A Model of Consumer Memory and Judgment

Thomas K. Srull, University of Illinois
ABSTRACT - One of the most perplexing problems in advertising research over the past several decades has been a failure to find any consistent relationship between memory for information presented in an ad and product evaluation. A general model of the relationship between recall and judgment is outlined, and several empirical tests of the model are reported.
[ to cite ]:
Thomas K. Srull (1986) ,"A Model of Consumer Memory and Judgment", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 643-647.

Advances in Consumer Research Volume 13, 1986      Pages 643-647

A MODEL OF CONSUMER MEMORY AND JUDGMENT

Thomas K. Srull, University of Illinois

ABSTRACT -

One of the most perplexing problems in advertising research over the past several decades has been a failure to find any consistent relationship between memory for information presented in an ad and product evaluation. A general model of the relationship between recall and judgment is outlined, and several empirical tests of the model are reported.

The purpose of the present paper is to consider carefully the relationship between consumer memory and judgment. Many psychological theories assume implicitly that there should be a close correspondence between the specific facts that can be recalled about an object and any global evaluation that is made of that object. This is true in most theories of attribution, impression formation, and other areas of social judgment. In general, the person is assumed to recall specific episodic events involving the target, and then determine the implications of these events for the judgment to be made. Theorists differ on how elaborate the intervening combinatorial process is hypothesized to be, but there is almost universal agreement that there should be a strong relationship between the evaluative implications of whatever events are recalled and the extremity of any evaluative judgment that is made. The extent of this agreement is so strong that Hastie and Carlston (1980) have referred to this as the "traditional model."

While theorists have been nearly universal in stating what should happen, empirical examinations have painted quite a different picture. For example, there are approximately fifty published experiments in the psychology literature that have examined the relationship between memory and judgment (Hastie and Park 1985). The typical procedure is to present subjects with information, wait for some period of time, and then ask them to recall as much of the information as possible and make some type of judgment of the target. The average evaluative implication of each item recalled is then correlated across subjects with the judgments that are made. Do subjects who recall the most favorable information make the most favorable judgments? Usually not; the correlations described above are invariably quite small, and seldom statistically different from zero.

It is somewhat surprising from current psychological theory that one would ever find such a lack of correspondence between the specific facts that are recalled and the judgments that are made. It is even more surprising how general and robust this finding appears to be. For example, in a classic impression formation study, Anderson and Hubert (1963) provided subjects with trait adjectives. Their essential finding was that there are strong primacy effects in impression formation, but reasonably large recency effects in recall. The authors concluded from this that the impression judgments and memory for the trait adjectives must be independently stored and accessed in memory. In a later study, Dreben, Fiske, and Hastie (1979) replicated these effects, and also found that temporal delays had a large effect on recall but very modest changes in impression ratings. They also concluded from this that there must be some independence between episodic memory and abstract evaluative impressions (see also Riskey 1979).

There is also a long line of persuasion studies that have found a very weak relationship between memory for information in a persuasive communication and attitude formation and attitude change. A good illustration of this is a classic set of studies reported by Greenwald (1968). Subjects were presented with arguments concerning two separate controversial issues. Subjects first received an opinion pretest followed by the presentation of the communication. Upon reading the communication subjects were asked to write a one-sentence reaction to the main point of the paragraph. Subjects' opinions were again assessed followed by an unexpected recall test. Following a one week interval, the opinion postest and the recall test were unexpectedly readministered again. Greenwald found no correlation between postest opinion and retention of communication content at either of the delay intervals. He concluded from this that memory for information in a persuasive communication is not a mediating variable of subject attitude change. A number of other studies are also consistent with this general conclusion (see e.g., Insko 1964; Miller and Campbell 1959; watts and McGuire 1964). These failures to find the predicted relationship represent the type of concern that led to the "cognitive response" approach to the study of persuasion that is so dominant today.

The same essential conclusion has been reached in the study of attributions of causality. For example, Fiske, Taylor, Etcoff, and Laufer (1979) conducted a series of studies examining whether imagery accounts for the effects of empathy on attributions. No correlation between what was remembered and the attributions that were made was observed in any of these studies. Similar to many other experiments, there was no evidence for any recall-attribution link. Fiske et al. concluded that these results parallel those found in the other domains, a conclusion that has also been reached in more recent reviews (Hastie and Park 1985; Lichtenstein and Srull 1985).

A very similar situation exists in the psychology of advertising (Beattie and Mitchell 1985; Lichtenstein and Srull 1985). Most theorists have assumed that there should be a close correspondence between the specific facts that are recalled about a product and any general evaluative judgment that is made of that product. Because of this, one of the most popular measures of advertising effectiveness is recall. Again, however, the empirical evidence is quite different. There have recently been four separate reviews of the relationship between recall and judgment within an advertising context. In each case, the relationship has been found to be very small, and only in rare cases was it significantly different from zero (Gibson 1983; Ross 1982; Stout 1981; Young 1972).

It is seldom that theory and data stand in such stark contrast. Is it possible for so many theorists to be misguided on such a fundamental issue? It may be possible, but it is also doubtful. In the following sections of the paper, recent advances in cognitive psychology are used to explicate the conditions under which a strong correspondence between memory and judgment should and should not be expected on conceptual grounds. In the process, a general model that can be used to conceptualize the judgment process will be outlined. Several lines of empirical evidence bearing on the model will then be summarized.

The model proposed is based on a more comprehensive theory of information processing developed by Wyer and Srull (1980; 1985). The Wyer and Srull theory is relatively complex and well beyond the scope of the present discussion. However, two aspects of it are of particular importance. First, the theory suggests that the mental representation of product information will often differ as a function of the initial processing objectives of the consumer. The importance of differing mental representations in understanding the recall-judgment relationship will soon become apparent.

The second noteworthy aspect of the Wyer and Srull theory concerns its broad scope. Because it is intended to trace the flow of information from the initial encoding stage to the final response stage, it makes specific statements about how any given mental representation should be reflected in various performance measures. Although different representations may "look the same" on the basis of some dependent measures (e.g., overall levels of recall) they can be differentiated on the basis of others (e.g., probe reaction time or the types of judgments that are made). As just one example, Lichtenstein and Srull (1986) have used the theory to generate predictions for six separate measures of performance: (1) overall levels of recall, (2) the extremity of evaluative judgments, (3) the relationship between the episodic facts that are recalled and the judgments that are made, (4) serial position order effects in recall, (5) serial position order effects in judgment, and (6) the relationship between spew recall order and judgment. The underlying assumption of this work is that our understanding of the recall-judgment relation will be best advanced by simultaneously considering all six aspects of performance in conjunction with one another.

An additional virtue of the Wyer and Srull theory is that it incorporates the distinction between "retrieval" and "computational" processes. Imagine a consumer is asked, "Is the Ford Thunderbird a luxury automobile?" A retrieval model would suggest that, at least for some people, the answer to such a question has already been determined and stored in memory. Thus, one simply needs to "retrieve" it from memory in order to answer such a question.

In contrast, a computational model would suggest that, at least for some consumers, the answer to such a question has not already been determined. That is, the mental representation of Ford Thunderbird does not contain any propositional knowledge indicating a value along some luxury dimension (or membership in some luxury-related category). In order to answer such a question, one would need to retrieve as much information about the Ford Thunderbird as possible, compare it to one's referent for "luxury automobile," and then "compute" or figure out an answer on the spot. Of course, if one were asked the same question again he/she would no need to re-compute an answer but simply retrieve the previous judgment (Wyer and Srull 1985).

Although the distinction between retrieval and computational processes is not often discussed, a number of researchers in both social cognition (e.g., Allen and Ebbesen 1981; Ebbesen and Allen 1979) and consumer behavior (e.g., Bettman 1579; Burke 1980; Brucks and Mitchell 1981; Mitchell and Smith 1982; Smith, Mitchell, and Meyer 1982) have recently found it to be useful. This distinction turns out to be crucial in understanding when there will and will not be a relationship between recall and judgment. In particular, the model proposed suggests that the consumer's information processing objectives or "goals" are a critical mediating variable that determines the nature of the relationship between recall and judgment. The reason for this is that such processing objectives often determine whether product evaluations have been pre-stored or need to be computed on the spot.

The model postulates that when a consumer acquires ad information with the (implicit or explicit) objective of making an evaluative judgment of the product, the global evaluation will be made at the time of information acquisition and stored in memory separately and independently from the specific episodic facts that are learned. If the consumer is later asked to make a specific judgment, the evaluation will have already been "computed" and will simply be accessed at that time. This is consistent with a large body of literature that suggests self-generated evaluations are much more accessible in memory than externally presented information (Johnson and Raye, 1981; Ostrom, Lingle, Pryor and Geva 1980; Slamecka and Graf 1978). Thus, under these conditions, there is no reason to expect any strong relationship between the specific facts that are recalled and the global evaluation that is made. One can see that this is a very straightforward retrieval model in which the previous judgment and the specific episodic facts on which it is based are independently stored and accessed. It is a process that is consistent with the conceptualizations outlined by Anderson and Hubert (1963) and Dreben et al.

There are occasions in which an alternative process will apply. Specifically, when a person acquires ad information with no specific objective in mind, or only a very general objective such as to comprehend the information being presented, a global evaluation of the product will not be made at the time of information acquisition. This, no evaluation will be incorporated into the resulting mental representation. If later asked to make a specific judgment, the consumer will be forced to retrieve the previously acquired information, or some subset of it, and use it as a basis for his/her evaluative judgment of the product. In other words, a judgment will have to be computed on the spot. Under these conditions, a strong relationship would therefore be expected between the global evaluation and the evaluative implications of the information that is recalled.

As an initial test of the model, Lichtenstein and Srull (1985) presented undergraduate students with print ads that: (a) were as complex as possible in the sense that they contained a large amount of attribute information, and (b) pertained to products with which subjects were likely to have little prior familiarity. Three separate stimulus replications were used. Half of the subjects, who participated in what is typically called an on-line processing condition, were told to read the ad with the purpose of forming an evaluation of the product so that they would later be able to judge how desirable it would be relative to other competing brands. The other half, participating in a memory-based processing condition, were told that the ad was written by an undergraduate advertising major. Their task was to read the ad in order to judge how grammatical, coherent, and interesting it was.

After reading the ad, there was either a 5-minute or 48 hour delay. Then subjects were asked to recall as much of the information as possible and make a general evaluative judgment of the product, half of the subjects completing the recall task first and the other half beginning with the judgment task.

The typical way to analyze such data is to have a separate group of subjects rate the evaluative implication of each proposition or "idea unit" in the ad. Then the mean evaluative rating for each proposition recalled by any given subject is determined. These values are then correlated, across subjects, with the product evaluation ratings.

Lichtenstein and Srull found in 12 out of 12 independent comparisons that the correlation between recall and judgment was higher in the memory-based than on-line conditions. Moreover, the correlations in the memory-based conditions were universally large and statistically different from zero. The mean correlation across 12 conditions was .64 in the memory-based condition and .22 in the on-line condition.

More recently Hastie and Park (1985) replicated these effects by using a slightly different paradigm. The on-line condition was very similar to that described earlier. However, for a memory-based condition, they had subjects anticipate making one judgment, but later asked them for a different, unrelated judgment. Across four separate experiments, the average correlation between recall and judgment was .51 in the memory-based condition and .16 in the on-line condition. Srull (1985) also reported several conceptual replications using slightly different orienting tasks.

These findings are very strong and replicate across a variety of laboratories, stimulus sets, content domains, and delay intervals. They suggest that past studies have consistently found only a weak relationship between recall and judgment because subjects have tended to use some type of on-line processing strategy. In the few studies that have included memory-based conditions, however, a strong correspondence between recall and judgment has indeed been found. For example, Sherman, Zehner, Johnson, and Hirt (1983) provided subjects with current information about two historically rival football teams. In one case the goal was to remember the information as well as possible. In another case, the goal was to form an impression of the potential outcome of an upcoming game between the two teams. Sherman et al. found a very weak relationship between memory for the information and subjects' judgments of the outcome when they had formed their impressions at the time of information acquisition. However, this relationship was quite strong among subjects who had initially been given a memory set. The authors suggest that the "availability" of information in memory will only effect judgments when they are initially unanticipated. A similar finding has recently been reported by Reyes, Thompson, and Bower (1980).

Although the model proposed predicts that the level of correspondence between consumer memory and judgment will vary considerably as a function of the person's initial processing objectives, much more subtle tests of the model are also possible. Srull (1985) reported one experiment that was specifically designed to tap into the intervening processes that are postulated to occur. The presumed difference between a "product evaluation" group (who theoretically are forming their evaluations on-line) and a "comprehension" group (who theoretically are not) is in the amount of computational activity required at the time of judgment. This was tested by examining the amount of time required to make a judgment.

Subjects were presented with a series of 12 print ads. One group was told to read the ads and form an evaluation of each product described. Another group was told to simply comprehend the information presented. The judgment task was then administered either 5 minutes or 48 hours later. The subject sat n front of a computer screen and was probed with "Please rate the desirability of each of the following products." On each of the 12 trials, the screen would display after a rest interval the brand name and a 10-point rating scale, ranging from 0 ("very undesirable") to 9 ("very desirable"). The time interval between stimulus onset and response was recorded on each trial.

The mean reaction time for the product evaluation group was approximately 5 seconds in the 5-minute delay condition, and approximately 6 seconds under the 48-hour delay condition. The mean reaction time for the comprehension group was approximately 10 seconds in the 5-minute delay condition and nearly 20 seconds in the 48-hour delay condition.

The most parsimonious interpretation of these results is that the on-line processing group formed their evaluations of the product at the time of information acquisition. At the time of judgment, they simply needed to access the appropriate evaluation in memory. It took them slightly longer after 48 hours than after 5 minutes. But they were very rapid in both cases. In contrast, the memory-based group must retrieve the episodic facts that were learned, or at least some subset of them, integrate the items that are retrieved into a global evaluative judgment, and finally make an overt response. These subjects took twice as long as the on-line group to make their judgments after 5 minutes, and nearly four times as long after 48 hours.

The most diagnostic test of the process model proposed in the present paper was reported by Lichtenstein and Srull (1985). The experiment was based on the fact that prior retrieval attempts typically aid subsequent recall. For example, consider a group of subjects who attempt to recall information at Time 1 and then again at Time 2. Their performance at Time 2 will generally be better than that of a control group who only recalls at Time 2. This is a very robust phenomenon (see e.g., Modigliani 1976), and it is usually interpreted to mean that prior attempts to recall strengthen the memory traces of the retrieved items.

Lichtenstein and Srull used this phenomenon as a diagnostic tool to determine what information is retrieved from memory under various conditions. Three separate experimental groups were created, each of which was exposed to the same sequence of print ads. One hour after receiving the ad information, one third of the subjects attempted to recall the information in as much detail as possible. Then, after a 48 hour delay, they tried to recall the information again. A second group made a global evaluative judgment after one hour. Then, after a 48 hour delay, they also tried to recall the information in as much detail as possible. Finally, as a control condition, one third of the subjects were not asked to complete any task after one hour but were asked to recall as much of the original information, in as much detail as possible, after 48 hours. The dependent variable of theoretical interest is the number of attribute units recalled by subjects after 48 hours.

One can consider the group that had no intervening task as a baseline control condition. Lichtenstein and Srull report that levels of recall were essentially the same for on-line and memory-based subjects in this condition. However, subjects who recalled after one hour showed much higher levels of recall after 48 hours than did the control subjects, providing a replication of earlier findings reported in the literature. Moreover, this was true for both the on-line and memory-based subjects and, as before, there was very little difference between them in terms of overall levels of recall.

The most important condition in terms of the underlying theory is that where intervening judgments were made after a one hour delay. The model suggests that memory-based subjects will need to retrieve the original information in order to make such a judgment. If this is true, the same type of recall advantage after 48 hours as is seen with the "recall" group should be observed. In other words, the model suggests that the ultimate recall of memory-based subjects in the "intervening judgment" condition should be: (a) greater than that in the "no intervening activity" condition, and (b) not significantly different than that in the "intervening recall" condition. This is exactly what was found.

The model also makes specific predictions about the online subjects in the "intervening judgment" condition. The model suggests that the on-line subJects will have already formed the evaluation at the time of information acquisition. Thus, after one hour, they need not review the original information in order to make the judgment, but simply retrieve their original evaluation from memory. In this process occurs, however, the ultimate memory performance of such subjects should not be as great as that of subjects in the "intervening recall" condition. This was also found.

The results of this experiment are subtle and provide strong converging evidence in support of the model proposed. The fact that intervening product evaluations increase levels of recall (compared to a no-intervening-task control group) for memory-based subjects but not for on-line subjects is difficult to explain without postulating different retrieval mechanisms.

One of the most perplexing problems in advertising research over the past several decades has been the general failure to find any consistent relationship between memory for information presented in an ad and product evaluation. These failures have beer met with surprise, discouragement, and some have even used them to question why researchers need to be concerned with memory processes in the first place. The research presented here should help to frame the problem in a somewhat new perspective.

There are three assumptions that are central to the present approach. First, it is assumed that different types of mental representations can only be distinguished by simultaneously examining different aspects of performance. The reason for this is that different underlying representations will often react to experimental manipulations in the same way. Thus, they may appear to be the same even though they are not. In one study reported by Lichtenstein and Srull (1985), for example, the overall levels of recall and the overall levels of judgment were identical for on-line and memory-based subjects. It is only by looking at a range of measures that one can discover converging evidence for how the underlying mental representations must be different from one another.

The second assumption is that theories must be developed to the point at which they have implications for multiple dependent measures if they are going to be distinguishable empirically (see Srull 1984). While plausible alternative interpretations of any single effect can be numerous, building alternative, internally-consistent theories that are capable of accounting for a large array of data is much more difficult. Moreover, even when such alternative accounts are devised, subsequent attempts to distinguish between them are nearly certain to produce substantial theoretical growth (see Srull and Wyer 1984 for an elaboration of this point).

Finally, the model outlined in the present paper is a very general one, within which a variety of more specific theoretical commitments can be made. For example, the exact format of knowledge representation has been left unspecified in the present report. Similarly, there is no commitment to a particular type of retrieval process. At this level of analysis, for these purposes, such commitments are unnecessary. However, as new constraints are added to the model, more sensitive predictions will emerge. Future work should be directed toward examining some of the more subtle implications of the model that emerge when a more specific set of theoretical assumptions are imposed.

REFERENCES

Allen, R.3. and Ebbesen, E.B. (1981), "Cognitive Processes in Person Perception: Retrieval of Personality Trait and Behavioral Information," Journal of Experimental Social Psychology, 17, 119-141.

Anderson, N.H. and Hubert, S. (1963), "Effects of Concomitant Verbal Recall on Order Effects in Personality Impression Formation," Journal of Verbal Learning and Verbal Behavior, 2, 379-391.

Beattie, A.E. and Mitchell, A.A. (1985), "The Relationship Between Advertising Recall and Persuasion: An Experimental Investigation," in Psychological Processes and Advertising Effects, ed. L.F. Alwitt and A.A. Mitchell, Hillsdale, NJ: Erlbaum.

Bettman, J.R. (1979), An Information Processing Theory of Consumer Choice, Reading, MA: Addison-Wesley.

Brucks, M. and Mitchell, A. (1981), "Knowledge Structures, Production Systems, and Decision Strategies," Advances in Consumer Research, 8, 750-757.

Burke, R. (1980), "A Preliminary Model of Consumer Cognition," Unpublished manuscript, Center for Consumer Research, University of Florida, Gainesville.

Dreben, E.K., Fiske, s.'r., and Hastie, R. (1979), "The Independence of Item and Evaluative Information: Impression and Recall Order Effects in Behavior-Based Impression Formation," Journal of PersonalitY and Social Psychology, 37, 1758-1768.

Ebbesen, E.B. and Allen, R.B. (1979), "Cognitive Processes in Implicit Personality Trait Inferences," Journal of Personality and Social Psychology, 37, 471-488.

Fiske, S.T., Taylor, S.E., Etcoff, N.L., and Laufer, J.K. (19?9), "Imaging, empathy, and Causal attribution," Journal of Experimental Social Psychology, 15, 356-37?.

Gibson, L.D. ( 1983), "Not recall," Journal of Advertising Research, 23, 39-46.

Greenwald, A.G. (1968), "Cognitive Learning, Cognitive Response to Persuasion, and Attitude Change," in Psychological Foundations of Attitudes, eds. A.G. Greenwald, T.C. Brock, and T.M. Ostrom, New York: Academic Press.

Hastie, R. and Carlston, D. (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, N.J.: Erlbaum.

Hastie, R. and Park, 3. (1985), "The Relationship Between Memory and Judgment Depends on Whether the Judgment Task is Memory-Based or On-Line," Unpublished manuscript, Northwestern University.

Insko, C.A. ( 1964), "Primacy Versus Recency in Persuasion as a Function of the Timing of Arguments and Measures," Journal of Abnormal and Social Psychology, 69, 381-391.

Johnson, M.K. and Raye, C.L. (1981), "Reality monitoring," Psychological Review, 88, 67-85.

Lichtenstein, M. and Srull, T.K. (1985), "Conceptual and Methodological Issues in Examining the Relationship Between Consumer Memory and Judgment," in Psychological Processes and Advertising Effects: Theory, Research, and Application, eds. L.F. Alwitt and A.A. Mitchell, Hillsdale, N.J.: Erlbaum.

Lichtenstein, M. and Srull, T.X. ( 1986), "Processing Objectives as a Determinant of the Relationship Between Recall and Judgment," Unpublished manuscript, University of Illinois.

Miller, N. and Campbell, D.T. (1959), 'Recency and Primacy in Persuasion as a Function of the Timing of Speeches and Measurements," Journal of Abnormal and Social Psychology, 59, 1-9.

Mitchell, A.A. and Smith, T.R. (1982), "The Applicability of Computational Process Models for Representing Consumer Behavior," Advances in Consumer Research, 9, 125-131.

Modigliani, V. (1976), "Effects on a Later Recall by Delaying Initial Recall," Journal of Experimental Psychology: Human Learning and Memory, 2, 609-622.

Ostrom, T.M., Lingle, J.H., Pryor, J.B., and Geva, N. (1980), "Cognitive Organization of Person Impressions," in Person Memory: The Cognitive Basis of Social Perception, eds. R. Hastie, T.M. Ostrom, E.B. Ebbesen, R.S. Wyer, D.E. Hamilton, and D.E. Carlston, Hillsdale, N.J.: Erlbaum.

Ostrom, T.M., Pryor, J.B., and Simpson, D.D. (1981), "The Organization of Social Information," in Social Cognition: The Ontario SymPosium, eds. E.T. Higgins, C.P. Herman, and M.P. Zanna, Hillsdale, N.J.: Erlbaum.

Reyes, R.M., Thompson, W.C., and Bower, G.H. (1980), "Judgmental Biases Resulting from Differing Availabilities of Arguments," Journal of Personality and Social Psychology, 39, 2-12.

Riskey, D.R. (1979), "Verbal Memory Processes in Impression Formation," Journal of Experimental Psychology: Human Learning and Memory, j, 271-281.

Ross, H.L. (1982), "Recall Versus Persuasion: An Answer," Journal of Advertising Research, 22, 13-16.

Sherman, S.J., Zehner, K.S., Johnson, J., and Hirt, E.R. (1983), "Social Explanation: The Role of Timing, Set, and Recall on Subjective Likelihood Estimates," Journal of Personality and Social Psychology, 44, 1127-1143.

Slamecka, N.J. ar,d Graf, P. (1978), "The Generation Effect: Delineation of a Phenomenon," Journal of ExPerimental Psychology: Human Learning and Memory, 4, 592-604.

Smith, T.R., Mitchell, A.A., and Meyer, R. ( 1982), "A Computational Process Model of Evaluation Based on the Cognitive Structuring of Episodic Knowledge," Advances in Consumer Research, 9, 136-143.

Srull, T.X. (1984), "Methodological Techniques for the Study of Person Memo-y and Social Cognition," in Handbook 04 Social Cognition., Vol. 2 , eds . R. S. Wyer and T.K. Srull, Hillsdale, N.J.: Erlbaum.

Srull, T.K. (1985), "The Relation Between Memory and Product Evaluation," Paper presented at the Fourth Annual Advertising and Consumer Psychology Conference, Needham Harper Worldwide, Chicago.

Srull, T.K. and Wyer, R.S. (1984), "Progress and Problems in Cognitive Social Psychology," in Annals of Theoretical Psychology, Vol. 1, eds. J.R. Royce and L.P. Mos, New York: Plenum Press.

Stout, R.G. (1981), "Copy Testing is Only Part of Advertising Research," Paper presented at the Twelfth Annual Conference of the Association for Consumer Research.

Watts, W.A. and McGuire, W.J. (1964), "Persistency of Induced Opinion Change and Retention of the Inducing Message Contents," Journal of Abnormal and Social Psychology, 68, 233-241.

Wyer, R.S. and Srull, T.K. (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, N.J.: Erlbaum.

Wyer, R.S. and Srull, T.K. (1985), "Human Cognition in its Social Context," Unpublished manuscript, University of Illinois.

Young, S. (19(2), "Copy Testing Without Magic Numbers,' Journal of Advertising Research, 12, 46-63.

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