New Directions in Attitude Research: a Critical Evaluation

Joel B. Cohen, University of Florida
[ to cite ]:
Joel B. Cohen (1978) ,"New Directions in Attitude Research: a Critical Evaluation", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 370-376.

Advances in Consumer Research Volume 5, 1978      Pages 370-376

NEW DIRECTIONS IN ATTITUDE RESEARCH: A CRITICAL EVALUATION

Joel B. Cohen, University of Florida

I have been asked to discuss the four papers presented at the session, "New Directions In Attitude Research." I suppose it would be appropriate to start off with the question, "So what is new?" While I'm not certain any of the authors would claim to have started us down a truly virgin path, consumer researchers can, I think, hearken back to four identifiable research orientations which have been given a shot in the arm by these papers. All have been underrepresented in previous consumer research. I have labeled these four major research thrusts as: (1) an information processing approach to attitude formation and change, (2) the importance of time in studying cognitive organization and change, (3) explicit consideration of the motivational bases for attitudes and attitude change, and (4) attitudes as interpersonal communications. I'd like to discuss how each of these has been treated by the authors who have addressed these issues.

ATTITUDES AS INTERPERSONAL COMMUNICATIONS

I begin with this topic because it is probably the newest of the attitude research orientations to be represented within the consumer behavior area. The notion of products as identity-relevant symbols, of course, is one of the oldest and best enduring concepts in our field. From the social class and reference group-oriented approaches of the 1950's (cf. Rainwater, et. al., 1959; Levy, 1959) through interpersonal response trait-oriented research (cf. Cohen, 1967; Cohen and Barban, 1970) to self-concept based research (cf. Dolich, 1969; Ross, 1971), consumer researchers have long recognized the importance of identity enhancing social communications to product and brand choice.

Schlenker's "social identity theory" provides a unifying conceptual structure within which to examine self-presentation strategies and behaviors. Attitude expressions and statements are viewed as interpersonal communications--thus, behaviors to be studied in their own right. According to Schlenker, "Even 'private' attitudes, expressed only to oneself, occur in the context of imaginary social settings and serve the same functions in our private worlds." Viewed within the framework of social identity theory, then, a person's attitudes and expressions of attitudes represent a deliberate attempt to create a particular identity, an impression management strategy.

Knowing Barry Schlenker as well as I do, I simply cannot believe that he is expressing favorable attitudes toward his social identity theory merely or primarily to gain respect and admiration or to otherwise stroke his self-concept. No, I am certain he believes very strongly in the worth of this orientation and, in fact, would evidence considerable enthusiasm for it were he isolated on a desert island with no praise in sight. Indeed, were Barry physically restrained from communicating about this topic to any other human being, it would probably play the same guiding and organizing role in his beliefs about human behavior.

Rich Lutz might want us to analyze Barry's attitudes toward social identity theory in terms of a "functional" approach, which I suspect might go heavy on utilitarian and particularly knowledge functions. In fact, what better example of an attitude object based on the knowledge function can be found than a theory in the hands of a dedicated scientist? Such a theory helps the scientist to find meaning out of seemingly disconnected observations, to organize relevant aspects of human behavior and thereby to provide clarity and consistency to his thought processes. We can think of this in terms of the observations Jerry Olson and Phil Dover have made with respect to the organizing aspects of memory schemata.

The functional approach argues for balance, for a willingness to see attitudes as serving multiple and diverse needs and goals. Social identity theory promises to enrich our understanding of attitudes which, at least in part, serve value expressive needs. Such attitudes, whether expressed in the form of volunteered statements about a particular product, responses to a questionnaire or in overt purchase behavior convey an association between aspects of a desired self-image and certain product characteristics. Under what conditions (e.g., situations, types of products, the nature of the "audience") a consumer's attitudes will be most likely to reflect impression management goals would appear to be a highly relevant, and eminently researchable, topic area.

MOTIVATIONAL BASES FOR ATTITUDES

One of the interesting theoretical issues involved in all of the papers is the motivating role of cognitive consistency in human behavior. To what degree is the establishment and maintenance of cognitive consistency a major driving force, goal or function? Rich Lutz's paper addresses this issue by viewing knowledge, cognitive organization and consistency needs as one of the major bases underlying the existence of attitudes in the first place: to provide meaning and structure and as a guide for action. The disruption of cognitive consistency, in this sense, could produce considerable ambiguity, uncertainty and behavioral conflict. Restoration of consistency, such as through exposure to information favorable to one's initial position, should be favorably regarded and conducive to the acceptance of such material. Yet, conflicts frequently exist between one's desire for consistency and one's adjustment needs. Only by being open to new information and being willing to modify existing beliefs as well as behaviors can the individual grow and adapt successfully to a dynamic environment.

Under conditions other than those which threaten to disrupt cognitive systems which the individual is committed to maintain (e.g., central beliefs about one's self, beliefs whose disruption would so disorient the individual as to impact on his ability to act), knowledge and organization-based cognitive consistency processes, while fundamental and pervasive, are probably not dominant.

Sources of Cognitive Inconsistency

Barry's paper makes the point (which other critics have also made) that the concept of cognitive inconsistency as a drive-like state (which then "pushes" dissonance reduction processes) is not well conceptualized or defined by its advocates. Is cognitive inconsistency really all that important? This criticism is, I believe, well founded and stems in large part from the failure of dissonance theorists to come to grips with and spell out both the source and the nature of such drive-like states.

One could, I suppose, attempt to argue that critics have all too often ignored the clearly stated proposition that dissonance theory requires an assessment of the number and importance of dissonant cognitions, and thereby a determination of the magnitude of the dissonance present, before rushing off to make predictions regarding dissonance reduction processes. Clearly, some cognitive inconsistencies appear to be trivial and others quite important. But, frankly, the theory's advocates have tended to skirt this important issue, and the critics have been left with an inviting (if imprecise and ambiguous) target. Psychological commitment to an action or position, degree of choice, choice incentives and foreseeability, while eminently researchable correlates of the magnitude of dissonance, do not provide a very satisfying conceptualization of the nature and strength of the dissonant state.

One's response to this state of affairs could be to look elsewhere, to turn to other "competing" models in the hope they will be at least as productive and a good deal more precise. Incentive theory (Elms, 1967), self-perception theory (Bem, 1972) and most recently social identity/impression management theory (Tedeschi, et. al., 1971; Schlenker, 1977) represent well-conceived efforts in this direction. While each of these (and particularly social identity theory) can be seen to offer compelling alternative explanations for dissonance-like effects, particularly in the typical counterattitudinal behavior paradigms, I do not find any of these formulations to be as fundamental theoretically as dissonance theory. The potential integrative strength of dissonance theory has not been realized, in part because of a failure to recognize and combine both information processing and self-concept related underpinnings.

What I would like to propose is that there are two major sources of dissonance, in the sense of dissonance being a drive-like state resulting from cognitive inconsistency. The first of these is clearly related to Katz's (1960) "knowledge function" which Lutz has described in his paper. There is a need for meaningful and consistent cognitive organization in order to deal effectively with the external world, without this, the world would be a mass of disconnected stimuli. In response to such needs, we see the development of cognitive structure, which provides an interpretative and integrative function influencing all information processing activities.

Clearly, there must be an adaptive balance between openness to reality (even if it be counter to our prior beliefs and expectations) and the maintenance of an organized system of cognitive categories (which we take to be our knowledge of what is and what is not, what things go together, etc.). Experiments such as those which confront subjects with a deck of playing cards containing a black queen of hearts or a red king of spades illustrate a principle which has been termed "assimilation to the typical instance" (Jones and Gerard, 1967) and which is of considerable importance for social perception (e.g., stereotyping). Milton Rokeach has emphasized this dimension by defining resistance to change of a belief "solely in terms of connectedness: the more a given belief is functionally connected or in communication with other beliefs, the more implications and consequences it has for other beliefs and, therefore, the more central the belief" (1968, p. 5).

Very few dissonance studies have specifically examined the drive-like effects of a disruption in one of a person's central beliefs. It is not an easy matter to design a study which would seriously challenge an important personal belief. There can be no doubt, however, that the dissonance caused by strongly believing A implies B and then learning that A implies not B would be drive-like: some resolution of this cognitive dilemma must be found to restore order and meaningfulness. (See Abelson, 1959 for a discussion of denial, bolstering, differentiation and transcendence in resolving belief dilemmas.) Carried to an extreme, "It may be supposed that any inexplicable disruption of these taken-for-granted constancies, physical or social or self, would lead one to question the validity of one's own senses, one's competency as a person who can cope with reality, or even one's sanity" (Rokeach, 1968, page 7). Field studies which have the potential to create powerful disconfirmation of beliefs (supported by behavioral commitment) as when one who is committed to a course of action is suddenly and dramatically confronted by evidence that his beliefs and course of action are wrong (see for example Festinger, et. al., 1956) may provide a more appropriate setting in which to examine the power of cognitive inconsistencies than the usual types of scenarios presented to subjects in laboratory studies.

The major source of dissonance is probably inconsistencies relating to the self-concept. Here the issue is not coping with reality and cognitive organization, per se, but rather what Katz (1960) has termed "value expressive'' needs. It is quite true, of course, that self-concept related beliefs may be important anchoring points for a large number of related beliefs, and hence play an important organizing function. Such beliefs should be quite resistant to change, and behavior seen as inconsistent with them should provoke considerable dissonance. It is with self-concept related beliefs that dissonance theory and social identity theory may be inextricably linked. Whether or not social identity theory is a competitor or serves to extend and clarify may be more a matter of judgment than an empirical question.

Another reaction may be that what is proposed here is no longer quite dissonance theory or at least is a substantial modification of it (perhaps extending Aronson's, 1968, line of thinking). The basic point, though, is that the theory itself is unusually silent on such a key issue as the nature and identification of the "importance" of a cognitive element. This was acknowledged as early as 1962 (Deutsch, et. al. ) when the following proposition was advanced: "Thus, we offer the hypothesis that a chooser will experience post-decisional dissonance only when he perceives his choice in a given situation to be inconsistent with the conception of some aspect of himself which he tries to maintain (for himself or for others) in that situation."

Dissonance Reduction Versus Self Enhancement

What then is the motivating force, inconsistency among cognitive elements one or both of which serve important self-defining or organizational functions, or (per Schlenker) a desire to act so as to increase personal association with desirable attributes and outcomes and to decrease such association with undesirable attributes and outcomes? Social identity theory sees the degree of fit between the identity the individual is trying to project and some action, attitude, or choice object as the key element. Dissonance theory would conceptualize this fit or association in terms of cognitions of varying importance and consistency.

Paired in this way, the commonality among these formulations may be more important than the differences. Social identity theory greatly extends, clarifies and refines the nature of attitude-behavior relationships which involve social identities and related behavior.

Schlenker and Schlenker (1975, p. 963) state, "The greater the degree of personal responsibility an individual perceives for an action which generates negative consequences, the greater his degree of personal dissatisfaction, anxiety, and attendant social punishment." The dissonance theory position proposed earlier leads to a very similar hypothesis, since negative consequences of one's behavior should be strongly inconsistent with a very important set of cognitions: those which define and proclaim one's self-concept. The Schlenkers, in an insightful overview of the counterattitudinal behavior paradigm so often used in dissonance and self-perception theory research, point out that these "have placed individuals in positions where their self-image and public image is jeopardized..." (p. 964). While they maintain that "attitudinal discrepancy is not a necessary antecedent to attitude change," I would argue that the discrepancy between self-concept attitudes and inconsistent behavior is precisely the nature of the individual's predicament. Social identity theory, to its credit, goes on to identify tactics to be used by the individual to maintain and enhance his self-image and presentation of self. These same tactics, of course, serve to provide the individual with cognitions regarding his behavior which are more consistent with his self-concept.

The conceptualization of self-image maintenance and enhancement tactics appears to be quite similar to that adopted by Lutz in his treatment and proposed operationalization of the functional approach. By positioning the functional approach within an expectancy-value framework, Lutz develops the similar (to Schlenker) notion that attitude change attempts (whether externally or internally generated) must effect either the perceived association (i.e., perceived instrumentality) between the individual and the action (e.g., taking personal responsibility in Schlenker's system) or the value of the consequences (e.g., minimizing the amount or existence of harm in Schlenker's system). Perception of choice manipulations are a favorite device used to impact on the amount of responsibility (perceived instrumentality) the experimental subject perceives he had for whatever consequences developed. When personal responsibility is heightened, and the individual is therefore unable to excuse or justify his actions, he will frequently attempt to alter his perception of the resulting consequences (e.g., No "real" harm occurred; At the time I said, wrote or did X I really believed it was the good, true, or the right thing to do").

The same type of expectancy-value analysis has been applied by Schlenker and Schlenker (1975) to the situation in which positive rather than negative consequences result. Specifically, an average to dull person was told that she was quite attractive under either high or low choice (responsibility) conditions. S's subsequent scores were more positive than those of a control group only when there was low choice and S's expected future interaction during the study with the other person. A social identity theory expectancy-value analysis explains this result quite nicely in terms of S's exhibiting favorable attitudes toward the other person in order to associate themselves with and claim responsibility and credit for the positive consequences stemming from their (experimenter imposed) communications to the other person.

Schlenker and Schlenker feel that attitude change, which has the effect of establishing an internal locus of responsibility, is counter to dissonance theory predictions (i.e., S's having no choice and producing positive consequences should, they claim, experience no dissonance).

The problem, however, is that dissonance theory is ambiguous in terms of the source of the cognitive inconsistency. If, as the Schlenkers acknowledge, the dissonant cognition may relate to the self-concept (e.g., "I'm a friendly and helpful person. Yet I'm coming across as if I might not be that way at all because the experimenter wouldn't let me exhibit these qualities--I was given no choice."), then dissonance reduction is a viable alternative explanation. The Schlenkers' unwillingness to accept this explanation seems to be more a matter of overall discontent with dissonance theory's lack of precision (or an inability to pin it down, if you like) than the responsiveness of this particular explanation to their data.

Actually, there may be an even simpler dissonance theory interpretation: S's were merely bringing their questionnaire responses into line with their anticipated behavior. The no-choice, future interaction S's probably anticipated behaving in a positive and favorable manner toward the other experimental subject when the experimenter brings them together. Such S's will also know that the experimenter has told the other person that the favorable communication was assigned rather than chosen. These S's should have a greater reason to show both the experimenter and the other person that their attitudes and subsequent behavior are, in fact, consistent than S's whose free choice has already conveyed this.

Now, whether this preference for consistency stems primarily from a desire for cognitive order (i.e., to eliminate confusions that might interfere with one's ability to behave clearly and convincingly-- see Jones and Garard, 1967, for a discussion of the value of an "unequivocal behavioral orientation."), or a desire to appear consistent and genuine (to be otherwise would be inconsistent with important self-concept attitudes) can't be definitively resolved in this study. Even worse, it looks like it is going to be very difficult indeed to design a study in which dissonance reduction and the maintenance and enhancement of social identity (as alternative explanations) are not somehow confounded.

Concerning the need for jousting with dissonance theory, my recommendation to Barry would be to let the dragon slumber, it's been pretty quiet of late and doesn't seem to be up and about too often. There isn't really any need to take it on. The issues, settings, and most importantly the interpersonal goals which are directly addressed by social identity theory are not directly captured or represented by the more cognitively oriented dissonance theory.

Thus, even though dissonance theory may provide a somewhat more elegant and over-arching theoretical position which can incorporate virtually any set of data or conceptualization which possesses consistency-striving characteristics (whether to an actual or ideal self-image or presentation), it offers comparatively few insights into the self presentation and impression management areas when compared to social identity theory. While theoretical validity, comprehensiveness and parsimony may be ultimately satisfying to the soul of any behavioral scientist, the usefulness, adequacy and operational vitality of a conceptual framework may be a good substitute for the working social psychologist. On this score, and given the area on which Barry has chosen to focus, social identity theory appears to be off to a flying start. That one person or another may explain the same percentage of variance in one's data by some simple or "oblique" rotation of conceptual axes in and of itself does not establish the "truth" of that conceptual system.

Social identity theory is at one and the same time both less and a good deal more ambitious than any generalized consistency theory. It is less ambitious because it presents no claims on areas of inquiry which do not reside within its narrower domain (e.g., Olson and Dover's concern with information processing aspects of cognitive consistency). It is more ambitious because it does seek to provide a much more adequate and complete description and explanation of the many facets of human behavior concerned with social identity and self-presentation. This is a difficult and challenging area and one of great relevance to social psychologists and consumer behaviorists alike. There is no doubt that we shall be hearing a great deal more about social identity theory in the near future.

Status of the Functional Approach

I'd like to turn now to a consideration of Rich Lutz's excellent review and analysis of the so-called "functional approach to the study of attitudes." There are a number of reasons for advocating a categorization scheme similar to the one Rich has been working on. First, of course, such a scheme has the potential for shaking us out of any intellectual lethargy we may have fallen into in thinking too narrowly about such things as the antecedents, support for and conditions for changing attitudes and behaviors. There is also a specificity problem inherent in theories conceptualized at a highly abstract level (e.g., learning theory, dissonance theory), especially when considering applications to an area like consumer behavior, although such theories are generally considered to be more fundamental as well as parsimonious.

The general notion of differing functional bases seems sound enough. Attitudes and behavior expressive of prejudice toward a minority group, or for that matter, of liking for a dashing new sports car, may have very different functional bases and goals. Lutz's extension and modification of Katz's (1960) categorization scheme for the consumer behavior area may well help us to keep this fact firmly in our minds.

Another reason for advocating such a categorization scheme is as a kind of motivational check list (e.g., before leaving on a long trip, stop the newspapers, call the post office, lock the doors and windows, etc.). It is here that Rich may run into some trouble. At what level of specificity should functions be stated so as to be useful, say, to those interested in changing existing attitudes? My check list of things to do before leaving on a long trip is quite specific and at an actionable level. In addition, I have some confidence that these are, in fact, the things to do and that there aren't any major omissions. Can the same be said of Lutz's listings? Kiesler, Collins and Miller (1969) had earlier leveled this type of criticism at the functional approach.

If what we are after is a check list which is both more complete and more precise, we might begin with a review of the literature pertaining to personality needs and goals, from the pioneering efforts of McDougall (1908) through Murray (1938) to the recent day factor theories (see, for example, Cattell and Scheier, 1958). Clearly, the number and diversity of human needs, motives and goals can approach staggering proportions if our goal is to be exhaustive or purely descriptive (as in psychographic research).

It seems clear that Rich must be striving for something different than an exhaustive check list, some midpoint at which the structure provided attains that happy balance between pointing and naming, on the one hand, and abstract conceptualization on the other. While researchers have never been able to reach a consensus on how many functions, or needs or goals is the "right" number, not to mention what these are, that probably should not foreclose efforts at developing the sort of categorization which Rich believes would be useful. It seems clear, however, that "merely" developing satisfactory operational measures for the functions he advocates does not address the more basic question regarding the adequacy of the conceptualization of needs and goals represented by the classification scheme.

In summary, it appears as though the jury is still out, and a verdict cannot yet be given as to the usefulness of the functional approach in consumer behavior. If it does nothing more than alert us to the need to look beyond utilitarian product attitude in assessing consequences of using (or not using) a particular product, it will have contributed to our understanding of consumer attitudes and attitude change. We will look forward to greater operational specificity and empirical research on the part of those who advocate this approach.

INFORMATION PROCESSING ASPECTS

The Olson and Dover paper is important for consumer researchers particularly because it raises a set of issues which cannot be ignored by those who would focus on the outcome of information processing and decision making activities extending over several days, weeks or even months.

Changes in Cognitive Structure Over Time

Time (that bane of a researcher's existence, because it forces us out of one-shot measurement paradigms) is the catalyst. It is inconceivable that we will be able or willing to generalize our findings regarding information processing and decision-making very far without adequately investigating changes in cognitions and cognitive organization that occur over time.

Olson and Dover go on to analyze the impact of time on cognitive structure by looking at data from a longitudinal study designed to investigate disconfirmation of expectations. Obviously, some compromise must be made whenever the data one uses are not exactly what one would have gathered for the present purpose, and the authors acknowledge this limitation.

A major result, which Olson and Dover seek to understand and explain, is the increasing consistency among elements of cognitive structure over time. One suggestion that could be made with respect to research of this kind is that cognitive structure is a property of an individual's cognitive system, and hence the degree of consistency should be studied on an individual rather than group level. I'm more than a little uncomfortable with group means and correlations as indicants of changes in an individual's cognitive structure over time.

Evaluative and Probabilistic Consistency

An important distinction with respect to inference processes has been highlighted by the authors who call attention to the parallel concepts of evaluative and probabilistic consistency. The latter refers to the degree to which two concepts seem to be logically related to one another based on a combination of learning and stimulus cues so that inferences regarding an object's possession of other attributes follow along the lines of a conditional probability statement. Inferences based on evaluative consistency, on the other hand, seem not to be as concept-driven as they are affect-driven. That is, I may attribute positive qualities such as intelligence and warmth to a person because he is portrayed positively, not because I believe there is an associative link between the two attributes. Aspects of this process have been referred to as a "halo effect." These, then, are alternative mechanisms for filling in the gaps in one's belief systems, and since all information acquisition is necessarily incomplete and leaves gaps, these are very important processes.

Olson and Dover seek to disentangle the results of these two processes by looking at the increasing consistency they observe among biei elements. I would recommend looking at the bi elements by themselves since the concept of evaluative consistency pertains to the basis for the inference process (i.e., a positively regarded object is thought to possess positive attributes), the result of the inference being the association of the object with some attribute. Once again, though, it is hard to see how this can be done on anything other than an individual basis. In addition, disentangling these processes will require a study specifically designed for this purpose. While the authors suggest that evaluative consistency is operating in their study, this appears to be highly speculative in the absence of data specifically establishing the probabilistic linkage among the belief elements (i.e., the probability of A given B) as well as an independent evaluation of attributes.

Their group data (Figure 1) reveals wide fluctuation in attribute evaluations until people first tasted the bitter coffee, at which point four of the five attribute evaluations moved in the same direction (i.e., became more negative). Then, after an extended trial of the unadulterated coffee, four of the five attribute evaluations became more positive. The authors, looking at intercorrelations among attribute evaluations at the end of the study, concluded that the three attributes most closely related to overall product attitudes were also closely related logically (i.e., probabilistically). While these findings are quite interesting, the data used is simply not adequate to the delineation of inference mechanisms. It is simply not sufficient to estimate the degree of internal consistency among attribute evaluation scores. This sheds no light on the process that led to such a level of consistency.

Separating Out Sources of Impact

One potentially troublesome research issue suggested by the use of multiple measurements in a within-subjects longitudinal design is that our attribute-specific measures may be consistency-inducing. Wyer (197 6) has referred to this not uncommon phenomenon as the "Socratic effect." That is, we may be making a person more aware of the probable existence of an attribute-based belief structure and its presumed "logical" relationship to one's overall attitude. In so doing, we may encourage him to think through any inconsistencies in his own belief structure and, particularly for insightful experimental S's, to resolve these in the direction that would allow us to perceive them as more logical.

The authors make the important observation that beliefs may not be well-formed or stable after an initial exposure to information about an object or even the object itself: it may take several such experiences with the object. They argue for the direct measurement of S's confidence in their belief ratings. Here again, a within -subjects design may cause some problems. It may be that S's will believe that they should become more confident as a result of increased information and/or experience.

Olson and Dover report that the evaluation associated with beliefs regarding bitterness was, by the end of the study, the most strongly related attribute to overall attitude. This may, of course, simply be the result of demand effects since there were three separate communications designed to convince people that the coffee was not bitter followed by a "bitter" manipulation, and S's may have gotten the idea that bitterness was somehow important, at least to the researchers.

It is interesting, however, that the correlation between the evaluation of this attribute and overall evaluation was not particularly strong nor showing any particular trend until people actually experienced the bitter coffee. Perhaps this is still another illustration of the difference in impact between communications as information and experience as information. It may be something to think about in designing experimental manipulations, not to mention the various reasons to be interested in attitude and behavior change strategies! Perhaps one explanation for the greater impact, following the thrust of Olson and Dover, may be the greater "depth of processing" that is probably associated with the increased involvement overt behavior typically generates (See Nisbett, et. al., 1976, for a related viewpoint).

This particular piece of data also reminded me of the unique problems (e.g., long hours spent trying to communicate to respondents followed by blank stares) involved in the association between the absence of a negative state (i.e., this coffee is not bitter) and one's resulting positive evaluation. This double negative leading to a positive may not be as easily learned as a positive state leading to a positive evaluation.

There are, after all, a very large number of attributes an object doesn't possess, most of which we are not aware of to begin with or ignore. Saying that an object doesn't have something may actually communicate less information than saying it does have something, when the attribute in question is not highly salient to begin with. Thus, communications which instruct me that a particular brand of coffee does not possess an attribute I hadn't given much thought to will not produce very much cognitive work: there is nothing more I must deal with. By comparison, instructing me that the coffee does possess an attribute I hadn't given much thought to provides me with information which I need to integrate with existing information. The informational act of tasting the bitter coffee and experiencing its consequences should have given people a much better reason to link the absence of that attribute with their overall evaluation of the coffee.

Time and the Decay of Advertising Effects

Olson and Dover have enlightened us as to the substantial changes which may take place in cognitive structure over time, and Sawyer and Semenik, likewise, stress the need for researchers to examine delayed as well as immediate effects of advertising. The major thrust of the paper has to do with the decay of corrective advertising effects. Sawyer and Semenik invoke Jost's Second Law of Learning to explain why a corrective ad may appear to be more effective in reversing the effects of a deceptive campaign than it really is. According to Sawyer and Semenik, this translates to mean that "the effect of the older of the two (stimuli) will decay less rapidly - reaching some asymptotic level higher than that of the newer campaign."

The rate of decay (i.e., memory based) argument is quite similar to the explanation frequently given for observed recency effects when dependent measures are administered shortly after the second communication and some interval of time has elapsed between the first and second communications.

I have some difficulty with the higher asymptotic level of the older stimulus without knowing a great deal more about the reward and practice schedules which might so forestall further decay as to give it absolute superiority over a newer stimulus. Phrased more colloquially, Sawyer and Semenik's Figure 2A suggests to me that the older stimulus is comparable to a George Allen established performer and newer stimulus to an unproven rookie. A specific a priori determination of relative belief strength toward to the two stimuli, in my opinion, will not be derivable from any law of learning which fails to specify the exact relationships between each stimulus and other pertinent beliefs together with the value of the respective items of information to the individual. This is a tall order!

Another compelling interpretation may be that processing of the information in the second communication may be incomplete if measures are administered immediately thereafter, with the result that a "play back" of the position advocated in this communication (assuming it has been presented in a convincing manner) is a "sensible" response to the questions we are asking. With continued processing of the message, the individual should have a much greater opportunity to compare both its conclusions and premises with related stored information. If aspects of the new information are not as consistent with previously well-supported beliefs as it appeared on first contact, it may then be treated with less than perfect certainty or otherwise qualified. In any event, both memory-based and information processing-based arguments support the taking of delayed measures.

Time and Cue-Dissociation

I have related problems with the "discounting cue hypothesis" which assumes that a dissociation of an inhibiting cue (the authors look upon a corrective ad as an inhibiting cue) over time will reinstate the effects of the original stimulus. First, I have difficulty with translating a sometimes empirical observation into an explanation. Some information had less effect over time. Why? Because people discounted it. Very enlightening!

Perhaps a closer look should be taken at the authors' descriptions of the characteristics of information that was apparently "discounted" in other research. This might provide some idea as to why people might behave in this fashion. Let me throw out just one possibility. Whereas the non-discounted information seems to combine both conclusions and supporting arguments, the discounted information appears only to have offered a summary blanket refutation. Unless this refutation succeeded in removing the previous information from memory storage, further reality and consistency testing of that information was possible, indeed likely because of the conflicting conclusions which had been transmitted. No such opportunity was apparently present for arguments underlying the refutational conclusion, because none were presented. So a much greater opportunity seems clearly present, then, for people to make a subsequent assessment as to whether the "positive" conclusion "rings true."

Additionally, I'm not sure that Sawyer and Semenik have made a very good case for applying such concepts directly to the corrective advertising area. As I tried to indicate above, we need to take a long and hard look at both the concepts advocated and how they are operationalized. A concept invariably contains surplus meaning beyond a given operationalization. Perhaps we ought to be a little more conservative and claim no more territory than we have empirically staked out. In a sense, this is another argument for the importance of construct validity in behavioral research. To the extent that the corrective advertisement operationalizations are quite different from the research paradigms on which the concepts are based, a healthy skepticism seems in order. Identification of the major differences between the two contexts and settings may be a necessary and important step.

Let's just take one more brief example which spotlights the nature of the problem. Sawyer and Semenik present some findings indicating that qualified messages (stated more equivocally and with more reservations) persist better than unqualified messages. One set of now familiar explanations is that the qualifications and reservations are forgotten, discounted or dissociated from the main conclusions over time.

But what if this effect is due instead to the greater perceived credibility, expertise and trustfulness of a communicator who appears to be trying to be fair to both sides as well as more objective? It certainly looks like that person has been very thorough and is clearly well informed (more so that I could be on the topic). I might tend to believe his conclusions to a far greater extent than those of the communicator who has presented an unqualified message. Sawyer and Semenik seek to equate qualified messages with the format of corrective advertisements. But, if the communicator credibility notion is correct, and we assume that the credibility of an FTC-initiated conclusion is quite high, making the conclusion more equivocal and tentative may not only be unnecessary but a serious mistake.

Unfortunately, the ambitious study designed by Sawyer and Semenik appears to have misfired with respect to the stimulus materials themselves (i.e., there were no significant differences on the target belief between a control group of people who were not exposed to any message and S's exposed to either of the experimental messages). It would not appear to be particularly useful, therefore, to attempt to interpret some of the other results (a number of which are non-obvious) in the light of the difficulties and suggestions for improvement the authors have themselves raised in the paper.

REFERENCES

E. Aronson, "Dissonance Theory: Progress and Problems," in R. P. Abelson, E. Aronson, W. J. McGuire, T. M. New-comb, M. J. Rosenberg and P. H. Tannenbaum (Eds.), Theories of Cognitive Consistency: A Sourcebook, (Chicago: Rand McNally), 1968.

D. J. Bem, "Self-Perception Theory," in L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 6), (New York: Academic Press), 1972.

R. B. Cattell and I. H. Scheier, "The Nature of Anxiety: A Review of Thirteen Multivariate Analyses Comprising 816 Variables," Psychological Reports, 4(1958), p. 351-388.

J. B. Cohen, "An Interpersonal Orientation to the Study of Consumer Behavior," Journal of Marketing Research, 4(Aug., 1967) p. 270-278.

J. B. Cohen and A. Barban, "An Interactive Consumer-Product Typological System: A Progress Report and Partial Evaluation," Working Series in Marketing Research, Number 12 (College of Business Administration, The Pennsylvania State University), 1970.

M. Deutsch, R. M. Krauss, and N. Rosenau, "Dissonance or Defensiveness," Journal of Personality, 30(March, 1962), p. 16-28.

I. J. Dolich, "Congruence Relationships Between Self-Images and Product Brands," Journal of Marketing Research, 6(Feb. 1969), p. 80-84.

A. C. Elms, "Role-playing, Incentive, and Dissonance." Psychological Bulletin, 68(1967), p. 132-148.

L. Festinger, H. Riecken, and S. Schachter, When Prophecy Fails. (Minneapolis: University of Minnesota Press), 1956.

E. E. Jones and H. B. Gerard, Foundations of Social Psychology, (New York: Wiley), 1967.

D. Katz, "The Functional Approach to the Study of Attitudes." Public Opinion Quarterly, 24(1960), p. 163-204.

C. A. Kiesler, B. E. Collins and N. Miller, Attitude Change, (New York: Wiley), 1969.

S. J. Levy, "Symbols for Sale," Harvard Business Review, 37(July-August, 1959), p. 117-124.

W. McDougall, An Introduction to Social Psychology, (London: Methane and Company), 1908.

H. A. Murray, Explorations in Personality, (New York: Oxford), 1938.

R. E. Nisbett, E. Borgida, R. Crandall and H. Reed, "Popular Induction: Information is not Necessarily Informative," in J. S. Carroll and J. W. Payne (Eds.), Cognition and Social Behavior, (Hillsdale, N. J.: Lawrence Erlbaum Assoc.), 1976.

L. Rainwater, R. P. Coleman and G. Handel, Workingman's Wife, (New York: Oceana Press), 1959.

M. Rokeach, Beliefs, Attitudes and Values, (San Francisco: Jossey-Bass), 1968.

I. Ross, "Self Concept and Brand Preference," Journal of Business, 44(Jan. 1971), p. 38-50.

B. R. Schlenker and P. A. Schlenker, "Reactions Following Counterattitudinal Behavior Which Produces Positive Consequences," Journal of Personality and Social Psychology, 31(1975), 962-971.

B. R. Schlenker, "Translating Actions to Attitudes," unpublished research proposal (University of Florida, Department of Psychology), 1977.

J. T. Tedeschi, B. R. Schlenker and T. V. Bonoma, "Cognitive Dissonance: Private Ratiocination or Public Spectacle?" American Psychologist, 26(1971), p. 685-695.

R. S. Wyer, Jr., Cognitive Organization and Change: An Information Processing Approach, (Potomac, Md.: Lawrence Erlbaum Assoc.), 1974.

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