Attitude Structure and Function

Anthony R. Pratkanis, Steven J. Breckler, Anthony G. Greenwald, Anthony R. Pratkanis, Steven J. Breckler, Anthony G. Greenwald

• Publisher: Psychology Press
• Year: 2014
• ISBN:
  9781317766575
• Edition: 1
• Topic: Psicología
• Subtopic: Historia y teoría en psicología

Utilizing "new wave" research including new psychological theories, new statistical techniques, and a stronger methodology, this collection unites a diversity of recent research perspectives on attitudes and the psychological functions of an attitude. The objective of the editors was to bring together the bits and pieces of validated data into one systematic and adequate set of general principles leading to the view of attitudes as predictions. As the volume reformulates old concepts, explores new angles, and seeks a relationship among various sub-areas, it also shows improvements in the sophistication of research designs and methodologies, the specifications of variables, and the precision in defining concepts.

1

Why Are Attitude Important?

Anthony G. Greenwald

University of Wasington

Before reading beyond this paragraph, the reader might try to answer the question stated as the title of this chapter. Some relevant background starts with Allport’s (1935) declaration that attitude is “social psychology’s most indispensable concept.” Allport apparently regarded the importance of attitudes as being so evident that it was not necessary to detail the basis for his assertion of its importance. Subsequent reviewers have often followed Allport’s lead, resting the case for importance of the attitude construct chiefly on its great popularity (e.g., DeFleur & Westie, 1963; Doob, 1947; Fishbein & Ajzen, 1975; McGuire, 1969). However, if the construct of attitude is indeed of major importance, then there must be some important phenomena of social behavior that cannot be explained (or, at least, cannot easily be explained) without appealing to attitudes. But what are the phenomena that would be difficult to explain if attitude were stricken from the psychologist’s vocabulary? (Here is where the reader can try to answer the question, before reading further.)

WHAT ANSWERS HAVE BEEN OFFERED?

An explanation of the importance of attitudes is not readily found in scholarly reviews or texts. More accurately, the four types of answers that one finds in the literature turn out to be unsatisfying. These are:

Attitudes Are Pervasive

This observation is accurate, as can be verified by noticing (a) the ease with which people report evaluative reactions to a wide variety of objects, (b) the difficulty of identifying categories of objects within which evaluative distinctions are not made, and (c) the pervasiveness of an evaluative component in judgments of meaning (Osgood, Suci, & Tannenbaum, 1957). Nevertheless, the pervasiveness of attitudes is itself not a reason for concluding that attitudes are important in explaining social behavior. As Bem (1967) suggested, attitudes might be cognitive illusions that are constructed after the fact of behavior.

Attitudes Predict Behavior Toward Their Objects

An important, early critique of the usefulness of attitudes in predicting behavior was given by LaPiere (1934). (Problems with LaPiere’s critique are reviewed later in this chapter.) Thirty years later, Festinger (1964) critically noted the lack of published support for the reasonable expectation that changes in attitudes should lead to changes in behavior toward their objects. Subsequently, Wicker (1969) reviewed a body of research that revealed only weak correlations between measures of attitudes and measures of behavior toward their objects.

In the 1970s and 1980s two major programs of research succeeded in clarifying attitude-behavior relations. The first of these, directed by Martin Fishbein and lcek Ajzen (e.g., 1974; see chap. 10 in this volume), demonstrated that attitude and behavior are correlated (a) when the observed behavior is judged to be relevant to the attitude, (b) when attitude and behavior are observed at comparable levels of specificity, and (c) when mediation of the attitude-behavior relation by behavioral intentions is taken into account. The second major program, directed by Russell Fazio (e.g., 1986; see chap. 7 in this volume), showed that attitude and behavior, and changes therein, are correlated (a) when the attitude is based on direct experience with the attitude object, and (b) to the extent that the attitude is cognitively accessible.

Although the successful Fishbein–Ajzen and Fazio research programs have established that attitudes can and do predict behavior toward their objects, these programs have also placed important qualifying conditions on the attitude-behavior relationship. Attitude-behavior relations do not appear to be sufficiently powerful or robust to establish the importance of attitude as a theoretical construct. (Further discussion of attitude-behavior relations is found in chap. 3, 7, and 9 of this volume.)

Attitudes Are a Selective Force in Perception and Memory

It has long been supposed that perceptual and cognitive processes are guided by attitudes. The two most-often-stated principles regarding attitude-guided information processing are that persons selectively (a) seek information that agrees with their attitudes while avoiding disagreeing information (e.g., Festinger, 1957), and (b) remember attitude-agreeable information in preference to disagreeable information (e.g., Levine & Murphy, 1943). However, the empirical basis for both of these hypothesized distortions of perception and memory was sharply questioned in the 1960s (e.g., Freedman & Sears, 1965; Greenwald & Sakumura, 1967; Waly & Cook, 1966). It now appears that these selective effects on information seeking and memory occur only under rather limited circumstances (see Wicklund & Brehm, 1976). Consequently, these phenomena do not establish the importance of attitude as a theoretical construct. (Chap. 4 in this volume gives a more detailed review of the role of attitudes in cognitive processes, including evidence for substantial effects on cognitive processes more complex than the seeking and remembering of agreeable information.)

Attitudes Serve Various Psychic Functions

The most direct attention to the importance of attitudes was given in the functional analyses of Smith, Bruner, and White (1956) and Katz (1960); they proposed that attitudes serve functions designated by labels such as utilitarian, social adjustment, object appraisal, knowledge, value expression, and ego-defense. Because these functional theories generated little research, claims for functions of attitudes remain largely unsubstantiated. (The poverty of empirical support for attitude functions is only recently beginning to be addressed, with the initiation of research programs such as those described in chap. 4, 6, 12, 13, 14, and 15 in this volume.)

WHY HAS IT BEEN SO DIFFICULT TO DEMONSTRATE THE IMPORTANCE OF ATTITUDES?

Answering this question depends on understanding the relation of attitudes to behavior. Of the three answers to be suggested here, only the first encourages satisfaction with the current understanding of attitudes in relation to behavior.

Difficulty 1: Ordinary Situations Are Attitudinally Complex

In expecting attitudes to predict behavior toward their objects, researchers have often assumed that only a single attitude should be operative in the situation on which their research focused. This assumption is often implausible. LaPiere’s (1934) research, which played an important role in criticism of the attitude construct, is used here to illustrate a setting complicated not only by uncertain identification of the focal attitude object, but also by multiple attitude objects beyond the one focal for the researcher.

Uncertain Identification of the Focal Object

In LaPiere’s study, he and a young Chinese couple traveled widely in the United States, seeking accommodation at many hotels and restaurants while observing the hotel and restaurant proprietors’ attitudes toward Chinese (assessed with a mailed questionnaire) and their behavior of providing accommodations or service to the Chinese couple. LaPiere assumed that the salient attitude object was “members of the Chinese race.” However, the couple (who were described as “personable” and “charming”) could also have been identified as customers, as middle-class persons, as a young married couple, and so forth. There is little justification for assuming that the only (or even the most) salient attitude-object identification was “members of the Chinese race.”

Multiple Attitude Objects

A restaurant proprietor might be concerned that an unpleasant scene with the young Chinese couple could intrude on the meals of other patrons or harm the reputation of the restaurant. The proprietor’s behavior toward the Chinese couple might therefore be as much (or more) influenced by attitudes toward those other objects (i.e., other patrons, the restaurant) as by attitudes toward the young couple. When, as in this situation, additional objects are important, attitude toward the presumably focal object should not dominate the prediction of behavior.

LaPiere’s research is not an isolated example of the problems that (a) objects of behavior are difficult to identify in compact verbal labels, and (b) multiple attitude objects are potentially salient.¹ When these problems characterize a research situation, the attitude measured by the researcher should predict behavior only weakly, if at all. Some solutions are to limit research on attitude-behavior relations to objects that are easily identifiable verbally, and to use behavior assessments that lack multiple potential attitude objects. The research successes of Fazio (1986) and of Fishbein and Ajzen (1974) were achieved in large part through such limitations.

Importance of a Phenomenon in Relation to the Difficulty of Demonstrating It

Because strong attitude-behavior correlations are difficult to produce, it may appear that attitudes are only weakly connected to behavior. On the contrary, however, the need for well-controlled research settings to demonstrate strong attitude-behavior relations may mean only that the influence of attitudes on behavior is so pervasive that it is difficult to observe the isolated effect of a single attitude. (An example of a parallel point is the difficulty of observing a classically conditioned response in isolation; in ordinary situations, such as eating a meal, classically conditioned responses are certainly important but are not easily observed due to masking by multiple other conditioned responses.) In general, the difficulty of demonstrating a phenomenon in research is irrelevant to a conclusion about its importance; the difficulty may mean only that the phenomenon is typically embedded in an obscuring degree of complexity.

Difficulty 2: The Concept of Attitude Needs to Be Refined

The Conception of “Attitude Object”

Collectively, and for the most part also individually, attitude researchers have treated virtually any nameable or describable entity as an attitude object. One can find studies of attitudes toward (a) sensory qualities (colors, odors, textures), (b) concrete objects (animals, persons, places, foods), (c) abstract concepts (personality traits, subjects of academic study), (d) verbal statements (beliefs, opinions, policies), (e) systems of thought (aesthetic styles, ideologies), (f) actions (e.g., drinking alcohol, sexual behavior), and even (g) attitudes (e.g., an attitude toward prejudice). The conceptual tolerance represented by this breadth is surely to be encouraged in the early stages of a concept’s development. However, the present breadth of the attitude concept may now be an obstacle to theoretical development. That is, the cost that is exchanged for this benefit of breadth may be a lack of precision.

The Three-Component Definition

In noting the variety of definitions that have been offered for attitude, previous reviewers have often been reluctant to suggest that one definition is superior to others. Accordingly, many reviewers have supported definitions that (a) permit a broad array of research operations for attitude measurement, and (b) put no apparent boundaries on the sort of entity that can be regarded as the object of an attitude (e.g., Allport, 1935; DeFleur & Westie, 1963; Greenwald, 1968b). The definition that has been most attractive to social psychologists, perhaps because of both its breadth and its ancient philosophical roots, conceives attitude as having three components—affective, cognitive, and conative (or behavioral).

We here indicate that attitudes are predispositions to respond to some class of stimuli with certain classes of responses and designate the three major types of response as cognitive, affective, and behavioral. (Rosenberg & Hovland, 1960, p. 3)

… attitudes [are] enduring systems of positive or negative evaluations, emotional feelings, and pro or con action tendencies with respect to social objects. (Krech, Crutchfield, & Ballachey, 1962, p. 139)

The three-component definition has achieved widespread adoption and almost no criticism. The one active line of criticism has questioned the nature of relationships among the three hypothesized components (cf. Breckler, 1984; Kothandapani, 1971; Ostrom, 1969). This gentle treatment notwithstanding, a harsh evaluation of the three-component definition may be warranted—it appears to have bred confusions that have weakened the attitude construct. Chief among these confusions is that associated with investigations of attitude-behavior relationships. (See Zanna & Rempel, in press, for a similar conclusion about the three-component definition.)

Consider that the following four types of operations involving action in relation to an attitude object can serve equally either to measure the conative (behavioral) component of an attitude or to measure behavior that is presumably under the control of that attitude component: (a) observations of overt action, (b) verbal self-report of past action, (c) self-report of intentions regarding action, and (d) endorsement of statements about hypothetical actions. With this range of operations, a single research investigation can serve to test (a) the attitude-behavior relationship, (b) relations of the conative to other attitude components, or (c) the relation between behavior and the conative component of attitude. By affording this multiplicity of interpretations, the three-component definition appears to permit too broad an array of interpretations for a given set of data. Additionally, the three-component definition implicitly promotes the (below questioned) supposition that the chief behavioral impact of an attitude should be on behavior toward the attitude’s object. (Further discussion and critique of the three-component definition is found...