Retrieval Processes in Consumer Evaluative Judgment Making: the Role of Elaborative Processing, Context, and Retrieval Goals

ABSTRACT - A stage model specifying the processes occurring over time between the acquisition of information and its use in judgment/choice consumer situations is discusses and three factors believed to have a significant impact on memory retrieval are singled out. An experiment where the factors are manipulated is presented. The results support the proposition that elaborative processing enhances recall and the likelihood of information utilization.

Citation:

Alain d'Astous and Marc Dubuc (1986) ,"Retrieval Processes in Consumer Evaluative Judgment Making: the Role of Elaborative Processing, Context, and Retrieval Goals", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 132-137.

Advances in Consumer Research Volume 13, 1986      Pages 132-137

RETRIEVAL PROCESSES IN CONSUMER EVALUATIVE JUDGMENT MAKING: THE ROLE OF ELABORATIVE PROCESSING, CONTEXT, AND RETRIEVAL GOALS

Alain d'Astous, Universite de Sherbrooke

Marc Dubuc, National Bank of Canada

[The authors would like to thank Dipankar Chakravarti for helpful comments on an earlier draft of this paper. This research was supported by a grant from Marketix Inc.]

ABSTRACT -

A stage model specifying the processes occurring over time between the acquisition of information and its use in judgment/choice consumer situations is discusses and three factors believed to have a significant impact on memory retrieval are singled out. An experiment where the factors are manipulated is presented. The results support the proposition that elaborative processing enhances recall and the likelihood of information utilization.

INTRODUCTION

The issue of how consumers arrive at evaluative judgments and choices has always been central to consumer research. Until recently however the role played by memory processes in these types of situations has been somewhat neglected. This is unfortunate since simple introspection suggests that judgment and choice processes always rely on both information retrieved from memory and information externally available. For instance, consider the case where information about a brand's position on several relevant dimensions or attributes is provided. Even in such a stimulus-based context, consumers may have to search their memory to find information about the relative weights that should be applied to the various attributes in order to generate an evaluative judgment or to make a choice.

As Lynch and Srull (1982) and others have pointed out, a consideration of retrieval processes is fundamental to consumer research. There is a vast literature in cognitive psychology that has been concerned with this problem. Over the years, a number of general principles of information retrieval, based on various experimental findings, have emerged (see e.g. Glass, Holyoak and Santa 1979). In general however, cognitive psychologists have limited their focus on retrieval per se, without worrying too much about information utilization. While it may be of interest to inquire about the process of accessing information in memory, it must be recognized that in most natural situations information is retrieved in order to be used. Thus, consumers generally do not search their memory for facts simply to comply to someone's request; most of the time they do so in order to use the retrieved information in typical tasks such as making choices, evaluative judgments, inferences, communications, etc. Therefore, it seems important to look at the problem of consumer information retrieval within the perspective of the utilization of recalled knowledge for achieving specific objectives.

This paper discusses aspects of information retrieval in consumer judgment/choice situations and presents an experiment designed to provide answers to some research questions that seem of relevance to this area.

BACKGROUND DISCUSSION

Stage Analysis

An understanding of the factors that may impact on one's ability to access a given piece of information in memory and use it for some purposes is contingent upon an understanding of the processes that occur prior to the retrieval attempt. For instance, failure to access the information may simply be due to its unavailability. in which case an examination of the conditions prevailing at the encoding stage is warranted. It may also be due to the utilization of an inappropriate search path, in which case the compatibility between the retrieval specifications and the contextual elements surrounding the needed record must be investigated. Therefore, it seems useful to consider the problem of information retrieval and use in the context of a framework that specifies which processes occur over time between the acquisition of some target information and its use. Although simplistic in nature, this stage analysis (Crowder 1976, see also Hastie and Carlston 1980, Tulving 1976) provides some structure upon which various research questions can be formulated.

A stage model is presented in Figure 1. Briefly, the process begins with the individual's initial goals. These will eventually influence the way the target information is encoded. Once encoded, that information is stored and retained for a given period of time until it is retrieved and used for the purpose of making a judgment or a choice. Some issues related to this motel are now discussed in more detail.

FIGURE 1

A STAGE MODEL FOR EXAMINING ISSUES IN INFORMATION RETRIEVAL FOR CONSUMER JUDGMENT AND CHOICE

Goals

There is research evidence that one's objectives (e.g. to make an evaluative judgment, to select an alternative, to learn some information) will influence the strategies used to process incoming information and consequently the nature of the information subsequently available in memory (see e.g. Biehal and Chakravarti 1982, 1983, Cohen and Ebbesen 1979, Johnson and Russo 1981). Individuals differing initially with respect to their goals may exhibit quite different remembrances after having been exposed to the same set of stimuli. Ultimately however, all other things being equal, any disparity in recollection follows from the performance of qualitatively and/or quantitatively different mental operations at the encoding stage (Glass, Holyoak and Santa 1979); differences in goals are likely to have an effect on resulting memory structures to the extent that processing strategies are affected accordingly. For example, Biehal and Chakravarti (1982) 'nave shown that directed learning of product information versus learning incidental to a choice task lead to different retrieval processes. In the first case, retrieval operations (as inferred from free-recall protocols) were predominantly brand-based whereas in the other case higher levels of attribute-based processing were observed. According to the authors, the learning goals had a different effect on memory structure because they led to different processing activities during encoding.

Encoding

At some point, information presented to the individual is encoded, i.e. put in a form that is comprehensible to the human information processing system (Klatzky 1980). Actually, the encoding stage, as it is conceptualized here, involves more than a transformation process but refers to the total interpretation that is taking place (see also Wyer and Srull 1980). Several implications for retrieval follow from an examination of what can happen during this stage.

Encoding. A trivial observation but one that has rather serious implications is that the to-be-retrieved information might not get encoded. The consequences for retrieval need not be elaborated in great detail. From a methodological standpoint however, this suggests the necessity of making sure that the individual allocates some time to information processing.

Prior knowledge. A well-documented result in memory research is that the outcomes of the encoding process are partially determined by what one already knows about the stimulus information (e.g. Bransford 1979, chapter 5). In addition to providing some initial structure that will allow the system to prime those stored concepts that may facilitate comprehension (especially when dealing with ambiguous material), prior knowledge triggers inference making processes. Of the many inferences that are made, one of particular concern here has to do with the believability of a target information. The "S said X is O versus X is O" belief distinction (Fishbein and Ajzen 1975) is crucial as far as retrieving information for making judgments or choices is concerned. Two strategies for enhancing the believability of communicated information appears possible. One is to minimize the effect of prior knowledge by selecting a somewhat unfamiliar experimental topic. The other is to attribute the information to a credible source. Both strategies are implemented in the present study.

Mental operations. The type of mental operations performed at encoding is likely to make a difference at retrieval time. If the allocation of processing capacity to the target information is low, the resulting memory traces are less likely to become associated with potential retrieval cues. Mental activities such as rehearsal and elaborative processing; (e.g. making relations with prior knowledge, associating the information with other facts) have been repeatedly shown to substantially improve one's memory. The usual explanation is that it increases the number of retrieval paths and consequently the likelihood of accessing the elaborated upon information (Anderson 1980).

Retrieval Specification

The search for the target information may be initiated and pursued with the help of cues coming from the environment, the retrieval goals (e.g. instructions, task objectives) or memory (Norman and Bobrow 1979). Each of these information suppliers is briefly discussed below.

Environmental cues. In many instances the environment may offer indications that can be helpful for locating information in memory. Consider the example of a consumer in; store trying to remember which of two brands was rated as being of higher-quality in a recently read Consumer Reports article. Examining the packages may remind him of attributes that the article mentioned as being particularly important in the quality rating, allowing him/her to recover the name of the winning brand.

Often, the mere fact that the encoding and retrieval environments are the same will suffice to significantly increase the probability of retrieval. An experiment conducted by Godden and Baddely (1975) illustrates the phenomenon. Divers were asked to learn a list of unrelated words either twenty feet under the sea or on the shore. They were subsequently tested in the same environment or in the other one, resulting in completely crossed 2 X 2 factorial design. As expected, the interaction effect between learning and recall environment was significant, i.e. recall in the same environment (dry or wet) was superior to recall in the different environment. These results car. be taken as one manifestation of the principle of encoding specificity (Tulving and Thomson 1973) which asserts that the representation scored in memory is a function of the information present at encoding. Therefore, according to this position, in order to be effective, retrieval cues must be compatible with encoding inputs.

Retrieval goals. Clearly, one's objectives play a prominent role in guiding the retrieval process. They can be specific, as when one is asked "Who's Canada's prime minister?". But, in many cases, the individual's goals are broad, sometimes vague and necessitate his accessing different types of information of which the target information may only be a small subset. As an example relevant to this study, suppose one is asked to evaluate the goodness/badness of a set of profiles systematically defined using specific orthogonal dimensions, like in conjoint analysis (see Green and Srinivasan 1978), and that the research interest is in finding if some previously presented (target) information on the relative importance of these dimensions is retrieved and used by the individual to accomplish this task. The overall goal here is well defined, but the person has to come up by him/herself with some strategies to attain it and these in turn imply the formation of various retrieval goals, one of them being related to the weights that should be mentally assigned to the dimensions when the evaluative judgments are made. The degree of freedom enjoyed by the individual to perform this task makes it difficult to predict if the target information will be retrieved at all. But, if the goal is further specified, e.g. by cuing the person on the utility of considering the relative importance of the dimensions, one can be more confident that the relevant memories will be accessed and utilized.

Information from memory. Frequently, retrieval does not proceed directly. Rather, on the basis of some initial specification, memory is searched and the outcomes serve as inputs for further search until a match is obtained or a decision is made to stop. Memory then supplies itself with informational cues that constantly provide more specification to the retrieval process.

Research Propositions

Some summary statements about information retrieval in judgment and choice situations can now be made.

(1) The direct retrievability of some specific information relevant to some judgment or choice is contingent upon its availability in memory. [Obviously there are situations where a piece of information is not available in memory, but can be deduced from other elements. Such would not be a case of direct fact retrieval, however. For example, it is probable that the results of dividing the number 207 by 36.2 is not available in most people's mind. Yet, it can be computed with appropriate procedural knowledge.]

(2) Previously encoded information pertaining to some judgment or choice may be retrieved but will be used only if the individual agrees with the conclusions implied by this information.

(3) Elaborative activities performed on some target information at the encoding stage will increase the likelihood that it will be retrieved and used.

(4) The more similar the retrieval conditions are to the encoding conditions, the better the chances of recovering the target information and using it.

5) The level of retrieval specification, i.e. the preciseness of the incitation to search for the target information, is an important determinant of retrievability; the more specified is the retrieval goal, the better the chances of accessing the information and using it.

METHOD

Overview

The experiment basically follows the sequence of events displayed in the stage model of Figure l. Subjects were resented with some target information of relevance for an valuative judgment, namely which attribute a credible source considered to be the most important aspect in the valuation of a given consumer product. After a retention period of one week, they were given the opportunity o retrieve that information and use it in the context of judgment task.

Preliminary Study

Prior to the experimental phase of the research a small scale study was conducted in order to identify a consumer product with which subjects would have minimal familiarity n terms of the dimensions to use for evaluation and their relative importance. In addition, it hat to be possible D adequately predict subjects' evaluative judgments with few non-interactive and approximately equally weighted attributes. A group of business students took part in a conjoint analysis study involving three products: binoculars, smoke detectors and stereo headphones. For each product a set of 18 profiles varying according to three attributes in a 3 X 3 X 2 complete factorial arrangement as constructed. Students rated the profiles on nine-point good buy-poor buy scales_ After the task, they also reported how knowledgeable they were about the importance of the attributes. For this latter measure, the lowest mean was obtained with binocuLars. Furthermore, the statistical anaLysis of the conjoint ratings for binoculars showed no interactions between attributes and very compatible magnitudes of main effects. Results for the other two products were much less interesting (some interactions, large differences between magnitudes of main effects). Thus, on the basis of these results, binoculars were chosen as experimental product.

Exposure Stage

Objects were undergraduate business students at the university of Sherbrooke. In the first stage of the experiment, they participated in a study of "reading comprehension'', supposedly conducted by a graduate student in psychology. Their task consisted in reading and answering various questions about three excerpts from recently published magazine articles. One excerpt was presented as coming from Consumer Reports and contained the target information. It mentioned that consumers in the process of looking at market offerings of binocuLars should give more importance to one attribute: field of view. The choice of this attribute was based on a real article published in Consumer Reports (CR).

Elaboration manipulation. One of the research propositions considered earlier asserts that elaborative processing of incoming information helps memory by the fact that it allows the creation of additional usefuL retrieval routes. A two-level between-subjects factor is incorporated in this experiment. In the elaboration condition, the interaction with the target information was as follows. First, a four-minute delay was given to read the 150-word CR excerpt and answer four general questions about it (e.g. what title would you give to this excerpt?). Then a paragraph from the excerpt in which the target information appeared was made available for a period of ninety seconds during subjects had to answer four questions related to the information (What product? What attributes? Which is most important? Why?). After this elaboration task, subjects were handed a blank sheet and given two minutes to write down the paragraph as they recalled it. In the no elaboration condition, the experimental sequence was identical except that subjects were not asked any questions during the ninety-second presentation of the paragraph.

Judgment Stage

One week after having been in what they thought was a study of reading comprehension, subjects participated in a "decision making" experiment conducted by a different researcher. [There is no general rule as to how long a retention period should be, although a one-week delay is common (see e.g. Bradshaw and Anderson 1982, Wyer and Hartwick 1984).] The experiment consisted in going through conjoint analysis tasks with the same consumer products employed in the preliminary study. Within-subjects evaluative judgments of profiles constructed by considering all possible combinations of the attributes associated with the products were collected. Table l displays the attributes and levels defining the profiles of binoculars. In addition, subjects were asked to rank the binoculars profiles in order of preference and provide ratings and rank orders of perceived importance for all attributes.

TABLE 1

ATTRIBUTES AND LEVELS OF BINOCULARS

Environmental context manipulation. Reviving the environmental encoding context has been shown to facilitate recall in many memory experiments (see Smith, Glenberg and Bjork 1978 for a review). Consistent with the principle of encoding specificity, reinstatement of the encoding context at the retrieval stage makes it easier to trace the needed information by maximizing the congruence between the retrieval and the previously encoded cues. In this experiment, context effects were investigated by having subjects provide the conjoint ratings either in the room where the reading comprehension study took place or in another room. The rooms were chosen so as to be distinctive, i.e. either a large lecture room with windows, blackboards, etc. or a small seminar room with a few chairs around a large table and no windows. The assignment of subjects to either type of room and experimental condition was random.

As Bettman (1979) has indicated, environmental context effects have been neglected in consumer research; yet, their relevance to marketing situations seems non negligible. For example, matching information context in ads with the in-store context:

"... advertisements present information in a particular context which very often does not match the in-store context. Perhaps information usage, usage of particular attributes as criteria, or even recognition of brands is influenced by the degree to which the context posed in the ad is present in the actual choice situation" (Bettman 1979, p. 46, underlining added).

In this study, the context manipulation was meant to simulate the same type of situation described above, although advertising issues were not of particular concern.

Retrieval specification manipulation. It was suggested earlier that broad goals such as forming an evaluation of a product do not insure that information potentially relevant to the attainment of that goal will be retrieved and utilized. What is needed to direct the system to some specific information are specific retrieval goals. Therefore, one additional manipulation in this study involved informing half of the subjects of the usefulness of making a priori decisions about the relative importance of the attributes that composed the profiles to be evaluated. This retrieval specification was made orally by the experimenter immediately before proceeding with the conjoint task.

Control group. In order to provide a basis upon which to compare the experimental results, an additional group of subjects provided the conjoint and attribute importance data. These respondents did not participate in the reading comprehension study and were not subjected to either the environmental context or the retrieval specification manipulations.

Experimental Camouflage

One concern with this study was that subjects might see a link between the exposure and judgment stages. In order to minimize the likelihood of subjects guessing that the experiments were in any way related, they were first contacted by the researcher supposedly in charge of the second experiment (judgment stage). Two or three days later, the other researcher, acting as if he had learned by chance that they had agreed to participate in some experiment, asked for another commitment to a study taking place one week earlier. This simple trick was meant to prevent subjects from making causal inferences regarding the relationship between the experiments too easily. Subjects were paid three dollars for their participation in the first studY.

Postexperimental Survey

A few days after the judgment stage, subjects were contacted individually for a postexperimental interview. The objectives were to make some checks on the success of the retrieval specification manipulation and to verify if they had perceived a link between the two experiments. Those interviews confirmed that none of the subjects made connections between the studies.

RESULTS

Manipulation Checks

Retrieval specification. During the postexperimental interview, all subjects were asked if they recalled having been told by the experimenter to think about the importance of the attributes composing the conjoint profiles before proceeding with the task. Twenty-nine of the thirty-one subjects in the retrieval specification (i.e. 90.6 percent) recalled the experimenter's intervention. Only one of the thirty-three subjects in the no retrieval specification condition thought he hat heard the retrieval verbal incitation.

Elaboration. To verify the success of the elaboration manipulation, subjects' performance on the memory test given immediately after the exposure to the target information can be examined in both elaboration treatment conditions. The proportion of subjects having written down properly the target information on the provided sheet is 0.64 in the no elaboration condition as compared to 0.84 in the other condition. The difference is statistically significant (Z = 1.83, p = 0.0336, one- tailed test).

Dependent Variable

In this study, the interest is in finding if the target information was retrieved and used at judgment time and in determining the impact of elaboration, context and retrieval specification on the process. As shown in Figure l, it is assumed that use of the target information will quite automatically follow its retrieval from memory. To assess subjects' use of the target information during the evaluative processing of the binoculars, a measure of relative importance of the field of view attribute must be considered. The following analysis strategy was adopted. For each subject, the conjoint data (ratings and rankings) pertaining to binoculars were analyzed through ordinary least-squares methods in order to verify that an additive model was appropriate and that, consequently, the importance of the field of view attribute could be measured independently of the other attributes. The minimum proportion of total variation explained by additive effects only was arbitrarily fixed at 0.70 as a cut-off for inclusion in the final sample. Only one subject had more than thirty percent of the variation in his data explained by interactions and, as a result, was excluded. Overall, the mean proportion of variation due to interactions was very low (about 0.10).

For each subject in the final sample two measures of the relative importance of field of view were computed: (l) mean square associated with the field of view main effect over the sum of the mean squares of all three attributes, and (2) absolute difference between the marginal means of the field of view attribute over the sum of these differences for all attributes. These scores were obtained with both ratings and ranking data and the final dependent variable was defined as the sum of the four scores. [Other ways of operationalizing the dependent variable were also considered (e.g. absolute mean squares and utility ranges), but the pattern of results has remained essentially the same.]

FIGURE 2

GROUP MEANS IN THE EXPERIMENTAL CONDITIONS

Design and Analysis

The experimental design is a 23 between-subjects factorial arrangement with one control group where the two-level factors are elaboration, context and retrieval specification. The field of view relative importance group :means are plotted in Figure 2. Since the seventy-four subjects composing the final sample are distributed unevenly across the experimental conditions (seven, eight or ten subjects per treatment condition), the factorial design is unbalanced. Accordingly, Appelbaum and Cramer's (1974) general linear model approach was applied in the computation of the different sums of squares. By using this model comparisons procedure neither the three-way nor any of the two-way interactions are statistically significant by usual scientific standards (i.e. p < .05). The only interaction effect which stands out is that involving the retrieval specification and context factors with a p-value of 0.1371; on a descriptive basis, retrieval specification leads to greater weight given to the field of view attribute when the retrieval context is different from the encoding context and lower weight when both contexts are the same. As Table 2 reports, the only statistically significant main effect is associated with the elaboration factor; subjects who engaged in elaborative activities at the encoding stage gave more importance to the field of view attribute at judgment time than subjects who did not. The differences in marginal means for the other independent variables are in the wrong predicted direction and not statistically significant. A comparison between the control group and the eight experimental groups as a whole confirmed that exposure to the target information leads to more importance attributed to field of view during the conjoint task (t = 1.96, p = 0.0273, 65 df, two-tailed test).

TABLE 2

ANALYSIS OF VARIANCE RESULTS

DISCUSSION

The only statistically significant results of this study are those associated with the elaboration manipulation. The results support the proposition that elaborative processing of stimulus information leads to better memory ant, consequently, better chances of later retrieving that information and using it in some evaluative judgment task. It must be noted, however, that the elaboration manipulation, as operationalized is this experimental study, involved a question that required subjects to retrieve the target information specifically, viz. "Which attribute is most important?". This form of manipulation might then be more appropriately termed elaborative rehearsal, since it did not directly incite subjects to create new memory associations. The superior memory performance of elaboration subjects for the target information observed during the interpolated memory test suggests however that these associations were nevertheless created.

Although it did not reach statistical significance, the observed cross-over interaction between the retrieval specification and room context factors (see Figure l) warrants some discussion, given the small sample sizes and the resulting low statistical power. It seems interesting to speculate about what may have caused this interaction. As indicated earlier, when encoding and retrieval room contexts differ, the retrieval specification manipulation leads to greater weight given to the target attribute. When contexts match, the effect of retrieval specification is opposite. As Figure 2 shows, the interaction seems to be concentrated in the conditions where the target information was elaborated upon. One explanation for the interaction is that the conjunction of retrieval specification and identical environmental context led subjects to question the validity of the target information, thereby reducing the importance given to the field of view attribute in the evaluation of the binoculars. Perhaps these two facilitating conditions occurring at the same time resulted in the retrieval of additional information that cast doubt upon the truth of the target belief, while each condition taken separately was less likely to activate this information. This argument is best conveyed using a spreading-activation memory network representation (e.g. Anderson 1980). Such a representation in the context of this study is depicted in Figure 3. As shown, the target information is assumed to be associated with the room context, the concept "Binoculars Attribute Importance", and some additional information processed during the encoding stage. Examples of additional informational units are the experimental research situation in which the target information was learned, the CR excerpt context, the other attributes that were mentioned in the excerpt, the pro-and counter-arguments generated during the learning episode, and so on.

FIGURE 3

NETWORK REPRESENTATION OF SUBJECTS' ASSOCIATIVE MEMORY STRUCTURE

Now, assuming only the room context mode is activated, accessing the target information should be easier since, contrary to other related information, it has been elaborated upon and the experimental procedures have made it more distinctive. Similarly, assuming only the "Binoculars Attribute Importance" concept note is activated (e.g. using retrieval specification), accessing the target information should be easier for the same reasons. Note that these two scenarios are consistent with the observed interaction. In Figure 3, the greater potency for memory access to the target information is represented by the wider paths connecting the nodes.

Consider the case where both nodes are activated (corresponding to the retrieval specification-same context experimental condition); again the target information should be easily accessed. However, the chances of accessing the other informational units should increase since activation now emanates from two concept nodes. Thus, for example, subjects might come to think about the research study context or the attributes mentioned in the CR excerpt or even self-generated counter-arguments. Such thoughts could have a significant impact on the decision to make use of the target information during the evaluation task. One possible consequence could be a decrease in the importance given to the target attribute.

Obviously, the above discussion has been highly speculative and other explanatory schemes could be proposed. One interesting aspect of this discussion though is that it stresses the distinction between information retrieval and information use and suggests that retrieval of some particular information does not necessarily lead to an unqualified utilization in some subsequent task (see also Gardner 1983).

CONCLUSION

Information retrieved from memory plays a significant role in the inference making process. Therefore, research on the conditions under which memory retrieval is facilitated must be extended to judgment or choice situations. The results presented in this paper suggest that engaging in elaborative mental activities at the encoding stage may increase the likelihood of retrieving and using the encoded information in a subsequent consumer judgment task. Although the positive impact of elaboration on memory is a well-documented finding in cognitive psychology, its extension to information utilization appears to have been neglected so far. Hopefully, the present study will have stimulated research into extending similarly other cognitive psychology findings.

REFERENCES

Anderson, J.R. (1980), Cognitive Psychology and Its Implications, San Francisco: Freeman.

Appelbaum, M.I., and E.M. Cramer (1974), "Some Problems in the Non-orthogonal Analysis of Variance," Psychological Bulletin, 81, 335-43.

Bettman, J.R. (1979), "Memory Factors in Consumer Research: A Review," Journal of Marketing, 43, 37-53.

Biehal, G., and D. Chakravarti (1982), "Information-Presentation Format and Learning Goals as Determinants of Consumers' Memory Retrieval and Choice Processes," Journal of Consumer Research, 8, 431-441.

Biehal, G., and D. Chakravarti (1983), "Information Accessibility as a Moderator of Consumer Choice," Journal of Consumer Research, 10, 1-14.

Bradshaw, G.L., and J.R. Anderson (1982), "Elaborative Encoding as an Explanation of Levels of Processing," Journal of Verbal Learning and Verbal Behavior, 21, 165-74.

Bransford, J.D. (1979), Human Cognition: Learning, Understanding and Remembering, Belmont, CA: Wadsworth.

Cohen, C.E., and E.B. Ebbesen (1979), "Observational Goals and Schema Activation: A Theoretical Framework for Behavior Perception," Journal of Experimental Social Psychology, 15, 305-29.

Craik, F.I.M. (1979), "Human Memory," Annual Review of Psychology, 30, 63-102.

Crowder, R.G. (1976), Principles of Learning and Memory, Hillsdale. NJ: Lawrence Erlbaum Associates.

Fishbein, M., and I. Ajzen (1975), Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research, Reading, MA: Addison-Wesley.

Gardner, M.P. (1983),-"Advertising Effects on Attributes Recalled and Criteria Used for Brand Evaluations," Journal of Consumer Research, 10, 310-318.

Glass, A.L., K.J. Holyoak, and J.L. Santa (1979), Cognition, Reading, MA: Addison-Wesley.

Godden, D.R., and A.D. Baddeley (1975), "Context-Dependent Memory in Two Natural Environments: On Land and Underwater," British Journal of Psychology, 66, 325-31.

Green, P.E., and V. Srinivasan (1978), "Conjoint Analysis in Consumer Research: Issues and Outlook," Journal of Consumer Research, 5, 103-23.

Hastie, R., and D.E. Carlston (1980), "Theoretical Issues in Person Memory," in Person Memory: The Cognitive Basis of Social Perception, eds. R. Hastie, T.M. Ostrom, E.B. Ebbesen, R.S. Wyer, D.L. Hamilton, and D.E. Carlston, Hillsdale. NJ: Lawrence Erlbaum Associates, pp. 1-53.

Johnson, E.J., and J.E. Russo (1981), "Product Familiarity and Learning New Information," in Advances in Consumer Research, Vol. 8, et. K. Monroe, Ann Arbor, MI: Association for Consumer Research. pp. 151-5.

Klatzky, R.L. (1980), Human Memory: Structures and Processes. Second Edition, San Francisco: Freeman.

Lynch, J.G., Jr., and T.K. Srull (1982), "Memory and Attentional Factors in Consumer Choice: Concepts and Research Methods," Journal of Consumer Research, 9, 18-37.

Norman, D.A., and D.G. Bobrow (1979), "Descriptions: An Intermediate Stage in Memory Retrieval," Cognitive Psychology, 11, 107-23.

Smith, S.M., A. Glenberg, and R.A. Gjork (1978), "Environmental Context and Human Memory," Memory and Cognition, 6. 342-53.

Tulving, E. (1976), "Ecphoric Processes in Recall and Recognition," in Recall and Recognition, ed. J. Brown, New York: John Wiley & Sons.

Tulving, E., and D.X. Thomson (1973), "Encoding Specificity and Retrieval Processes in Episodic Memory," Psycho- logical Review, 80, 352-73.

Wyer, R.S., and J. Hartwick (1984), "The Recall and Use of Belief Statements as Bases for Judgments: Some Determinants and Implications," Journal of Experimental Social Psychology, 20, 65-85.

Wyer, R.S., and T.K. Srull (1980), 'The Processing of Social Stimulus Information: A Conceptual Integration," in Person Memory: The Cognitive Basis of Social Perception, eds. R. Hastie, T.M. Ostrom, E.B. Ebbesen, R.S. Wyer, D.L. Hamilton, and D.E. Carlston,Hillsdale, NJ: Lawrence Erlbaum Associates, pp. 227-300.

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