Attitudes and Behavior: the Cognitive-Affective Mismatch Hypothesis

Murray G. Millar, University of Wisconsin-Parkside
Abraham Tesser, University of Georgia
ABSTRACT - Thought about one's attitudes has inconsistent effects on the attitude-behavior relationship: Sometimes thought increases the correlation between attitudes and behaviors and sometimes thought decreases the correlation between attitudes and behaviors. One way to understand this inconsistency is to assume that the attitude reports are based on whatever aspect of the attitude is salient when the report is given. If affect is salient, then the attitude report will reflect one's feelings about the attitude object. If cognitions are salient, then one's attitude report will reflect one's beliefs about the attitude object. In short one's self-reported attitude will differ as a function of what is salient. At the same time, one's behavior may be more or less affectively or cognitively driven. Consummatory behaviors, behaviors that are engaged in for their own sake, are likely to be affectively driven. On the other hand, instrumental behaviors, behaviors intended to accomplish a goal which is independent of the attitude object, are likely to be cognitively driven. Thus, if one's affect is salient when the attitude report is given, then the attitude report will do a better job of predicting consummatory behaviors than instrumental behaviors; If one's cognitions are salient when the attitude report is given, then the attitude report will do a better job of predicting instrumental behaviors than consummatory behaviors. Evidence for these hypotheses was reviewed, and the implications of this work for a definition of attitude are discussed.
[ to cite ]:
Murray G. Millar and Abraham Tesser (1990) ,"Attitudes and Behavior: the Cognitive-Affective Mismatch Hypothesis", in NA - Advances in Consumer Research Volume 17, eds. Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, Provo, UT : Association for Consumer Research, Pages: 86-90.

Advances in Consumer Research Volume 17, 1990      Pages 86-90

ATTITUDES AND BEHAVIOR: THE COGNITIVE-AFFECTIVE MISMATCH HYPOTHESIS

Murray G. Millar, University of Wisconsin-Parkside

Abraham Tesser, University of Georgia

ABSTRACT -

Thought about one's attitudes has inconsistent effects on the attitude-behavior relationship: Sometimes thought increases the correlation between attitudes and behaviors and sometimes thought decreases the correlation between attitudes and behaviors. One way to understand this inconsistency is to assume that the attitude reports are based on whatever aspect of the attitude is salient when the report is given. If affect is salient, then the attitude report will reflect one's feelings about the attitude object. If cognitions are salient, then one's attitude report will reflect one's beliefs about the attitude object. In short one's self-reported attitude will differ as a function of what is salient. At the same time, one's behavior may be more or less affectively or cognitively driven. Consummatory behaviors, behaviors that are engaged in for their own sake, are likely to be affectively driven. On the other hand, instrumental behaviors, behaviors intended to accomplish a goal which is independent of the attitude object, are likely to be cognitively driven. Thus, if one's affect is salient when the attitude report is given, then the attitude report will do a better job of predicting consummatory behaviors than instrumental behaviors; If one's cognitions are salient when the attitude report is given, then the attitude report will do a better job of predicting instrumental behaviors than consummatory behaviors. Evidence for these hypotheses was reviewed, and the implications of this work for a definition of attitude are discussed.

THE COGNITIVE-AFFECTIVE MISMATCH HYPOTHESIS

The effects of thought on both attitudes and the attitude- behavior relation has received considerable research attention (e.g., Carver & Scheier, 1981; Fazio, Chen, McDonel, & Sherman, 1982; Tesser, 1978; Wilson, Dunn, Kraft, & Lisle, 1989). In this research thought prior to an assessment of an attitude has produced inconsistent effects on the attitude-behavior relationship. That is, sometimes thought increases the correlation between attitudes and behaviors and sometimes thought decreases the correlation between attitudes and behaviors. In a series of recent studies, Wilson and his colleagues (e.g., Wilson & Dunn, 1986; Wilson, Dunn, Bybee, Hyman, & Rotondo, 1984; Wilson, Kraft, & Dunn, 1987), have amassed considerable evidence that thought prior to an assessment of an attitude decreases the attitude's ability to predict behavior. For example, Wilson et al. (1984) required half of his participants to analyze their reasons for liking five puzzles, and the other half were not given any instructions. The participants' attitudes expressed subsequent to this procedure were correlated with behavioral measures of liking. The correlation between attitudes and behavior in the thought group was significantly lower (r = .17) than in the no-instruction group (r = .54).

Wilson et al. (1989) has explained this effect by suggesting that thinking about reasons causes persons to change their minds about how they feel, but does not change their behavior. Wilson's explanation suggests that when persons are asked to explain their feelings they feel compelled to generate reasonable sounding answers despite the fact they often do not know exactly the reasons for their feelings. Consequently, the reasons generated may be incorrect or only a biased subset of the actual reasons underlying their attitude, i.e., persons will tend to produce reasons easiest to verbalize or most available in memory. When persons are required to state their attitudes subsequent to this process they are influenced by this biased set of reasons and change their attitude in the direction of the biased sample. On the other hand, according to Wilson, the biased set of reasons has only a short term effect on behavior. That is, the person's behavior is initially influenced by the biased set then "snaps back" to the person's original position as the person's affective response reasserts itself.

In contrast to Wilson's findings, other research has suggested that thought about an attitude increases its ability to predict behavior. For example, Wicklund (1982) in his work on self-awareness, suggested that persons who are focused on themselves (i.e., presumably thinking about their internal feelings) have better access to their attitudes than persons who are not self-focused. Wicklund, unlike Wilson et al., predicted that highly self-focused persons should be less likely to be misled about their "true" attitude. Consequently, their attitudes should be good predictors of their behavior. Consistent with Wicklund's hypothesis Scheier, Buss, and Buss (1978) have found that persons dispositionally high in self-awareness exhibit higher attitude-behavior correlations than do persons low in self-awareness. Also Pryor, Gibbons, Wicklund, Fazio, and Hood (1977) reported higher attitude-behavior correlations for persons in the presence of a mirror (high self awareness) than for persons in the absence of a mirror. Similarly, Snyder has argued and found evidence that instructing persons to think carefully about an attitude increases the attitudes ability to predict behavior (Snyder & Kendzierski, 1982; Snyder & Swann, 1976). Snyder suggested that thought operates to increase both the accessibility and strength of that attitude.

Even research that used a research paradigm similar to that used in Wilson's work has produced conflicting results. For example, Fazio, Zanna, and Cooper (1978), who used the same puzzles as Wilson et al. (1984), had participants view other people performing the puzzles. One half were asked to think about how they would feel about each type of puzzle, and the other half were not given these directions. After this procedure, all of the participants were given a 15 minute play period to work on the puzzles. The attitudes of the participants who were asked to think about the puzzles were better predictors of play period behavior than were the attitudes of those not asked to think about their feelings.

Attitude Components and Thought

In order to reconcile these conflicting findings, we (Miller & Tesser, 1986, 1989) examined the relationship between thought and the different attitude components. Although, there is no universally agreed upon definition of the attitude construct, attitudes are usually conceptualized as multi-component entities (e.g., Kramer, 1949; Ostrom, 1969; Thurstone, 1928). We regard attitudes as encompassing a global evaluation based on both cognition and affect (see Zanna and Rempel (1988) for a similar conceptualization). The cognitive component is generally conceived of as containing the encoding of attributes and beliefs about the attitude object and the affective component of the attitude as containing the encoding of emotions and feelings associated with the object (Fleming, 1967). Considerable theoretical and empirical support for this dichotomy between cognition and affect has been offered by Zajonc (1980, 1984), Breckler (1984), and Breckler & Wiggins (1989)

It was our contention that attitude reports (global evaluations) are based on whatever aspect of the attitude is salient when the report is given. If affect is salient, then the attitude report will reflect one's feelings about the attitude object. If cognitions are salient, then one's attitude report will reflect one's beliefs about the attitude object. Thought prior to making an evaluation has the potential to make either the affective or cognitive component of the attitude salient and more important in the global evaluation. For example, Wilson's procedure of requiring persons to think about reasons for liking or disliking an object would tend to make the cognitive component of the attitudes salient, i.e., beliefs about the object's attributes would be salient. Alternatively, requiring persons to think about the feelings they experience in the presence of the object would make the affective component of their attitude salient. Consequently, if the affective and cognitive components of an attitude are not in perfect evaluative agreement, it is possible for thought to produce different attitude reports about the same object.

Attitude Components and Behavior

In addition, we proposed that the decision to engage in a behavior may be based more or less on the cognitive or affective component of the attitude rather than a global evaluation. That is, some types of behavior may be more cognitively driven and other types more affectively driven. There are probably a number of dimensions that would make either the attributes of the object or the feelings the object evokes more important in directing behavior. For the present work we distinguished between instrumental and consummatory behaviors. Behaviors intended to accomplish a goal which is independent of the attitude object (instrumental behavior) are likely to be cognitively driven. For example, a person who performs a puzzle in order to develop analytic ability would primarily be interested in various attributes of the puzzle and how they affect analytic ability, not how the puzzle makes him or her feel. Alternatively, behaviors engaged in for their own sake (consummatory behavior), are likely to be affectively driven. For example, a person who performs a puzzle simply to please himself, should be primarily interested in how the puzzle makes him feel, not in the attributes of the puzzle.

Mismatch Hypothesis

Since one's global evaluations or attitude report can be more or less influenced by the affective or cognitive component and since behavior can be more or less driven by the affective and cognitive components, it was hypothesized that a match between the attitude component emphasized by thought and the attitude component driving behavior would lead to a strong attitude- behavior relation and a mismatch between components would lead to a weak attitude-behavior relation.

When the previous literature dealing with the effects of thought on the attitude-behavior relation is reexamined within this framework, many of the apparently contradictory findings can be resolved. The reasons-analysis procedure used by Wilson made the cognitive component of the attitude salient, producing global evaluations based on this component, whereas the puzzle-playing behavior in Wilson's free-play situation was probably a consummatory behavior (i.e., there was little instrumental value in playing the puzzles). The mismatch between the attitude component influencing the global evaluation and the component- driving the behavior resulted in lower attitude-behavior correlations. Alternatively, the directions of Fazio et al. (1978) to concentrate on feelings probably made the affective component salient when the global evaluation was formed, whereas the puzzle-playing behavior remained a consummatory behavior. The match between the attitude components driving the behavior resulted in higher attitude-behavior correlations.

Millar and Tesser (1986) performed a preliminary test of the mismatch hypothesis by measuring the relationship between attitudes formed after either an affective or cognitive focus procedure and behavior performed under instrumental and consummatory conditions. It was expected that attitudes formed when affect was salient would predict consummatory behavior better than instrumental behavior and alternatively attitudes formed when cognition was salient would predict instrumental behavior better than consummatory behavior.

To test this participants were informed that the purpose of the study was to evaluate the effectiveness of five types of analytic puzzles that were designed to increase analytic ability. To create instrumental behavior conditions half the participants were informed that at the end of the study they would be receiving a test of their analytic ability. Participants in the instrumental conditions should view work on the puzzles as a way to improve their performance on the upcoming test, i.e., they had a goal independent of the attitude object. To create consummatory behavior conditions the other half of the participants were informed that they would receive an unrelated test of their social sensitivity. Participants in the consummatory conditions should not have this motive and engage in the behavior for its own sake. After five minutes to familiarize themselves with the puzzles, both groups were required to report their attitude about each of the five puzzles. During this five minute period, one half of the participants were given the reason- analysis directions that required the participants to analyze why they felt the way they did about each of the puzzles types. These directions were similar to those used by Wilson et al. (1984) and were designed to make the cognitive component of the attitude salient. The other half received affective-analysis directions that required the participants to analyze how they felt while performing each type of puzzle. These directions were designed to make the affective component of the attitude salient. Finally, all of the participants were given a seven minute free-play period to work on any of the puzzles they desired. During this period three behavioral measures were recorded: the time they spent working on the puzzles in the freeplay period, the order in which they choose the puzzles, and the proportion of each type of puzzle they attempted.

The results provided strong support for the hypothesis. Three indexes of attitude-behavior consistency were constructed by computing the rank-order correlation of each participant's evaluation of the puzzles with the three behavioral measures. When these within subject correlations were analyzed within an analysis of variance framework the predicted interaction between the type of thought prior to the attitude report (affective vs. cognitive) and the type of behavior performed (instrumental vs. consummatory) was obtained. Thought emphasizing the affective component produced attitude reports that predicted consummatory behavior (affectively driven behavior) better that instrumental behavior (cognitively driven behavior). Alternatively, thought emphasizing the cognitive component produced attitude reports that predicted instrumental behavior better than consummatory behavior.

Moderating Role Of Affective-Cognitive Consistency

Having obtained preliminary support for the model a further investigation was conducted to examine the moderating role of affective-cognitive consistency in the relationship between global evaluation and behavior (cf. Norman, 1975; Rosenberg, 1960). From the present perspective, if the affective and cognitive components are in good evaluative agreement then thought emphasizing either component should lead to a similar global evaluation. If this is the case, then global evaluations should relate to consummatory or instrumental behavior in the same manner regardless of what component is made salient by thought. Overall we would expect highly consistent attitudes to predict both consummatory and instrumental behavior uniformly well. On the other hand, if the cognitive and affective components are not in agreement then thought emphasizing different components would lead to different global evaluations. Consequently, if thought operates in the manner suggested by-the model, the match and mismatch effects should occur when attitudes are characterized by low affective-cognitive consistency and become attenuated when there is high affective-cognitive consistency.

To test this hypothesis Millar and Tesser (1989) attempted to measure affective-cognitive consistency directly and demonstrate that the match and mismatch effects occur only under low consistency. Each participant was required to familiarize himself/herself with different types of puzzles by performing two examples of each puzzle under affective thought conditions, i.e., think about how the puzzle makes you feel, and two examples under cognitive thought, i.e., think about why you feel the way you do about the puzzle, conditions. During both the affective focus and cognitive focus, participants were required to write down their responses to the puzzles. These responses were used to produce affective and cognitive liking scores for each type of puzzle and calculate affective-cognitive consistency. In order to manipulate affective and cognitive focus the measurement of affective and cognitive components was conducted so that half the participants finished the procedure with an affective focus and half finished with a cognitive focus. Immediately after this procedure participants were required to evaluate each of the puzzles. Following the evaluation of the puzzles, participants were allowed to play with the puzzles under the instrumental or consummatory conditions described in the first study.

We expected that when affective-cognitive consistency was low there would be an interaction between Focus of Thought and Behavior Type. That is, when there is a match between the attitude component emphasized by thought and the component driving behavior, the evaluation-behavior relation should be strong; when there is a mismatch between the focus of thought and the component driving behavior the relation between evaluation and behavior should be weak. Alternatively, when affective-cognitive consistency is high this interaction should disappear.

The results provided strong support for the predicted moderating role of affective-cognitive consistency in the attitude-behavior relation. For each participant a measure of affective-cognitive consistency was constructed by computing the rank-order correlation of his/her cognitive liking scores for each type to puzzle to his/her affective liking score for each type of puzzle. When there was low affective-cognitive consistency a match between the attitude component emphasized by thought and the component driving behavior resulted in higher evaluation-behavior correlations than a mismatch between components. Alternatively, when there was high affective- cognitive consistency the match and mismatch effects disappeared.

The results also provide support for the hypothesized relation between liking expressed by the attitude components (affective and cognitive) and behavior type (consummatory and instrumental). As we expected the liking expressed by the affective component tended to predict consummatory behaviors better than the liking expressed by the cognitive component and, alternatively, the liking expressed by the cognitive component tended to predict instrumental behavior better than consummatory behaviors.

Comparison to Cognitization Hypothesis

Wilson and his colleagues (e.g., Wilson & Dunn, 1986; Wilson, Dunn, Kraft, & Lisle, 1989) have also noted the discrepancy between their findings on the effects of thought on the attitude-behavior relation and earlier findings. Similar to our explanation Wilson and Dunn (1986) have argued that discrepancy between the thought procedures used in their work and in earlier work was responsible for the conflicting findings. That is, the reasons analysis procedure used in their work created a cognitive biased attitude whereas the thought procedures used in earlier work did not create a cognitive bias, e.g., the procedure used by Fazio, Zanna, and Cooper (1978) that required participants to think about how they feel. In Wilson's explanation if thought fails to produce a cognitive bias it will also fail to change the attitude and, consequently, the attitude and behavior will remain consistent.

This explanation brings out some important differences between our understanding and Wilson's understanding of the effects of thought on the attitude-behavior relation. Wilson's model focuses primarily on the effect of thought on affective attitudes. That is, the original attitude to be biased by cognition must be affective. Presumably attitudes based on cognition would be less susceptible to obtaining a cognitive bias than attitudes based on affect. In addition Wilson's model focuses primarily on the prediction of affectively driven behaviors. That is, except for a short time when the behavior is influenced by the cognitive bias, behavior is dependent upon the person's affect. Our model, on the other hand proposes that thought may make salient either affective or cognitive responses to the attitude object, i.e., create either affective or cognitively based attitudes, and that behavior may either be affectively or cognitively driven. Consequently, our model allows us to integrate the effects of thought on both affectively' and cognitively based attitudes and behaviors.

Having emphasized the differences it should be said that overall Wilson's and our models are not inconsistent. It seems likely that both processes occur. To distinguish the two processes future research will need to identify the type of attitudes discussed in our model (i.e., attitudes that have both affective and cognitive components) and the type of attitudes discussed in Wilson's model (i.e., attitudes that are affectively based with little or no cognition) .

Conclusion

The present model not only allows us to integrate previous attitude-behavior consistency literature but also has important implications for our understanding of the attitude construct. Many researchers have conceptualized attitudes as relatively stable an_lasting unitary evaluations (e.g., Ajzen, 1984). In ''contrast the present conceptualization suggested that at any particular moment, one has the potential for a number different attitudes toward an object. Some will be based on feelings towards the object and some based on beliefs about the object. In addition the present conceptualization emphasized the importance of immediate and changing environmental factors in determining which of the attitudes held about the object is reported. That is, different attitudes may be obtained depending on whether current environmental conditions make affect or cognition salient.

REFERENCES

Ajzen, I. (1984). Attitudes. In R.J. Corsini (Ed.), Wiley Encyclopedia of Psychology, (Vol. 1, pp. 99-100). New York: Wiley.

Breckler, S.J. (1984). Empirical validation of affect, behavior, and cognition as distinct components of attitude. Journal of Personality and Social Psychology, 47, 1191 -1205.

Breckler S.J. & Wiggins, E.C. (1989). Affect versus evaluation in the structure of attitudes. Journal of Experimental Social Psychology, 25, 253271.

Carver, C.S., & Scheier, M.F. (1981). Attention and self-regulation: A control-theory approach to human behavior. New York: Springer-Verlag.

Fazio, R.H., Chen, J., McDonel, E.C., & Sherman, S.J. (1982). Attitude accessibility, attitude-behavior consistency, and the strength of the object-evaluation association. Journal of Personality and Social Psychology, 18, 339-357.

Fazio, R.H., Zanna, M.P., & Cooper, J. (1978). Direct experience and attitude-behavior consistency: An information processing analysis. Personality and Social Psychology Bulletin, 4, 48-52.

Fleming, D. (1967). Attitude: The history of a concept. Perspectives in American History, 1, 287-365.

Kramer, B. M. (1949). Dimensions of prejudice. Journal of Psychology, 27, 389-451.

Millar, M.G., & Tesser, A. (1986). Effects of affective and cognitive focus on the attitude-behavior relation. Journal of Personality and Social Psychology, 51, 270-276.

Millar, M.G., & Tesser, A. (1989). The effects of affective-cognitive consistency and thought on the attitude-behavior relation. Journal of Experimental Social Psychology, 25, 189-202.

Norman, R. (1975). Affective-cognitive consistency attitudes, conformity, and behavior. Journal of Personality and Social Psychology,32, 83-91.

Ostrom, T.M. (1969). The relationship between the affective, behavioral, and cognitive components of attitude. Journal of Experimental Social Psychology, 5, 12-30.

Pryor, J.B., Gibbons, R.X., Wicklund, R.A., Fazio, R.H., & Hood, R. (1977). Self-focused attention and self-report validity. Journal of Personality, 45, 514-527.

Rosenberg, M.J. (1960). A structural theory of attitude dynamics. Public Opinion Quarterly, 24, 319-341.

Scheier, M.R., Buss, A.H., & Buss, D.M. (1978). Self-consciousness, self-report of aggressiveness, and aggression. Journal of Research in Personality, 12, 133-140.

Snyder, M., & Kendzierski, D. (1982). Acting on one's attitudes: Procedures for linking attitude and behavior. Journal of Experimental Social Psychology, 18, 165 - 183.

Snyder, M., & Swann, W.B., (1976). When actions reflect attitudes: The politics of impression management. Journal of Personality and Social Psychology, 34, 1034-1042.

Tesser, A. (1978). Self-generated attitude change. In L. Berkowitz (Ed.), Advances in experimental social psychology, (Vol 11, pp. 229-338), New York: Academic Press.

Thurstone, L. L. (1928). Attitudes can be measured. American Journal of Sociology, 33, 529-544.

Wicklund, R.A. (1982). Self-focused attention and the validity of self-reports. In M.P. Zanna, E.T. Higgins, & C.P. Herman (Eds.), Consistency in social behavior: The Ontario Symposium, (Vol. 2, pp .1 49 - 172). Hillsdale, NJ: Erlbaum.

Wilson, T.D., & Dunn, D.S. (1986). Disruptive effects of explaining attitudes: Moderating effects of knowledge about the attitude object, and mediating effects of affective-cognitive discrepancies. Journal of Experimental Social Psychology, 22, 249 -263.

Wilson, T.D., Dunn, D.S., Bybee, J.A., Hyman, D.B., & Rotondo, J.A. (1984). Effects of analyzing reasons on attitude-behavior consistency. Journal of Personality and Social Psychology, 47, 4-16.

Wilson, T.D., Dunn, D.S., Kraft, D., & Lisle, D.J. (1987). Introspection, attitude change, and attitude-behavior consistency: The disruptive effects of explaining why we feel the way we do. In L. Berkowitz (Ed.), Advances in experimental social psychology New York: Academic Press.

Wilson, T.D., Kraft, D.S., & Dunn, D.S. (1987). Disruptive effects of analyzing reasons: The basis of attitudes as a moderating variable. Unpublished manuscript, University of Virginia.

Zajonc, R.B. (1980). Feeling and Thinking: Preferences need no inferences. American Psychologist, 35, 151-175.

Zajonc, R.B. (1984) On the primacy of affect. American Psychologist, 36, 117-123.

Zanna, M.P., & Rempel, J.K. (1988). Attitudes: A new look at an old concept. In D. Bar-Tal, & A. Kruglanski (Eds.), The social psychology of knowledge. New York: Cambridge University Press.

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