination nay be entirely independent of their past customs or experiences.

In a more general sense, we propose that the relative roles of past and present social position are a function of how "unique" the particular decision is. Where many decisions are highly habitual and based on past patterns (e.g., consumption decision relating to child-rearing practices), other decisions are unique and require the construction or creation of choice procedures (e.g., the choice of whether to try an innovative birth control method).

This logic is similar to Bettman and Zin's (1977) distinction between stored rules and constructive mechanisms, where the latter is likely employed when a choice is made for the first time or when a changed situation is encountered. We hypothesize that the more unique or constructive the decision, the less the role of past social class as a determinant of current consumption activity. In such a situation, the indexing procedure should assign less weight to past social class. The proposition is portrayed in the Figure.

The empirical upshot is the need for a procedure to identify the relative influence of past and present social classes in context of the consumption decision at issue. Though obviously not a simple task, it would be possible to devise a series of questions to get at the extent to which a family-of-procreation's decision is based on a constructive mechanism, or, instead, results from stored rules acquired during husband's and wife's respective families-of-orientation.

We close this section by suggesting that the empirical difficulties are not as insurmountable as might appear. This is predicated on our view that social class should in the first place be employed as an explanation only in those situations where the consumption activity is truly influenced by cultural class antecedents (Carman 1965). Provided that research is restricted to such situations, then adults in families-of-procreation should be capable of providing reasonably accurate responses concerning the types of behavior performed by their parents during the period of initial consumption socialization in their families-of-orientation.

FIGURE 1

TYPE OF DECISION AND RELATIVE INFLUENCE OF PAST AND PRESENT SOCIAL CLASS

DATA SUPPORTING THE NEED FOR EXTENDED CONCEPTUALIZATIONS OF SOCIAL CLASS

Implicit in the foregoing arguments are two crucial assumptions. One is that husbands' and wives' social classes are indeed inconsistent in a relatively large number of households: If the amount of discrepancy were minimal, the combination of their separate classes would be moot, since the husband-only method would capture all the relevant information. A second assumption is that intergenerational mobility is sufficiently extensive to warrant the effort to devise a procedure for combining past and present classes.

A study was performed to test these assumptions. A state-wide sample of approximately 800 South Carolina households received a questionnaire in 1979; measures of social class were included among many other questions. The initial mailing and a follow-up generated an 80% response rate. Respondents were somewhat upscale, mostly white, and predominantly from urban locales.

They were asked to describe as specifically as possible the occupation and education of: the husband, the wife, the husband's father, and wife's father. They were instructed to identify their fathers' occupations and highest educational achievement at that period when they (respondents) were "growing up," i.e., "when you were in elementary school." The intent was to capture fathers' social classes at the time of respondents primary consumption socialization.

Hollingshead's two-factor index was used to assign each household's responses to four possible social classes associated with husband, husband's father, wife, and wife's father. Incomplete data were extensive due to such reasons as inability or unwillingness of respondents to designate their father's occupation or education; respondents were retired or had student status, and thus no current occupation could be assigned, a spouse was not employed outside the home; respondent was unmarried; etc.

One person coded all the questionnaires on two separate occasions, with a minimum separation of one month between recoding the same questionnaire. The agreement was substantial. When coding differences could not be reconciled, the questionnaire was removed from further analysis.

Congruity of Husbands' and Wives' Social Classes

The relationship between husbands' and wives' social classes is presented in Table 1. These data represent only those households where both husband and wife were employed outside the home. A total of 311 cases, or almost exactly one half of the total responses, provided this data.

The main diagonal in Table 1 reflects instances of status congruency--a total of 46.9% of the 311 cases. Entries above the diagonal indicate instances where the husband's social class exceeds his wife's. There were 29.3% such cases. Though not surprising, it is noteworthy that the higher the husband's social class, the greater the probability that the wife's class was lower than his, with the obverse holding as well.

TABLE 1

CONGRUENCY BETWEEN HUSBANDS' AND WIVES' SOCIAL CLASSES

The most revealing result concerns the prevalence of cases where the wife's social class exceeds her husband's. There were a total of 23.S% entries, or nearly one-fourth of all cases, below the diagonal. Considering the upscale sample involved in this analysis, it is likely that much greater frequency of wife' s-class-greater-than-husband's would be found in a more representative sample. The implication is clear: The traditional, husband-only measure of social class probably under-represents the true social class and purchasing power of a large percentage of families.

Intergenerational Mobility

Table 2 presents data indicating the relationship between the social class of husbands and wives in families-of-procreation with their respective father's social class. There were a total of 457 cases where data were available on husband's and husband's-father's social classes, and a total of 325 cases of wife's and wife's-father's social classes.

Entries on the main diagonal reflect the absence of inter-generational mobility, i.e., children in families-of-procreation possess the same social class as their fathers. The proportion of such cases is essentially equivalent for husbands and wives, 26.3% for the former and 24.9% for the latter. In other words, approximately three-fourths of all families-of-procreation have experienced status modification.

Upward mobility is indicated in the entries below the diagonal, which reflect fathers' social classes as lower than children's. In fact, 65.2% of all the husbands experienced upward mobility, while a slightly larger 67.1% of the wives experienced this mobility.

Entries above the diagonal reveal relatively few instances of downward mobility; there were only 8.5% cases of this nature for husbands, and 8% such cases for wives. These data are harmonious with previous results (e.g., Blau and Duncan 1967) indicating the extensive upward mobility in American society.

TABLE 2

RELATIONSHIP BETWEEN FAMILY-OF-PROCREATION AND FAMILY-OF-ORIENTATION SOCIAL CLASS

DISCUSSION

These data have obvious limitations; foremost, perhaps, is the non-representative sample. However, the data are presented for illustration purposes only, and are not intended as definitive. Limitations aside, the data clearly manifest extensive intergenerational mobility as well as significant discrepancies between the social classes of husbands and wives in those households where both partners have employment outside the home.

The upshot is that the traditional approach to social class research in consumer behavior is conceptually and empirically flawed. Measuring only the husband's social class and then attempting to predict choice behavior based on this information alone assumes that the husband's class is the sole indicator of a family's class standing. This assumption is untenable in view of the fact that women from approximately one-third of all households are in the work force, and many possess indicators of higher class than their husbands'. We have proposed a method which takes into consideration the social class of both husband and wife, and combines the two by weighting each by the degree of influence each partner has in the particular decision process that is subject of inquiry.

It is also unreasonable to assume that current behavior is solely a function of a household's present class standing, A present-only operationalization disregards past socialization influence. The empirical problem resulting from this is that a sample of households assigned to the same social class by a present-only measure may be quite heterogeneous with regard to values, motives, and other real determinants of consumption behavior.

It is one thing to recommend the indexing of past-and-present social classes into an overall conjoint scale. Doing so is another matter. We have conceptualized a general procedure for determining the relative weights of past and present classes, but specific procedures await future research.

The procedure for combining the separate social classes of husbands and wives is feasible, however; it is merely a matter of doing the research to ascertain whether a combined husband-and-wife index provides superior predictions to the simple, husband-only procedure that has dominated past research.

REFERENCES

Bettman, James R., and Zins, Michel A. (1977), "Constructive Processes in Consumer Choice," Journal of Consumer Research, 4, 75-85.

Blau, Peter M., and Duncan, Otis D. (1967), The American Occupational Structure, New York: John Wiley & Sons.

Carman, James M. (1965), The Application of Social Class in Market Segmentation, Berkeley: Institute of Business and Economic Research.

Carman, James M. (1978), "Values and Consumption Patterns, A Closed Loop" in Advances in Consumer Research, ed., H. Keith Hunt, Ann Arbor: Association for Consumer Research.

Coleman, Richard P. (1960), "The Significance of Social Stratification in Selling," in Marketing: A Maturing Discipline, ed., Martin L. Bell, Chicago: A.M.A.

Davis, Harry L. (1976), "Decision Making within the Household," Journal of Consumer Research, 2, 241-260.

Davis, Harry L. and Rigaux, Benny P. (1974), "Perception of Marital Roles in Decision Processes," Journal of Consumer Research, 1, 51-61.

DeJong, P. Y., Brewer, M. J., and Robin, S. S. (1971), "Patterns of Female Intergenerational Occupational Mobility," American Sociological Review, 36, 1033-42.

Dominquez, Luis V., and Page, Albert L. (1978), "Stratification in Consumer Behavior Research: A Re-examination," Case Western Reserve Univ. Working Paper.

Foxall, Gordon R. (1975), "Communication Note: Social Factors in Consumer Choice: Replication and Extension," Journal of Consumer Research, 2, 60-64.

Haley, Russel, Overholser, C. C., and Associates (1975), "Purchase Influence: Measures of Husband/Wife Influence on Buying Decisions," New Canaan, CT: Haley, Overholser, and Associates, Inc.

Haug, Marie R. (1973), "Social Class Measurement and Women's Occupational Roles," Social Forces, 52, 86-98.

Haug, Marie R. (1977), "Measurement in Social Stratification," Annual Review of Sociology, 3, 51-77.

Haug, Marie R. and Sussman, Marvin B. (1971), "The Indiscriminate State of Social Class Measurement," Social Forces, 48, 549-63.

Hawkins, Del I., Coney, Kenneth A., and Best, Roger J. (1980), Consumer Behavior: Implications for Marketing Strategy, Dallas: Business Publications, Inc.

Hisrich, Robert D., and Peters, Michael P. (1974), "Selecting the Superior Segmentation Correlate," Journal of Marketing, 38, 60-63.

Hunt, Shelby D. (1976), Marketing Theory: Conceptual Foundations of Research in Marketing, Columbus, OH: Grid.

Jackson, E. F., and Curtis, R. F. (1968), "Conceptualization and Measurements in the Study of Social Stratification," in Methodology in Social Research, eds., H. M. Blalock and A. B. Blalock, New York: McGraw-Hill.

Jain, Arun K. (1975), "A Method for Investigating and Representing an Implicit Theory of Social Class," Journal of Consumer Research, 2, 53-59.

Kassarjian, Harold H. (1971) , "Personality and Consumer Behavior: A Review," Journal of Marketing Research, 8, 409-19.

Kernan, Jerome B. (1977), "Retrospective Comment on Martineau's "Social Classes and Spending Behavior," in Classics in Consumer Behavior, ed. Louis E. Boone, Tulsa, OK: The Petroleum Publishing Co.

Levy, Sidney J. (1966), "Social Class and Consumer Behavior," in On Knowing the Consumer, ed. Joseph W. Newman, New York: John Wiley & Sons.

Martineau, Pierre (1958), "Social Classes and Spending Behavior," Journal of Marketing, 23, 121-30.

Mathews, Herbert L., and Slocum, John W. (1969), "Social Class and Commercial Bank Credit Card Usage," Journal of Marketing, 33, 71-78.

Moschis, George P., and Churchill, Gilbert A. (1978), "Consumer Socialization: A Theoretical and Empirical Analysis," Journal of Marketing Research, 15, 599-609.

Myers, James H., and Mount, John F. (1973), "More on Social Classes Vs. Income As Correlates of Buying Behavior," Journal of Marketing, 37, 71-73.

Myers, James H., Stanton, Roger P., and Haug Arne F. (1971), "Correlates of Buying Behavior: Social Class Vs. Income," Journal of Marketing, 35, 8-15.

Nicosia, Francesco M., and Mayer, Robert N. (1976), "Toward a Sociology of Consumption," Journal of Consumer Research, 3, 65-75.

Peters, William H. (1970), "Relative Occupational Class Income: A Significant Variable in the Marketing of Automobiles,'' Journal of Marketing, 34, 74-77.

Pralle, Michael E. (1980), "Survey Dispels Myth About Growing Dual-Earner Market," Marketing News, May 16, 17-18.

Prasad, V. Kanti (1975), "Socio-economic Product Risk and Patronage Preferences of Retail Shoppers," Journal of Marketing, 39, 42-47.

Rich, Stuart U., and Jain, Subhash C. (1968), "Social Class as Predictor of Shopping Behavior," Journal of Marketing Research, 5, 41-49.

Sheth, Jagdish N. (1977), "Demographics in Consumer Behavior,'' Journal of Business Research, 5, 129-38.

Slocum, John W., and Mathews, H. Lee (1970), "Social Class and Income as Indicators of Consumer Credit Behavior," Journal of Marketing, 34, 69-73.

Ward, Scott (1974), "Consumer Socialization," Journal of Consumer Research, 2, 1-16.

Ward, Scott, Wackman, Daniel B. and Wartella, Ellen (1977), How Children Learn to Buy, Beverly Hills: Sage Publications.

Wasson, Chester R. (1969), "Is It Time to Quit Thinking of Income Classes," Journal of Marketing, 33, 54-57.

Zaltman, Gerald, and Wallendorf, Melanie (1977), "Sociology: The Missing Chunk or How We've Missed the Boat," A.M.A. Educators' Proceedings, Series #41, 235-38.

Zaltman, Gerald (1979), Consumer Behavior: Basic Findings and Management Implications, New York: John Wiley & Sons.

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