Group Differences in the Construction of Consumption Sets

ABSTRACT - The assortment or set of products a consumer owns has begun to assume an important role in consumer research. This paper proposes a construct, the consumption set, to operationalize sets. Exploratory research presented describes a way to elicit consumption sets from consumers. Interviews were conducted to determine whether consumption sets constructed would differ significantly by occupation group. In addition to significant relationships between occupation and set construction, relationships were found between household life cycle variables and sets. The authors argue, given the group differences found in consumption sets, this construct is a useful theoretical tool to include macro-variables in consumer behavior models.


Kathleen M. Rassuli and Gilbert D. Harrell (1996) ,"Group Differences in the Construction of Consumption Sets", in NA - Advances in Consumer Research Volume 23, eds. Kim P. Corfman and John G. Lynch Jr., Provo, UT : Association for Consumer Research, Pages: 446-453.

Advances in Consumer Research Volume 23, 1996      Pages 446-453


Kathleen M. Rassuli, Indiana-Purdue University Fort Wayne

Gilbert D. Harrell, Michigan State University


The assortment or set of products a consumer owns has begun to assume an important role in consumer research. This paper proposes a construct, the consumption set, to operationalize sets. Exploratory research presented describes a way to elicit consumption sets from consumers. Interviews were conducted to determine whether consumption sets constructed would differ significantly by occupation group. In addition to significant relationships between occupation and set construction, relationships were found between household life cycle variables and sets. The authors argue, given the group differences found in consumption sets, this construct is a useful theoretical tool to include macro-variables in consumer behavior models.


The marketing literature contains numerous instances of scholars suggesting a shift from research on single products to sets of products (e.g., Wind 1977; Sheth 1979; 1992). There are at least two benefits that can arise from such a shift. First, we know that the products consumers own differ by culture, geographic location, and social class (e.g., Douglas and Isherwood 1978; Hirschman 1986). Yet, these variables are often modeled as exogenous variables. Douglas and Isherwood (1978) contend that there is a code locked in the goods people own; the code relates specific goods to specific cultures. In a similar vein, Hirschman (1986) suggests that common threads run through the lives of one group of consumers she studied. McCracken (1988) and other anthropologists, as well as consumer behaviorists (e.g., Solomon and Assael 1987) have proposed that we must consider the totality of products consumers own C the set. Since set contents are material artifacts of culture (McCracken 1986), sets might provide a mechanism for observing the influence of, heretofore, macro exogenous variables on consumer decisions.

Second, from a marketing standpoint, the set also may shed a new light on consumer decision making. If consumers are creating sets, then consumers may be seeking products for their capacity to initiate or to complete a set. In this set completion mode, a product would be purchased to complement other products in existing sets (Green, Wind and Jain 1972). Choice, then, is not only the output of a discovery and evaluation process, but also an input to the process of building a set. Bandura (1978) believes, "people create and activate environments." Sets of products become part of the consumer's environment (Solomon 1983).

There is a small, but growing, literature that broaches the topic of the influence of sets in consumer behavior (e.g., McCracken 1989; Solomon and Assael 1987; Johnson 1989). In this paper we offer a formal construct for sets and one technique to operationalize the concept. The proposed conceptualization of sets, the consumption set, merges the notion of product sets (found in the literature mentioned above) with the standard marketing treatment of products as bundles of attributes. We suggest that attributes might be a common link between goods in a set and with groups of people. Finally, we test whether consumers are able to construct the theorized sets and whether there are group differences in consumption set construction.


The marketing literature provides a theoretical foundation upon which to build an operational definition of a consumption set. Typically, products are defined as bundles of attributes (Kotler and Armstrong 1991). However, the role of products and their attributes has been expanded by several scholars. McAlister and Pessemier (1982) conceived of products as being stores of attributes. According to the Lancastrian view, products (goods) produce attributes (Lancaster 1966). Thus, one might construe that consumers acquire products for the attributes they store and/or produce.

The idea that products belong to a set implies some type of relationship among them. This relationship suggests an extended role for attributes. To develop this role, one might build on the work of Alderson. He observed that products are not useful in themselves; utility arises in an assortment of complementary goods (Alderson 1957, pp. 198-99). In other words, product purchases are inter-dependent. One might hypothesize that complementary products are those which possess complementary attributes. (Here we would also expect consumption sets to include substitutes and unrelated elements). Furthermore, according to the literature, products can be expected to have both physical (tangible) and perceived (intangible) attributes. Many scholars are in agreement on this point (Wilkie and Pessemier 1973; Lilien and Kotler 1983; Hirschman 1980; Johnson 1988). [For simplicity sake, we only dealt with broad dimensions of attributes. Physical and perceived attributes can be broken down into more elementary components, e.g., brass. Chemicals, compounds, and minerals create brass.] Therefore, we define a consumption set as the entire assortment or portfolio of complementary, substitute and unrelated attributes and attribute combinations that a consumer holds at a particular time. A consumption set is the entire universe of products which surrounds a consumerCexpressed in attribute form. If, indeed, the attribute composition of consumption sets differs among groups, then perhaps we would be able to use consumption sets to decode consumption messages. Attributes would provide a taxonomical tool for classifying, comparing and evaluating consumption sets.

Consumption sets are created, formed, built and shaped by consumers. Conceivably consumption sets are first created in the consumers mind, and then, take shape in reality. The building process occurs over a period of time. Each choice a consumer makes can be expected to have an impact on the set. Consumption sets can be decomposed into consumption subsets or subgroupings of goods and attributes used together as a system, such as a living room, pantry, wardrobe, laundry, and so forth. [The attribute conceptualization of consumption sets is for operational purposes. The ultimate consumer does not typically enter the market to buy attributes. However, as shown in this study, consumers are aware of the attributes they use to produce consumption sets.]




Two macro variables were used to test for group differences in construction of the hypothesized consumption sets. First, to the extent that individuals are members of a group, similarities in consumption sets should be evident. Relying on Hirschman (1986) and Laumann and House (1970), we hypothesized that the consumption sets individuals createCthat is, the attribute profile or compositionCwould differ by social strata. Thus,

H1: The composition of consumption sets is similar for individuals within a group (e.g., occupation, subculture, culture) and different between groups.

To test this hypothesis, occupation groups were used.

Household life cycle (HLC) is another important macro-variable. Past research on household life cycles shows that as individuals advance in age, get married, and have children, and as the occupational status of the husband and wife change, household purchases of products (particularly durables) change accordingly. Since we believe that a person's consumption set is created and shaped over time, we expected differences depending upon a subject's stage in the HLC. Given an open-ended task of creating any consumption set, we expected consumers early in the HLC to create a set which resembled their ideal set. Having had time to accumulate a consumption set, we expected persons later in the HLC to create their actual set for the test.

H2: Consumers earlier in the household life cycle will create their ideal consumption set, while persons later in the life cycle will reconstruct their actual set.


In order to explore whether it is possible to recover consumption sets and to test for group differences, four groups of consumers were asked to construct a living room set and to complete a lengthy personal questionnaire. A living room consumption subset was chosen for several reasons. We needed to narrow the scope of the test to make the task manageable for subjects. Previous research on living rooms provided some a priori expectations about the products and attributes that might be included in living room sets (Csikszentmihalyi and Rochberg-Halton 1981; Laumann and House 1970; McCracken 1989). Living rooms are used for display purposes (Laumann and House 1970), and if, as Douglas and Isherwood (1978) state, people use goods to "signal membership" in groups, living rooms should provide evidence of group membership.

Instrument Subjects responded to a mail questionnaire that had been pre-tested on a sample of 30 subjects. First, respondents were instructed to create a living room set (write out the contents) based on physical and perceived attributes they selected from a list. The second section asked respondents to identify the living room style and to indicate whether the living room that was created was closer to their "actual" living room or an "ideal" living room. Demographic questions were asked at the end.

The main manipulation was a page with seven blank boxes across the top and a list of physical and perceived attributes down the side. Subjects were instructed to fill in each box with any product s/he desired and check off the attributes the product would have. Two stipulations applied: (1) respondents were instructed that the set must be within his/her present income (to prevent outlandish creations that would be out of reach for the subject's occupation group), (2) subject were asked to fill-in at least five of the seven products (to help reduce the number of blank questionnaires). To reduce fatigue, respondents were instructed that they did not have to construct each product in infinite detail. A number of popular magazines were content analyzed to produce a list of physical and perceived attributes shown in Table 1.

Subjects Groups of subjects were chosen from three occupations. Since the purpose of the research is to establish the existence of group differences in the attribute composition of consumption sets, the choice of groups was not essential. Occupation is an important indicator of social status; Hollingshead's two-item index of social status is composed of occupation and education, with occupation weighted more heavily. 443 questionnaires were sent to people employed in various white and blue collar occupations; 246 were returned. Population size (at a particular occupation site)/sample size/number of questionnaires returned (respectively) are as follows: physicians 270/110/42; college professors 307/200/82; firefighters 400/100/79; and a convenience sample of 43 from a mix of occupations. Ph.D.'s, M.D.'s, and men are over-represented in the sample. Although sample characteristics limit generalizability, they do not affect the major tests regarding differences in consumption set construction across groups.

Partitioning the Data Across the population one should witness heterogeneity in consumption sets composed of attributes, while finding homogeneity within groups. Hierarchical cluster analysis was used to assign each individual to a given cluster. Cluster analysis requires the selection of relatively independent variables, and the choice of a similarity measure, clustering method and the appropriate number of clusters. An examination of the correlation matrix shows that the measurement instrument provided relatively independent attributes. Most correlations were in the single digits with only a few exceptions. The highest correlation, 0.41, is between S5 (authentic) and S8 (classic). Hair, et al. (1987) indicate that attributes which fail to differentiate between or among groups will diminish the quality of the cluster solution. Objective attributes T1 (wood), T3 (fabric), and T5 (brass), as might be expected, were used by 99 percent, 95 percent, and 74 percent, respectively, of all groups. These attributes were not used in the cluster analysis. Square Euclidean distance was chosen for the similarity measure. Punj and Stewart (1983) note that choice of similarity measure "does not appear to be critical...." Ward's minimum variance method was selected as the clustering method. Punj and Stewart (1983) argue that Ward's method is among the better performing, except when outliers are present. We standardized the data in this study to reduce the outlier problem. Finally, a number of methods were employed to ensure the correct number of clusters were chosen. A plot of the coefficients against the number of clusters (scree diagram) yielded four clusters. Using the "mixture model approach" the results of the cluster analysis using Ward's method and the complete linkage method were compared. The cluster analysis was run using 75 percent of the sample, and then the remaining 25 percent were reclustered. This provided a test of the stability of the cluster solution, or evidence of convergence. Discriminant functions were derived, and observations were reclustered on the basis of those functions.


Four clusters emerged from the analysis. Each cluster had a highly unique profile in terms of the mean inclusion of attributes in the consumption sets constructed (Figure 1). A MANOVA routine was performed on the attributes by cluster. The univariate F-tests for each attribute across clusters showed that all perceived attributes were significantly different across clusters at the 0.0001 level, except S18 (impeccable). The only physical attributes that was significant was T8 (wicker).

Cluster Interpretation and Profiling Ten percent of the observations were classified in Cluster 1 C the "harmonious" cluster. For these individuals, living room sets were characterized by five attributes: simple, practical, natural, understated, harmony. While three of four clusters included the attribute comfort in their sets, Cluster 1 had the highest mean use of this attribute. Cluster 2 contained 16.9 percent of the respondents C the "distinctives." As Figure 1 shows, individuals in this cluster were characterized by three attributes: distinctive, dramatic, and futuristic. The inclusion of the attribute futuristic is interesting because other clusters avoided it's use. Cluster 3 contained 43.95 percent of the sample C the "practicalists." The attributes included most in the living room sets created by individuals were: practical, cozy, and comfortable. This cluster also had the largest mean inclusion of rustic. These individuals used fewer attributes per product and also developed fewer products. [The effect of the heavy use of attributes, by other clusters, would have been minimized by the process of standardization mentioned earlier.] They are perhaps described better by attributes they did not include in their set: gracious, understated and futuristic. Cluster 4 comprised 29.03 percent of the sample C the "classicists." These subjects developed sets composed mainly of three attributes: tradition, classic, and charming. In general, members of this cluster tended to be more expressive when describing the products they created. Figure 1 shows relatively wide use of all attributes. Consumption sets for members of Cluster 4 were not practical or comfortableCattributes used by more frequently by other clusters.

The four clusters were found to be significantly different from one another based on a MANOVA performed on the original clustering attributes. Since clusters were formed using Ward's minimum variance method, the MANOVA provides confirmation of a significant difference in the between, versus within, group variances. [MANOVA was merely used as a check, however, its distinctiveness in profiling variables is somewhat of an overstatement.] Wilk's Lambda was 0.099, with a significance of better than 0.001.

Profiling on variables not used in the clustering procedure. Descriptive statistics for the four clusters for variables other than those used in the clustering procedure are given in Table 2. MANOVA was used to compare the four clusters on income, education and occupation of the household head, and whether this was their actual or ideal set; F-statistics for the first three variables were 6.956, 12.295 and 9.232 (all significant at the .000 level or better). For actual versus ideal, the F-statistic was 2.488 (significant at better than .06). There were no significant differences in style of living room (not shown in Table 2); styles varied a great deal within each cluster.

Demographic Profiles. In Cluster 1, sixty percent of the heads of household were employed as professionals. More than two-thirds of these people said they had created their own (actual) living room set. Members of Cluster 2 were describing their "ideal set" more often than members of other clusters. Cluster 3 had more non-professionals, fewer college degrees compared to other clusters. More than four-fifths (83%) of Cluster 4 worked in professional occupations. This cluster had the highest mean income ($65,000+).

Product Use by Cluster. While products were not the focus of this study, some interesting findings appear. The basic products used to create a living room did not differ by cluster: a sofa/couch, a chair or two, an end table, carpet and a fireplace showed no significant differences (Table 3). But, there were significant differences for accessory products. Coffee tables, desks, pianos, televisions, stereos, and paintings were included by some clusters and not by others. For example, Cluster 2 did not include desks, but they did include stereos and paintings. Cluster 3 included televisions, but few pianos. It was in Cluster 4 where pianos were found in living room consumption sets along with paintings.

Internal and external validation. Internal validation was carried out by the split sample validation technique (Punj and Stewart 1983). Seventy-five percent of individuals were reclustered; the same clusters emerged. Then the classification of individuals into clusters was checked to determine whether they were clustered into the same groups as they previously had been. The analysis showed that only 21.7 percent, or 43 out of 198 individuals, were misclassified. [Twenty-eight of the 43 (65%) were original members of Cluster 3 who were misclassified into Cluster 2. When we tested the three cluster solution (earlier in the paper), we found that Cluster 2 was combined with Cluster 3. Therefore, this result might be expected.] This clearly supports the internal validity of the procedure. For external validity, one asks whether the solution is useful (Punj and Stewart 1983, p. 146). An overwhelming majority of respondents (88%) answered in the affirmative to the question of whether they had a picture in mind as they went through the exercise. This is strong support for external validity.





Tests of Hypotheses. Hypothesis 1. We believed that there should be consistency within sets of attributes created by occupation groups of individuals. MANOVA was conducted on a randomized block design, with the physical and perceived attributes as dependent variables and the four original groups as the independent variable. Two blocks of 50 percent of the subjects were chosen at random from the occupational subsamples. The purpose of the blocks was to test for the presence of any effects due to sample stratification. The results of the test show that Wilk's Lambda for the occupation treatment was 0.47; this was significant at better than .001. The blocking variable was insignificant; Wilk's Lambda was .90 with only a 63% level of significance. As part of the output of MANOVA, discriminant weights for the functions that differentiate clusters are produced. (See Table 4, for the standardized discriminant weights and discriminant functions.) Using these weights for classification purposes, a discriminant analysis was performed on a random sample of 90 percent of the total observations. 88.9 percent of grouped cases were correctly classified. All members of the holdout sample were correctly reclassified. Therefore, the results of the test support the belief that, given the opportunity to create a living room consumption set of their choice, individuals within occupational groups will create consumption sets of attributes that were more alike than those across groups. Comparing the cluster results to the original occupation groups, physicians mainly fell into Cluster 2, firefighters mainly into Cluster 3 and professors mainly into Clusters 1 and 4 (Table 2).



Hypothesis 2. The second research question deals with the issue of an actual versus an ideal set. Respondents were asked whether the consumption set they were describing was their actual set, their ideal set, or something else (other). While only those three options were provided, respondents often wrote "both" in the category marked "other." A content analysis of respondents answering "both" reveals that they often said "everything was mine except...," and the exception was "the big screen television" or "the baby grand piano."

Across all clusters, 55 percent of respondents created their own "actual" set. Slightly more than one-third said they had created their ideal. These averages can be broken down by cluster. In Cluster 1, 68 percent created their actual set, 20 percent their ideal. In Cluster 2, 43 percent created their actual set and 45 percent their ideal. For Cluster 3, 48 percent created their actual set and 41 percent their ideal. Sixty-one percent of Cluster 4 created their actual set and 29 percent their ideal. The creation of an actual set may be related to the ability to use more specific attributes (characteristics of Clusters 1 and 4) and/or to Cluster 3's inability to be descriptive.

A second discriminant analysis was performed in an effort to explore the impact of stage in the household life cycle on the creation of an actual, versus an ideal, set. In line with Hypothesis 2, the dependent variable, for the discriminant analysis, was the categorical variable, "actual or ideal," with independent variables C age, marital status, and presence of children 6 years of age and younger, 6B17 years of age, and 18 years of age and older (typical household life cycle variables). All variables had a significant contribution in discriminating between actual sets and ideal sets, except the presence of children under the age of six. The signs of all the variables were as expected. The function correctly discriminated between actual and ideal sets 60 percent of the time. Wilk's Lambda was 0.9378 (significant at the 0.05 level). The chance of being classified in a group is 50 percent; the model improved on chance to some extent.


The instrument was designed to explore only one part of a consumption set, the living room subset. Given the limited scope, there is room for future refinement. Other subsets may be less prone to group differences than the living room. A limited number of occupation groups were included in the study. Perhaps other occupations would exhibit similar differences. Finally, the sample was over-representative of persons with higher education and of males. One must wonder whether a sample that consisted mostly of females would have developed significantly different living room consumption sets. However, the purpose of the study was not to generalize to other groups, but to show that group differences can be detected in consumption sets and to suggest that perhaps this result can be generalized to other macro influences.


The authors set out to develop a tool sensitive enough to detect group differences in the construction of consumption sets. Such a tool would allow researchers to make comparisons across many types of macro groupings in society.

Taken as a whole the findings suggest that consumption sets specified in attribute form do, indeed, offer such a tool. Different occupational groups did have distinct attribute profiles within the consumption sets they constructed. This research suggests that differences in perceived attributes occur across groups (physical attributes did not). There was no significant difference in the basic products included in a consumption set across groups, although some of the accessory products differ. And it is fascinating to note that the overall style of living room consumption sets differed a great deal within a cluster (group). Yet, the attributes used to create these varied styles did not differ. In other words, group members use the same attributes to achieve different styles.



The research reported here implies that attributes might be used to conceptualize the common threads that run through a consumption set (cf., Hirschman 1986). Furthermore, attributes may be a first step in uncracking the group code locked in consumer goods (cf., Douglas and Isherwood 1978). For future research, one might expect the impact of other macro variables to show up in the composition of consumption sets. For example, while basic products composing a consumption set do not differ across groups within the American culture, one might expect them to differ from culture to culture. The product composition of a consumption set may differ due to physiology, geography and/or the unique nature of resource availability. Objective attributes may follow a similar logic. Moreover, some perceived attributesCpracticality and comfortCappeared in the consumption sets of all occupation groups under study. The incidence of these attributes in all consumption sets may be the result of a macro influence broader than occupation group, such as culture. Consumers from other cultures might develop different consumption sets of perceived attributes.


The research is intended to provide a starting point to demonstrate how consumption sets can be recovered from consumer responses and to explore macro differences in consumption set construction. The findings lend credence to the present conceptualization of consumption sets. Given an open-ended task, consumers can create consumption sets consisting of products and attributes. Furthermore, the composition of those consumption sets differs by group membership. Occupation groups were used to test the hypothesis. There were clear differences in the way consumers from different occupation groups combined attributes to create consumption sets. This research is a starting point. The consumption set construct may serve as a conceptual basis for classifying the contents of consumption sets. Once the contents of consumption sets have been inventoried, comparisons could be made across other macro groups C culture, social class, and reference groups. Clear patterns may emerge. A wealth of information may be contained in an analysis of the contents of consumption sets.


Alderson, Wroe (1957) Marketing Behavior and Executive Action, Homewood, IL: Irwin.

Bandura, A. (1978), "The Self System in Reciprocal Determinism," American Psychologist, 33 (April), pp. 344-58.

Csikszentmihalyi, M. and E. Rochberg-Halton (1981), The Meaning of Things: Domestic Symbols and the Self, Cambridge, England: Cambridge University Press.

Douglas, Mary and Baron Isherwood (1978), The World of Goods, Harmondsworth, Middlesex, England: Penguin Books Ltd.

Frank, R. E. and P. E. Green (1968), "Numerical Taxonomies in Marketing Analysis: A Review Article," Journal of Marketing Research, Vol. IX (February), 83-98.

Green, Paul E., Yoram Wind, and Arun K. Jain (1972), "Preference Measurement of Item Collections," Journal of Marketing Research, Vol. IX (November), 371-77.

Hair, Joseph F., Jr., Rolph E. Anderson, and Ronald L. Tatham (1987), Multivariate Data Analysis, New York: Macmillan Publishing Company.

Hirschman, Elizabeth C. (1986), "Humanistic Inquiry in Marketing Research: Philosophy, Method, and Criteria," Journal of Marketing Research, Vol. XXIII (August), 237-49.

Hirschman, Elizabeth C. (1980), "Attributes of Attributes and Layers of Meaning," Advances in Consumer Research, Vol. 7, ed. Jerry C. Olson, Provo, UT: Association for Consumer Research, 7-12.

Johnson, Michael D. (1989), "The Differential Processing of Product Category and Non-comparable Choice Alternatives," Journal of Consumer Research, 16 (December), 300-9.

Johnson, Michael D. (1984), "Consumer Choice Strategies for Comparing Non-comparable Alternatives," Journal of Consumer Research, Vol. 11 (December), 741-53.

Kotler, Philip and Gary Armstrong (1991), Principles of Marketing, Englewood Cliffs, NJ: Prentice-Hall, Inc.

Lancaster, Kelvin J. (1966), "A New Approach to Consumer Theory," Journal of Political Economy, Vol. 74, 2, 132-157.

Laumann, Edward O. and James S. House (1970), "Living Room Styles and Social Attributes: The Patterning of Material Artifacts in a Modern Urban Community," Sociology and Social Research, 54 (April), p. 326.

Lilien, Gary L. and Philip Kotler (1983), Marketing Decision Making, NY: Harper and Row.

McAlister, Leigh and Edgar Pessemier (1982), "Variety Seeking Behavior: An Interdisciplinary Review," Journal of Consumer Research, Vol. 9.

McCracken, Grant D. (1989), "Homeyness: A Cultural Account of One Constellation of Consumer Goods and Meanings," in Interpretive Consumer Research, ed., Elizabeth C. Hirschman, Provo, UT: Association for Consumer Research, 168-83.

McCracken, Grant D. (1988), Culture and Consumption, Bloomington: Indiana University Press.

McCracken, Grant D. (1986), "Culture of Consumption: A Theoretical Account of the Structure and Movement of the Cultural Meaning of Consumer Goods," Journal of Consumer Research, Vol. 13 (June), 71-84.

Morrison, D. G. (1967), "Measurement Problems in Cluster Analysis," Management Science, Vol. 13, B-775-80.

Punj, Girish and David W. Stewart (1983), "Cluster Analysis in Marketing Research: Review and Suggestions for Application," Journal of Marketing Research, Vol. 20 (May), 134-48.

Sheth, Jagdish N. (1979), "The Surpluses and Shortages in Consumer Behavior," Journal of the Academy of Marketing Science, 7 (Fall), 414-26.

Sheth, Jagdish N. (1992), "Acrimony in the Ivory Tower: A Retrospective on Consumer Research," Journal of the Academy of Marketing Science, 20 (4), 345-53.

Solomon, Michael R. (1983), "The Role of Products as Social Stimuli: A Symbolic Interaction Perspective," Journal of Consumer Research, Vol. 10 (December), 319-29.

Sheth, Jagdish N. and Henry Assael (1987), "The Forest or the Trees?: A Gestalt Approach to Symbolic Consumption," Marketing and Semiotics: New Directions in the Study of Signs for Sale, ed. Jean Umiker-Sebeouk, New York: Mouton de Gruyter, 189-217.

Wilkie, William L. and Edgar A. Pessemier (1973), "Issues in Marketing's Use of Multiattribute Attitude Models," Journal of Marketing Research, Vol. X (November), 428-41.

Wind, Jerry (1977), "Toward a Change in the Focus of Marketing Analysis: From A Single Brand to an Assortment," Journal of Marketing, 41 (October), 12, 143.



Kathleen M. Rassuli, Indiana-Purdue University Fort Wayne
Gilbert D. Harrell, Michigan State University


NA - Advances in Consumer Research Volume 23 | 1996

Share Proceeding

Featured papers

See More


F8. Dual Routes for Consumer Responses to Corporate Social Responsibility: The Role of Positive Moral Emotions, Attitudes, and Empathy

Chunyan Xie, Western Norway University of Applied Sciences
Richard P. Bagozzi, University of Michigan, USA

Read More


D10. It's Meant for Me: When Serendipity Increases Word-of-Mouth

Colleen Patricia Kirk, New York Institute of Technology
Joann Peck, University of Wisconsin - Madison, USA
Claire Hart, University of South Hampton, UK
Constantine Sedikides, University of South Hampton, UK

Read More


N13. Smaller Self but Larger Tips? The Effect of Awe on Consumers’ Tipping Intention

Ran Li, Chinese University of Hong Kong, China

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.