A Typology of Consumer Knowledge Content

ABSTRACT - It is well accepted that stored knowledge about a domain affects the processing of new information about that domain; however, there is little agreement among consumer researchers on how to measure knowledge. It is argued here that consumer knowledge can be classified and measured by its content. A typology of consumer product knowledge content is proposed, based on a review of the relevant literature. An empirical study finds that this typology is comprehensive, reliable, and able to classify knowledge into empirically distinguishable categories.

Citation:

Merrie Brucks (1986) ,"A Typology of Consumer Knowledge Content", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 58-63.

Advances in Consumer Research Volume 13, 1986      Pages 58-63

A TYPOLOGY OF CONSUMER KNOWLEDGE CONTENT

Merrie Brucks, University of North Carolina

ABSTRACT -

It is well accepted that stored knowledge about a domain affects the processing of new information about that domain; however, there is little agreement among consumer researchers on how to measure knowledge. It is argued here that consumer knowledge can be classified and measured by its content. A typology of consumer product knowledge content is proposed, based on a review of the relevant literature. An empirical study finds that this typology is comprehensive, reliable, and able to classify knowledge into empirically distinguishable categories.

INTRODUCTION

Stored knowledge about a domain affects the processing of new information about that domain. In one of the earliest studies to demonstrate this effect, de Groot (1965, 1966) found that expert chess players were able to recall mid-ga e chess positions better than novice chess players. Later research (Chase and Simon 1973a, 1973b) indicated that the experts' superior memory for chess positions was based on their knowledge of prototypical game patterns as opposed to a superior memory in general. Analogous conclusions have been drawn in at least eight other domains: physics, electronics, music, baseball, programming, and the games of Go, Gomoku, and bridge (Chi, Glaser, and Rees 1981). All of these findings indicate that differences in the content and organization of domain-specific knowledge between experts and novices result in differences in the processing and recall of domain-related information.

Recently, several studies have examined the effect of variables related to knowledge (e.g., product class experience, familiarity) on various consumer information processing activities (e.g., Alba 1983; Bettman and Park 1980; Bettman and Zins 1977; Brucks 1985; Edell and Mitchell 1978; Johnson and Russo 1984; Park 1976; Punj and Staelin 1983; Srull 1983; Sujan 1985). With few exceptions, the consumer behavior literature has treated knowledge as a unidimensional variable. However, recent research indicates that knowledge is a complex construct that is characterized by the structure and the content of information stored in memory (Brucks and Mitchell 1981; Hutchinson and Alba 1985; Kanwar, Olson, and Sims 1981). The structure of knowledge refers to how knowledge is represented and organized in memory. The reader is referred to Hutchinson and Alba (1985) for an outstanding review of this topic. The content of knowledge refers to the subject matter of information stored in memory. For example, three types of knowledge content are knowledge about terminology, knowledge about specific brands, and rules for evaluating a new brand. This paper reviews some previous typologies of knowledge content, and proposes a new, more comprehensive typology, which is empirically examined.

An understanding of the composition of knowledge content will aid researcher in developing objective tests of knowledge by indicating what questions need to be asked. Furthermore, a multi-dimensional account of knowledge content may provide a better understanding of consumer behavior if different types of knowledge content affect behavior in different ways.

BACKGROUND

There are many possible ways to classify knowledge by its content, and in fact, several typologies have been proposed. One early typology was proposed by educational psychologists in an effort to help teachers construct exams that tap all types and levels of knowledge (Bloom et al. 1956). This comprehensive classification system, while developed for an entirely different application, provides many useful ideas for developing a typology of consumer product knowledge. The relevant portion of this hierarchy is Presented in Table.

TABLE 1

TYPOLOGY OF KNOWLEDGE CONTENT FROM EDUCATIONAL PSYCHOLOGY (BLOOM ET/AL. 1956)

1. Knowledge of Specifics

1.1 Knowledge of terminology: knowledge of the referents for specific verbal and non-verbal symbols.

1.2 Knowledge of specific facts: knowledge of dates, events, persons, places, sources of information, etc.

2. Knowledge of Ways and Means of Dealing With Specifics

2.1 Knowledge of conventions: knowledge of characteristic ways of treating and presenting ideas and phenomena.

2.2 Knowledge of trends and sequences: knowledge of the processes. directions, and movements of phenomena with respect to time.

2.3 Knowledge of classifications and categories: knowledge of the classes, sets. divisions, and arrangements that are regarded as fundamental or useful for a given subject field. purpose, argument, or problem.

2.4 Knowledge of criteria: knowledge of the criteria by which facts. Principles, opinions, and conduct are tested or judged.

2.5 Knowledge of methodology: knowledge of the methods of inquiry, techniques, and procedures employed in a particular subject field as well as those employed in investigating particular problems and phenomena.

3. Knowledge of the Universals and Abstractions in a Field.

3.1 Knowledge of principles and generalizations: knowledge of particular abstractions that summarize observations of phenomena.

3.2 Knowledge of theories and structures: knowledge of the body of principles and generalizations together with their interrelations that present a clear, rounded. and systematic view of a complex phenomenon, problem, or field.

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

More recently, other ways to classify knowledge by its content have been proposed. One useful and well-accepted dichotomy distinguishes declarative knowledge from procedural knowledge (Anderson 1976). Declarative knowledge refers to knowledge about concepts. objects, or events. Procedural knowledge, on the other hand. refers to knowledge of rules for taking action. Declarative knowledge is currently believed to be stored in an associative network and organized into schemas. whereas procedural knowledge is believed to be stored and organized into production systems.

The content of knowledge about physics has been examined for physics exports and novices (Chi. Glaser, and Rees 1981). In this study, two graduate physics students (experts) and two undergraduate physics students (novices) were as}od to tell everything they knew about a selected sample of 20 physics concepts. The protocols generated from this task were then analyzed on the basis of knowledge content. The analysis indicated. first, that both experts and novices possessed fundamental knowledge about the basic properties of physics objects (such as "inclined plane"). but the experts possessed a more complete. or detailed. knowledge of these properties, as well as additional knowledge based on the major physical laws affecting the object (such as "conservation of energy"). Second, the experts closely linked explicit problem-solving procedures with declarative knowledge of physical objects in long term memory. Novices showed almost no such linkages. And third, the experts had such more knowledge about the. conditions under which the major physical laws are applicable to the object.

To generalize these results to the domain of consumer products, one might expect that both experts and novices possess knowledge about basic properties of a product class, possibly knowledge about which attributes characterize the product class. This knowledge would. however, be more detailed in experts. possibly including knowledge of how attributes are related to each other and how attribute types and levels affect performance. Second, we might expect that experts would have more knowledge about how to choose and buy a product in the domain. And third. we might expect that experts would have more knowledge about how the conditions under which the product will be used would affect the important attributes for evaluation. For example. compared with a novice. a stereo expert would be more likely to know that requirements for sensitivity and selectivity in stereo tuners and receivers differ between rural and urban settings. The discussion above suggests that detailed knowledge of attributes be classified separately from knowledge of the existence of attributes. and that classifications are necessary for procedural and situational knowledge.

Finally, this discussion of possible knowledge content classifications concludes with some notable research from the consumer behavior literature. Hastie (1982) makes a distinction between generic product knowledge and individual product knowledge. Generic product knowledge includes !'general information about classes of products. instances exemplifying the products. the existence of different types of products (implying knowledge of correlations between product attributes, such as luxuriousness and gas consumption of an automobile), and information about the attributes or dimensions that are relevant and important in making decisions concerning the products (e.g., the distribution of existing product types along the price dimension).'

Individual product knowledge includes "information such as prices, color, taste, durability. features, etc., of each product. Furthermore. this knowledge structure would include information about relationships among the products. For example. that product X is more expensive than product Y, or that product A and product E are manufactured by the same company."

An approach that focuses on "individual product knowledge is advocated by Russo and Johnson (1980). They propose that product knowledge be classified by level of inference and whether it is linked by attribute or by brand. Level of inference refers to the degree to which information available in the environment has been processed by the individual and retained in memory. The type of linkage refers to whether individuals link attribute values for each brand by brand or by attribute. In other words, individuals may link all the attribute values for a single brand together or they may link all the brands together for a single attribute. Russo and Johnson's classification scheme appears to be useful in that 80 percent of knowledge statements that had been generated by a knowledge probe fit into the scheme. Bettman and Park (1980) developed a coding scheme for classifying the use of prior knowledge during decision-making. This coding scheme includes brand evaluations. attributes of specific brands, and evaluations of particular levels of attributes.

TYPOLOGY OF KNOWLEDGE CONTENT

In this section of the paper a comprehensive typology for consumer product knowledge content is proposed. The typology has continuously evolved over time. An early version, presented in Brucks and Mitchell (1981), was modified after some initial exploratory research. This initial research consisted of obtaining responses from a convenience sample of subJects to the probe Write down everything you know about [product class]." Products used were microwave ovens, sugar, and aspirin. Analysis of the responses indicated that the typology was not sufficiently comprehensive, i.e., many statements of knowledge did not fit in any of the categories.

Based on this research, a new typology was proposed and a coding scheme developed. As the research described in the following section progressed, it became apparent that some minor changes were needed in this coding scheme. Modifications to the coding scheme (and therefore the typology) were made to satisfy three objectives:

1) The typology and coding scheme should be easy to use and seem logical to people who are using the coding scheme.

2) The typology should cover as many of the subjects' statements as possible while remaining relatively parsimonious.

3) The categories in the typology should be as distinct from each other as possible.

The following typology was developed in an attempt at satisfying the above objectives. It also incorporates many of the ideas from the research previously cited.

1. Terminology refers to knowledge of the meanings of terms used within a domain, for example, knowing that "pronation" refers to heels that turn in while running.

2. Product Attributes refers to knowledge of which attributes are available for evaluating a brand. It includes knowledge of attributes that a person would use in making a decision and also those that she would not use but is aware of their existence. For example. the statement "Some people consider light weight essential in a new running shoe" indicates knowledge of the attribute "weight" whether or not the individual thinks this a relevant attribute for her own decision making.

3. General Attribute Evaluation refers to knowledge of the overall evaluation for an attribute or an attribute level, for example, "I like waffle soles" or "I don't like a heavy shoe."

4. Specific Attribute Evaluation on the other hand, refers to knowledge of specific criteria used to evaluate an attribute, i.e., cut-off points or reference points used to Judge how satisfactory an attribute level is, for example, "I won't spend more than 30 dollars." It also refers to how an attribute (or a specific level of an attribute) is related to other attributes and/or performance criteria, for example, "A lightweight shoe doesn't last very long."

5. General Product Usage refers to knowledge of how the product can be used, what characteristics of the usage situation are relevant when a purchase is being considered, and which product characteristics are affected by these usage situation characteristics. Included in this category are normative rules for usage, for example, "You should not wear running shoes to play tennis," and knowledge of hour the product class can be categorized based on usage, for example. "There are several types of running shoes: competition shoes. training shoes, track shoes. and cross-country shoes."

6. Personal Product Usage knowledge includes memories of usage experiences, knowledge of memories of usage experiences of personal acquaintances. and knowledge of the salient characteristics of one's own usage situation and the usage situations of personal acquaintances. For example, knowledge of characteristics of personal usage situations includes such knowledge as "I don't run long distances" or "My sister once wore out a pair of shoes in 3 weeks.'

7. Brand Facts refers to knowledge of how brands "score" on an attribute, overall evaluations of a brand, and other brand facts such as comparisons between brands on an attribute. Example are "I'd never wear Adidas "or" Adidas have an antimicrobial sole."

8. Purchasing and Decision Making Procedures refers to knowledge about the purchasing process. Included in this category are memories of personal purchase experiences and also normative models of the purchase process. An example of the former is "I bought my last pair of shoes from Mitchell's Attic." An example of normative knowledge is. "You shouldn't buy the first thing you see."

Notice that categories 1 through 7 versus category 8 represent the distinction between declarative and procedural knowledge (although category 5 also contains some procedural knowledge). Declarative knowledge is further classified as either general product class knowledge (categories 1 through 6) or individual product knowledge (category 7). within the declarative, general product class knowledge category, the distinctions made were inspired by Bloom et al. (1956). Chi. Glaser. and Rees (1981). and Pretest results.

The usefulness of the proposed typology depends on how well these categories characterize a person's total knowledge structure. how well these categories can be measured. to what extent they exist independently of each other. and to what extent they have some differential impact on behavior. The study presented in the following section addresses the first three of these issues.

EMPIRICAL EXAMINATION OF TYPOLOGY

Specifically, the objectives of this study are:

1) to provide a test of the comprehensiveness of the typology,

2) to assess the inter-judge reliability of the coding scheme, and

3) to provide an estimate of the inter-correlations between the amount of knowledge in each of the categories.

Since this research is concerned with naturally occurring knowledge structures, there were no experimental manipulations of product knowledge. Rather, memory probes were used to elicit existing product knowledge. A relatively unstructured technique was chosen to measure the subjects' knowledge in order to minimize any biases that would result from imposing a preconceived structure. Thus, a free recall procedure was employed using the following memory probe: "I want you to think about purchasing a [product]. Now. tell me everything that comes to dad about [product]." The intent of this probe was to encourage the subjects to produce as much knowledge about the product class as they could. In this regard. the probe is similar to those used by Chi. Glaser, and Rees (1981) and Russo and Johnson (1980).

It was judged unlikely. however. that subjects would recall everything they knew in response to a single probe. Therefore. a double-layered series of probes was used in which subjects were asked to elaborate on each of the points that they had mentioned in response to the general probe. No estimates exist for the reliability of this procedure. The procedure used in this study was modeled after a multi-layered probe method described by Olson (1979). Olson's procedure was designed to measure the structure of stored knowledge in memory, thus his subjects were supposed to verbalize knowledge as it had been stored in memory. In the present study, however, subjects were not limited to reporting knowledge as it exists in memory. Rather, they were allowed to make inferences and verbalize them during the reporting process. Such inferences have been termed "constructive recall." It was decided to allow constructive recall as well as stored knowledge in this study because it seems to better represent the knowledge that people actually use during decision making than stored knowledge alone. She statements elicited by these probes were classified according to the type of knowledge content as previously defined.

Subjects

The thirty-one subjects participating in this experiment were undergraduate college students. They agreed to participate in return for a three dollar cash payment. Subjects participated in the study individually, and the experimental sessions lasted 50 to 60 minutes on average, although only 20 to 30 minutes were devoted to the data collection described here.

Products

The product chosen was running shoes. This product showed substantial variance in self-reported expertise among a pretest population of 86 students. Running shoes were also considered complex enough to warrant study of knowledge, but not so complex as to create a frustrating task for the more knowledgeable subjects.

Experimental Procedure

At the beginning, of each experimental session, the subject was given a practice verbalization task about a different product class--bicycles. This task was used to familiarize tho subject with verbalizing her/his knowledge about a product class. She experimenter provided feedback to help the subject understand the desired breadth and depth of the response.

At this point, the tape recorder was turned on and the subject was asked, "I want you to think about purchasing a pair of running shoes. Now, tell me everything, that comes to mind about running shoes." When the subject had finished, the subject was asked to elaborate on each of the points he or she had made (which the experimenter had recorded on paper and then read back to the subject one at a time). She entire verbalizing, process, including the elaboration, took an average of about ten minutes. Because of tape machine failure, data from 5 subjects were unusable, resulting in a final sample size of 26. The author numbered the subjects' statements on the transcriptions of the audiotapes, and a research assistant classified these statements according, to the coding scheme (which may be found in Brucks 1984). The author also coded the first 8 tapes that were transcribed in order to provide an estimate of inter-judge reliability.

Interjudge Reliability

The eight transcribed tapes that were coded by both judges yielded a total of 542 statements and 765 classifications. These 8 tapes represented 42 percent of the statements recorded. The judges agreed on seventy-two percent of the classifications. indicating that the elements in the proposed typology can be consistently identified.

In order to rely exclusively on the coding of one judge, it should be demonstrated that the coding of this judge does not systematically differ from the coding of the other judge. s repeated measures ANOVA was performed in order to test whether the judges systematically used the categories differently (i.e., whether they "favored" different categories). The factors (within-subject) were categories and judges, and the dependent variable was the number of coded pieces of knowledge. Since there were two judges. ten categories, and eight subjects, the ANOVA was computed based on 160 observations on the dependent variable. ^ significant Category X Judge interaction would imply that the judges were systematically assigning category codes differently from each other. As Table 2 shows, this effect was not significant.

TABLE 2

ANALYSIS OF VARIANCE FOR NUMBER OF STATEMENTS CODED

Fiske and Cox (1979) suggest a measure of interjudge reliability that may be derived from this analysis. They point out that as the Judge X Category mean square error goes to zero, the following expression approaches unity, thus providing a measure of reliability:

SSE (Categories) - MSE (Judges x Categories)

SSE (Categories)

This formula yields an estimate of reliability of .934 for the present study. This means that although the judges disagreed on category classifications twenty-eight percent of the time, these differences were not systematic.

In order to investigate the possibility that the judges were systematically assigning category codes differently from each other for the most infrequently used categories (which might not be revealed by the above ANOVA), the total number of assigned statements per category for each judge was compared) At most, the student judge's assignments differed from the author's by 29 percent (for General Attribute Evaluation). Based on these results. it was decided to use the coding of the student judge exclusively.

Data Description

In total, 1295 statements were elicited from the 26 subjects. Of these. 65 (or 5.0 percent) were judged by the coder to be irrelevant to running shoes. A typical example of an irrelevant statement is "My parents took my brother to the doctor because he bad really bad fallen arches." Eliminating these statements reduced the total number of statements to 1230. the number of relevant statements elicited for an individual varied from 15 to 130. The average was 47.

Composition of Elicited Knowledge

Over 98 percent of the relevant statements were classified into one of the eight categories of knowledge, clearly demonstrating that the typology is sufficiently comprehensive. As the data in Table 3 indicate, knowledge content was more likely to fall into some categories than others. In fact, over 60 percent of the statements were classified as Attribute or Attribute Evaluation knowledge. Almost 40 percent were classified as Specific Attribute Evaluation, indicating a possible need for subdivision within this category. On the other hand, less than one percent of the statements were classified as Terminology knowledge. In contrast to Russo and Johnson (1980). only a small proportion of elicited statements dealt with specific brands. This difference is probably due to the different probes used in the two studies to elicit product knowledge. As Rossiter (1980) noted, the Johnson and Russo probe emphasized brand knowledge.

TABLE 3

COMPOSITION OF KNOWLEDGE FOR RUNNING SHOES

It may be erroneous to conclude, however, that the frequencies in Table 3 represent the actual amount of knowledge that people have stored in memory in each of these categories. First. the general probe may have produced biases in the types of knowledge elicited. She accessibility of a piece of knowledge is related to how strongly it is associated with the probe (Collins and Loftus 1975). Consequently, a biased assessment of the composition of knowledge may result from the use of a probe that differs in its strength of association with the various types of knowledge content. For example, running shoe terminology may not be as closely associated with buying running shoes as evaluations of running shoe attributes are. Second, the number of statements elicited in a category may not represent the amount of knowledge in that category if information is "chunked" in memory, i.e., if a single statement summarizes a lot of detailed knowledge. This problem was somewhat reduced by the use of the double-layered probe technique and the coding instructions. Specifically, a statement could receive more than one code if it was judged to contain more than one piece of knowledge.

Relationships Between The Knowledge Categories

If the levels of knowledge (i.e., numbers of statements) in each category are highly correlated with each other, the proposed typology would be very limited in usefulness. First. high correlations would signify that a unidimensional measure would be sufficient to assess knowledge content. Second, high correlations would indicate that little benefit would be gained from isolating the effects of each type of knowledge on decision making behavior (since the knowledge variables would rarely occur in isolation). Of course, the knowledge categories are not conceptually independent, so a moderate level of correlation is to be expected.

Table 4 displays the correlations between the number of statements in each category. Of these, nineteen are below .3 and four are between .3 and .5. Only five correlations are higher than .5. the results indicate that these categories of knowledge content are, in general, fairly distinct. Since a few of the correlations are very large, however, it is useful to analyze the pattern of correlations in order to gain some insight into the relationships between the measures of the different categories .

Factor analysis was used to examine these relationships. Since there is no reason to expect the factors to be uncorrelated, an oblique rotation was used. Only those factors with eigenvalues greater than one were selected, resulting in a three factor solution. Direct oblimin was selected as the rotation procedure. A moderate degree of correlation between the factors was specified by the delta parameter. The sensitivity of the solution to this parameter was investigated by altering the delta parameter and refactoring the data. She resulting solutions led to the same conclusions as the solution presented here. The factor loadings are displayed in Table 5. (An orthogonal solution was also obtained, again with very similar results.)

TABLE 4

CORRELATIONS BETWEEN TYPES OF KNOWLEDGE FOR RUNNING SHOES

TABLE 5

FACTOR PATTERN

TABLE 6

CORRELATIONS BETWEEN FACTORS

Three factors are identified by this analysis. The first factor can be interpreted as a "product attributes" factor, including knowledge of the existence of product attributes and knowledge of how to evaluate specific attribute values. Knowledge of Attributes, General Attribute Evaluation, and Specific Attribute Evaluation load highly on this factor. Three knowledge types load highly on the second factor: Specific Attribute Evaluation, Terminology, and General Usage. These three types are useful to a serious runner who needs to be able to communicate with other runners and choose shoes that will perform well in the various running situations that the serious runner encounters. Specific Attribute Evaluation and General Usage correspond to some of the differences in knowledge between experts and novices found by Chi, Glaser, and Rees (1981). The third factor represents knowledge of a personal nature, the kind of knowledge that is likely to arise directly out of personal experience. Knowledge of Purchasing Procedures and Personal Usage load highly on this factor. Brand Pacts did not load highly on any of the factors.

The results raise some interesting questions concerning how knowledge develops and how the knowledge content of experts differs from that of novices. It appears that consumers learn about product attributes somewhat independently of personal experience with the product. Only serious users of the product say acquire much knowledge of terminology and of the relationship between product usage situations and the appropriate product characteristics. Future research is needed to further explore these issues.

SUMMARY AND CONCLUSIONS

She coding scheme for knowledge content was found to be sufficiently reliable and comprehensive. In general, the knowledge categories were distinct from each other although there were some notable exceptions. She factor analysis indicates that, for running shoes, there say be fewer than eight empirically separable dimensions of knowledge content. This analysis is limited in generalizability, however, because of the small sample, the limitations of the free-elicitation measurement method, and the characteristics of the specific context of running shoes.

This research indicates that knowledge content is multidimensional, which has two important implications for future research on consumer product knowledge. First, measures of consumer knowledge should include a variety of questions in order to capture the full range of the knowledge construct. For example, Brucks (1985) developed structured and open-ended questions to measure product knowledge based on a modified version of this typology. Second, the different types of knowledge content say have different effects on decision-making and purchasing behavior. For example, knowledge of "Brand Facts" may be used to substitute for external information search, (thus decreasing search) while knowledge of "Terminology" may facilitate external information search (thus increasing search). Futuro research is needed to address the effects different types of knowledge content on decision-making behavior.

REFERENCES

Alba, Joseph W. (1983), "The Effects of Product Knowledge on the Comprehension, Retention, and Evaluation of Product Information," in Richard P. Bagozzi and Alice M. Tybout, eds., Advances in consumer Research, Volume 5, Chicago: Association for Consumer Research, 577-580.

Anderson, John R. (1976), Language, Memory, and Thought, Hillsdale, NJ: Lawrence Erlbaus Associates.

Bennett, Peter D. and Robert S. Mandell (1969). "Prepurchase Information Seeking Behavior of New Car Purchasers - The Learning Hypothesis," Journal of Marketing Research, 6 (November), 430-433.

Bettman, James R. and C. Whan Park (1980), "Effects of Prior Knowledge and experience and Phase of the Choice Process on Consumer Decision Processes," Journal of Consumer Research. 7 (December), 234-248.

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

Bloom, B.S., M.D. Englehart, E.J. Furst, W.H. Hill, and D.R. Krathwohl (1956), Taxonomy of Educational Objectives: The Classification of Educational Goals. Handbook I: Cognitive Domain, New York: David McKay Company, Inc.

Brucks, Merrie (1984), "The Effects of Product Class Knowledge on Information Search Behavior," unpublished doctoral dissertation, Graduate School of Industrial Administration, Carnegie-Mellon University, Pittsburgh, PA 15213

Brucks, Merrie (1985), "The Effects of Product Class knowledge on Information Search Behavior," Journal of Consumer Research, 12 (June), 1-16.

Brucks, Merrie and Andrew A. Mitchell (1981), "Knowledge Structures, Production Systems and Decision Strategies", in Kent B. Monroe, ed., Advances in Consumer Research, Vol. VIII, Ann Arbor: Association for Consumer Research, 750-757.

Chase, William G. and Herbert A. Simon (1973a), "Perception in Chess," Cognitive Psychology, 4 (January), 55-81.

Chase, William G. and Herbert A. Simon (1973b), "The Mind's Eye in Chess," in William G. Chase, ed., Visual Information Processing, Hew York: Academic Press.

Chi, Michelene, T.H., Robert Glaser, and Ernest Rees (1981). "Expertise in Problem Solving," in Sternberg, ed., Advances in the Psychology of Human Intelligence, Hillsdale. W : Lawrence Erlbaus Associates.

Collins, Allan H. and Elizabeth F. Loftus (1975), "A Spreading Activation Theory of Semantic Processing," Psychological Review. 82 (6), 407-428.

de Groot, A. (1965), Thought and Choice in Chess, The Hague: Mouton.

de Groot, A. (1966), "Perception and Memory Versus Thought: Some Old Ideas and Recent Findings," in B. Kleinmuntz, ed., Problem Solving, New York: Wiley.

Edell, Julie A. and Andrew A. Mitchell (1978), "An Information Processing Approach to Cognitive Response," in S.C. Jain. ed., Research Frontiers in Marketing: Dialogues and Directions, Chicago: American Marketing Association. 178-183.

Fiske, Susan T. and Martha G. Cox (1979), "Person Concepts: The Effect of Target Familiarity and Descriptive Purpose on the Process of Describing Others," Journal of Personality, 47, 136-161.

Hastie, Reid, "Comment: Consumers' Memory for Product Knowledge," in Andrew A. Mitchell, ed., Advances in Consumer Research, Vol. II, Ann Arbor: Association for Consumer Research. 72-73.

Hutchinson, J. Wesley and Joseph W. Alba, (1985) "Framework for Understanding Consumer Knowledge: I. Structural Aspects and Information Acquisition," Working paper. University of Florida.

Johnson, Bric and J. Edward Russo (1984), "Product Familiarity and Learning New Information." Journal of Consumer Research, 11 (June), 542-550.

Kanwar, Rajesh, Jerry C. Olson, and Laura S. Sims, (1981), "Toward Conceptualizing and Measuring Cognitive Structures," in Kent B. Monroe, ed., Advances in Consumer Research, Vol. VIII, Ann Arbor: Association for Consumer Research, 122-127.

Olson, Jerry C. (1979), "The Importance of Cognitive Processes and Existing Knowledge Structures for Understanding Food Acceptance," Working Paper # 89, College of Business Administration, Pennsylvania State University.

Olson, Jerry C. and Aydin Muderrisoglu (1979), "The Stability of Responses Obtained by Free Elicitation: Implications for Measuring Attribute Salience and Memory Structure," in William L. Wilkie, ed., Advances in Consumer Research, Vol. VI, 269-283.

Punj, Girish X. and Richard Staelin (1983), "A Model of Consumer Information Search Behavior for New Automobiles," 9 (March), 366-380.

Park, C. Whan (1976), "She Effect of Individual and Situation-Related Factors on Consumer Selection of Judgmental Models," Journal of Marketing Research, 8 (hay), 144-151.

Rossiter, John R. (1980), "Representation of Information in Memory: Some General Issues," in Jerry C. Olson, ed., Advances in Consumer Research, Vol. VII. Ann Arbor: Association for Consumer Research. 429-430.

Russo, J. Edward and Johnson, Eric J. (1980), "What Do Consumers Know About Familiar Products?" in Jerry C. Olson, ed., Advances in Consumer Research, Vol. VII, Ann Arbor: Association for Consumer Research, 417-423.

Srull, Thomas K. (1983), "The Role of Prior Knowledge in the Acquisition, Retention, and Use of new Information," in Richard P. Bagozzi and Alice S. Tybout. eds., Advances in Consumer Research, Chicago: Association for Consumer Research, 572-576.

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

Tulving, Endel (1972), "Episodic and Semantic Memory" in Endel Tulving and Wayne Donaldson, eds., Organization of Memory, New York: Academic Press.

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