Evoked Set Formation and Composition: the Learning and Information Processing Hypotheses

Frederick E. May, University of Missouri-St. Louis
ABSTRACT - This discussion paper examines the controversy between learning theory and information processing theory in terms of their abilities to explain evoked set composition and formation. These two hypotheses appear to lead to contradictory expectations with respect to the composition and formation of evoked sets. The report examines existing data on evoked set formation in an attempt to reconcile these two approaches.
[ to cite ]:
Frederick E. May (1979) ,"Evoked Set Formation and Composition: the Learning and Information Processing Hypotheses", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 222-226.

Advances in Consumer Research Volume 6, 1979      Pages 222-226

EVOKED SET FORMATION AND COMPOSITION: THE LEARNING AND INFORMATION PROCESSING HYPOTHESES

Frederick E. May, University of Missouri-St. Louis

[Thanks to Johann Arndt, and Ronald P. May for their helpful suggestions.]

ABSTRACT -

This discussion paper examines the controversy between learning theory and information processing theory in terms of their abilities to explain evoked set composition and formation. These two hypotheses appear to lead to contradictory expectations with respect to the composition and formation of evoked sets. The report examines existing data on evoked set formation in an attempt to reconcile these two approaches.

INTRODUCTION

This report is concerned with Howard's (1977) concept learning theory and McGuire's (1976) information processing theory in relation to consumer choice of evoked sets. For the purpose of this discussion the evoked set is "defined as those brands the buyer considers when he (or she) contemplates purchasing a unit of the product class." (Howard and Sheth, 1969; Glossary of Terms: Definitions of Central Concepts, p. 416).

The evoked set is important because it is a subset of existing brands in the market, and it is a subset of the brands of which the buyer is aware. The process of forming an evoked set from these broader sets is therefore of interest to both consumer researchers and managers. [McGuire defines evoked set as the available options that are generated as a result of the search strategy in the search and retrieval step in the information processing sequence. It is assumed that the options correspond to the brands considered by the buyer.]

McGuire postulates that the choice of the evoked set is an intermediary decision toward the brand choice and that the process of identification is different from the process of choosing a brand within the set.

Ideally, one would like to identify by name the brands that the consumer selects for the evoked set. However, a more practical approach is to identify brands in the evoked set by the consumer's prior experience with them. Thus the dependent variable becomes the following typology of evoked sets: those evoked sets composed solely of untried brands, those composed of some tried and some untried brands, and finally, those composed solely of tried brands. [Howard and Sheth (1969) develop a typology of the evoked set based on the consumer's familiarity with the brands considered in order to predict the amount and duration of search that would follow the formation of the evoked set. Here, one is interested in using a similar typology (the operational equivalent of the familiarity dimension) in order to analyze the factors leading to the formation of the evoked set.]

The composition of the evoked set as a dependent variable permits a comparison of the learning and information processing approaches to evoked set formation. Learning theory predicts that all consumers progress from untried sets through mixed sets to tried sets as they acquire experience with the product. Information processing theory states that the composition of the consumer's evoked set depends on the social and personal characteristics of the consumer and the decision rules used to identify the evoked set.

MCGUIRE'S INFORMATION PROCESSING MODEL

McGuire's (1976) model of directive aspects in personality structure is shown in Figure I. There are eight stages in the information processing sequence. In each stage McGuire postulates two or three constructs. McGuire's entire paradigm is too massive to deal with in this report. This report focuses on two steps---exposure to information, and information search and retrieval.

FIGURE I

INFORMATION PROCESSING IN CONSUMER DECISION MAKING

There are two constructs presented in search and retrieval. The available options (evoked set) are identified or attained by search strategies. The structure of the cognitive storage systems refers to tree diagrams, matrices, and push down lists.

McGuire presents two constructs in the exposure to information stage. For the social structuring of information exposure the adoption and diffusion of innovation literature is relevant. This is recently reviewed by Engel, Blackwell, and Kollat (1978). The cognitive style literature reviewed by Pinson (1975) provides a valuable discussion of the personal characteristics affecting exposure to information.

Social and Personal Characteristics

The social characteristics of the consumer which direct the consumer's exposure to information here refer to the consumer's position in the social structure. An operational measure of social position is occupational prestige. An analysis of variance shows a significant relationship between evoked set composition and occupational prestige in a study of new automobile buying (May, Homans, and Maddox, 1977b). The untried evoked sets are correlated with the highest occupational prestige and the tried sets with the lowest. Income was found to be a weak predictor of evoked set composition.

Personal characteristics here refer to information processing complexity of which one important indicator is education. Pinson (1975) describes the three basic aspects of information processing complexity:

"'Differentiation', 'Discrimination', and 'Integration'. Broadly defined, 'differentiation' refers to the number of dimensions used by an individual in processing information. 'Discrimination' refers to the number of separated conceptual categories on a dimension. Finally, 'integration' refers to the degree of interrelatedness of elements within a particular cognitive domain."

Although studies have not linked education directly to information processing complexity, education is a frequent and strong correlate in studies of the amount of information search (Newman, 1977). Amount of information search has in turn been shown to be correlated with information processing complexity in studies reviewed by Pinson (1975).

Education and occupational prestige correlates with evoked set composition and size in a canonical correlation analysis of new car buyers (Homans, Maddox, and May, 1977). Evoked set size, which tends to be correlated with composition, has been shown to be related to education in studies of Norwegian and United States car buyers (Maddox, Gr°nhaug, Homans, and May, 1978). Brand loyalty is inversely related to education across grocery product categories (Frank, Douglas, and Polli, 1968). These data, however, speak only indirectly to the composition of the evoked set.

Consumer Product Specifications

Product specifications are a third major factor that determines the composition of the evoked set. Product specifications are the appropriate decision rules for forming the evoked set. In the study of industrial buyer behavior the use of specifications is well known. In fact, the terminology used here comes from the industrial buying literature (Lamar and Dobler, 1971). However, the notion that the consumer constructs specifications as a search strategy is new to the consumer behavior literature (Bettman and Zins, 1977).

May, Homans, and Maddox (1978) find four types of product specifications: 1) single, and 2) multiple brand category specifications, 3) attribute specifications, and 4) performance specifications. New car buyers were asked: "When you first decided to buy what were your specific ideas about the type of car you wanted?"

The answers were classified in the following manner: the respondent mentioned the name of a single brand (single brand category specifications); the respondent mentioned the name of more than one brand (multiple brand specifications); the respondent mentioned attributes of the car wanted (attribute specification). For example, the respondent may have mentioned that he wanted a small car, a heavy car, a four door sedan.

If the respondent mentioned a function or activity for which the car is to be used and related it to a necessary attribute he was placed in the performance specification category. For example, the respondent may have expressed the desire for an economical car for city driving, or a station wagon for a big family and hauling things, or a heavy car for highway driving(May and Homans, 1977a).

The product specifications range over a dimension of complexity from the simplest (single brand) to the most complex (performance) specifications. The choice of product specification has been shown to be a function of personal and social characteristics, of which education correlates most highly with the complexity of specifications (May, Homans, and Maddox, 1978).

The different specifications are likely to produce different types of evoked sets. To illustrate: when brand specifications are used the most frequent outcome is a tried evoked set, and when attribute specifications are used, the most frequent outcome is a mixed evoked set. Finally, when a performance specification is used the most frequent outcome is an untried set.

The product specifications appear to correspond to three types of memory structures upon which the consumer draws to construct the evoked set. These structures are utilized in the search and retrieval step of McGuire's information processing model.

Memory Structures

McGuire's hypothesized memory structures provide a conceptual explanation of the link between product specifications and evoked set composition. The three memory structures are shown in Figure II. The hierarchical network product concept (tree-diagram structure) encompasses the entire product class which is usually labeled by nouns and sub-product classes that are labeled by using adjectives. For example, the superordinate product class concept of automobiles is subordinated by such classes as compact cars, station wagons, sports cars, etc. Each of the sub-product classes is usually designated by an adjective and has a distinguishable function. For example, compacts are for economy, station wagons are for large families, and sports cars are for fun and pleasure.

Matrix product concept structures refer to brands and their attributes. Such structures are likely to be used for one product or sub-product class. Finally, push down lists (or preference lists) are structures which contain a list of preferred or acceptable brands. The list is a kind of collapsed matrix. The consumer buys the first preference if it is available at the right price. If not available, the second preference is selected, and so forth.

The particular memory structure in which the consumer stores relevant information about the product influences the type of information which the consumer seeks about the product in the process of forming the evoked set. To illustrate, the most appropriate memory structure from which to draw performance specifications is a network structure because it defines the domain (product class) in functional terms. Following logically, the most appropriate memory structure for the attribute specifications is the matrix. The matrix structure can be a sub-structure in the network structure. Finally, the push down list is the most appropriate memory structure for a single or multiple brand product specification.

FIGURE II

NETWORK PRODUCT CONCEPT STRUCTURES

Need For a Dynamic Theory

One problem with the information processing paradigm is that it assumes a static consumer who will, once his social and personal characteristics are determined, always follow the same pattern. The highly educated consumer, for example, can be expected to always construct network memory structures, performance specifications, and untried evoked sets. How does the highly educated consumer ever come to examine tried sets and how does the poorly educated consumer ever come to examine untried sets? Stated differently, how does the poorly educated consumer adopt a new brand or product? To answer this question we need a theory of dynamics or learning.

HOWARD'S THEORY OF CONCEPT LEARNING

Howard (1977) distinguishes three concept learning phases in the adoption process. (See Figure III). In the first phase the consumer forms the concept of a new product (or sub-product) by comprehending its distinctive functions and attributes. Howard calls this phase concept formation. Concept formation is here defined as the learning process that results in untried sets.

In the second phase of new product adoption the consumer extends the product concept by comparing untried brands to previously tried brands. Howard calls this process concept attainment. However, psychologists generally do not discriminate between concept formation and attainment. Therefore, the term concept extension is more appropriate for this phase of the adoption process. Concept extension results in mixed evoked sets.

FIGURE III

MODEL OF EVOKED SET COMPOSITION

In the third phase of the adoption process the consumer has completely learned the product concept, and begins to utilize it. In this phase the consumer chooses only tried brands in the evoked set. Howard calls this phase concept utilization.

To predict the frequency distribution of concept learning situations, Howard introduces the product life cycle. In the introduction stage most consumers are in the concept formation phase and, therefore, untried evoked sets are most common. In the growth stage most consumers are in the concept extension phase and mixed sets of some tried and untried brands are most frequent. In the maturity stage most consumers are in the concept utilization stage, and tried sets are most common.

There is some evidence that consumers proceed through the three adoption phases and form evoked sets according to the theory. For example, the mean age and the mean number of previous purchases for the different types of evoked sets appear to vary as predicted (May, Homans, and Maddox, 1977b). However, variance analyses indicates that the age and purchase distributions do not differ significantly.

One can conclude that age and the number of prior purchases are confounded by education and occupational prestige factors. That is, given the information processing complexity of consumers the learning hypothesis appears to apply.

Howard's product life cycle hypothesis predicts the frequency distribution of evoked sets correctly. New car buyers considered tried evoked sets most frequently, and untried sets least frequently (May, Homans, and Maddox, 1977c). The finding is somewhat surprising in view of Howard's argument that durable goods buyers are unlikely to reach the concept utilization phase even in the maturity stage of the product life cycle.

DISCUSSION

Table I summarizes typical consumer behavior according to the two theoretical approaches. Howard's three stage process corresponds in certain respects to the three levels of information processing complexity. The concept formation phase requires the highest level of information processing complexity, and the utilization phase requires the simplest level of information processing.

TABLE I

TYPICAL CONSUMER BEHAVIOR IN INFORMATION PROCESSING STEPS AND CONCEPT LEARNING SITUATIONS

Howard's theory assumes that the accumulation of product knowledge and experience changes the consumer's information processing behavior. Learning simplifies the information processing task.

In contrast to learning theory, information processing theory assumes that the consumer starts at a given level of complexity and remains there. The theory implies that the type of product specifications used by consumers do not change as the consumer accumulates purchase experience. So far unpublished data support the latter hypothesis. The proportion of brand specifiers remains constant, regardless of the number of prior new car purchases.

The implication for the frequency distribution of evoked sets is as follows. Information processing theory predicts that the frequency distribution of evoked sets remains constant as long as the frequency distribution of the complexity of the buyers of the product remains constant. As the product diffuses through the social structure one would expect the frequency of evoked set compositions to change as less complex buyers adopt the product.

Learning theory predicts that the frequency distribution of evoked sets remains constant as long as the familiarity of the buyers with the product remains constant. As the product progresses through the life cycle one would expect the distribution of evoked sets to change with increasing buyer familiarity with the product.

One may conclude that both theories are necessary to predict and explain evoked set composition and formation.

REFERENCES

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

Engel, ,J. F., Blackwell, R. D., and Kollat, D. T. (1978) Consumer Behavior, Hinsdale, IL: Dryden Press.

Frank, R. E., Douglas, S. P. and Polli, R. E. (1968) "Household Correlates of 'Brand Loyalty' for Grocery Products," Journal of Business, 41, 237-245.

Homans, R. E., Maddox, R. N., and May, F. E. (1977) "Correlates of the Level of Decision Processing for Automobiles," Proceedings of the American Institute of Decision Sciences.

Howard, J. A. (1977), Consumer Behavior: Application of Theory, New York: McGraw-Hill.

Howard, J. A., and Sheth (1969) The Theory of Buyer Behavior, New York: Wiley & Sons, Inc.

Lamar, L. Jr. and Dobler, D. W. (1971) Purchasing and Materials Management, New York: McGraw-Hill.

McGuire, J. (1976) "Some Internal Psychological Factors Influencing Consumer Choice," Journal of Consumer Research, 2, 302-319.

Maddox, R. E., Gr°nhaug, K., Homans, R. E. and May, F. E. (1978) "Correlates of Information Gathering and Evoked Set Size for New Automobile Purchasers in Norway and the U.S." in Advances in Consumer Research, Volume V, Hunt, K. ed., Ann Arbor: Association for Consumer Research.

May, F. E., and Homans, R. E. (1977a) "Evoked Set Size and the Level of Information Processing in Product Comprehension and Choice Criteria," in Advances in Consumer Research, Vol. 4, ed. William D. Perreault, Jr., Ann Arbor: Association for Consumer Research.

May, F. E., Homans, R. E., Maddox, R. N. (1977b) "Dimensions of Problem Solving Behavior in New Automobile Purchases," in Proceedings of 4th International Research Seminar in Marketing, Senanque Abbey, France: Institut D'Administration des Enterprises.

May, F. E., Homans, R. E., and Maddox, R. N. (1977c) "Evoked Sets and Memory Systems" in Contemporary Marketing Thought 1977 Educators Proceedings, Greenberg and Bellenger, eds. Chicago: American Marketing Association.

May, F. E., Homans, R. E., and Maddox, R. N. (1978) "Evaluative Criteria and Cognitive Style," Proceedings of the Southwestern Marketing Association, Dallas, Texas.

Newman, J. W. (1977) "Consumer External Search: Amount and Determinants," in Consumer and Industrial Buying Behavior, eds. Woodside, A. G.; Sheth, J. N.; Bennett, P. D., Amsterdam: Elsevier North-Holland, Inc.

Pinson, C. (1975) "Consumer Cognitive Styles: Review and Implications for Marketers," Research Paper No. 160 INSEAD The European Institute of Business Administration.

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