Applications and Extensions of Categorization Research in Consumer Behavior

Mita Sujan, Penn State University
Alice M. Tybout, Northwestern University
ABSTRACT - A special topic session was organized to explore recent applications and extensions of categorization research in consumer behavior. The purpose of this paper is to summarize the session by: 1) providing a brief overview of categorization theory and consumer researchers' recent contributions to this topic, 2) abstracting the four papers presented, and 3) highlighting comments made by the discussants that integrated the papers and suggested directions for future research.
[ to cite ]:
Mita Sujan and Alice M. Tybout (1988) ,"Applications and Extensions of Categorization Research in Consumer Behavior", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 50-54.

Advances in Consumer Research Volume 15, 1988      Pages 50-54


Mita Sujan, Penn State University

Alice M. Tybout, Northwestern University


A special topic session was organized to explore recent applications and extensions of categorization research in consumer behavior. The purpose of this paper is to summarize the session by: 1) providing a brief overview of categorization theory and consumer researchers' recent contributions to this topic, 2) abstracting the four papers presented, and 3) highlighting comments made by the discussants that integrated the papers and suggested directions for future research.


The basic premise underlying categorization theory is that people naturally divide the world of objects around them into categories in order to achieve efficient understanding and processing of their environment (Rosch 1975; Rosch and Mervis 1975; Rosch et al. 1976). Categorization allows people to react to new stimuli as members of previously defined categories stored in memory rather than having to formulate unique responses when new stimuli are encountered.

Categorization theory is particularly relevant to understanding consumer behavior because consumers face a complex choice environment replete with brands having both shared and unique features. Consumers may use categorization to simplify and structure their environment. For example, they may group brands of, say, cars by categories such as sports cars, family cars and sub-compacts. This allows reactions to a new brand/car to be based, at least in part, on the category in which it is classified.

This simple observation about how people structure their environment has increasingly and profitably been applied to extend understanding of how consumers perceive and react to marketing stimuli. For example, the notion that consumers can transfer affect from the general product category to a specific exemplar or brand has been employed to explain consumers' use of evaluation strategies that are not based on analytic information integration approaches (e.g., Cohen 1983; Sujan 1985). The idea that information about product categories is likely to be organized hierarchically into broad product class categories, below which are nested more specific or "basic" product type categories, with brand level categories serving as the most specific level of categorization has been used to explicate market structure issues (e.g., Day, Shocker and Srivastava 1979), to understand product evaluations (MeyersCLevy and Tybout 1987), and to determine contingencies under which comparative advertising will differ from noncomparative advertising (Sujan and Dekleva 1987). More recently, exploration of categories idiosyncratic to the marketing domain has begun. For example, some research has examined the categories or "schemas" consumers have about advertising (Wright 1986). Other research has examined categories of noncomparable products (e.g., vacations and home improvement expenditures) that consumers often must choose from (Johnson 1984;Bettman and Sujan 1987).Finally, consumer researchers have explored how differences in consumer expertise with a product category affects categorization and judgment processes (Alba and Hutchinson 1987; Sujan 1985). Thus, significant extensions of the current theorizing in psychology are being made by consumer researchers. Many of these insights are a result of consumer researchers' focus on complex "real world" stimuli and their interest in differences in the ability and motivation to process product information among groups of consumers.

The four papers presented in this special topic session add to the growing body of consumer research examining categorization issues. The first paper addresses the issue of how consumers categorize. Specifically, it reports research examining two alternative processes that have received much attention in the literature: analytic versus holistic, and provides insight about how expertise/training influences which of these processes is employed. The second paper also examines the categorization process but takes a developmental perspective. The process by which children of varying ages categorize products is examined and it is observed that young children operate in a manner conceptually equivalent to inexperienced or "novice" consumers for a product category. The third paper discusses mood as an antecedent to categorization and provides evidence that individuals in a positive mood show greater flexibility, adaption, and creativity in their categorization strategies than do individuals in a neutral mood. The final paper explores the effect that categorization may have on product evaluations and judgments and reports evidence suggesting that products that are moderately incongruent with their claimed category receive more favorable evaluation than either congruent or extremely incongruent products. Thus, the four papers together examine the various stages in the categorization process: the antecedent to categorization, the process of categorization itself and some important consequences of categorization for judgment. What follows are more detailed abstracts of each paper that were provided by the paper authors.




J. Wesley Hutchinson, University of Florida

Joseph W. Alba, University of Florida

When a new product category emerges or an old one is sampled for the first time, consumers must determine its structure and learn how to discriminate between subclasses within it. Even when a product category is very familiar, a new entrant will prompt classification behavior. Once classified, decision making is simplified. In some cases, however, producers complicate the categorization process by intentionally imitating competing brands, particularly on perceptual dimensions.

Alba and Hutchinson (1987) suggest that expert consumers will be more likely than novices to discriminate between artificially similar products due to experts' greater propensity to engage in analytic rather than holistic processing. In the context of categorization, analytic behavior refers to the ability to ignore irrelevant information and classify products solely on the basis of features that are diagnostic of category membership. It should be noted that expertise is conceptually distinct from product familiarity, which is defined in terms of the number of product related experiences. That is, the ability to be analytic develops as a function of the qualitative nature of one's experiences.

The hypothesized relationship between expertise and categorization was explored experimentally by providing subjects with several types of experience and observing the effects of experience on subsequent categorization performance. Specifically, during an initial training phase, 180 subjects were shown ten pairs of drawing of fictitious stereo speakers. The speakers in each pair always differed with respect to two category assignments (i.e., high priced vs. low priced and Brand A vs. Brand B). These category assignments were designated by arbitrary cues beneath each picture. The speakers were completely described by five visual dimensions: shape (trapezoidal vs. rectangular), size (tall vs. short), speaker location (top vs. front), speaker cover (solid vs. mesh) and pedestal size (full vs. recessed). However, the categorization rule was defined in terms of only one of these features (e.g., tall = high priced; short = low priced). The remaining attributes were predictive of category membership, but not perfectly. That is, they correctly predicted category membership 70% of the time. Thus, for these training stimuli the criterial and irrelevant attributes generally agreed in their prediction about category membership. ID the subsequent test phase, test pairs were constructed such that the criterial attribute indicated one category but, based on correlational information in the training experience, the remaining attributes indicated the alternate category. The primary dependent measure was the percentage of test pairs properly categorized. Good performance would indicate use of the correct analytic rule. Poor performance would indicate use of irrelevant informationCeither an incorrect rule or a holistic strategy wherein categorization is based on the overall similarity of the test stimulus to the training stimuli from that category.

The critical manipulations involved the type of experience given to subjects during the training phase. Several factors were varied orthogonally. First, the difficulty of learning the criterial attribute was varied by having the two speakers in each training pair differ along all five dimensions (Hard condition) or on only two dimensions (Easy condition). Second, some subjects were explicitly instructed to learn the rule for discriminating between the two categories of speakers. For half of these subjects, the categories that were learned during training were the same as those that were subsequently tested (Intentional condition); for the other half, training and test categories were different (Incidental condition). The remainder of the subjects were told to examine the speakers and indicate which one was more visually appealing (Preference condition). Finally, the criterion feature itself was varied systematically across subjects. Five features were used, and the feature that defined category membership ranged from very salient (i.e., shape) to very subtle (i.e., pedestal size).

The results showed that, as expected, subjects in the Easy learning condition were significantly more analytic than those in the Hard condition. Also, the level of analytic behavior varied as a function of the criterial attribute. Shape was the easiest to learn, pedestal size was the most difficult, and the remainder did not differ from each other. Surprisingly, however, there was no effect of intentionality. This outcome was hypothesized to be due to the level of attention paid to the two categories by subjects in the Incidental and Preference conditions and to the perceptual salience of the attributes. Thus, a follow-up experiment was run that omitted all reference to price or brand in the training instructions to the Preference group. The results of this study supported the interpretation of the initial experimental results. Subjects in the Intentional condition were significantly more analytic than those in the Preference condition.

Subjects were also asked to rate their confidence in each of their categorization responses. Analysis of these responses revealed an interesting effect. First, subjects who used a holistic strategy became more confident of their (incorrect) responses as the stimulus become more visually similar to the prototype of the incorrect category. Moreover, subjects who learned the criterial attribute and performed perfectly in the test phase nonetheless became less confident of their responses as the test stimuli became less similar to the correct prototype. Thus, even when attributes can be ignored for purposes of categorization, irrelevant information can have a residual effect OD one's beliefs. This suggests that even experts may be influenced by such information in some situations. Such an effect might be expected under adverse processing conditions such as when time pressure or information load are high or involvement is low.



Deborah Roedder John, University of Minnesota

To date, categorization research has focused primarily on examining the processes employed by adults. However, considerable insight into adult processing can be gained by taking a developmental perspective and examining how children of different ages categorize products. In addition to advancing understanding of categorization in general, exploring children's processing has the benefit of enhancing understanding of a segment of consumers that is important to many marketers.

A starting point for developing hypotheses about children's categorization processes is the literature on differences between expert and novice adults. Research suggests that novices tend to use "surface structure" attributes (attributes that are readily observable) to categorize, whereas experts use "underlying attributes(attributes related to the functional meaning of the category) to categorize. Further, it is argued that even if these underlying attributes are known to novices, and therefore are available for making decision, these attributes are not readily accessible. Therefore, it is possible that novices can be made to behave more like experts by priming or cuing these underlying attributes.

Drawing a parallel between expertise-related differences and age-related differences, two hypotheses are proposed. First, it is anticipated that even after controlling for product experience, younger children will be more dominated by visual salience (e.g., package, color) in categorizing products than will older children, whereas older children will be more likely to use underlying product characteristics (e.g., taste) in categorizing products than will younger children. Second, it is expected that "cued" rather than "free" categorization tasks (e.g., asking children to categorize based on choice preferences rather than to merely group products without specifying a "criterion" for categorization) will eliminate some of the observed age differences (see John 1981 for a review of the literature that provides further support for these hypotheses).

To test the hypotheses, a study was conducted with approximately 120 children ranging in age from 4 to 10 years. Children were divided into three groups based on age: below 6, 6-8, and 9 and over. Two product categories were selected for study: beverages and cereals. Brands familiar to all age groups were used. Children were presented with three brands from a product category at a time (e.g., Cheerios, Apple Jacks and Apple Squares). The brands were clearly identified by the experimenter and familiarity with the brands ascertained. Subjects were asked to group two of the three brands together and separate from the third and to verbalize their basis for doing so. Subjects sorted several triads and either no basis for categorization was specified (free sort) or a variety of cued bases for categorization were provided.

The results supported the hypotheses. As predicted by the first hypothesis, in the free categorization task, the use of taste (an underlying feature) as a basis for sort increased with age. Also, in the free categorization task, both the youngest and middle age groups were more likely to use visual product and package attributes to sort compared to the oldest age group. In the cued categorization task, however, there was some indication that the middle age- group behaved more like the oldest age group in their use of underlying and visual product attributes. Thus, support for the second hypothesis suggests that providing bases for categorization can alleviate some age differences. These results have implications for both categorization theory (especially for expert-novices differences in categorization), and public policy.



Noel M. Murray, Penn State University

Harish Sujan, Penn State University

Mita Sujan, Penn State University

Edward R. Hirt, Penn State University

Recent research has demonstrated that mood can affect categorization. Isen and Daubman (1984) found that subjects in a positive mood tended to create and to use categories more inclusively than did subjects in a neutral mood condition. Isen hypothesizes that positive mood causes subjects to see more interconnections between items and therefore to categorize at a superordinate level. Although Isen does not commit herself to an explanation for why subjects in positive moods see more interconnections, one possibility that she offers is that people tend to learn information in a positive mood and therefore reinstating a positive mood makes these interconnections more available.

An alternative explanation of the effect of mood on categorization can be developed by examining parallels between the effects of positive mood and the effects of expertise on processing. Both factors appear to increase cognitive flexibility or creativity in approaching tasks (Showers and Cantor 1985). Thus, positive mood subjects, like experts may be more adept at using their knowledge structures. This could explain why positive mood subjects can categorize at superordinate levels if the task requires them to do so (as it did in Isen's study) and it leads to the additional inference that positive mood subjects also should be able to categorize at subordinate levels if the task can best be performed at more specific levels.

Two studies were conducted to test this cognitive flexibility hypothesis. The first varied Mood and Processing Goal and observed the effects on categorization. Positive mood subjects adapted their categorization scheme to the processing goal to a greater extent than other subjects thereby providing evidence for their cognitive flexibility. Specifically, contingent upon task demands, positive mood subjects were both more inclusive and less inclusive than other subjects.

A second study pursued the cognitive flexibility hypothesis by examining subjects' bases for categorization. Positive mood subjects were found to employ creative and non-obvious ways to categorize at both the superordinate level, where they found non-obvious-or novel interconnections between items, and at subordinate levels, where they developed novel distinctions between items. Implications of these findings for consumer choice and evaluation strategies were discussed.



Laura Peracchio, Northwestern University

Alice M. Tybout, Northwestern University

Mandler (1982) suggests that categorization processes may have consequences for judgment. More specifically, he hypothesizes that the match or congruity between a category activated in memory and a new object determines the extent and direction of the processor's elaboration. This cognitive elaboration, in turn, influences judgments such that moderate incongruity leads to more favorable evaluation than either complete congruity or extreme incongruity.

Moderate incongruity is argued to lead to more favorable evaluation than congruity because encountering an object/product that is unexpected and puzzling over it is thought to be inherently satisfying, provided that the incongruity can somehow be accommodated within the existing cognitive structure. Moderate incongruity also is expected to lead to more favorable evaluation than extreme incongruity because the thought stimulated by extreme incongruity often fails to result in a satisfactory resolution.

Mandler's hypothesis regarding the relationship between level of congruity and evaluation and the role that cognitive elaboration plays in this relationship were examined experimentally. In this research, Rosch's notions of hierarchical category structure were used to operationalize levels of congruity (Rosch 1975; Rosch and Mervis 1975). Congruity was assumed to occur when a new product conformed to the attributes associated with an activated product category. Moderate incongruity was operationalized by a new product that possessed attributes that not completely matching those associated with the activated product category, but the incongruity could be resolved by moving to the next lower level in the product hierarchy. Finally, extreme incongruity was argued to occur when the attributes of a new product neither matched those associated with an activated product category, nor did they match those associated with categories at the next lower level in the product hierarchy.

The design included eight treatments that were analyzed as three overlapping 2 x 2 factorial designs. The results of Design One supported the hypothesis that moderate incongruity leads to more favorable evaluation than either congruity or extreme incongruity. These findings replicated previous studies (MeyersCLevy and Tybout 1987; Lehtisalo 1985) and demonstrated the robustness of the phenomenon by employing a new product category.

The results of Design Two provided some evidence for the role that cognitive elaboration plays in the relationship between level of congruity and evaluation. Specifically, the effects observed in Design One were only obtained when comparative cognitive elaboration was encouraged. No effect of the level of congruity on evaluation was observed when subjects were instructed to evaluate a product without reference to category knowledge stored in memory.

Design Three demonstrated that the advantage of moderate incongruity could be reinstated by using a cue other than instruction set to stimulate comparative processing. Specifically, when the manufacturer was incongruent with the product category, moderate incongruity was once again observed to lead to more favorable evaluation than congruity.

On the basis of this research, it was argued that Mandler's hypothesis may help to explain how consumers form judgments. However, it is important to note that Mandler's theorizing only pertains to the affect stimulated by the process of responding to different levels of match/mismatch between a category and a new object. It is likely that in some consumer judgments this affect will be overwhelmed by the affect associated with the category or with the attributes of the object.


Following the presentations, two discussants, John Carroll of M.I.T. and Ann Beattie of New York University, offered their perspectives on the topic of categorization in general and the papers presented in particular. Their insights are briefly summarized below.

John Carroll observed that the theme of individual differences, particularly differences in expertise, ran through many of the papers presented. Hutchinson and Alba explicitly examined this variable through varying subjects' training experience, John used children of different ages to represent degrees of expertise and Murray et al. drew an analogy between mood and expertise. (Although the observation was not made, even the Peracchio and Tybout paper might be argued to be related implicitly to expertise because experts and novices may have different thresholds for recognizing incongruity as a function of the development of their existing knowledge structure.) Carroll encouraged consumer researchers to pursue this line of research with an emphasis on the learning process itself (e.g. How do people learn?).

In conducting further research, he cautioned that it would be important to recognize that no one style of processing would be optimal for all situations. Although Hutchinson and Alba provide evidence that holistic processing is less effective than analytic processing in detecting the difference between similar products, holistic processing is not an inherently inferior strategy; it can be quite sophisticated and, in some situations, a superior approach. The overriding criterion for success is adapting one's strategy to the task demands, as the Murray et al. paper suggested.

Carroll encouraged John's strategy of using children as one means of understanding the learning process. He noted that when using children as subjects, the dependent measures must control for age differences in children's abilities to verbalize the strategies they employ. For this reason, he suggested that it may be necessary to rely more on nonverbal or behavioral bases rather than verbal self-reports as a basis for inferring the categorization strategies used by children. John's strategy of obtaining both verbal and behavior measures allows assessing the extend to which the age differences in verbalization confound interpretation.

Carroll expressed interest in mood effects on categorization and suggested that one direction for future research in this area would be to explore more extreme levels of mood. He speculated that the relationship between favorableness of the mood and cognitive flexibility in categorization might be nonmonotonic rather than linear, at extreme levels positive mood might interfere with categorization. He also wondered whether children of different ages might spontaneously experience different moods when aware they were participating in an experiment.

Finally, Carroll suggested that, in contrast to the problems with verbalizations when studying children, verbal reports such as cognitive response measures might provide additional insight into the process hypothesized to underlie the findings reported by Peracchio and Tybout. He also noted that, when examining questions about how people learn, it would be especially worthwhile to explore the Rosch-type hierarchical structures used by Peracchio and Tybout with particular attention to how consumers form basic level categories and how these differ across segments.

Ann Beattie structured her remarks around three aspects of categorization; the formation of categories, the process of categorizing, and the effect of categorization. Echoing Carroll's call for further research examining the process by which categories are formed, Beattie argued that a better understanding of how people come to appreciate covariation between attributes or features is needed. Hutchinson and Alba's work, which included a manipulation of the ease of detecting covariation, was viewed as an appropriate step in this direction. It also was suggested that future research pursue the ways in which experience and knowledge affect the formation of categories.

In discussing ongoing categorization tasks, Beattie noted that a better understanding of how context affects categorization processes is needed. She observed that all of the papers in the session made some contribution to this important topic by examining contextual factors such as processing goals and ease of categorization, and she encouraged consumer researchers continue their focus on such variables.

Beattie indicated that fur her work examining the effect of categorization on evaluation also would be desirable. She suggested that it would be worthwhile to examine how various levels of abstraction in categorization (i.e. levels in a category hierarchy) influence evaluation. She also observed that the role of novelty in stimulating elaboration and thereby influencing evaluation, as proposed by Peracchio and Tybout paper, is an interesting avenue for future research.

Finally, Beattie noted that the papers in the session explored a number of individual differences (e.g expertise, age, mood). She observed that such a focus on individual differences can help achieve general understanding of categorization processes and attention would continue to be directed to such variables.

In their summary remarks, both discussants, who are psychologists by training, observed that consumer researchers are not simply borrowing categorization theory developed in psychology and applying it in consumption situations. Instead Carroll and Beattie contended that consumer researchers were giving back to -psychology theoretical contributions to understanding Categorization processes. Moreover, the rapid evolution of categories in areas such as high tech products may uniquely position consumer researchers to provide further advances in the field of categorization.


The session co-chairs thank the paper presenters, discussants, and audience for their thoughtful and stimulating contributions to this session.


Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research, 13 (March), 411-454.

Bettman, James R. and Mita Sujan (1986), "Effects of Framing on Evaluation of Comparable and Noncomparable Alternatives by Expert and Novice Consumers," Journal of Consumer Research, 14 (September), 141-154.

Cohen, Joel B. (1982), "The Role of Affect on Categorization: Towards a Reconsideration of the Concept of Attitude," in Advances in Consumer Research, Vol. 9, ed. Andrew A. Mitchell, Ann Arbor, MI: Association for Consumer Research, 94-100.

Day, George S., Allan D. Shocker and Rajendra K Srivastava (1979), "Customer-Oriented Approaches to Identifying Product Markets," Journal of Marketing, 43 (Fall), 8-19.

Isen, Alice M., and Kimberly a. Daubmar (1984), "The Influence of Affect on CategorizationS" Journal of Personality and Social Psychology, 47 (December), 1206-1217.

John, Roedder Deborah (1981), "Age Differences in Children's Responses to Television Advertising: An Information Processing Approach," 8 (September), 144-53.

Johnson, Michael D. (1984), "Consumer Choice Strategies for Comparing Noncomparable Alternatives," Journal of Consumer Research, 11 (December), 741-753.

Lehtisalo, Marjo-Ritta (1986), The Influence of Price on Product Evaluation: An Information Processing Perspective, Northwestern University, unpublished doctoral dissertation.

Mandler, George (1982), "The Structure of Value: Accounting for Taste," Affect and Cognition: The 7th Annual Carnerge Symposium, Margaret S. Clarke and Susan T. Fiske (eds.), Hillsdale, NJ., Erlbaum, 3-36.

Myers-Levy, Joan and Alice M. Tybout (1987), "Schema Congruity As a Basis for Natural Object Evaluation," manuscript under review.

Rosch, Eleanor (1975), "Cognitive Representation of Semantic Categories," Journal of Experimental Psychology: General, 104 (September), 192-233.

Rosch, Eleanor and Carolyn B. Mervis (1975), "Family Resemblance Studies in the Internal Structure of Categories," Cognitive Psychology, 7 (October), 573-605.

Rosch, Eleanor, Carolyn B. Menis, Wayne D. Gray, David M. Johnson, and Penny Boyes-Braem (1976), "Basic Objects a in Natural Categories," Psychology, 8 (July), 382-439.

Showers, Carolyn and Nancy Cantor (1985), "Social Cognition: A Look at Motivated Strategies," Annual Review of Psychology, 36, 275-305.

Sujan, Mita (1985), "Consumer Knowledge Effects on Evaluation Strategies Mediating Consumer Judgments," Journal of Consumer Research, 12 (June), 31-46.

Sujan, Mita and Christine Ann Dekleva (1987), "Product Categorization and Inference Making: Some Implications for Comparative Advertising," Journal of Consumer Research, 14 (December), 372-378.

Wright, Peter L. (1986), "Schemer Schema: Consumers' - Intuitive Theories About Marketers' Influence Tactics," in Advances in Consumer Research, Vol. 13, Richard J. Lutz, ed. Ann Arbor, MI: Association for Consumer Research, 1-3.