• Tidak ada hasil yang ditemukan

CONCLUSION

Dalam dokumen Attitudes, Personality and Behavior (Halaman 155-160)

CONCLUSION 143

and measured at various levels of generality or specificity. Aggregation of responses across time, contexts, targets, or actions - or across a combination of these elements - permits inferences of dispositions at varying levels of generality. Inferred dispositions can range from the tendency to perform a single action (over time) to the tendency to engage in a broad range of actions, as reflected in a multiple-act aggregate. Even when they address the same content domain, two measures can be considered indicators of the same disposition only if they correspond in their levels of generality.

And it is only in the presence of such compatibility that behavioral con­

sistencies manifest themselves reliably. The realization that measures of global attitudes and personality traits, obtained by means of responses to questionnaires, are compatible only with equally general, broadly aggre­

gated measures of other types of responses has helped to clarify much of the initially baffling lack of predictive validity.

It is no longer very meaningful to ask whether attitudes and personality traits predict behavior - they clearly do. Nor does the crucial issue have to do with the conditions under which attitudes and personality traits are related to behavior. Instead, the literature poses and provides answers to three interrelated questions. First, is there consistency between different observations of behavior? Second, do verbal responses predict nonverbal behavior? Finally, are general behavioral dispositions related to specific response tendencies?

BEHAVIORAL CON S I S TENCY

The answer to the question of behavioral consistency across observations is closely tied to the principle of aggregation. Generally speaking, observa­

tions of single actions on individual occasions do not correlate well with each other. Too many factors unique to a given occasion prevent emergence of a clear response tendency. However, by aggregating observations of a given behavior across occasions we obtain a stable measure of the dis­

position to perform the behavior in question. Temporal stability is in fact found to become quite high with aggregation over a sufficient number of observations.

There is also evidence for consistency between behavioral measures that aggregate across different actions, so long as each aggregate assesses the same broad underlying disposition. We can infer broad dispositions from representative samples of behaviors performed in a variety of situations, and multiple measures of this kind tend to correlate highly with each other.

The aggregation solution to the consistency dilemma of course limits the explanatory and predictive utilities of traits and attitudes to broad classes of responses; it does not provide a means for the prediction of tendencies to engage in specific behaviors. This limitation, however, is per­

haps of more concern in applied attitude research, where the aim often is to predict specific actions, than it is in personality research, which deals largely with broad response tendencies. The principle of compatibility overcomes the limitation in the attitude domain by permitting attitudes to be reduced to the level of individual behaviors.

GEN ERAL DI S POSITI O N S A N D S PECI F I C ACTI O N S

A s a general rule, broad response dispositions are poor predictors o f specific actions. This is perhaps the most important lesson to be learned from the prolonged consistency controversy, but also perhaps the most difficult to accept. It would indeed be very convenient if we could measure general attitudes or personality traits and use the resulting scores to predict any behavior that appears relevant to the disposition in question.

Unfortunately, both theory and empirical findings negate this possibility.

The attempt to link broad behavioral dispositions to specific response tendencies by means of moderating variables has produced some interest­

ing studies, but future progress along these lines faces serious difficulties.

One drawback of this approach is the sheer number of personal and situational factors that can potentially moderate the effect of attitudes or personality on behavior, as well as the complications introduced by higher-order interactions among these factors. In the attitudes domain, this problem was alleviated in part by development of the MODE model (Fazio 1990a; Fazio and Towles-Schwen 1999) which links the effects of many moderating variables to the concept of attitude strength or accessibility in memory. According to this view, only strong, highly accessible attitudes are likely to be guide behavior. Variables such as self-monitoring tendency, need for cognition, involvement with the attitude object, confidence in one's attitude, and direct experience with the attitude object are assumed to moderate the relation between attitudes and behavior because they influence the attitude's accessibility in memory.

The MODE model offers a useful conceptual framework to think about the effects of general attitudes on specific behaviors and about the role of moderating variables. However, the moderating variables approach must ultimately end in failure because it leads to the unavoidable conclusion that behavior can be predicted from attitudes or personality traits only for some individuals under a limited set of circumstances. The multitude of conditions that moderating variables place on prediction of specific responses from broad dispositions severely limits the practical utility of this approach.

VERBAL AND NONVERBAL RESPONSES

What people say and what they do are not always the same (Deutscher 1966, 1973) . In part, this is a problem of measurement validity. The validity of verbal responses has often been questioned because of the possible presence of social desirability biases, acquiescence tendencies, strategic biases, and so forth. The current popularity of implicit measurement tech­

niques, such as the Implicit Association Test and sequential priming (Fazio and Olson 2003) attest to the continuing concern with measurement validity. Somewhat less attention has been given to the measurement implications of the fact that observed actions may be equally biased to

CONCLUSION 145

create favorable impressions, to avoid conflict, or to gain an advantage by means of ingratiation. The consequence of such biases is to invalidate the measures from which behavioral dispositions are inferred. If the biases associated with a verbal response differ greatly from the biases operating on the physical action, correlations between the two measures will necessarily deteriorate.

The potential for biased responding does not, however, doom efforts to predict nonverbal from verbal behaviors. Many situations provide little incentive for strong biases, and tendencies toward biased responding can be further reduced by careful application of appropriate measurement procedures. The question therefore remains, what is the relation between valid, relatively unbiased verbal and nonverbal responses? The answer to this question is related to the principle of compatibility, and it has nothing to do with the fact that one indicator of the disposition is verbal and the other nonverbal. Instead, the answer revolves around the generality or specificity of the measures involved. As noted above, verbal measures of broad attitudes or personality traits have been shown to predict equally broad, multiple-act measures of overt behavior. However, as a general rule, they do not predict specific responses, whether nonverbal or verbal. To predict single actions, we must turn to dispositional measures that deal specifically with those actions. The concept of intention appears to be a useful starting point. Intentions represent behavioral dispositions that conceptually are very closely tied to the corresponding behavior, and there is good evidence to show that many behaviors are sufficiently under volitional control to be predictable from people's intentions. Barring unforeseen events, people tend to carry out their behavioral plans. Added difficulties arise when achievement of a behavioral goal is at least partly determined by internal or external factors over which a person has only limited control. The theory of planned behavior described in Chapter 6 represents an attempt to account for the formation of intentions and the achievement of behavioral goals. Attitudes toward the behavior, subjective norms, and perceived behavioral control are the three primary deter­

minants of intentions. Their formation is traced, respectively, to beliefs about the behavior's likely outcomes, beliefs about the expectations of important others, and beliefs about factors that may facilitate or hinder performance of the behavior. When people are aware of potential diffi­

culties, they are assumed to plan their actions accordingly. The theory of planned behavior is thus designed to permit prediction and explanation of behavioral achievement by taking into account motivational antecedents, reflected in intentions, as well as other factors that are only partly under volitional control, factors that are reflected in perceived behavioral control.

A great deal of research over the past 20 years has provided good support for the theory.

This book ends the way it began, with the dispositional prediction of human behavior. It should be clear by now that dispositional concepts are not only useful, they are indispensable tools at the disposal of the behavioral scientist. While measures of behavioral dispositions cannot be used indiscriminately, when appropriately employed they yield highly valuable information. As intuitive observation would suggest, people are quite consistent in the patterns of behavior they exhibit. They act in ways

that cannot be described as capricious, nor would it be accurate to claim that their behavior is controlled by external forces. Instead, human action is found to follow reasonably and consistently from relevant behavioral dispositions.

Dalam dokumen Attitudes, Personality and Behavior (Halaman 155-160)

Dokumen terkait