An Empirical Update and Extension of Patronage Behaviors Across the Social Class Hierarchy

ABSTRACT - This study provides an updated assessment of the extent to which social class differentiates patronage behaviors. After briefly reviewing literature concerning class and patronage, a study is described in which Coleman's (1983) objective measure of social class is used to differentiate patronage. In a second stage of analysis, this measure is embellished by encompassing class consciousness and relative class income within objective class levels. The results indicate that combining measures yields improvement in predicting department store patronage, and that significant differences in patronage occur within objective levels of class according to consumers' class identification and relative class income.


Scott Dawson, Bruce Stern, and Tom Gillpatrick (1990) ,"An Empirical Update and Extension of Patronage Behaviors Across the Social Class Hierarchy", in NA - Advances in Consumer Research Volume 17, eds. Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, Provo, UT : Association for Consumer Research, Pages: 833-838.

Advances in Consumer Research Volume 17, 1990      Pages 833-838


Scott Dawson, Portland State University

Bruce Stern, Portland State University

Tom Gillpatrick, Portland State University


This study provides an updated assessment of the extent to which social class differentiates patronage behaviors. After briefly reviewing literature concerning class and patronage, a study is described in which Coleman's (1983) objective measure of social class is used to differentiate patronage. In a second stage of analysis, this measure is embellished by encompassing class consciousness and relative class income within objective class levels. The results indicate that combining measures yields improvement in predicting department store patronage, and that significant differences in patronage occur within objective levels of class according to consumers' class identification and relative class income.


Social class refers to a group of people who share commonality in such social characteristics as prestige, education, occupation, social skills, status aspirations, community participation, family history, recreational habits, and physical appearance (Coleman 1983). Relations between such groups in terms of inferiority, equality, or superiority provide the basis for a social hierarchy. Coleman and Rainwater (1978) conceive of an American class structure which includes seven groups that can be described in descending order as "old family names" (Upper-upper class), "accepted new money" (Lower-upper), "collegiate credentials expected" (Upper-middle), "white-collar associations" (Middle), "blue-collar life style" (Working), "definitely below the mainstream" (Working poor), and "the welfare world" (Under-class). Marketers have found social class useful since it manifests a distinctiveness in life style (Myers and Gutman 1974) and, in turn, consumption (Sobel 1981). For example, social class is a natural variable for segmenting product-markets that satisfy needs related to such life style expressions as housing, travel, sports, and entertainment. Specific product categories in which consumption has been found to be distinctive across social classes include furniture, clothing, and housing (Laumann and House 1970; Schaninger 1981).

Differences in patronage behaviors across social classes were first empirically observed in proprietary studies conducted by Levy and Martineau. These and other researchers arrived at the general conclusion that social class is a salient dimension of store image and that patronage is the result of consumers seeking congruence between self-image and store image (Martineau 1958; Levy 1966). Studies have confirmed that department and specialty store patrons primarily are from upper social classes (Rich and Jain 1968; Hirschman 1980), while discount stores tend to attract lower class consumers (Rich and Jain 1968). This pattern is more distinctive for purchases such as furniture or clothing (Schaninger 1981) which evoke higher social risk (Prasad 1975).

The goal of this study was to ascertain the extent to which an updated social class measure Coleman's (1983) Computerized Status Index (CSI) discriminates patronage in the current social milieu. An additional motivation of the study was to assess the degree to which relative income and class consciousness are factors which affect patronage within levels of objective social class. Scholars have found that relative occupational class income - income relative to the median within a class level significantly differentiates consumption of automobiles (Peters 1970) and coffee (Klippel and Monoky 1974). While occupation and education strongly determine the commonality of values which typify social classes, variations in number of wage earners and age may produce disparate levels of income. A related phenomenon occurs when two individuals with the same approximate incomes are objectively measured as middle and working class, but spend their incomes quite differently. Relative class income, however,. has not been investigated in the retailing literature. Overprivileged members of the working class may exhibit a higher level of patronage of mass merchandise stores (e.g., Sears, Montgomery Ward)-than their underprivileged counterparts. Conversely, overprivileged members of the middle class may exhibit their emulation of the upper class by shopping at higher prestige specialty and department stores.

Class consciousness may also affect behaviors within levels of objective class. Though class consciousness tends to be a strong predictor of attitudes and behaviors related to political ideology (Centers 1949; Curtis and Jackson 1977), it may also help in better understanding patronage differences within objective levels of social class. The presence of factors such as prestige or status in a store's image may lead to patronage differences according to the saliency of these factors to consumers based on their class identification. Similarly, consumers who are objectively middle class but perceptually upper class may demonstrate patronage behaviors of a more upper class nature than individuals who are middle class both objectively and subjectively. In sum, patronage may differ between members of the same objective social class and consideration of class consciousness and relative income should yield an improved understanding of these differences.



The population for this study was defined as adults residing in all private households (from single and multiple unit dwellings) in three metropolitan areas of the Pacific Northwest. The data were collected by trained interviewers through in-home, personal interviews. Eligible respondents were defined as the person renting, owning, or buying or otherwise residing in and responsible for the dwelling unit. In households where more than one adult met this criterion, interviews took place with the adult employed full-time. In the event that more than one adult was employed full time, the first adult contacted was the respondent. A multi-stage area probability sampling procedure was used to select the households. Up to three callbacks were employed to reach target households where adults were not at home. The households who refused to participate and who were not available after three callbacks were replaced in a random manner.

The final sample numbered 997 households. The majority of respondents within these households were married (78.6%). Comparison with census data showed that the sample is not significantly different in terms of age, gender, or income, but is marginally higher in level of education (p<.10).


Given the results of previous studies, social class should be associated positively with patronage of higher status stores. Nine well-known stores were selected in the market area from which interviews took place, including 3 discount stores, 3 mass merchandise stores, 2 department stores, and 1 specialty store. Patronage was measured using a 4-point scale ranging from "frequently" to "never."

As noted above, the objective measure of social class used in the study was Coleman's (1983) Computerized Status Index (CSI). Based on measures of five household characteristics, the CSI provides an efficient and effective method for estimating social class in studies which involve large samples (see Coleman 1983, p. 267 for further explanation). Coleman (1983) claims that at least 75% of households are correctly classified by the CSI, where "correct" is defined as an expert's opinion after evaluating 10-times more information than is used in the CSI index. Overall, the CSI departs from early class measures in several respects. First, education's influence on household status is captured by including levels of education for both husband and wife. If the respondent is not married, education is doubled. Second, CSI reflects the strong impact of occupation on social class (cf., Snipp 1985; Ciddens 1973) by doubling the weight of the head of household's occupation. Finally, the index includes total family income, rather than only husband's income. An interviewer's evaluation of residential area (i.e., "slum" area to "wealthy or society" area) is the final component of the index. The five items are then added and households are classified as Upper American, Middle Class, Working Class, or Lower American (see Coleman 1983, p. 276).

Class consciousness is operationalized here as is common in the sociological literature and in a manner similar to that followed in studies of status crystallization; as one's self-designated social class (cf, Curtis and Jackson 1977). In order to facilitate comparability with Coleman's class structure, the question used here stated: "Which socioeconomic group or class would you say best describes you and your family?" Interviewers then read alternative responses including upper class, upper middle class, middle class, working class, and lower class. The top two categories were combined for data analysis in order to produce a structure comparable to Coleman's.

Comparison of the social structures produced from the CSI and self-report measures below shows that, though they are far from independent (Chi square = 247.7, p<.0001), the two distributions share about 17% of common variance (Kendall's Tau B = .42, P<.001). This amount of shared variance is close to the correspondence Gilbert and Kahl (1982, p. 267) believe ought to exist between self-report and objective measures of class. Not surprisingly, a substantial number of objectively-assigned working class respondents appear to identify with the middle class as reflected by the self-report measure. This occurrence may be attributable to a myriad of factors including a resurgence of conservative, traditional and materialistic American values, a shift toward a greater percentage of white collar occupations, a negative stigma associated with lower than middle class identification, and a decline in the membership and power of working class unions.



The presentation of results examines differences in patronage, first across objective social class groups, and then through a combination of class consciousness and relative income within objective class. Analysis of variance showed that eight out of the nine stores produce statistically significant differences between social classes in frequency of shopping (see Table 1). Examining the mean scores shows that patronage differences between social classes tend to occur as predicted in the case of higher status specialty and department stores. The r-square values for these three stores are encouraging from a pragmatic standpoint and suggest that social class is a useful segmentation variable for higher end retailers. Post-hoc contrasts show that upper class respondents shop significantly more at the specialty store featured in the study than any of the remaining class groups (t-9.15, p<.001). Other significant contrasts indicate that upper and middle class respondents shop more frequently at both department stores 1 (t=9.43,p<.001) and 2 (t=8.29,p<.001) than working and lower class respondents. The absence of significant contrasts involving mass merchandise or discount stores suggests that social class is less effective in explaining patronage of these stores than in earlier studies (Rich and Jain 1968; Levy 1966). With one exception (mass merchandise store 3), no social class group differs substantively in patronage of mass merchandise stores. This result testifies to the expansive market to which mass merchandise stores appeal.



A cross-tabulation between objective and self-report measures of class produced the ten class combinations seen in Table 2. The presence of differences in perceived social class versus objective class is apparent from the relatively high numbers in the non-matched categories. Not surprisingly, a high number of individuals feel they are middle class, though they are objectively categorized as working class. The tendency of some respondents to downgrade their class from a higher objective category may occur because of under-privileged income status (as shown below), or may indicate a problem with the objective class measure. Recall that the self-report "Upper Middle Class" group was previously combined with the "Upper Class" group. Without this collapsing the discrepancy between the upper class self-report and objective categories would have been even greater.

Turning to patronage differences between the self-report objective class groups, analysis was performed by first-averaging frequency of shopping within each of the four store categories. The trend in means indicates that patronage in both specialty and department stores drops in a near linear fashion as both objective and perceptual social class declines. Mass merchandise patronage remains virtually flat across the ten class groups, while discount store patronage rises somewhat as class decreases from upper to middle class. Comparing Tables 1 and 2 shows that the expanded class model does explain approximately 8% more of the variation in department store patronage than the reduced class model. However, no differences in explained variation occur for the remaining three store categories.

The value of this analysis is perhaps best illustrated by examining patronage differences between class identification groups within levels of objective class. Within the objective upper class, patronage of department stores is significantly higher (F=6.17, p<.05) among those individuals identifying with the upper class than those identifying with the middle class. Several differences emerge within the objective middle class group as well. Both specialty store and department store patronage drop significantly (F=6.39, p<.001; F=11.85, p<.001, respectively) as class consciousness decreases within the objective middle class. Clearly, patronage within Coleman's middle class designation is quite heterogeneous and is better differentiated by considering individual's perceived social class. The effect of class identification is also apparent from the higher specialty and department store patronage by objective middle class respondents identifying with the upper class, than respondents who are objectively upper class but perceptually middle class. Further breakdown of the objective working class yields improvement in understanding patronage of department stores as well as mass merchandisers. Within the objective working class, patronage of department stores drops significantly (F=5.68, p<.01) as perceived class drops from middle to lower. Also, mass merchandise patronage is significantly lower (F=14.12, p<.001) among those respondents identifying with the lower class than those identifying with either the working or middle class.



Coleman's (1983, p. 274) suggestion for designation of relative income groups within social classes was used to develop an expansive relative income class structure. This structure (not shown) is highly correlated with the class consciousness structure just described (Kendall's Tau B=.80), which suggests that relative class income is a salient determinant of how people form perceptions of their class positions. As judged by the primary reason for undertaking this analysis - explication of patronage across social classes - the relative income breakdown yields a very similar set of findings. These two analyses produce close to identical results for overall F and R-square values, and for the pattern of significant differences within objective class groups.


Although this study did benefit from a large, representative sample, several limitations are notable. First, patronage of a relatively small number of stores was measured, especially in the case of specialty stores, which may have biased the results. Some consumers form intense loyalty to certain smaller and unique boutique-type shops, and such patronage was not measured here. Second, patronage itself was measured using a "frequently" to "never" scale, which is less valid than an exact measure of dollars spent over a period of time. However, the findings do suggest several conclusions regarding social class in the 80s. Clearly, frequency of shopping at specialty and department stores increases with one's social class position. However, objective social class by itself is unable to predict patronage of mass merchandise or discount stores. These findings reflect the multitude of strategies pursued by retailers during the 1980s. Many specialty and department stores have carefully engineered a clear and stable store image targeted at upscale consumers. On the other hand, many mass merchandisers have attempted to make strategic adjustments in pursuit of this same upscale market, and in doing so have blurred their image. This has been particularly true for some of the major retailers. At the same time, many discount stores have traded up and other types of institutions, such as off-price retailers offering nationally-known durable goods and fashion labels, have emerged and gained popularity with both the middle and upper echelons of society.

Comparing the current value of social class as a retail segmentation variable with its value in earlier studies, and comparing Coleman's updated measure against earlier measures, is nearly impossible due to the absence of reported effect sizes in earlier studies. For example, although Martineau (1958) and Levy (1966) found patronage patterns across social classes similar to those reported here, neither provided any indication of explained variance. Other studies have tested for class differences in patronage-related behaviors and attitudes, but have used different dependent variables and not reported effect sizes (Prasad 1975; Schaninger 1981; Rich and Jain 1968; Foxall 1975). While multivariate models using gravitational and store image measures are able to account for roughly 40% of the variation in actual store choice (Nevin and Houston 1980), the results here show that from 10% to 15% of the variation at higher prestige stores is explained by social class alone. Further, the present study- did not consider situational or risk factors associated with purchase occasions. Based on previous research, the inclusion of these factors would undoubtedly increase the extent to which social class explains patronage above that found here.

The usefulness of social class as a predictive variable in retailing is limited by the size of middle America. Over 80% of respondents in this study were objectively classified as middle or working class, making differentiation of mass merchandise patronage based on class particularly difficult. However, the results here also show that relative income or perceived class identification - when considered with objective social classes - do produce differences in patronage. The close correspondence between the expanded hierarchies suggests that relative class income may affect class identification. The embellishment of the objective class hierarchy with either of these criteria improves prediction of overall department store patronage. Further, a number of differences in patronage were identified within objective social classes. While these results were' most present in the department store category, both relative class income and class consciousness differentiate patronage of mass merchandise stores within the working class. Given the difficulty of identifying distinct consumer segments that shop in mass merchandise stores, this result is encouraging. The expansive approach to the American class structure taken here may prove useful to specialty, department, or mass merchandise retailers.


Centers, Richard (1949), The Psychology of the Social Classes: A Study Class Consciousness, Princeton, N.J.: Princeton Press.

Coleman, Richard P. (1983), "The Continuing Significance of Social Class to Marketing", Journal of Consumer Research, 10 (December), 265-280.

Coleman, Richard P. and Lee P. Rainwater with Kent A. McClelland, (1978), Social Standing in America: New Dimensions of Class, New York: Basic Books.

Curtis, Richard F. and Elton F. Jackson (1977), Inequality in American Communities, New York: Academic Press.

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

Giddens, Anthony (1973), The Class Structure of the Advanced Societies, New York: Harper & Row.

Gilbert, Dennis and Joseph A. Kahl (1982), The American Class Structure, Homewood, IL: Dorsey Press.

Hirschman, Elizabeth C. (1980), "Women's Self-Ascribed Occupational Status and Retail Patronage", in Advances in Consumer Research, Kent Monroe (ed), Vol. 8, Ann Arbor: Association for Consumer Research, 648-54.

Klippel, R.E. and J.F. Monoky (1974), "A Potential Segmentation Variable for Marketers: Relative Occupational Class Income", Journal of the Academy of Marketing Science, 2 (Spring), 351-56.

Laumann, Edward O. and James S. House (1970), "Living Room Styles and Social Attributes: The Patterning of Material Artifacts in a Modem Urban Community", Sociology and Social Research, 54, 32142.

Levy, Sidney J. (1966), "Social Classes and Consumer Behavior", in Knowing the Consumer, ed. J.W. Newman, New York: John Wiley, 146-160.

Martineau, Pierre (1958), "Social Classes and Spending Behavior", Journal of Marketing, 23 (October), 121 -130.

Myers, James H. and Jonathan Gutman (1974), "Life Style: The Essence of Social Class", in Life Style and Psychographics, William D. Wells, ed, Chicago: American Marketing Association, 235-256.

Nevin, John R. and Michael J. Houston (1980), "Image as Component of Attraction to Interurban Shopping Areas", Journal of Retailing, 56 (Spring), 77-93.

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

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

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

Schaninger, Charles (1981), "Social Class Versus Income Revisited: An Empirical Investigation", Journal of Marketing Research, 18 (May), 192-208.

Snipp, C. Matthew (1985), "Occupational Mobility and Social Class: Insights From Men's Career Mobility", American Sociological Review, 50 (August), 475-493.

Sobel, Michael (1981), Lifestyle and Social Structure, New York: Academic Press.



Scott Dawson, Portland State University
Bruce Stern, Portland State University
Tom Gillpatrick, Portland State University


NA - Advances in Consumer Research Volume 17 | 1990

Share Proceeding

Featured papers

See More


The Dark Side of Luxury: The Social Costs of Conspicuous Consumption

Christopher Cannon, Northwestern University, USA
Derek Rucker, Northwestern University, USA

Read More


Love is Blind: How Sensory Liking Impacts Perceptions of Unbranded Products

Jennifer L Stoner, University of North Dakota
Maria A Rodas, University of Minnesota, USA

Read More


F5. Alternative Food Consumption (AFC) Adoption and Low SES Youth Food Well-Being: From Precontemplation to Maintenance

Wided Batat, American University Beirut

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.