Product and Brand User Stereotypes Among Social Classes

J. Michael Munson, University of Santa Clara
W. Austin Spivey, University of Santa Clara
ABSTRACT - This research investigated the relation of social class to product and brand-user stereotyping. The data base included over 200 females evaluating 48 brands in 8 product categories. As hypothesized, significant differences between upper, middle and lower social classes were observed for both product and brand-user stereotypes.
[ to cite ]:
J. Michael Munson and W. Austin Spivey (1981) ,"Product and Brand User Stereotypes Among Social Classes", in NA - Advances in Consumer Research Volume 08, eds. Kent B. Monroe, Ann Abor, MI : Association for Consumer Research, Pages: 696-701.

Advances in Consumer Research Volume 8, 1981      Pages 696-701

PRODUCT AND BRAND USER STEREOTYPES AMONG SOCIAL CLASSES

J. Michael Munson, University of Santa Clara

W. Austin Spivey, University of Santa Clara

ABSTRACT -

This research investigated the relation of social class to product and brand-user stereotyping. The data base included over 200 females evaluating 48 brands in 8 product categories. As hypothesized, significant differences between upper, middle and lower social classes were observed for both product and brand-user stereotypes.

INTRODUCTION

Marketers have given considerable attention to social class and opinion leadership. Both constructs are felt to play an important part in decisions about market segmentation and the selection of mass media aimed to reach specific target markets. Interestingly, a review of the marketing and communications research highlights the fact that these constructs have been studied independently. There is little, if any, explicit recognition of theoretic linkage. In addition, a review implies that the perceived usefulness of the constructs may vary depending upon specific product class, situation, and target market. Indeed, some seem to feel that one construct is "better" than the other.

Some researchers decry the difficulties associated with operationalizing opinion leadership. Rogers (1972) as well as King and Summers (1970) bemoan the problems in identifying and using opinion leaders directly. Robertson (1970) comments on the difficulty in simulating opinion leadership via testimonials, slice-of-life vignettes, and man-on-the-street recommendations. Loudon and Della Britta (1979) point out how hard it is to stimulate leaders to talk about a product and exert personal influence. There are conflicting findings about whether the influence of opinion leaders tends to be product specific (e.g., Robertson and Myers 1969; Silk 1966) or generalizable to overlapping product categories (e.g., King and Summers 1970).

Other researchers question the usefulness of social class. The "dated" nature of much social class research---most of the studies are 15 to 20 years old--leads some to question the relationship between social class and today's consumption patterns. Some believe that social classes are disappearing or becoming more homogeneous. Bieda and Kassarjian (1969) suggest that differences in product preferences among social classes may have disappeared because of common exposure to mass media. Their suggestion implicitly argues that mass media has had a leveling influence on the values and life style aspirations of all individuals. If this is true, the stereotypes of specific brand or product users should be relatively homogeneous across social strata.

In light of the weaknesses associated with both social class and opinion leadership, it seems incorrect to view either construct as better than the other, gather, it would seem advertisers should concentrate more effort on "integrating" the potential inherent in each. It may well be that many of the effects of opinion leadership are contingent upon social class considerations. Indeed, Klapper (1970) views mass media communications as operating through mediating factors--group membership, interpersonal contacts, selective exposure and defense mechanisms--so that it typically is a contributing agent but not the sole cause of audience reactions.

The desire to integrate the potential inherent in the two constructs provided the impetus for this research. Two key questions are addressed based upon Bieda and Kassarjian's (1969) suggestion that differences in product perception among social classes have disappeared:

(i)  do perceptions of brand and product user stereotypes differ across social classes?

(ii)  if so, for what types of product classes?

THE STUDY

If social class is to become a resurgent key to implementing opinion leadership, advertisers must answer a key question: to what extent do differential stereotypes exist across social classes at both the micro brand level and the macro level of product category. If common exposure to the mass media has acted as a leveling influence across all individuals, as Bieda and Kassarjian suggest (1969), than differences in product perception among social classes may have disappeared.

Given that prior marketing research on social class has focused on one of three areas---shopping, media and product class usage behavior (as opposed to brand usage)--little is known about how perceptions of specific brands may differ across social class. Moreover, much of the previous research is based on samples of university students which limits the generalizability of results.

Study Hypotheses

The need to fill in these gaps dictated that two working hypotheses be formulated to guide the study. Stated in alternative form, they are:

H1: Stereotypes of people who own or prefer a specific brand will differ across social classes.

H2: Stereotypes of people who own or prefer a specific category of product will differ across social classes.

Two corollaries (C1 and C2) were formulated from the working hypotheses. The first related to the magnitude of any observed differences among social classes:

C1: Any differences in stereotypes will be largest between lower and upper social classes.

The second corollary related to observed differences for a product category:

C2: Differences in user stereotypes will exist only for value-expressive product categories.

Specific support for H1, H2 and C1 comes from many sources. Warner (1960) suggested that the further apart two people are in class level, the further apart will be their usage and understanding of a product. Martineau (1957) noted social class helps to define the hierarchy of things on which people wish to spend money. The further apart two people are in social class, the greater the difference will be in perceived likelihood of purchase. Levy (1966) states "...As one looks at the social class ladder, the further he goes down it, the more difference he sees in life styles and values, and thus, the difference in consumption patterns becomes greater..."

C2 is based upon research by Locander and Spivey (1978). They discussed the basic differences between utilitarian and value-expressive products. Value-expressive products help an individual to give positive expression to their values and self concept. For these types of products, the key concern is the stereotype of the person who owns or prefers the product. On the other hand, for utilitarian products, user stereotyping is relatively unimportant compared to specific product attributes. Another point is noteworthy. Value-expression and utilitarianism are not opposite ends of a continuum; rather, they are independent dimensions (Spivey 1977). Therefore, a product quite possibly can be considered utilitarian and still provide an opportunity for self-expression. Generally, however, one would expect to observe few differences across social classes regarding the perceptions of utilitarian product users.

METHOD

Sample Design

In order to test the hypotheses and the corollaries, 203 women were interviewed in their homes. They were chosen by using a stratified, multi-stage probability sample with quotas. Approximately equal sample sizes were obtained for each social clue: 64, upper; 87, middle; and 52 lower.

The women were chosen from a city frequently used for test marketing. The sampling universe was the total number of city households. Initial stratification was done on the basis of income. Census tracts were used as the primary sampling units; each tract was designated as representative of either upper, middle, or lower social class. Interviewers were sent to random starting points and instructed to proceed in a clockwise direction around each block until a quota of six interviews per block had been completed. Block quotas were filled subject to two restrictions: (1) at least one and preferably two of each six respondents were employed; (2) no women under 18 years or no more than two women 65 years old or older be included so as to make the characteristics of the sample as similar as possible to the general population. Some call-backs were required to fulfill the quota within each block.

Product Typology

Across the entire sample of 203, a wide range of product categories and brand names were incorporated into the survey. Two primary criteria were used to select the products which were included in the study. First, the products had to represent categories which could be designated as convenience, shopping and specialty goods. Second, the products had to reflect a wide range with respect to their ability to communicate symbolically. Twelve judges familiar with these criteria were used to categorize 23 product categories. Of the 23, a smaller subset of 8 product categories which reflected the highest levels of inter-judge agreement were selected for inclusion in the final survey.

Two product categories high on the value-expressive dimension (VED) were selected: automobiles and magazines. Three categories moderately high on VED were chosen- washing machines, vacuum cleaners and gasoline. Three product classes low on VED were selected: brassieres, deodorants, and laundry detergents. Available research supports these categorizations. For example, Lessig and Park (1978) found automobiles to have the highest score, 3.65, on the value-expressive dimension among the 20 products studied; magazines ranked eight with a score of 3.08; laundry detergent ranked 17 with a score of only 2.42. Present research also suggests that for at least 6 of the 8 product categories, wives may be the dominant spouse in any decision processes involving their purchase (e.g., Fabian 1965, Davis and Riguax 1974). Within each product category six brands were selected for possible evaluation on the basis of greatest familiarity among a demographically representative pilot sample. Hence, the hypotheses were tested over a comprehensive product typology which included 8 product categories comprised of 48 individual brands.

Operational Definitions

Brand or product user stereotypes were obtained by having each respondent evaluate how she viewed "a person who owns or prefers" a particular brand over 16 semantic differential scales. These scales (e.g., successful-unsuccessful, informed-uninformed, etc.) have been shown in previous research to be well suited to assessing product and brand user stereotypes for various product categories (e.g., Ross 1971; Munson 1973; Munson and Spivey 1979). The scales represent major dimensions of how the individual views either himself and/or others and "...represent clear examples of self-concept definition" (Bills, et. al. 1951). These scale items have demonstrated adequate test-retest reliability over a three week interval, with reliability coefficients (r) ranging from r = .72 (p.<.01) to r = .68 (p.<.01). Additionally, Roan (1967) has undertaken specific converging operations in demonstrating the construct validity of the measure while others (Munson 1973 and Ross 1967) have established their freedom from social-desirability effects. Social class was determined for each respondent by using the Socioeconomic Index for Occupations (Duncan 1961). This index provides social class values for each of the detailed occupational classifications of the census based on a combination of occupational prestige ratings, educational status, income level, and age distribution.

Analytical Technique

The two hypotheses and the corollaries were tested by using one-way, univariate analysis of variance (ANOVA) coupled with a form of the chi-square statistic. For each ANOVA calculated the response variable was the sum of an individual's responses to all 16 semantic differential scales which indexed her stereotype of a specific brand user. Social class was the "treatment" >effect of interest. H1, about differences in perceptions for specific brands, was tested within the ANOVA framework by using the standard F-ratio computed or each brand (Hicks 1973). To test H2, the probability of differences in perception at the macro level of product category, a chi-square statistic was computed based upon the ANOVA results associated with each brand within a product category (Wiser 1962). The appropriate chi-square statistics were computed based upon contrasts calculated at the micro level of brand. The reader interested in a detail explanation of the relationship between ANOVA and the chi-square statistic used to test H2 should see Appendix 1 where the latter statistic for testing the difference in product category perceptions between lower and upper class is derived.

C1, about differences among social classes regarding perceptions of both brand and product class, was tested by using the principle of orthogonal contrasts (Hicks 1973). A linear and a quadratic contrast were considered simultaneously: the linear contrast considered the difference in mean perception between upper and lower social classes; the quadratic contrast considered the difference in mean perception between the middle class and the average of the upper and lower classes. C2, about differences among social classes existing only for perceptions about value-expressive products, was tested by noting the pattern of results. No specific test statistic was necessary since the hypothesis could be rejected simply by contradiction (Daniel and Terrell 1979).

RESULTS

A summary of the ANOVA and chi-square results is presented for all product classes and brands in Table 1. Note that only the linear contrasts and chi-square are shown; quadratic contrasts are omitted in the interest of clarity. In addition, note that the results are ordered: highly value-expressive products are mentioned first; more utilitarian products are mentioned last.

Value-expressive Brands and Products

Significant differences in brand-user stereotypes were found between upper and lower social classes for 9 of the brands tested (Table 1): Chrysler, Ford, Buick, Standard, Texaco, General Electric, Vogue, Ladies Home Journal and Playboy. These significant differences in brand user stereotypes were most pronounced between lower and upper class--as indicated by the significant linear F-ratios. In only one product category, washing machines, were significant quadratic effects observed. A significant difference in perception for Sears Kenmore washer was found between middle class compared to both lower and upper class. This indicates that the perceptions of lower and upper classes sometimes merge for this brand. Similar results were observed regarding differences among social classes about product categories. As shown by the chi-square statistics derived from the linear F-ratios, three of the eight categories--automobiles, gasolines, and magazines--were perceived differently by the lower social class compared to the upper class. In addition, middle class perceptions were different than the other classes with respect to washing machines as indicated by a quadratic contrast.

Utilitarian Brands and Products

Results for the utilitarian product categories--brassieres, deodorants, and laundry detergents--show relatively few significant differences in stereotypes at either the brand level or macro product category level. Only Right Guard and Salvo were perceived differently.

Conclusions About the Hypotheses

Generally, H1, H2, and C1 were supported; C2 was not supported.

The results for the individual product categories clearly support previous research. The findings for magazines are consistent with Warner and Lunt's (1941) earlier Yankee-City research, while those for automobiles are consistent with Coleman (1960), Peters (1970), and Akers (1970). Peters (1970), for example, found occupational class income (defined as a combination of family income and occupation) to be differentially related to automobile buying behavior for foreign economy, intermediate-sized, and compact cars.

Findings were as expected for automobiles and magazines because other research has shown that among women, these two product classes are perceived as more conspicuous and important than many others (Cohen and Barban 1970). Both may be perceived as having considerable potential to facilitate symbolic communication. The results for gasoline may, in part, be attributed to a type of complimentary relation they have with automobiles. These two products may represent logically associated categories within the consumer's cognitive structure. The findings were not quite as expected for washing machines. Given the similarity of this product to refrigerators, differences between social classes were expected based upon the results of Lessig and Park (1978). But the differences were expected between upper and lower social classes and not between middle class and both extremes.

IMPLICATIONS

Social Classes Have Different Perceptions

Overall, these results refute the contention of Bieda and Kassarjian (1969) that differences in social class have disappeared because of common exposure to mass media. True, as indicated by these results, there is unanimity of perception, but only for some products. Further, it is clear that differences in brand perception across social class are related to the generic product's ability to communicate symbolically. The highly value-expressive product categories had many brands with significant differences in user stereotypes while the utilitarian categories had few. Moreover, for highly value-expressive products, the product categories themselves were perceived as different by upper and lower social class.

The pattern of significant differences in user stereotype across social classes observed for value-expressive but not for utilitarian products has implications for researchers studying opinion leadership. To better understand the processes by which opinion leaders exert their influence, greater knowledge about their social class is essential. The results of this study imply that marketing efforts which attempt to employ either social class or opinion leadership in isolation will be less effective than efforts which consider both simultaneously. Focusing exclusively on one is an inadequate representation of a theoretic system in which both are interrelated. This is true for two reasons. First, the construct "opinion leadership" only assumes meaning within the context of some relevant social system in which the opinion leader is implicated. From a pragmatic viewpoint of identifying market segments and selecting media to reach target segments, social class is the most easily identifiable "social system" construct within which an opinion leader can operate. Second, where instances of opinion leadership occur, it would seen that complex social system and communication effects can intervene between both the opinion leaders' reception of the message from the mass media and his relaying of the message to others.

APPENDIX 1

Table 1A shows the derivation of the chi-square statistic derived from the linear contrasts associated with the various brands (makes) of automobiles. Brands within the product category can be regarded as a series of k experiments designed to test the some hypothesis (e.g., H2). The probabilities of the observed outcomes with respect to the common hypothesis tested can be designated P(1), P(2),... P(k). If the experimenter desires to make an overall probability statement, the following statistic may be used (Winer 1962):

X2 = 2Eu(i) where u1(i) = -1nP (i)   (1)

Under the hypothesis that the observed probabilities are a random sample from a population of probabilities having a mean of .50, the chi-square statistic in equation (1) has a sampling distribution which is approximated by the chi-square distribution having 2k degrees of freedom.

For example, the chi-square based on linear contrasts for automobiles in Table 1A may be determined as follows. The probabilities of obtaining the linear F ratio or one more extreme is given in the column headed Probability. The natural log of the probability with the sign changed is given in the next column. The numerical value of the chi-square statistic is equal to twice the sum of the negative natural log of each probability. The number of experiments (k) or brands equals six; the degrees of freedom is 2(k). An arbitrary alpha level of .05 was selected against which to test the null hypothesis. Hence, the probability of obtaining a chi-square statistic of 28.30 or larger is less than .005 when (chi-square .995 = 28.30) the mean probability in the population is .50.

Using the chi-square statistic to reject the null hypothesis is, in effect, accepting the series of individual brand-user stereotype evaluations as combined evidence indicating that the cosmos hypothesis (H2) in each of the experiments can be rejected if inferences are to be made with respect to the combined series of experiments.

TABLE I

DIFFERENCES IN BRAND AND PRODUCT USER STEREOTYPES ACROSS UPPER, MIDDLE AND LOWER SOCIAL CLASSES

TABLE A1

DIFFERENCES IN BRAND USER STEREOTYPES OF AUTOMOBILES BY SOCIAL CLASS AND CHI-SQUARE STATISTICS BASED ON LINEAR F RATIOS

REFERENCES

Akers, F. C., (1968), "Negro and White-Automobile-Buying Behavior: New Evidence," Journal of Marketing Research, 5, 283.

Bieda, J.C., and Kassarjian, H. H., (1969), "An Overview of Market Segmentation," in B. A. Morin ed., Marketing in a Changing World, Chicago: American Marketing Association, 249-253.

Bills, R. E., Vance, E. L., and McLean, O. S., (1951), "An Index of Adjustment and Values," Journal of Consulting Psychology, 15, 257-261.

Cohen, J. B., and Barban, A., (1970). "An Interactive Consumer Product Typological System, A Progress Report and Partial Evaluation," Working Series in Marketing Research, No. 12, College of Business Administration, Pennsylvania State University.

Coleman, R. P., (1970), "The Significance of Social Stratification in Selling," Proceedings of the American Marketing Association, ed., M. L. Bell, 171-184.

Daniel, W. W., and Terrell, J. C., (1979), "Business Statistics: Basic Concepts and Methodology," Boston: Houghton Mifflin Company

Davis, H. L., and Rigaux, B. P., (1974), "Perception of Marital Roles in Decision Processes," Journal of Consumer Research, 1.

Duncan, O. C., (1961), "Socioeconomic Index for All Occupations," in A. J. Reiss, Jr., Occupations for Social Status. New York: Free Press.

Fabian, L., (1965), "Expenditure Patterns of the American Family," National Industrial Conference Board, sponsored by Life Magazine.

Hicks, C. R., (1973), Fundamental Concepts in the Design of Experiments. New York: Holt, Rinehart and Winston.

King, C. W., and Summers, J. O., (1979), "Overlap of Opinion Leadership Across Consumer Product Categories," Journal of Marketing Research, 7, 43-50.

Klapper, J., (1970), The Effects of Mass Communication. Glencoe, Ill: The Free Press, 508.

Lessig, V. P., and Park, C. W., (1978), "Promotional Perspectives of Reference Group Influence: Advertising Implication," Journal of Advertising, 41-47.

Levy, S. J., (1959), "Symbols for Sale," Harvard Business Review, 37, 117-124.

Locander, W. B., and Spivey, W. A., (1978), "A Functional Approach to Attitude Measurement," Journal of Marketing Research, 15, 576-587.

London, D. L., and Della Britta, A. J., (1979), Consumer Behavior: Concepts and Applications. New York: McGraw Hill Book Co. 189-214, 263-288.

Martineau, P., (1957), Motivation in Advertising. New York: McGraw-Hill Book Company.

Munson, J. M., (1973), "An Investigation of Self Concept Congruity and Brand Preferences: Toward a Predictive Model," unpublished Ph.D. dissertation, University of Illinois, Champaign-Urbana.

Munson, J. M., and Spivey, W. A., (1979). "Assessing Self Concept," Proceedings Association for Consumer Research, Vol. VII, J.C. Olson, (ed.), 598-603.

Peters, R. A., (1970), "Perceived-Price Quality Relationship: Experimental Evidence," Journal of Marketing Research, 7, 525-528.

Robertson, R. S., (1970), Consumer Behavior. Glenview: Scott Foresman, 81-97.

Robertson, T. S., and Myers, J. H., (1969), "Personality Correlates of Opinion Leadership and Innovative Buying Behavior," Journal of Marketing Research, 6, 164-168.

Rogers, E. M., (1972), Diffusion of Innovations. New York: The Free Press, 1-42.

Ross, I., (1967), "The Relationship between Self Concept and Judgements Concerning Selected Consumer Products," unpublished doctoral dissertation, Ann Arbor, Michigan: University Microfilms.

Silk, A. H., (1966), "Overlap Among Self-Designated Opinion Leaders: A Study of Selected Dental Products and Services," Journal of Marketing Research, 3, 255-259.

Spivey, W. A., (1977), "An Experimental Evaluation of Attitude Change for Attitude Functions Serving Combinations of the Utilitarian and Value-Expressing Attitude Functions," unpublished Ph.D. dissertation, University of Houston.

Warner, W. L., and Lunt, P.S., (1941), "The Social Life of a Modern Community," Yankee City Series 1.

Warner, W. L., (1960), Social Class in America. New York: Harper and Row.

Winer, B. J., (1962), Statistical Principles in Experimental Design. New York: McGraw Hill Book Company.

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