Comments on Consumer Knowledge and Attribute Structures

Citation:

Thomas J. Page, Jr. (1987) ,"Comments on Consumer Knowledge and Attribute Structures", in NA - Advances in Consumer Research Volume 14, eds. Melanie Wallendorf and Paul Anderson, Provo, UT : Association for Consumer Research, Pages: 27-28.

Advances in Consumer Research Volume 14, 1987      Pages 27-28

COMMENTS ON CONSUMER KNOWLEDGE AND ATTRIBUTE STRUCTURES

Thomas J. Page, Jr., Michigan State University

INTRODUCTION

The three papers to be discussed below are all quite well done, and the authors are to be commended for the quality of work presented here. Thus, there are no major criticisms of the papers that need to be addressed. Instead, the major thrust of these comments concerns questions that should be answered in future research in each particular area. Also, the three papers have enough in common that areas of mutual benefit will be discussed.

"A SIMPLE REPRESENTATION OF THE CONTINGENT STRUCTURE OF KNOWLEDGE"

Deighton's paper presents a very interesting method for measuring peoples' knowledge structures. He demonstrates that for known groups, the technique produces plausible and meaningful knowledge structures. One minor point of concern with the paper, as initially presented, is that some readers may find it difficult to understand exactly how one would start with raw data and end up with a graphical representation of a group's knowledge structure. In other words, in order for others to make use of this technique, the method needs to be explained in more detail. However, in fairness to the author, the major point of the paper was to make others aware of the technique and demonstrate its usefulness, and given the space constraints for the paper, such information could easily be included in a longer version of the paper.

Two areas of concern need to be addressed in future research using this technique. First, in Deighton's example, the sample size was five. The obvious question is, what happens when the sample size becomes larger. Will the technique still be able to produce distinguishable and meaningful knowledge structures, or will they become indistinguishable and uninterpretable? If, for example, different groups are relatively homogeneous and distinct from each other, then the usefulness of the technique would not be likely to be affected by sample size. If, however, the groups lack homogeneity and they are not very distinct from each other, a large sample size may cause the representations of knowledge to become blurred or indistinguishable. This needs to be investigated in future research.

A second point, which is loosely related to the first, concerns the sensitivity of the technique in terms of being able to detect different knowledge structures of different groups. In Deighton's example, the two groups were general health foot consumers and a more radical group that espoused the Pritikin ideology. What needs to be determined is just how different these two groups are in terms of their belief patterns using some other method so that the sensitivity of Deighton's technique can be assessed. If it turns out that, through independent assessment, the two groups are not vastly different when compared to a general (non-health) food consumer, for example, then the technique may be fairly sensitive. On the other hand, should the two groups turn out to be as different from each other as they are from the general food consumer, then the sensitivity of the technique remains in question. This is especially important if the technique were to be used in assessing the knowledge structures of different segments, for example. If the technique lacked sufficient sensitivity to detect such differences, one would be lead to making a type II error and conclude that no significant differences in beliefs exist when in fact they do.

"THE STRUCTURAL CHARACTERISTICS OF CONSUMERS' KNOWLEDGE"

This paper teals with the knowledge structures of experts versus novices. In particular, knowledge structures are investigated in terms of their dimensionality, articulation, and abstraction. The authors' results showed that experts used significantly more dimensions and produced significantly more concrete attributes on the abstraction measure than did novices. All other results were in the expected direction, but were not significant.

While these results are interesting, and make a useful contribution to our knowledge in this area, the major question remains: "What do these results mean to consumer behavior?" Researchers have known for quite some time that experts and novices have different ways of thinking about a particular domain, but the implications of this have rarely been pursued. They were not pursued in the present paper more than likely due to space limitations.

Nevertheless, interesting questions concerning the implications of such research can be investigated in future research. For example, can these results be used to develop different communications mixes for experts and novices. The authors' results would imply that experts should be given more dimensions and more concrete attributes to evaluate products than should be given to novices. of perhaps their results could be applied to developing different communication mixes for different stages of the product life cycle. When a product is first introduced, everybody is a novice, so provide them with relatively few dimensions for evaluation. As the product matures, more people become familiar with it, and more dimensions could be provided for evaluation. In short, it seems that in this area, the discipline is now at a point where future research should concentrate on investigating the implications of such findings if further progress is to be mate.

"MEASURES OF THE ATTRIBUTE STRUCTURE UNDERLYING PRODUCT TYPICALITY"

This paper explores the determinants of typicality in a product category. The major contribution of the paper is the development of an alternative, and apparently superior, method of investigating typicality. The authors' results showed that their measure was more highly related to global measures of typicality than other existing measures, and was more highly related to attitude than other measures. The authors tit an especially good job of developing their measure based on multi-attribute attitude research, and using this research to develop their hypotheses.

One major point of concern needs to be investigated in future research. The authors mention that the concept of instantiation has been shown to be a determinant of typicality. Instantiation refers to the number of times an item is encountered as a member of a particular category. This concept was not controlled for in their research, but it appears that it could have a considerable impact on their results. Instantiation of a particular brand is likely to be highly influenced by advertising, so if familiarity were controlled for, how might their results have been affected.

Specifically, they found a correlation of .88 between typicality and attitude toward the brand. How much of that relationship is due to the possibility that the most heavily advertised brands become the most familiar (ant typical) and most well liked. Similarity, instantiation may also affect the .88 correlation between their attribute-structure measure and typicality. Subjects may not have known enough about all fourteen brands of shampoo to give reliable information about them, so the most heavily advertised brands become the ones that people know the most about and regard as typical.

OVERLAP OF THE DIFFERENT RESEARCH STREAMS

As stated at the outset of these comments, the papers have enough in common to allow them to build on each other in future research. Some of the broad areas that could be investigated in future research include, but are by no means limited to, the following. What would be fount if Deighton's technique for measuring knowledge structures was applied to the expert versus novice questions investigated by Walker et al? Would experts have more complex structures because they use more dimensions? Are their any differences in the way experts and novices judge typicality? How would the fact that experts use more dimensions affect what they regard as typical? How are Deighton's technique of measuring knowledge structures and Loken and Wart's attribute structure measure related? Do different knowledge structures affect typicality ratings, or does what people regard as typical affect their knowledge structure? How toes instantiation, or frequency of exposure affect knowledge structures? Many other such questions could serve as useful starting points for future research in this area.

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