Attitude “is probably the most distinctive and indispensable concept in contemporary American social psychology” (Allport 1985, p. 35). Hundreds of books and thousands of articles have been published on the topic. A review of this literature may be found in Eagly and Chaiken (1998). Despite this popularity, there is considerable disagreement about such basics as terminology. Several terms are frequently used as synonyms for attitude, including opinion and belief. Contemporary writers often distinguish attitudes from cognitions, which is broader and includes attitudes as well as perceptions of one’s environment. Most analysts distinguish attitude from value, the latter referring to a person’s ultimate concerns or preferred modes of conduct.
An attitude is a learned predisposition to respond to a particular object in a generally favorable or unfavorable way. Every attitude is about an object, and the object may be a person, product, idea, or event. Each attitude has three components: (1) a belief, (2) a favorable or unfavorable evaluation, and (3) a behavioral disposition. This definition is used by most contemporary writers. However, a small minority define attitude as consisting only of the positive or negative evaluation of an object.
A stereotype is one type of attitude. Originally, the term referred to a rigid and simplistic ”picture in the head.” In current usage, a stereotype is a belief about the characteristics of members of some specified social group. A stereotype may be positive (Asian Americans are good at math) or negative (women are bad at math). Most stereotypes are resistant to change.
Attitudes link the person to other individuals, groups, and social organizations and institutions. Each person has literally hundreds of attitudes, one for each significant object in the person’s physical and social environment. By implication, the individual’s attitudes should reflect his or her location in society. Thus, attitudes are influenced by gender, race, religion, education, and social class. Considerable research on the relationship between social position and attitudes has been carried out; this literature is reviewed by Kiecolt (1988).
ATTITUDE FORMATION
Many attitudes are learned through direct experience with the object. Attitudes toward one’s school, job, church, and the groups to which one belongs are examples. Attitudes toward the significant persons in one’s life are also learned in this way. More often, attitudes are learned through interactions with others. Socialization by parents, explicit teaching in educational and religious settings, and interactions with friends are important sources of attitudes. Research shows that children’s attitudes toward a variety of objects, including gender roles and political issues, are similar to those held by their parents.
Another source of attitudes is the person’s observations of the world. A topic of continuing interest is the impact of mass media on the attitudes (and behavior) of users. A thorough review of the literature on this topic (Roberts and Maccoby 1985) concludes that television viewing affects both children’s and adolescents’ attitudes about gender roles. Further, the viewing of programs intentionally designed to teach positive attitudes toward racial or ethnic minorities does increase children’s acceptance of such persons. With regard to adults, evidence supports the ”agenda setting” hypothesis; the amount and quality of coverage by the media (press, radio, and television) of an issue influences the public’s perception of the importance of that issue. The effects of mass media exposure on aggression are discussed by Felson (1996) and Geen (1998).
Stereotypes are also learned. A stereotype may arise out of direct experience with a member of the stereotyped group, for example, a person who encounters a musically talented black person may create a stereotype by overgeneralizing, inferring that all African Americans are gifted musically. More often, however, stereotypes are learned from those with whom we interact such as parents. Other stereotypes may be acquired from books, television, or film. Research indicates that television programming portrays women, the elderly, and members of some ethnic minorities in negative ways and that these portrayals create (or reinforce) misperceptions and negative stereotypes in viewers (McGuire 1985).
Social institutions influence the attitudes one learns in several ways. Adults’ ties to particular ethnic, religious, and other institutions influence the attitudes they teach their children. The instruction given in schools reflects the perspectives of the dominant political and economic institutions in society. The amount and quality of media coverage of people and events reflects the interests of particular groups in society. Through these mechanisms, the individual’s attitudes reflect the society, institutions, and groups of which she or he is a member.
Each attitude fulfills one or more of four functions for the individual. First, some attitudes serve an instrumental function: An individual develops favorable attitudes toward objects that aid or reward the individual and unfavorable attitudes toward objects that thwart or punish the individual. For example, a person who earns a large salary will have a positive attitude toward the job. Second, attitudes often serve a knowledge function. They provide the person with a meaningful and structured environment. Third, some attitudes express the individual’s basic values and reinforce self-image. Whites’ attitudes toward black Americans reflect the importance that whites place on the values of freedom and equality. Fourth, some attitudes protect the person from recognizing certain thoughts or feelings that threaten his or her self-image or adjustment.
Stereotypes also serve several functions. The act of classifying oneself as a member of a group (males, Republicans, whites) elicits the image of a contrasting group (females, Democrats, Latinos). Thus, stereotypes contribute to social identity. They also reduce the demands on the perceiver to process information about individual members of a stereotyped group; instead, one can rely on a stereotype. Finally, stereotypes may be used to justify the political and economic status quo.
ATTITUDE MEASUREMENT
Because attitudes are mental states, they cannot be directly observed. Social scientists have developed a variety of methods for measuring attitudes, some direct and some indirect.
Direct Methods. These methods involve asking the person questions and recording the answers. Direct methods include various rating scales and several sophisticated scaling techniques.
The three most frequently used rating scales are single item, Likert scales, and the semantic differential. The single-item scale usually consists of a direct positive or negative statement about the object, and the respondent indicates whether he or she agrees, disagrees, or is unsure. Such a measure is easy to score, but is not precise. A Likert scale typically involves several statements, and the respondent is asked to indicate the degree to which she or he agrees or disagrees with each. By analyzing differences in the pattern of responses across respondents, the investigator can order individuals from greatest agreement to greatest disagreement. Whereas Likert scales assess the denotative (literal) meaning of an object to a respondent, the semantic differential technique assesses the connotative (personal) meaning of the object. Here, an investigator presents the respondent with a series of bipolar adjective scales. Each of these is a scale whose poles are two adjectives having opposite meanings, for example, good-bad, exciting-boring. The respondent rates the attitude object, such as ”my job,” on each scale. After the data are collected, the researcher can analyze them by various statistical techniques.
A variety of more sophisticated scaling techniques have been developed. These typically involve asking a series of questions about a class of objects, for example, occupations, crimes, or political figures, and then applying various statistical techniques to arrive at a summary measure. These include magnitude techniques (e.g., the Thurstone scale), interlocking techniques (e.g., the Guttman scale), proximity techniques (e.g., smallest space analysis), and the unfolding technique developed by Coombs. None of these has been widely used.
Indirect Methods. Direct methods assume that people will report honestly their attitudes toward the object of interest. But when questions deal with sensitive issues, such as attitudes toward members of minority groups or abortion, respondents may not report accurately. In an attempt to avoid such reactivity, investigators have developed various indirect methods.
Some methods involve keeping respondents unaware of what is being measured. The ”lost letter” technique involves dropping letters in public areas and observing the behavior of the person who finds it. The researcher can measure attitudes toward abortion by addressing one-half of the letters to a prochoice group and the other half to a prolife group. If a greater percentage of letters to the latter group are returned, it suggests people have prolife attitudes. Another indirect measure of attitude is pupil dilation, which increases when the person observes an object she or he likes and decreases when the object is disliked.
Some indirect measures involve deceiving respondents. A person may be asked to sort a large number of statements into groups, and the individual’s attitude may be inferred from the number or type of categories used. Similarly, a respondent may be asked to write statements characterizing other people’s beliefs on an issue, and the content and extremity of the respondent’s statements are used to measure his or her own attitude. A third technique is the ”bogus pipeline.” This involves attaching the person with electrodes to a device and telling the person that the device measures his or her true attitudes. The respondent is told that some signal, such as a blinking light, pointer, or buzzer, will indicate the person’s real attitude, then the person is asked direct questions.
While these techniques may reduce inaccurate reporting, some of them yield measures whose meaning is not obvious or is of questionable validity. Does mailing a letter reflect one’s attitude toward the addressee or the desire to help? There is also evidence that measures based on these techniques are not reliable. Finally, some researchers believe it is unethical to use techniques that involve deception. Because of the importance of obtaining reliable and valid measures, research has been carried out on how to ask questions. This research is reviewed by Schuman and Presser (1996).
For a comprehensive discussion of attitude measurement techniques and issues, see Dawes and Smith (1985).
ATTITUDE ORGANIZATION
An individual’s attitude toward some object usually is not an isolated psychological unit. It is embedded in a cognitive structure and linked with a variety of other attitudes. Several theories of attitude organization are based on the assumption that individuals prefer consistency among the elements of cognitive structure, that is, among attitudes and perceptions. Two of these are balance theory and dissonance theory.
Balance Theory. Balance theory, developed by Heider, is concerned with cognitive systems composed of two or three elements. The elements can be either persons or objects. Consider the statement ”I will vote for Mary Sweeney; she supports parental leave legislation.” This system contains three elements—the speaker, P; another person (candidate Mary Sweeney), O; and an impersonal object (parental leave legislation), X. According to balance theory, two types of relationships may exist between elements. Sentiment relations refer to sentiments or evaluations directed toward objects and people; a sentiment may be either positive (liking, endorsing) or negative (disliking, opposing). Unit relations refer to the extent of perceived association between elements. For example, a positive unit relation may result from ownership, a relationship (such as friendship or marriage), or causality. A negative relation indicates dissociation, like that between ex-spouses or members of groups with opposing interests. A null relation exists when there is no association between elements.
Balance theory is concerned with the elements and their interrelations from P’s viewpoint. In the example, the speaker favors parental leave legislation, perceives Mary Sweeney as favoring it, and intends to vote for her. This system is balanced. By definition, a balanced state is one in which all three relations are positive or in which one is positive and the other two are negative. An imbalanced state is one in which two of the relationships between elements are positive and one is negative or in which all three are negative. For example, ”I love (+) Jane; Jane loves (+) opera; I hate (-) opera” is imbalanced.
The theory assumes that an imbalanced state is unpleasant and that when one occurs, the person will try to restore balance. There is considerable empirical evidence that people do prefer balanced states and that attitude change often occurs in response to imbalance. Furthermore, people maintain consistency by responding selectively to new information. There is evidence that people accept information consistent with their existing attitudes and reject information inconsistent with their cognitions. This is the major mechanism by which stereotypes are maintained.
Dissonance Theory. Dissonance theory assumes that there are three possible relationships between any two cognitions. Cognitions are consistent, or consonant, if one naturally or logically follows from the other; they are dissonant when one implies the opposite of the other. The logic involved is psycho logic—logic as it appears to the individual, not logic in a formal sense. Two cognitive elements may also be irrelevant; one may have nothing to do with the other.
Cognitive dissonance is a state of psychological tension induced by dissonant relationships between cognitive elements. There are three situations in which dissonance commonly occurs. First, dissonance occurs following a decision whenever the decision is dissonant with some cognitive elements. Thus, choice between two (or more) attractive alternatives creates dissonance because knowledge that one chose A is dissonant with the positive features of B. The magnitude of the dissonance experienced is a function of the proportion of elements consonant and dissonant with the choice. Second, if a person engages in a behavior that is dissonant with his or her attitudes, dissonance will be created. Third, when events disconfirm an important belief, dissonance will be created if the person had taken action based on that belief. For example, a person who buys an expensive car in anticipation of a large salary increase will experience dissonance if she or he does not receive the expected raise.
Since dissonance is an unpleasant state, the theory predicts that the person will attempt to reduce it. Usually, dissonance reduction involves changes in the person’s attitudes. Thus, following a decision, the person may evaluate the chosen alternative more favorably and the unchosen one more negatively. Following behavior that is dissonant with his or her prior attitude, the person’s attitude toward the behavior may become more positive. An alternative mode of dissonance reduction is to change the importance one places on one or more of the attitudes. Following a decision, the person may reduce the importance of the cognitions that are dissonant with the choice; this is the well-known ”sour grapes” phenomenon. Following disconfirmation of a belief, one may increase the importance attached to the disconfirmed belief. A third way to reduce dissonance is to change behavior. If the dissonance following a choice is great, the person may decide to choose B instead of A. Following disconfirmation, the person may change behaviors that were based on the belief.
Numerous books and hundreds of articles about dissonance theory have been published since it was introduced by Festinger in 1957. There is a substantial body of research evidence that supports various predictions from and elaborations of the theory. Taken together, this literature has produced a detailed taxonomy of situations that produce dissonance and of preferred modes of dissonance reduction in various types of situations.
ATTITUDE STABILITY AND ATTITUDE CHANGE
Both balance and dissonance theories identify the desire for consistency as a major source of stability and change in attitudes. The desire to maintain consistency leads the individual either to interpret new information as congruent with his or her existing cognitions (assimilation) or to reject it if it would challenge existing attitudes (contrast). This process is very important in preserving stability in one’s attitudes. At the same time, the desire for consistency will lead to attitude change when imbalance or dissonance occurs. Dissonance theory explicitly considers the link between behavior and attitudes. It predicts that engaging in counterattitudinal behavior may indirectly affect attitudes. This is one mechanism by which social influences on behavior may indirectly affect attitudes. This mechanism comes into play when the person experiences changes in roles and the requirements of the new role are inconsistent with his or her prior attitudes.
The classic perspective in the study of attitude change is the communication-persuasion paradigm, which grew out of the work by Hovland and his colleagues at Yale University. Persuasion is defined as changing the beliefs or attitudes of a person through the use of information or argument. Attempts at persuasion are widespread in everyday interaction, and the livelihood of advertisers and political consultants. According to the paradigm, each attempt involves source, message, target, and context. Thousands of empirical studies, many of them experiments done in laboratory settings, have investigated the influence of variations in these four components on the outcome of an attempt. In general, if the source is perceived as an expert, trustworthy, or physically attractive, the message is more likely to produce attitude change.
Thus, it is no accident that magazine and television commercials feature young, attractive models. Message variables include the extent of discrepancy from the target’s attitude, whether it arouses fear, and whether it presents one or both sides. Under certain conditions, highly discrepant, fear-arousing, and one-sided messages are more effective. Target factors include intelligence, self-esteem, and prior experience and knowledge. The most researched contextual factor is mood. For a review of this research, see Perloff (1993).
In the 1990s considerable work built upon the elaboration-likelihood model proposed by Petty and Cacioppo (1986). This model identifies two basic routes through which a message may change a target’s attitudes: the central and the peripheral. Persuasion via the central route occurs when a target scrutinizes the arguments contained in the message, interprets and evaluates them, and integrates them into a coherent position. This process is termed elaboration. In elaboration, attitude change occurs when the arguments are strong, internally coherent, and consistent with known facts. Persuasion via the peripheral route occurs when, instead of elaborating the message, the target pays attention primarily to extraneous cues linked to the message. Among these cues are characteristics of the source (expertise, trustworthiness, attractiveness), superficial characteristics of the message such as length, or characteristics of the context such as response of other audience members. Several factors influence whether elaboration occurs. One is the target’s involvement with the issue; if the target is highly involved with and cares about the issue addressed by the message, he or she is more likely to elaborate the message. Other factors include whether the target is distracted by noise or some other aspect of the situation, and whether the target is tired. Which route a message elicits in attitude change is important. Attitudes established by the central route tend to be more strongly held and more resistant to change because the target has thought through the issue in more detail.
The literature on attitude change flourished in the 1990s. The effects of many variables have been studied experimentally. A review of the research conducted in 1992 to 1995 concluded that any one variable may have multiple effects, depending upon other aspects of the persuasion attempt (Petty, Wegener, and Fabrigar 1997). For example, consider the effect of mood on attitude change. Assuming that happy people are more open to new information, one might predict that persuasive messages would lead to greater attitude change in happy persons than in sad ones. However, research indicates that happy people spend less time processing persuasive messages than persons in neutral moods, and so may be less influenced by them. The hedonic contingency hypothesis states that the effect of mood depends upon the hedonic tone of the message. Happy people want to maintain their happy mood, so they are likely to scrutinize and process happy messages but not sad ones. Sad people often want to change their mood, and most messages will improve it, so their processing is not affected by messages’ hedonic tone. The results of two experiments support the hypothesis (Wegener, Petty, and Smith 1995). Thus, the effect of mood on persuasion depends on whether the message is uplifting or depressing.
ATTITUDE-BEHAVIOR RELATION
The attitude-behavior relation has been the focus of considerable research since the early 1970s. This research has identified a number of variables that influence the extent to which one can predict a person’s behavior from his or her attitudes.
Some of these variables involve the measurement of the attitude and of the behavior. The correspondence of the two measures is one such variable: one can predict behavior more accurately if the two measures are at the same level of specificity. An opinion poll can predict the outcome of an election because there is high correspondence between the attitude (”Which candidate do you prefer for mayor in next month’s election?”) and the behavior (voting for a candidate in that election). The length of time between the measure of attitude and the occurrence and measure of the behavior is also an important variable. The shorter the time, the stronger the relationship. The longer the elapsed time, the more likely the person’s attitude will change, although some attitudes are stable over long periods, for example, twenty years.
The characteristics of the attitude also influence the degree to which one can predict behavior from it. In order for an attitude to influence behavior, it must be activated, that is, brought from memory into conscious awareness. An attitude is usually activated by a person’s exposure to the attitude object. Attitudes vary in accessibility, the ease with which they are activated. The more accessible an attitude is, the more likely it is to guide future behavior (Kraus 1995). Another variable is the source of the attitude. Attitudes based on direct experience with the object are more predictive of behavior. The certainty or confidence with which the person holds the attitude also moderates the attitude-behavior relationship.
The attitude-behavior relation is also influenced by situational constraints—the social norms governing behavior in a situation. An attitude is more likely to be expressed in behavior when the behavior is consistent with these norms.
An important attempt to specify the relationship between attitude and behavior is the theory of reasoned action, developed by Fishbein and Ajzen (1975). According to this theory, behavior is determined by behavioral intention. Behavioral intention is determined by two factors: attitude and subjective norm. Attitude is one’s beliefs about the likely consequences of the behavior and one’s evaluation—positive or negative—of each of those outcomes. Subjective norm is the person’s belief about other important persons’ or groups’ reactions to the behavior and the person’s motivation to comply with the expectations of each. One of the strengths of the theory is this precise specification of the influences on behavioral intention. It is possible to measure quantitatively each of the four components (likely consequences, evaluation, likely reactions, motivation to comply) and use these to make precise predictions of behavior. Many empirical studies report results consistent with such predictions. The theory applies primarily to behavior that is under conscious, volitional control.
On the other hand, researchers have shown that attitudes can affect behavior without being brought into conscious awareness (Bargh 1996). Attitudes toward objects influence our judgments and behavior toward those objects without conscious awareness or intent. Stereotypes of social groups are often activated automatically, as soon as an individual is perceived as a member of the group. Automatic processing is more likely when the individual experiences information overload, time pressures, or is not interested in engaging in effortful processing.
ATTITUDE AS INDICATOR
Increasingly, attitudes are employed as indicators. Some researchers use attitude measures as indicators of concepts, while others study changes in attitudes over time as indicators of social change.
Indicators of Concepts. Measures of specific attitudes are frequently used as indicators of more general concepts. For example, agreement with the following statement is interpreted as an indicator of powerlessness: ”This world is run by the few people in power, and there is not much the little guy can do about it.” Powerlessness is considered to be a general orientation toward the social world and is a sense that one has little or no control over events. Feelings of powerlessness may be related to such varied behaviors as vandalism, not voting in elections, and chronic unemployment.
Attitude measures have been used to assess many other concepts used in the analysis of political attitudes and behavior. These include the liberalism-conservatism dimension, political tolerance (of radical or unpopular groups), trust in or disaffection with national institutions, and relative deprivation. (For a review of this literature, see Kinder and Sears 1985.) Attitude measures are used to assess many other characteristics of persons. In the realm of work these include occupational values, job satisfaction, and leadership style.
A major concern when attitudes are employed as indicators is construct validity, that is, whether the specific items used are valid measures of the underlying concept. In the powerlessness example, the connection between the content of the item and the concept may seem obvious, but even in cases like this it is important to demonstrate validity. A variety of analytic techniques may be used, including interitem correlations, factor analysis, and LISREL.
Indicators of Social Change. Two methodological developments have made it possible to use attitudes to study social change. The first was the development of probability sampling techniques, which allow the investigator to make inferences about the characteristics of a population from the results obtained by surveying a sample of that population. The second is the use of the same attitude measures in surveys of representative samples at two or more points in time.
A major source of such data is the General Social Survey (GSS), an annual survey of a probability sample of adults. The GSS repeats a core set of items on a roughly annual basis, making possible the study of changes over a period of thirty years. Many of these items were drawn directly from earlier surveys, making comparisons over a forty- or fifty-year timespan possible. A published book describes these items and presents the responses obtained each time the item was used (Niemi, Mueller, and Smith 1989). Other sources of such data include the National Election Studies and the Gallup Polls.
This use of attitude items reflects a general concern with social change at the societal level. The investigator uses aggregate measures of attitudes in the population as an index of changes in cultural values and social institutions. Two areas of particular interest are attitudes toward race and gender roles. In both areas, efforts have been made to improve access to educational programs, jobs, and professions, increase wages and salaries, and provide greater opportunity for advancement. The availability of responses to the same attitude items over time allows us to assess the consistency between these social changes and attitudes in the population. Consider the question ”Do you think civil rights leaders are trying to push too fast, are going too slowly, or are moving at about the right speed?” This question was asked in surveys of national samples every two years from 1964 to 1976 and in 1980. The percentage of whites replying ”too fast” declined from 74 percent in 1964 to 40 percent in 1980 (Bobo 1988), suggesting increased white support for the black movement. In general, research indicates that both racial and gender-role attitudes became more liberal between 1960 and 1990, and this finding is consistent with the social changes in these areas. Other topics that have been studied include attitudes toward abortion, social class identification, and subjective quality of life.
There are several issues involved in this use of attitude items. The first is the problem of ”nonattitudes.” Respondents may answer survey questions or endorse statements even though they have no attitude toward the object. In fact, when respondents are questioned about fictional objects or organizations, some of them will express an opinion. Schuman and Kalton (1985) discuss this issue in detail and suggest ways to reduce the extent to which nonattitudes are given by respondents.
The second issue involves the interpretation of responses to items. In the example above, the analyst assumes that white respondents who reply ”Too fast” feel threatened by the movement. However, there is evidence that small changes in the wording of survey items can produce substantial changes in aggregate response patterns. This evidence and guidelines for writing survey items are discussed in Schuman and Presser (1996).
Finally, there is the problem of equivalence in meaning over time. In order to make meaningful comparisons across time, the items need to be the same or equivalent. Yet over time the meaning of an item may change. Consider the item ”Are you in favor of desegregation, strict segregation, or something in between?” This question was asked of national samples in 1964 and every two years from 1968 to 1978. From 1964 to 1970, the percentage of white, college-educated adults endorsing desegregation increased; from 1970 to 1978, the percentage decreased steadily. Until 1970, desegregation efforts were focused on the South; after 1970, desegregation efforts focused on school integration in northern cities. Evidence suggests that endorsement of desegregation changed because the meaning of the question for white adults changed (Schuman, Steeh, and Bobo 1985).
