Knowledge and Context Effects on Typicality and Attitude Judgements

Cynthia D. Huffman, University of Minnesota
Barbara Loken, University of Minnesota. Arizona State University
James Ward,
ABSTRACT - This paper examines the effects of prior knowledge and context on typicality, frequency of instantiation, and global attitude judgments made in contexts that differ in exemplar typicality. We present research that shows significant effects of context on typicality judgments and significant interaction effects of context and knowledge on typicality and frequency of instantiation judgments. In general, the findings suggest that the information available to consumers in the judgment context has effects that differ depending on the level of prior knowledge of the consumer.
[ to cite ]:
Cynthia D. Huffman, Barbara Loken, and James Ward (1990) ,"Knowledge and Context Effects on Typicality and Attitude Judgements", in NA - Advances in Consumer Research Volume 17, eds. Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, Provo, UT : Association for Consumer Research, Pages: 257-265.

Advances in Consumer Research Volume 17, 1990      Pages 257-265

KNOWLEDGE AND CONTEXT EFFECTS ON TYPICALITY AND ATTITUDE JUDGEMENTS

Cynthia D. Huffman, University of Minnesota

Barbara Loken, University of Minnesota

James Ward. Arizona State University

ABSTRACT -

This paper examines the effects of prior knowledge and context on typicality, frequency of instantiation, and global attitude judgments made in contexts that differ in exemplar typicality. We present research that shows significant effects of context on typicality judgments and significant interaction effects of context and knowledge on typicality and frequency of instantiation judgments. In general, the findings suggest that the information available to consumers in the judgment context has effects that differ depending on the level of prior knowledge of the consumer.

Studies of judgments and attitudes in consumer behavior typically assume a constant environment; that is, the context in which the judgments are made is often neglected as a source of possible impacts on the decision process of the consumer. However, the importance of environmental effects on judgments and consumer choice has emerged in the literature as an important topic of study. Specifically, among the issues that have been investigated are the effects of context on choice processes and outcomes (Bettman 1988; Klein 1989; Payne, Bettman, and Johnson 1988), the effects of context on choice models of noncompensatory processes (Johnson and Meyer 1984), the effects of framing and reference points on consumer decision-making processes (Klein and Oglethorpe 1987, Monroe and Chapman 1987, Rowe and Puto 1987), and effects of comparative versus noncomparative advertising on judgments of similarity, distinctiveness, ad informativeness, and polarization of attitudes (Sujan and Dekleva 1987). These studies strongly suggest that a product may be judged differently depending on the products with which it is encountered and the situation in which it is judged, but they only begin to detail the possible implications of the effects of context on other judgments and attitudes.

One of the most well-known treatments of the effects of context arises from social judgment theory (i.e. Sherif and Hovland 1961). According to social judgment theory, the extremity of a judgment with respect to a target depends in part on the reference point available - the standard of comparison. By manipulating the standard of comparison, the judgments elicited on aspects of a stimulus may be changed. The implications of social judgment theory and context effects for product judgments are important for product positioning and, more particularly, for advertising issues. For example, companies frequently use prototypical products or brand leaders as anchors or standards in comparative advertising .

However, with the exception of Sujan and Dekleva (1987), there has been no attempt to combine categorization judgment measures, such as typicality, with context effects. In addition, the interactions of knowledge with context have not been investigated. In this paper, we report an investigation of the effects of context and knowledge on judgments of prototypicality within a product class.

THEORY AND PRIOR RESEARCH

Contrast Theory

Sherif, Maub and Hovland (1958) found that when a person was asked to judge the heaviness of stimuli that were distributed over some range of weight with equal frequency and there was no explicit anchor or comparison stimulus in relation to which the stimuli were to be judged, the judgments tended to distribute themselves over the judgment categories approximately rectangularly. When a stimulus more extreme than any of the others was designated as the anchor stimulus from which to base judgments, other stimuli were judged to be further from the anchor stimulus than they would otherwise have been (the contrast effect). Applied to social stimuli, it has been argued that the perceiver apparently exaggerates the discrepancy between the stereotype (on a particular attribute) and the observed (Manis, Nelson, and Shedler 1988).

At least three distinct conceptualizations exist of how contrast effects might derive from some sort of comparison process (Manis, et al 1988). First, an explanation based on adaptation-level theory (e.g. Parducci, Calfee and Marshall 1960) would argue that contrast effects derive from comparison with the central tendency of the individual's internal representation of the category under evaluation. A second possibility, based on range theories, is that judgments are determined by the location of the stimulus to be evaluated within the range that is established by the most extreme of the relevant contextual stimuli. Third, a "rank based" hypothesis argues that assessment processes are based (in part) on the rank of a target stimulus relative to other relevant stimuli (i.e. How many are more extreme?). Regardless of the particular explanation, it appears that a comparison process takes place when consumers are asked to make judgments about products. Thus, a judgment is made in relation to other contextual stimuli rather than as an absolute value (see Smith, Diener and Wedell 1989 for further elaboration of range and other comparison processes).

In addition to providing a conceptual basis for hypothesizing effects on brand judgments, the context literature suggests that attributes may be judged differently in different contexts. Tversky (1977), for example, has shown that features may be weighted differently depending on the context.

Researchers have used a variety of procedures to manipulate the context in which objects are judged. Herr, Sherman, and Fazio (1983), for example, used a priming paradigm in which the category made accessible by the prime was used as the basis for contrast. In other cases, the stimuli provided have been the context that forms the basis for range and/or rank effects (e.g. Smith et al 1989). However the context is achieved, it is in effect a standard of comparison against which to assess the target(s).

Typicality Effects

People perceive exemplars of a category to vary in the degree to which they are representative, or typical, of a category (e.g. a robin is a more typical bird than a vulture). As discussed above in relation to social judgment theory, the context in which judgments are made may have effects on the extremity of those judgments. Specifically, different collections of exemplars may differentially influence judgments on a target stimulus. For example, a moderately typical exemplar embedded in a context of atypical exemplars may be judged to be typical, while the same moderately typical exemplar may be judged to be less typical when embedded in a collection of typical exemplars. The different collections of exemplars that comprise the context for the judgments on the target exemplar serve as standards of comparison in the same way that standards of comparison were used in the contrast studies discussed above (i.e. Smith et al 1989).

It seems likely also that a "hierarchy of contrast effects" may exist, where the contrast effects are strongest for typicality, next strongest for "surrogate" measures of typicality such as frequency of instantiation (Barsalou 1985), and least strong for attitude toward the target exemplar. The more closely the independent variable of context typicality is conceptually related to the dependent measure, the greater the effect predicted. These predictions also arise from measurement studies in categorization (i.e. Mervis and Rosch 1981, Barsalou 1985), where the category representation is argued to be most nearly approximated by typicality measures, next most closely by frequency of instantiation, etc.

The above arguments lead to the following predictions:

H1: A moderately typical target exemplar will be judged to be more typical in (a) a context with all atypical exemplars than in (b) a context with all typical exemplars. In (c) a context with a range of typicality of exemplars, the typicality judgments on the target exemplar will be somewhere in between the judgments from the other two contexts (context a and context b).

H2: A "hierarchy of contrast effects" will exist such that the above context effects on the target category member will be strongest for global typicality measures, less strong for frequency of instantiation, and least strong for attitudes toward the target category member.

Knowledge

Expertise or prior knowledge is associated with richer, more complete, and more detailed representations of a category in memory (Alba and Hutchinson 1987, Brucks 1986, Murphy and Wright 1984). Experts in a product category thus should have readily accessible knowledge about a wide range of the products in that category. Further, as Chi, Glaser and Farr (1988) argue, experts are less likely than novices to rely on less relevant surface features or literal objects stated in the problem description in making judgments. Both experts and novices use conceptual categories, but these categories are richer and more semantically-based for experts and more syntactically or surface-based for novices (Chi et al 1988; Walker, Celsi and Olson 1987). - .

More knowledgeable consumers are likely to know more about the various product category members and are more likely to have insight into the characteristics and criteria that differentiate one product from another. Novices should have less developed cognitive structures about specific members of a category and the characteristics that determine category structure. Thus, subjects with extensive accessible prior knowledge should be more likely than subjects with low prior knowledge to use that prior knowledge as their standard of comparison, in accordance with adaptation theory. In contrast, subjects with low prior knowledge should be more likely to use an external context for the standard of comparison, such that range theories would be appropriate in determining context effects. Since experts' perceptions of category representativeness may then be less likely to be influenced by external context than novices' perceptions, we expect to see the smallest effects of (external) context for subjects with high prior knowledge and larger effects of (external) context for subjects with low prior knowledge. That is:

H3: Subjects with low prior knowledge about the product category will show larger context effects than will subjects with high prior knowledge.

METHODOLOGY

Overview

Data collected from a laboratory study were used to test the above hypotheses. After reading through a list of 15 makes of automobiles, a sample of 81 college students completed measures of typicality, frequency of instantiation, attitude, and knowledge. The list of automobiles, which were also the ones rated on individual measures, defined the context manipulation. Each subject was randomly assigned to one of the three context conditions.

Context Stimuli Selection

Three contexts were used in testing the hypotheses in this study. Context 1 had fourteen automobiles that represented a range of typicality, based on typicality scores from a pretest (of 35 automobiles), relevant to the category of "economy cars". Context 2 had fourteen automobiles that were all previously rated as most typical of the category of economy cars. Context 3 had fourteen automobiles that were all rated as most atypical of the category of economy cars. In addition, each context included a target economy car (referred to here as Car "X"), which had been rated as typical of economy cars. (On scales ranging from 1 to 7, with greater numbers reflecting greater amounts of the dimension, the target car was rated as 5.86 for typicality and 4.67 for representativeness in a pretest; the target car was rated as one of the 4 most typical economy cars, out of a possible 35 cars).

For each context, half of the questionnaires had a randomly determined order and half had the reverse order. In each order, for each context, the target exemplar was the fifteenth automobile listed. This ensured that the context would be fully encoded by the time the target exemplar was rated and that comparisons of the target across contexts would not be confounded by order effects.

Measures

Measures of global typicality, frequency of instantiation, attitude and knowledge were taken, in that order. All subjects rated all fifteen automobiles in one context on all measures. However, only data pertaining to the target exemplar were used in subsequent analyses. The rationale for measuring all 15 automobiles was (1) to disguise the purpose of the study and (2) to strengthen the context manipulation.

Global typicality was measured on two single-item 7-point scales. The first was a "typicality" scale ranging from "very atypical" (1) to "very typical" (7), with instructions adapted from Hampton and Gardiner (1983). The second typicality scale was a 1 to 7 scale of "representativeness", ranging from "very unrepresentative" (1) to "very representative" (7). The correlation between the typicality and representativeness scales was 0.80 for the target exemplar, so the two scales were combined to form a summary scale. However, since initial pretest data suggested that the target exemplar yielded higher scores on the typicality scale than on the representativeness scale, these two measures were also treated independently in certain analyses. As subsequent analyses indicate, the representativeness scale yielded somewhat different findings than the typicality scale.

Attitude toward the target exemplar was measured on two 1 to 7 evaluative semantic differential scales (extremely bad (1) - extremely good (7) and extremely unfavorable (1) - extremely favorable (7)). The two scales correlated 0.90 and were summed to form the measure of attitude.

Frequency of instantiation was measured by asking subjects to rate how frequently they encountered each category member (automobile) - in store, ads,-etc. - as an instance of the category "economy cars". The frequency of instantiation (FOI) measure was a 7-point scale with endpoints "not at all frequently" (1) and "very frequently" (7). Instructions were based on Barsalou (1985).

Finally, knowledge about the category was measured in two ways. First, subjective knowledge was measured by a single-item 5-point scale, asking subjects to rate how knowledgeable they were about automobiles relative to other people, on a scale from "very familiar" to "very unfamiliar". A second 30-item objective knowledge scale was developed, some items of which were open-ended and some multiple choice. The open-ended questions encompassed terminology, personal experience, and industry knowledge. For example, one question was 'What does V-g engine mean?" Another asked subjects to give an example of each of eight types of automobiles (full size pickup truck, sedan, compact, etc.). The multiple choice questions tapped specific model knowledge, and included questions such as 'The has a 1.8 liter fuel injected engine" and "The Pulsar is made by _ ." The objective knowledge test was pretested and refined to enhance reliability (24 items remained). Cronbach alpha for the open-ended objective questions was 0.73; for the multiple-choice questions 0.71. The two scores (open and closed ended items) were summed for a total objective knowledge score. On the refined scale, the maximum possible score for total knowledge was 48. Subjective and objective knowledge correlated moderately high (0.74) but were analyzed separately in subsequent analyses.

RESULTS

Overview

The main hypothesis investigated was the effect of context on typicality judgments for the target exemplar (Car "X"). In addition, other measures taken were analyzed for differences across contexts to examine the hierarchy of effects proposition (hypothesis 2). Finally, the effect of prior knowledge on the other results was examined. The results of these analyses are reported below. In the analyses, all data from 7 point scales were recoded from -3 (atypical, unfavorable, not at all frequently, etc.) to +3 (typical, favorable, frequent, etc.). Data from the 5-point scales were coded from 0 (very unfamiliar) to 4 (very familiar).

To test the hypotheses, two-way analyses of variance were performed with each typicality measure separately as the dependent variable and the three levels of context as one independent variable. The second independent variable was knowledge (high or low, split at the median).

TABLE 1

MEANS FOR DEPENDENT MEASURE FOR EACH CONTEXT CONDITION

Effects of Context on Typicality Judgments

Results are shown in Table 1. The results for typicality show that, as predicted, judgments differ significantly (p < 0.05) by context. That is, in the all atypical exemplar context, the target exemplar was judged to be more typical (M = 1.54) than it was judged to be when it was in the context of all typical exemplars (M = 0.50). The mean of the typicality judgments for the target exemplar in the "range" context condition was somewhere in between the two context extremes (M= 1.22), as expected. The differences in means for the representativeness measure unexpectedly did not reach significance (p = 0.63), although the means are in the predicted direction.

Hierarchy of Contrast Effects

A hierarchy of effects of context was expected such that typicality and representativeness would show the largest effects of context, with FOI (frequency of instantiation) and attitude showing progressively smaller effects. Effects of context were not present for either FOI or attitude. However, as shown in Table 1, means were in the same direction as noted for typicality, and the F values were closer to significance for FOI than for attitude.

Effects of Knowledge on Typicality

The effects of both objective and subjective knowledge on typicality were also investigated as a function of context. As stated in hypothesis 3, an interaction between context and knowledge was expected such that subjects with low prior knowledge of automobiles would show larger effects of context than subjects with high prior knowledge. Results for typicality are shown in Table 2 and Figure A. The expected interaction effect is, indeed, significant for both objective (F = 6.273, p < 0.005) and subjective (: = 3.252, p < 0.05) knowledge, but the means are exactly opposite the predictions. As shown in Figure A, subjects with high knowledge were more likely than subjects with low knowledge to show the predicted contrast effects (i.e. greater perceived typicality of the target exemplar as the frequency of atypical exemplars increased in the context condition). Similar interaction effects, but less pronounced, were found for frequency of instantiation for objective (F = 7.516, p < 0.005) and subjective (F = 1.667, p < 0.20) knowledge, and effects were less evident for attitudes toward the target (F=2.347, p= 0.10 and F= 0.227, p= 0.63 for objective and subjective knowledge, respectively), as shown in Table 2 and Figures B and C.

In summary, in contrast to hypothesis 3, the effects of context were greater for high than for low knowledge subjects. A possible explanation for this effect is that only subjects with higher knowledge were able to appropriately categorize the context with respect to economy cars and use those categorizations to make typicality judgments. Further, the data seem to support the idea that the less knowledgeable subjects did not know about the characteristics of the stimuli since in the presence of the all atypical context they seemed to assume the target (typical) stimulus was atypical too, and thus their typicality rating for the target exemplar trended down. In other words, they seemed to know enough to realize that most of the stimuli in this context were atypical, but not enough to differentiate the target from this general judgment.

DISCUSSION

Summary Of Findings

Several interesting results occurred, not all of which were expected. First, as predicted, we found that subjects' judgments of typicality were affected by context. In particular, when the target exemplar was presented within a list of very atypical exemplars, it was rated as more typical than when presented within a list of very typical exemplars. This finding is consistent with social judgment theory (Sherif and Hovland 1961) as well as more recent theories of contrast (Manis et al 1988) and social comparison (Smith et al 1989). Not surprisingly, effects of context on frequency of instantiation (a "surrogate" measure of typicality) and attitude were less pronounced. That is, although means were in the same direction as judgments of typicality, subjects were not significantly more likely to judge the target as a familiar instance or as well-liked when the context included all atypical economy cars, than when the context included all typical economy cars.

TABLE 2

MEANS AND F TESTS FOR THE TARGET EXEMPLAR ON THE DEPENDENT VARIABLES FOR EACH KNOWLEDGE AND CONTEXT CONDITION

FIGURE A

MEAN TYPICALITY RATINGS FOR EACH CONTEXT AND KNOWLEDGE CONDITION

FIGURE B

MEAN FOI RATINGS FOR EACH CONTEXT AND KNOWLEDGE CONDITION

FIGURE C

MEAN ATTITUDE RATINGS FOR EACH CONTEXT AND KNOWLEDGE CONDITION

Finally, knowledge was found to interact with context such that the above effects of context on typicality occurred for high knowledge but not for low knowledge subjects. This finding was exactly counter to predictions. However, it does suggest that higher and lower knowledge subjects have differing abilities to integrate their internal category representations with the context presented. We had predicted, in essence, that low knowledge subjects' judgments would be affected by context due to their lack of internal standard of comparison in this product category and that high knowledge subjects' judgments would not be affected because they would use their internal knowledge as a standard of comparison instead of the external context. An explanation for the results may lie in the amount of information provided and the corresponding usefulness of the information to subjects with differing levels of knowledge.

In the questionnaires, no attribute value information was given for the automobile stimuli provided to subjects. Instead, automobiles were identified only by maker and model name (e.g. Chevy Nova, Ford Tempo). Providing only a name conveys, through associations and inferences, a lot of attribute value information to high knowledge subjects, but it provides little information to subjects who are not able to make the appropriate connections of attribute values to car name, such as our low knowledge subjects. Thus the results obtained may be explained through an information provision proposition which has yet to be tested: for context effects to occur for low knowledge subjects, attribute value information must be available in the context. For high knowledge subjects, because they are able to use associations and inferences from brand names, context effects will occur even in the absence of attribute value information .

The differences in context effects between typicality and attitude judgments may be due to differences in the use of context stimuli. For attitude judgments, experts may be less likely to rely on the attributes of the other category members in making an evaluation than for typicality judgments. If in fact they had already stored an evaluation of Car "X" or "X"-type cars (recall that Car "X" was the target exemplar), their evaluation would probably be less influenced by irrelevant contextual factors than the evaluations of the novices. This is one explanation of why the experts' attitude judgments do not seem to be affected by context. Note here that the arguments about attitude and typicality do not contradict each other if it is assumed that the typicality instructions oriented subjects more than the attitude instructions to use the category context as a reference. Finally, why were the non-experts' attitude judgments seemingly influenced by context while their typicality judgments were not? One explanation is that perhaps the non-experts did not have enough knowledge about the stimuli to form the context to influence the typicality judgment, which might require comparison on attributes, but did have enough overall evaluative knowledge (perhaps just overall stored affect about certain brands) that their attitude judgments were influenced by context.

The results found are interesting in that internal contexts (hypothesized to eliminate the contrast effects for high knowledge subjects) may not in some cases be responsible for contrast effects on judgments where an external context is available. High knowledge subjects' typicality judgments were affected by the context in which they were made. Research is needed on this question in order to clearly determine the conditions, if any, under which high knowledge subjects use their internal knowledge base as the context, or standard of comparison, against which they make their judgments. Finally, the unexpected results obtained indicate that the conclusions to be drawn are tentative. Further research is needed to test the explanations given and the conclusions drawn before any attempt to generalize is made.

CONCLUSIONS

These results, if supported by future research, have both managerial and theoretical implications. First, the results suggest that presentation of products to audiences may need to be adapted depending on the knowledge level of the targeted consumers and depending on the anticipated environment, or context, of presentation. For example, in a simple promotion presentation with minimal information given on context brands, context effects may occur only for individuals with high knowledge in the product category. Where attribute value information is available on context brands, context effects could occur for low knowledge individuals as well.

Context may be one avenue to help create a product image. On the surface, at least, it appears that if an "atypical" image is desired, the optimal context for presentation may be one with all typical brands. Of course, this assumes that the consumer will accept as credible the mix of brands presented by the advertiser. The results also suggest that much more subtle effects of context may appear with higher knowledge subjects due to their greater ability to sub-categorize context brands, with the sub-categorization subsequently affecting judgments.

Differences found between the effects of subjective and objective knowledge, or perceived and actual knowledge, although small, are interesting and may be integrated into other work in this area, such as Park, Gardner and Thukral (1989). However, as subjective knowledge was measured with a single scale only, the conclusions to be drawn here are necessarily limited.

This research is one step in determining how prior knowledge and experience fit into the judgment process. It is suggested here that high knowledge subjects may use their internal knowledge with appropriate cues, but they may not always use their category representations as the integrating schema for presented information in some conditions. One avenue of research suggested by this study may be the conditions under which prior knowledge is used in making judgments, and the detailing of what knowledge is actually used (general category knowledge versus brand-specific information).

REFERENCES

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

Barsalou, Lawrence (1985), "Ideals, Central Tendency, and Frequency of Instantiation a Determinants of Graded Structure in Categories," Journal of Experimental Psychology: Learning, Memory, and Cognition, 11, 629-654.

Bettman, James R. (1988), "Presidential Address: Processes of Adaptivity in Decision Making," in Advances in Consumer Research, vol. 15, ed. Michael J. Houston, Provo, Utah: Association for Consumer Research, 14.

Brucks, Merrie (1986), "A Typology of Consumer Knowledge Content," in Advances in Consumer Research, vol 139 el 1. Richard J. Lutz, Provo, UT: Association for Consumer Research.

Chi, Michelene T. H., Robert Glaser, and M. J. Farr (1988), The Nature of Expertise, Hillsdale, NJ: L. Erlbaum Associates.

Hampton, John and Margaret Gardiner (1983), "Measures of Internal Category Structure: A Correlational Analysis of Normative Data," British Journal of Psychology, 74, 491-516.

Herr, P. M., S. J. Sherman, and R. H. Fazio (1983), "On the Consequences of Priming," Journal of Experimental Social Psychology, 19, 323-340.

Johnson, Eric J. and Robert J. Meyer (1984), "Compensatory Choice Models of Noncompensatory Processes: The Effect on Varying Context," Journal of Consumer Research, 11 (June), 528-541.

Klein, Noreen M. and Manjit S. Yadav (1989), "Context Effects on Effort and Accuracy in Choice: An Enquiry into Adaptive Decision Making," Journal of Consumer Research, 15 (March), 411-421.

Klein, Noreen M. and Janet E. Oglethorpe (1987), "Cognitive Reference Points in Consumer Decision Making, in Advances in Consumer Research, vol. 14, eds. M. Wallendorf and P. Anderson, Provo, UT: Association for Consumer Research, 1 83-96.

Loken, Barbara and James Ward (1987), "Measures of the Attribute Structure Underlying Product Typicality," in Advances in Consumer Research, vol. 14, eds. M. Wallendorf and P. Anderson, Provo, UT: Association for Consumer Research, 22-26.

Manis, Melvin, Thomas E. Nelson, and Jonathan Shedler (1988), "Stereotypes and Social Judgment: Extremity, Assimilation, and Contrast," Journal of Personality and Social Psychology, 55 (1), 28-36.

Mervis, Carolyn B. and Eleanor Rosch (1981), "Categorization of Natural Objects," Annual Review of Psychology, 32, 89-115.

Monroe, Kent B. and Joseph D. Chapman (1987), "Framing Effects on Buyers' Subjective Product Evaluations," Advances in Consumer Research, vol. 14, eds. M. Wallendorf and P. Anderson, Provo, UT: Association for Consumer Research, 193-197.

Murphy, Gregory L. and Jack C. Wright (1984), "Changes in Conceptual Structure with Expertise: Differences Between Real-World Experts and Novices," Journal of Experimental Psychology: Learning, Memory, and Cognition, 10 (Jan), 144-155.

Parducci, A, R. C. Calfee, and L. M. Marshall (1960), "Context Effects in Judgment of the Mean, Mid-point, and Median of the Stimuli," Journal of Experimental Psychology, 60, 65-71.

Park, C. Whan, Meryl P. Gardner, and Vinod K. Thukral (1989), "The Role of Actual and Perceived Knowledge on Consumer InformationProcessing for Brand Choice Task," under review at Journal of Consumer Research.

Payne, John W., James R. Bettman, and Eric J. Johnson (1988), "Adaptive Strategy Selection in Decision Making," Journal of Experimental Psychology: Learning, Memory, and Cognition, 14 (July), 534-552.

Rowe, Debra and Christopher P. Puto (1987), "Do Consumers' Reference Points Affect Their Buying Decisions?" Advances in Consumer Research, vol. 14, eds. M. Wallendorf and P. Anderson, Provo, UT: Association for Consumer Research, 188-92.

Sherif, M. and C. I. Hovland (1961), Social Judgment: Assimilation and Contrast Effects in Communication and Attitude Change. New Haven-- Yale University Press.

Sherif, M., D. Taub, and C. I. Hovland (1958), "Assimilation and Contrast Effects of Anchoring Stimuli on Judgments," Journal of Experimental Psychology, 55, 150-55.

Smith, Richard H., Ed. Diener, and Douglas H. Wedell (1987), "Intrapersonal and Social Comparison Determinants of Happiness: A Range-Frequency Analysis," Journal of Personality and Social Psychology, 56 (3), 317-325.

Sujan, Mita and Christine Dekleva (1987), "Product Categorization and Inference Making: Some Insights for Comparative Advertising," Working Paper Number 162, College of Business Administration, The Pennsylvania State University.

Tversky, Amos (1977), "Features of Similarity In Transitivity of Preferences," Psychological Review, 84 (July), 327-352.

Walker, Beth, Richard Celsi and Jerry Olson (1987), "Exploring the Structural Characteristics of Consumers' Knowledge," in Advances in Consumer Research, vol. 14, eds. M. Wallendorf and P. Anderson, Provo, UT: Association for Consumer Research, 17-21.

Ward, James and Barbara Loken (1988), 'The Generality of Typicality Effects on Preference and Comparison: An Exploratory Test," in Advances in Consumer Research, vol. 15, ed. Michael J. Houston, Provo, UT: Association for Consumer Research, 55-61.

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