[advances in experimental social psychology] advances in experimental social psychology volume 31...

79
Charles G. Lord Mark R. Lepper REPRESENTATION I. Introduction “Attitude,” as Gordon Allport (1935) saw it in his classic chapter in the first Handbook of Social Psychology, was the “most distinctive and most indispensable concept in contemporary social psychology” (p. 798). Al- though one might quarrel today with the extremity of his claim, there is no doubt that the attitude concept has remained ubiquitous in, and characteristic of, social psychology in the intervening 60 years. From the early days of Allport and his colleagues through the glory days of Hovland’s Yale Attitude Change Program (e.g., Hovland, Janis, & Kelley, 1953; Kiesler, Collins, & Miller, 1969; McGuire, 1969) to the more complex and sophisticated models of more recent years (e.g., Ajzen & Fishbein, 1980; Eagly & Chaiken, 1993; Fishbein & Ajzen, 1975; Olson & Zanna, 1993; Petty & Cacioppo, 1981, 1986; Zanna & Rempel, 1988), attitudes have indeed remained a central topic in social psychology. Crucial to the overwhelming success of the attitude concept were two key ingredients. The first was the implied stability of most important social attitudes, the second the presumed generality of the evaluative tendencies these attitudes involved. Thus, basic social attitudes were commonly seen as likely to persist, at least in the absence of concerted efforts to change them, over time. Attitudes were therefore distinguished by early theorists from other more ephemeral influences on a person, such as passing mood states or the momentary salience of particular stimuli in a person’s thoughts or surroundings. Indeed, measures of temporal stability were often explicitly sought in judging the adequacy of various potential measures of social and political attitudes. Copyright 0 1999 by Academic Press. All rights of reproduction in any form reserved. 0065-2601/99 130.00 265 ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY, VOL. 31

Upload: charles-g

Post on 16-Mar-2017

213 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

Charles G. Lord Mark R. Lepper

REPRESENTATION

I. Introduction

“Attitude,” as Gordon Allport (1935) saw it in his classic chapter in the first Handbook of Social Psychology, was the “most distinctive and most indispensable concept in contemporary social psychology” (p. 798). Al- though one might quarrel today with the extremity of his claim, there is no doubt that the attitude concept has remained ubiquitous in, and characteristic of, social psychology in the intervening 60 years. From the early days of Allport and his colleagues through the glory days of Hovland’s Yale Attitude Change Program (e.g., Hovland, Janis, & Kelley, 1953; Kiesler, Collins, & Miller, 1969; McGuire, 1969) to the more complex and sophisticated models of more recent years (e.g., Ajzen & Fishbein, 1980; Eagly & Chaiken, 1993; Fishbein & Ajzen, 1975; Olson & Zanna, 1993; Petty & Cacioppo, 1981, 1986; Zanna & Rempel, 1988), attitudes have indeed remained a central topic in social psychology.

Crucial to the overwhelming success of the attitude concept were two key ingredients. The first was the implied stability of most important social attitudes, the second the presumed generality of the evaluative tendencies these attitudes involved. Thus, basic social attitudes were commonly seen as likely to persist, at least in the absence of concerted efforts to change them, over time. Attitudes were therefore distinguished by early theorists from other more ephemeral influences on a person, such as passing mood states or the momentary salience of particular stimuli in a person’s thoughts or surroundings. Indeed, measures of temporal stability were often explicitly sought in judging the adequacy of various potential measures of social and political attitudes.

Copyright 0 1999 by Academic Press. All rights of reproduction in any form reserved.

0065-2601/99 130.00

265 ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY, VOL. 31

Page 2: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

266 CHARLES G . LORD AND MARK R. LEPPER

Similarly, although people were clearly recognized to have attitudes about individual objects and persons, the prototypic social attitudes that gave the concept such breadth and utility were those concerning large and diverse groups of people or objects or policies-attitudes towards Blacks or Jews, Russians or communists, trade unions or integration. Not only were these attitudes presumed to indicate stable evaluative tendencies toward a single identical object at different times, they also implied cross-situationally con- sistent evaluative tendencies toward different members of a large class of specific attitude objects.

In short, the essential power of the attitude concept lay in the possibility that a person’s reactions to various different members of an object class, observed at different times and in an array of different settings, might all be influenced-and predicted-by that person’s global attitude toward that class of objects. Indeed, some early authors sought, through concepts like ethnocentrism and authoritarianism, to suggest that people’s attitudes on these issues might prove even more general-that attitudes towards Blacks and Jews, communists and Japanese, might all share some common ele- ments (e.g., Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950; All- port, 1954).

At the same time that these features helped to make attitudes a central and indispensable concept for social psychologists, they also set the stage for two early, and enduring, challenges to the attitude concept-challenges that were crucial in shaping the thinking that has gone into most of the research reported in this chapter. Let us turn to these two challenges, in turn.

A. THE ATTITUDE-BEHAVIOR PROBLEM

At roughly the same time that Allport (1935), at Harvard, was writing that first, influential Handbook chapter in praise of the attitude concept, Richard LaPiere, a sociologist at Stanford, was preparing one of the first important attacks on the generality of the attitude concept (LaPiere, 1934). His paper, the first to point to what has since become well-known as the problem of “attitude-behavior inconsistency,” began simply enough, with a series of automobile trips that LaPiere had taken, across the United States and up and down the West coast. On those trips, LaPiere had driven over 10,000 miles, a substantial itinerary in those early days of the automobile. In his travels, he had stayed at 66 hotels and motels and had eaten at 184 restaurants and cafes. Only at one hotel had he been refused lodging.

Such a record would, of course, have been entirely unremarkable had it not been the case that LaPiere had been accompanied on this trip by a young Chinese student and his wife, who were also offered food and lodging

Page 3: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 267

at all but one of the 251 establishments they approached. What made this observation of particular theoretical interest was the seeming inconsistency between the warm welcome his Chinese traveling companions received and the large research literature on social distance and racial prejudice that showed the powerful and pervasive negative attitudes that most white Americans held in those days about “Orientals.” To document this inconsis- tency further, LaPiere subsequentiy wrote to all of the establishments he had frequented, asking explicitly whether those establishments would “accept members of the Chinese race as guests in your establishment.” Of the 47 hotels and 81 restaurants that replied to his letter, only one answered with a straightforward “yes,” whereas over 90% answered with an unequivo- cal “no.”

Among several alternative hypotheses that LaPiere considered, in at- tempting to understand the striking apparent inconsistency of the proprie- tors of these businesses, one stood out to him-that the actual, well-dressed, highly educated, and clearly cultured Chinese couple who had accompanied LaPiere on his journeys may have borne little resemblance to the exemplars or stereotypes that may have come to the minds of the proprietors when they were asked an abstract question about purely hypothetical “members of the Chinese race.” Perhaps, he suggested, people’s responses to global attitude questions may fail to predict their overt actions toward specific members of the attitude category, when those specific representatives of the class fail to match the stereotypic representations called to mind by abstract and general attitude questions.

LaPiere’s (1934) demonstration of the attitude-behavior problem was followed by many similar critiques, each advancing a different explanation or explanations for the phenomenon (e.g., Ajzen & Fishbein, 1977; Camp- bell, 1963; Deutscher, 1973; Festinger, 1964; Kutner, Wilkins & Yarrow, 1952; McGuire, 1976; Wicker, 1969). This abiding theoretical interest in the attitude-behavior problem attests to how strongly it violates the layperson’s assumptions about being able to predict many facets of an individual’s behavior toward a stimulus or a class of stimuli merely from knowing that one crucial piece of information-the individual’s global attitude.

B. THE ATTITUDE-OBJECT PROBLEM

If the appearance of inconsistency in people’s thoughts and actions to- ward different members of an attitude category offered one central chal- lenge to the attitude concept, an equally important threat was posed by a second, quite different set of experiments reported by Solomon Asch (1940) just a few years after Lapiere’s study. Again, the basic phenomenon was

Page 4: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

268 CHARLES G. LORD AND MARK R. LEPPER

simple and straightforward. In particular, Asch asked college students to rank-order ten different professions in terms of characteristics like intelli- gence, character, and idealism. For one specific profession, politics, he also led the students to believe that 500 of their peers had previously rated this particular category as either the first, or the last, on the list of ten professions.

This normative information had a large impact on students’ rankings of politics. Those who had been led to believe that their peers had ranked politics first themselves ranked politicians an average of 4 to 5 ranks higher than those who had been led to believe that their peers had ranked politics last-a finding that seemed sharply at variance with the presumed stability of social attitudes. As Asch himself was quick to note, however, this effect was not simply the result of mindless conformity or of instant attitude change. Indeed, when he interviewed his participants, they strongly denied both of these accusations. They did, however, plead guilty, as it were, to a considerably more subtle and more interesting process.

What Asch argued, on the basis of these interviews, was that the students in the two groups had come to interpret the term “politics” rather differ- ently-to have had quite different politicians in mind when they answered Asch’s questions. Those who thought their peers had ranked politics at the top of the list, on the one hand, were more likely to have thought of great statesmen, such as Franklin D. Roosevelt, Churchill, Lincoln, or Washington when they considered the profession of politics. Those who thought their peers had ranked politics at the bottom of the list, on the other hand, were more likely to have thought of corrupt political bosses and party hacks, of Boss Tweed and Tammany Hall, when they considered the same profession of politics. It was as if, in Asch’s elegant and enduring phrase, there had been a “change in the object of judgment, rather than in the judgment of the object” (p. 458).

11. Attitude Representation Theory

It is these two challenges to the utility of the concept of general social attitudes, then, that have formed the primary stimuli leading to the program of collaborative research we report upon in this chapter. Indeed, it is in part from the speculations that both LaPiere (1934) and Asch (1940) offered in accounting for their respective findings that our own thinking about these problems has sprung. Both Asch and LaPiere offered, in slightly different ways, persuasive arguments for thinking about issues of attitude consistency and stability in terms of the construals and interpretations of the social perceiver.

Page 5: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 269

A. THE REPRESENTATION AND MATCHING POSTULATES

Our own model, referred to here as Attitude Representation Theory, formalizes and extends these theoretical ideas. The model seeks to help us to understand, in particular, when and where to expect consistency versus inconsistency in people’s responses to what is objectively the same class of stimuli-as a function of variations in the specific members of that class, their characteristics, the types of responses involved, and the contexts in which those responses are observed. It does so in terms of two basic postu- lates.

1. The Representation Postulate

Our first postulate, then, makes a simple point-that a person’s response to any attitude-relevant stimulus will depend not only on the perceived properties of that stimulus and the situation surrounding it, but also on the subjective representation of that stimulus by the person. This point, of course, has been made by other authors in a number of specific contexts (e.g., Schwarz, 1994; Tesser, 1978; Wilson & Hodges, 1992). We seek, how- ever, to present a more general version of this argument that we call the “representation postulate.” Like LaPiere and Asch, we will argue that generating more concrete representations of abstract categories is a wide- spread response when people are asked to make global verbal evaluations of various attitude classes. In addition, however, we will also argue that the manner in which people construct or construe attitude stimuli (Griffin & Ross, 1991; Ross, 1987; Ross & Ward, 1995) may be important not only when they are asked to answer abstract and general questions about broad categories of people or objects, but even when they are asked to respond to specific situations that involve particular instances or exemplars of these categories.

2. The Matching Postulate

Our second postulate makes an equally simple point-that the closer the match between the subjective representations and perceived immediate stimuli to which a person is responding in one situation and the subjective representations and perceived immediate stimuli to which the person is responding in a different situation, the more consistency there will be in that person’s responses. Again, this is a point that has been made by other authors, including Asch and LaPiere, in various specific contexts (e.g., Ajzen & Fishbein, 1980; Millar & Tesser, 1992), but once again we will propose a more general version of this claim that we call the “matching

Page 6: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

270 CHARLES G. LORD AND MARK R. LEPPER

postulate.” Thus, consistency across various opportunities for attitude- relevant responses should depend on the degree of match across an array of dimensions-not just the identity of the specific instances that come to mind, but also their assumed characteristics, the types of responses they are associated with, and the contexts in which they occur. The more two situations activate assumptions and perceptions that match across all of these relevant dimensions, the more response consistency or generality our model will predict.

In the sections to follow, we will begin by describing a process model of evaluative responses that incorporates both the representation postulate and the matching postulate. We will then review research that addresses each of the postulates as it pertains to this process model. Finally, we will examine issues raised by thinking about the relationship between Attitude Representation Theory and other theoretical and empirical contributions.

B. PROCESSES IN ATTITUDE REPRESENTATION THEORY

Figure 1 presents a process model of how people make evaluative re- sponses to different members of an object class, or even to the exact same attitude object at two separate times, or in two separate situations. According to the Attitude Representation Theory (ART) model, the same processes should occur at both times, but these processes can yield very different outcomes, depending upon the individual’s subjective representa- tion of the stimulus, as well as his or her perceptions of the immediate stimulus. These assumptions and perceptions, each weighted by its impor- tance, are then combined to produce an overall evaluation that can range from very negative to very positive. Unless other factors extraneous to the model intervene, the individual will respond on the basis of that overall evaluation. The two responses will be more similar when the overall evalua- tions, based on their assumptions and perceptions, match than when they mismatch.

As shown in Figure 1, the first step after identifying the situation as relevant to an attitude object is often to activate both a subjective represen- tation of the stimulus and, if appropriate, perceptions of its immediate properties. At Time 1 in LaPiere’s (1934) study, when the couple requested service, the attitude object was objectively the same as it would be later on the questionnaire-namely, “accepting members of the Chinese race as guests.” At Time 1, the proprietors might have noticed that the couple were Chinese and activated a subjective representation of “Chinese people” that carried with it various assumptions or “default values” about members of this category. In the 1930s, for instance, a typical proprietor might have

Page 7: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 271

attitude-relevant situation at

Time 1

attitude-relevant situation at

Time 2

assumptions perceptions from subjective from immediate representation

ARa, CFa

COMBINE

assumptions with perceptions to

reach an overall evaluation

2 wTla, + 2 wTlp, ZwTCa, + z wTCp, ZwARa, + 2 WARP, C wCFa, + Z wCFP,

Fig. 1.

ACTIVATE

assumptions from subjective representation

Tla TCa ARa2 CFa2

perceptions from immediate

stimulus

TIP 2

TCP 2

ARp2 CFD 2

Matchl + - - - b

Mismatch

COMBINE

assumptions with perceptions to

reach an overall evaluation

2 wTla, + XwTlp, 2 wTCa, + 2 wTCP, z wARa, + ZwARp, 2 wCFa, + 2 wCFp,

informed by

RESPOND 1 -~ Matchl

the overall informed by the overall evaluation Mismatch evaluation

+ + - - - -

kttitude Representation Theory (ART) model of the process that underlies attitude-relevant evaluative responses. TIa and TIp = assumptions about and perceptions of a target’s identity or identities. TCa and TCp = assumptions about and perceptions of a target’s characteristics. ARa and ARp = assumptions about and perceptions of the types of responses associated with a target. CFa and CFp = assumptions about and perceptions of contextual factors that surround responses to the target. w = weight given to each of these four aspects when they are combined to form an overall evaluation.

Page 8: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

272 CHARLES G. LORD AND MARK R. LEPPER

assumed that most Chinese people were coolie laborers (target identities), were ignorant and poor (target characteristics), were the “type of people I usually avoid and do not respect” (associated responses),’ and would be easy to eject because other patrons would applaud (contextual factors).2 These assumptions are represented in Figure 1 as TIal (assumed target identities at Time l), TCal (assumed target characteristics at Time l), ARal (other responses associated with the attitude object at Time l), and CFal (assumed contextual factors at Time 1).

With an actual Chinese couple standing before them, however, the propri- etors also had available to them a number of more direct perceptions. According to Lapiere’s (1934) account, the Chinese couple appeared to be academics or professionals (target identities). They were obviously well educated and well-to-do (target characteristics). They were in many respects the type of people the proprietors usually favored as paying customers (associated responses). In addition, they were accompanied by a Caucasian professor, and it may have appeared difficult to get rid of them without upsetting other patrons (contextual factors). These immediate perceptions are represented in Figure 1 as TIP,, TCpl, ARpl, and CFpI. To reach an overall evaluation of the attitude object (accepting these specific members of the Chinese race as guests), as Figure 1 shows, the proprietors combined the activated assumptions and perceptions, each of which carried both a valence and an importance weight.’ The overall evaluation, in turn, in- formed the proprietors’ responses.

Of course, when we speak of the contrast between “subjective representa- tions” and “assumptions” or “default values,” on the one hand, and “di- rect” or “immediate” perceptions, on the other, we must recognize that this distinction is actually a matter of degree, rather than a matter of kind (Bruner, 1957; McArthur & Baron, 1983). Certainly in the arena of social

‘ Assumptions about ‘ responses to the attitude past (e.g., Bern, 1972) or

‘associated responses” include cognitive, affective, and behavioral object (Eagly & Chaiken, 1998) that the individual has had in the imagines that he or she might have in the future (e.g., Ross, 1989).

Perceptions of “associated responses” include cognitive, affective, and behavioral responses in the immediate situation. One type of affective response in the immediate situation can be an automatic, spontaneous affective reaction to the attitude object (Fazio el a]., 1986; Bargh et al., 1992).

* “Contextual factors” are aspects of the social context that the individual assumes or perceives might facilitate or inhibit evaluating the attitude object with favor or disfavor (see Eagly & Chaiken’s, 1993, p. 1, definition of “attitude”).

Attitude representation theory thus incorporates several distinct elements of the theory of planned behavior (Ajzen, 1985)-subjective norms, perceived behavioral control (contextual factors), and behavioral intentions (associated responses)-within the assumptions and percep- tions that are activated and combined to reach an overall evaluation, which in turn informs an attitude-relevant response. (For details of possible combination rules and processes, see Eagly Chaiken. 1993, pp. 106-114.)

Page 9: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 273

attitudes, there are no “immaculate perceptions” uncolored by our beliefs and expectations (Higgins, 1996; Neisser, 1967). Instead, we know that even our immediate and direct perceptions of specific social actors standing next to us may also be influenced by our stereotypes, our prior knowledge, and the context in which we encounter them (Griffin & Ross, 1991; Rock, 1984; Ross, 1987). Thus Lapiere’s proprietors did not actually “see” a cultured, well-educated, affluent couple asking for service; those presumed attributes had to be inferred from the dress, accents, appearance, and remarks of LaPiere’s companions. It is only in order to simplify our presentation, then, that we will describe the perceiver’s recalled representation of the attitude object as “assumptions,” but the perceiver’s interpretations of the immedi- ate situation as “perceptions.”

In these terms, then, at Time 1 in Lapiere’s (1934) study (shown on the left side of Figure l ) , the proprietors might have relied almost entirely on their immediate perceptions of the specific Chinese couple facing them, which clearly contradicted most of the assumptions from their subjective representations. The target’s identity was presumably different from what the proprietors might have assumed, as were the target’s characteristics and the context. Only the type of response (accepting them as guests) was similar to the attitude-relevant situation that would later be identified in the questionnaire at Time 2 (shown on the right side of Figure 1). As a result, they chose to serve the Chinese couple.

By contrast, when the proprietors answered Lapiere’s attitude question- naire at Time 2, they were forced to rely entirely on assumptions from their subjective representations of Chinese people, because there was no spec@ target stimulus to be directly perceived. Whatever assumptions were activated, whether about the target’s identity, characteristics, associated responses, or contextual factors, had to carry all the weight in determining the overall evaluation and evaluative response. LaPiere himself character- ized the proprietors’ answers to his questionnaire as a “verbal response to a symbolic situation” (1934, p. 230). Given the negative stereotypes of the day, it is not surprising that when given three choices, the proprietors circled “no” rather than either “yes” or “uncertain; depend upon circumstances.” After all, how many reputable establishments would agree to serve a poor, ignorant, coolie laborer who could be easily ejected to loud applause?

As the terms “Matchhlismatch” at the bottom center of Figure 1 indi- cate, evaluative responses to seemingly related, or even objectively identical stimuli will sometimes match; but sometimes, as in LaPiere’s (1934) study, they will not. When they do not match, barring extraneous factors such as coercion, the lack of a match may frequently be due to differences between the subjective default assumptions and more immediate and direct percep- tions that were activated on the two occasions (and/or to differences in the

Page 10: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

274 CHARLES G. LORD AND MARK R. LEPPER

weights that those assumptions and perceptions carried in determining the separate evaluations).

In sum, laypeople in LaPiere’s (1934) and Asch’s (1940) era, as well as in the present, expect attitudes to be stable (expressed consistently across time) and general (expressed consistently across situations). The “attitude- object problem” arose because empirical results suggested that attitude- relevant self-reports are often unstable, leading Asch (1940) to suggest that even when an attitude object is objectively the same at two separate times, people might have in mind subjectively different instances or exemplars. Our more general “representation postulate” holds that evaluative re- sponses of all types, not just answers to general attitude questions, can involve both perceptions of the immediate situation and representation- based assumptions.

Similarly, the “attitude-behavior problem” arose because empirical re- sults suggested that attitude-relevant responses often do not generalize from one situation to the next. LaPiere (1934) suggested that the consistency between verbal reports and overt actions may depend on a match between the assumed and the actual attitude object’s identity. Our more general “matching postulate” suggests that attitude generality depends not just on the matching of assumed and perceived identities, but also on the matching of representation-based assumptions and more immediate perceptions on a number of relevant dimensions.

These two basic premises-the “representation postulate” and the “matching postulate”-were then incorporated into what we call Attitude Representation Theory-a general model of the process by which people make evaluative responses. The previous section provided a broad overview of this model. In the two sections to follow, we describe the specifics of this model, by concentrating first on research relevant to the representation postulate and then on research relevant to the matching postulate.

111. Research Relevant to the Representation Postulate

The starting point for ART, then, is the postulate that attitude-relevant responses typically involve the activation of a subjective representation of the stimulus. An attitude-relevant response, whether to a direct question or to a behavioral situation that concerns “members of the Chinese race,” for instance, might activate a default representation of a Chinese person or persons. Although ART allows for many different types of representational formats, such as associative networks, schemas, exemplars, and parallel distributed processing systems (Smith, 1998), much of the supportive evi-

Page 11: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 275

dence comes from investigations of the activation of exemplars. Thus, in one set of studies, we tested whether exemplars are selectively activated when people evaluate a general stimulus given only its name. In a second set of studies, we tested whether people who activate the same exemplar to represent a stimulus on separate occasions also report more stable or consistent attitudes. Third, we tested whether consistent exemplar activa- tion lends resistance to attitude change. Fourth, we tested whether attitudes can be altered by making different exemplars temporarily accessible. Fifth, we tested the role of abstract and specific exemplar representations in Allport’s (1954) social contact hypothesis. Finally, we tested the importance of exemplar representativeness in determining the generalization of attitude change that derives from equal status contact. Together, these studies pro- vided convergent evidence for, and explored the implications of, the repre- sentation postulate.

A. SELECTIVE ACTIVATION OF REPRESENTATIONS

After identifying a situation as relevant to an attitude object, as shown in Figure 1, some representation of the attitude object is activated. When responding to a questionnaire about attitudes toward “politicians,” for example, Asch’s (1940) participants might have activated a representation of the relevant category. Such cognitive representations will often entail category exemplars, such as Franklin Roosevelt or Winston Churchill (e.g., Smith, 1998). Indeed, Asch (1940) asked participants which politicians they had in mind when they made their attitude judgments and many of them named Roosevelt or other famous “statemen.” Unfortunately, we cannot necessarily assume from their self-reports that these participants were able to report accurately on their cognitive processes (Nisbett & Wilson, 1977).

Considerable research has shown, however, that exemplars are frequently used for other, nonattitudinal social judgments. Judgments about social categories, for instance, often depend on the specific exemplars that an individual activates on that occasion (Smith, 1992; Smith & Zarate, 1992). In one relevant study, sportswriters and coaches rated very highly a college football player who happened to have the same hometown as a well-known professional star (Gilovich, 1981). Novel members of a social category may also be evaluated in terms of their similarity to the “default values” assumed for a typical category member (Rothbart & John, 1985). Black men are categorized primarily as “Black” and White women are categorized primar- ily as “women” in a culture that takes “White men” as the default value (Zarate & Smith, 1990).

Page 12: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

276 CHARLES G. LORD AND MARK R. LEPPER

Moreover, exemplars can be activated and thus have an impact on judg- ment without being accessible to conscious recall or recognition (Smith, 1998). Participants in several studies have been better able to complete a fragmented word such as -SS -SS-- even though they could not recall or recognize having earlier read the word (assassin) on a list (e.g., Roediger, 1990; Tulving, Schacter, & Stark, 1982). Similarly, participants in another study avoided interacting with an experimental assistant who happened to have the same hairstyle as an experimenter who had previously insulted them, even though they staunchly denied being influenced by that earlier encounter (Lewicki, 1986).

More generally, studies of repetition priming, word-fragment completion, and savings in relearning have all demonstrated types of implicit learning that seem qualitatively different from performance on semantic tasks such as defining or comprehending concepts (Carlston & Skowronski, 1994; Smith, Stewart, & Buttram, 1992; Tulving et al., 1982; for a recent review see Smith, 1998). Moreover, studies of halo effects, mere exposure, subliminal attitude conditioning, and context effects in survey research have likewise shown implicit social cognition processes that do apply more specifically to attitudes (Edwards, 1990; Greenwald, 1990 Schwarz & Clore, 1983; Thorndike, 1920; Zajonc, 1980; for a recent review see Greenwald & Ba- naji, 1995).

If attitude judgments about social categories are similar to other types of judgments about social objects, then, attitude assessments might rely heavily on exemplars that are activated from a representation of the attitude object or stimulus, just as LaPiere (1934) and Asch (1940) suggested. Fur- thermore, people might use exemplars in assessing their attitudes even if they cannot accurately report having done so (Tesser & Martin, 1996). One way of gaining evidence about the process of attitude assessment that does not rely completely on the validity of self-reports is derived from a well- established principle of mental operations-that people selectively activate, from what they know about a topic, those pieces of information that are most useful for the specific judgment at hand. Thus selective activation implies that a concept is used in a judgment process (Anderson, 1983; Anderson & Pichert, 1978; Higgins & King, 1981; Newell, Rosenbloom, & Laird, 1989; Wyer & Srull, 1986).

Because it is often impossible (or inefficient) when making a judgment “to engage in an exhaustive search of all relevant information stored in memory, the outcome of a search will be influenced by the relative accessi- bility of the relevant information” (Strack, 1992, p. 256). More specific to attitudes, several models hold that “people often have a large, conflicting ‘data base’ relevant to their attitudes on any given topic, and the attitude they have at any given time depends on the subset of these data to which

Page 13: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 277

they attend” (Wilson & Hodges, 1992, p. 38; see also Tesser, 1978). That is one reason why attitude measurement is so susceptible to subtle context and framing effects (Ottati, 1997; Schwarz, 1994; Tourangeau & Rasin- ski, 1988).

Sia, Lord, Blessum, Thomas, and Lepper (in press) used three “process” measures to investigate Asch’s (1940) claim and the ART prediction that people selectively activate relevant exemplars when asked to indicate their general social attitudes. One such process measure was based on the well- documented principle that the activation of cognitive elements may be detected by asking participants to interpret ambiguous stimuli. In one classic study of construct accessibility, for example, students who had recently memorized positive trait words such as “adventurous” interpreted a fic- tional character’s exploits more positively than did students who had re- cently memorized negative trait words such as “reckless” (Higgins, Rholes, & Jones, 1977). Since then, dozens of studies have shown that recently activated constructs can color the interpretation of subsequently presented ambiguous information (Stangor, 1990; Sedikides, 1990; Srull & Wyer, 1979). Similarly, recently primed or activated constructs are more likely to be “found” in the jumbled letters of anagrams (Roediger, 1990), even when explicit measures such as recognition or free recall remain unaffected (Rajaram & Roediger, 1993). Assessing one’s attitude toward talk show hosts might thus increase the likelihood of solving the anagram HAPOR as OPRAH rather than HARPO.

In their first experiment, therefore, Sia and colleagues (in press) had college students complete one of two booklets. Students who completed an attitudes booklet wrote out their attitudes toward several social categories, including politicians, basketball players, talk show hosts, and televangelists. Students who completed a definitions booklet wrote brief definitions, in- stead of attitudes, for the same social categories. Immediately after answer- ing each question, participants in both groups were asked to solve an “anagram,” such as rearranging letters from CILRNOPPROTUN to form a seven-letter word or name. The only difference between the conditions, therefore, was that some participants wrote their attitude toward a category before solving each anagram, whereas other participants wrote a definition of the category before solving the same anagrams.

Readers may have noticed that seven of the letters in CILRNOPPRO- TUN-which occurred directly after a question asking for the students’ attitudes toward politicians in the attitudes condition and directly after a question asking for a definition of politicians in the definitions condition- could be rearranged to form either “Clinton” or “corrupt.” In fact, in a pretest in which participants merely free-associated to the word “politi- cians,” 37% had mentioned “Clinton” and 37% had mentioned “corrupt.”

Page 14: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

278 CHARLES C . LORD AND MARK R. LEPPER

We know, therefore, that the nonexemplar associate “corrupt” and the exemplar “Clinton” were equally likely to be activated by the mere name of the category, when no judgment was required. Because using the concept to solve an ambiguous anagram is a measure of activation during a judgment, the ART prediction was that students would find “Clinton” more often than “corrupt” after an attitude than a definition question about politicians. Similarly, they should find “Shaq” more often than “Shoe,” “Oprah” more often than “smile,” and “Baker” more often than “money” after attitude questions than after definition questions about basketball players, talk show hosts, and televangelists, respectively, even though in each case the alterna- tives had been mentioned by an equal percentage of pretest participants who were asked to free-associate to the category namee4

Figure 2 shows the significant Preceding Question X Type of Solution interaction for the number of anagram solutions (of a possible four) that students produced. After semantic, definition questions, students found an equivalent number of nonexemplar and exemplar solutions. After attitude questions about the same categories, they found significantly more exemplar than nonexemplar solutions. The result, derived from process measures rather than self-reports, clearly supported Asch’s (1940) contention and the ART hypothesis that attitude assessment selectively activates exemplars from a representation of the stimulus, presumably because concrete exem- plars are very useful for making an evaluative response when given only the name of the attitude object. One’s opinion of Newt Gingrich, for in- stance, seems more relevant to assessing one’s attitude toward “politicians” than does one’s opinion of “people who wear suits,” even though they might be equally associated with the category in a strictly semantic sense.

Sia, and her associates’ (in press) second experiment yielded similar results, using another well-researched indicator of concept activation, namely word-fragment completions (see Tulving et al., 1982; Tulving & Schacter, 1990), instead of anagram solutions. A preceding question about attitudes toward, rather than definitions of, basketball players, for instance, increased the likelihood of completing --RD-N as “Jordan” rather than as “garden.” Likewise, in a third experiment Sia and associates (in press) borrowed a response latency technique that had previously been used to disconfirm a “common-sense’’ hypothesis about the concepts that are acti- vated when people assess their own personality traits (Klein, Loftus, & Burton, 1989). The technique relies on the principle that cognitive processes occur faster if one of the pieces of information used in a judgment has recently been activated (e.g., Bargh, 1992; Collins & Quillian, 1969; Devine, 1989; Dovidio, Evans, & Tyler, 1986; Klein & Loftus, 1993; Klein et al.,

“Bakker” was misspelled by all participants.

Page 15: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

1.6 - 1.4 - 1.2 -

REPRESENTATION THEORY

Nonexemplar solutions ~

279

1 - .- : CI a - 5: 0.8 -

s

r

L 0

2 0.6 - 2

0.4 - 0.2 -

0 7

Semantic Attitude

Question preceding the puzzle

Fig. 2. Mean number of exemplar and nonexemplar anagram solutions that participants used after answering either a semantic or an attitude question about each of four social categories. (Data from Sia, Lord, Blessurn, Thomas, & Lepper, in press, Experiment 1.)

1989; Lord, Desforges, Chacon, Pere, & Clubb, 1992; Macht & O’Brien, 1980; Macht & Spear, 1977; Niedenthal, 1992: Perdue & Gurtman, 1990; Smith et al., 1992). Indeed, for many purposes, response latency measures have proven more reliable than a perceiver’s (often erroneous) impressions of his or her own cognitive processes (Nisbett & Wilson, 1977).

College undergraduates (Sia et al., in press, Experiment 3), then, were asked to answer yes or no to questions about various social categories on a computer screen. For eight of the categories, the students answered (at least 10 questions apart) both an attitude question and a definition question. Immediately following each of these two types of judgments, they answered a category-inclusion question such as “Is Bill Clinton a politician?” The hypothesis was that, if attitude assessment selectively activates exemplars, participants would be faster to answer the category-inclusion question after an attitude question than after a comparable definition question. Figure 3 shows mean response latencies for the category-inclusion question following either an attitude question or a definition question for a representative four of the eight categories. (The other four categories showed the same pattern.)

Page 16: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

280 CHARLES G. LORD AND MARK R. LEPPER

2500

Q 2000

.- t

6

U C 0 0

- - ‘i 1500 c .- - C

# 1000

B - Q u)

3 K 500

0

Semantic question Attitude question

Politicians Talk show Televangelists People with hosts AIDS

Social category

Mean response latency (in milliseconds) for a category-inclusion question that followed either a semantic or an attitude question about each of four social categories. (Data from Sia, Lord, Blessurn, Thomas, & Lepper, in press, Experiment 3.)

Fig. 3.

As predicted, students were significantly faster to judge whether an exem- plar belonged to the category after an attitude question than after a defini- tion question, as would be the case if the attitude question had indeed selectively activated exemplars. This result held when differences in re- sponse latency for the first question (attitude or definition) were statistically controlled. It held also whether answers to the category-inclusion question were “yes” or “no,” presumably because even a “no” category-inclusion answer (e.g., “Is Rochester the capital of New York?”) is facilitated by having correct exemplars activated in advance.

The evidence regarding concept activation during attitude-relevant judg- ments thus supported the “activation” step in the ART model as depicted in Figure 1. Admittedly, actual exemplars might not be activated for social categories that have no named exemplars, such as “homeless people” or “student dissidents,” but both exemplar theory (Smith, 1998) and associated

Page 17: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

AITITUDE REPRESENTATION THEORY 281

systems theory (Carlston, 1994) argue that exemplars are often imaginary, nameless mental images such as a bag lady we saw on the street or a young man standing in front of a column of tanks. Similarly, exemplars might also not be selectively activated for attitudes toward other types of attitude objects, such as specific persons (e.g., Clinton), social policies (e.g., capital punishment), abstract concepts (e.g., religion), or activities (e.g., going to the movies). Finally, some attitudes might be “known directly” from a spontaneous affective response rather than inferred from any other acti- vated elements (Bargh, 1996; Bargh, Chaiken, Govender, & Pratto, 1992; Fazio, Sanbonmatsu, Powell, & Kardes, 1986), though response latencies of approximately 2000 ms, as shown in Figure 3, suggest that spontaneous affective reactions, which occur in less than 500 ms (Logan, 1989; Neely, 1977), might well be integrated with other pieces of information, such as default values from activated assumptions about likely exemplars, before people arrive at a final attitude-relevant response.

B. CONSISTENT ACTIVATION AND RESPONSE STABILITY

One central concern of the ART model presented in Figure 1, is, of course, the relationship between attitude-relevant responses that occur at two separate times. In such cases, people’s responses sometimes remain consistent, but sometimes they do not. ART addresses at least one reason for attitude stability versus instability. If no useful information has been added in the interval between two attitude-relevant responses, one might assume that the two responses should be identical. Suppose, for instance, that an individual answers the question “What is your attitude toward politicians?” at two separate times, one month apart. Because there is no immediate stimulus or specific attitude object present on either occasion, the individual is presumed to respond both times on the basis of attitude representation assumptions shown in Figure 1 as TIa, TCa, ARa, and CFa, which suggests that “Match/Mismatch” in the figure should automatically be “Match.” ART predicts, in contrast, that the match is unlikely to be perfect, because, even with no intervening information and no change in the content of the underlying attitude representation from Time 1 to Time 2, it is possible that the assumptions activated at Time 1 might not be the same as the assumptions activated at Time 2.

To understand this prediction, recall that representation-derived assump- tions depend in part on which particular aspects of a representation are activated (Anderson, 1983; Strack, 1992; Tesser, 1978; Wilson & Hodges, 1992). An individual’s representation for politicians, for instance, might contain many exemplars, not just Bill Clinton. At any given time, 37% of

Page 18: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

282 CHARLES G. LORD AND MARK R. LEPPER

respondents might activate Clinton as their first exemplar (Sia et al., in press), but not all of those respondents would necessarily activate Clinton first if they were to assess their attitudes again at a later date. According to data reported by Bellezza (1984), when people are asked twice, one month apart, to name an exemplar of a natural object category such as vegetables, they have only a .69 probability, not a 1.0 probability, of naming the same exemplar (e.g., carrots or corn) on both occasions.

Sia, Lord, Blessum, Ratcliff, and Lepper (1997) reasoned that similar probabilities might pertain to social categories such as politicians or talk show hosts. An important tenet of ART, suggested by Figure 1, is that the stability of evaluative responses, even using the exact same response format, will depend on the match between the attitude representations and assump- tions that are activated at the two separate times when people assess their attitudes. With all else held constant, people who activate the same exem- plar when assessing their attitudes at two separate times have a higher probability of making the same evaluative response than do people who activate different exemplars.

To test this prediction, Sia and her colleagues (1997, Experiment 1) asked students, for each of seven social categories, to assess their attitudes toward the categories, to name the first exemplar that came to mind to represent the categories, and then to name other exemplars that they thought might occur to them first on other occasions. Approximately one month later, the students completed exactly the same questionnaire. At each of the two sessions, a randomly selected subset of students reported their attitudes before their exemplars, and the others reported their exemplars before their attitudes.

Figure 4 shows that students who named the same first exemplar at Time 2 as they had at Time 1 (for whom the assumed target identity at Time 1 matched the assumed target identity at Time 2) had greater attitude stability than did students who named different first exemplars (for whom the as- sumed target identity at Time 1 mismatched the assumed target identity at Time 2). The figure shows attitude-attitude correlation coefficients for each of four social categories. (The other three categories showed the same pattern.) Across the seven categories, students who named the same exemplar displayed attitude stability of .74, whereas students who named different exemplars showed attitude stability of .53. Also, when each stu- dent’s seven attitudes were divided into those for which the student had named the same exemplar and those for which the student had named a different exemplar, attitude stability was greater for the former categories than for the latter, regardless of the order in which students had named attitudes and exemplars. Furthermore, naming the same rather than a differ-

Page 19: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 283

0.9] 1

0.8

0 Named different exemplar Named same exemplar

Politicians Talk show Televangelists Homosexuals hosts

Social category

Fig. 4. Attitude stability coefficients for participants who named the same versus a different exemplar, one month apart, for each of four social categories. (Data from Sia. Lord, Blessum, Ratcliff, & Lepper, 1997, Experiment 1.)

ent exemplar was category-specific, not an individual difference that applied across categories.

Other analyses suggested that participants were likely to change their attitudes only when they named a different exemplar that they reported liking more, or less, than the one they had named first during the initial attitude assessment, not when they merely switched to naming a different exemplar that they liked or disliked equally with the first one. Furthermore, the students who named a different first exemplar tended to switch to one they had said might occur to them on a different occasion or to an exemplar that they had not named but presumably knew (e.g., Newt Gingrich), as suggested by the fact that many other participants had named that person.

These results supported ART by showing that response stability de- creases when the elements activated from a representation of the attitude object are different on separate occasions. For ART, moreover, it is reassur-

Page 20: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

284 CHARLES G. LORD AND MARK R. LEPPER

ing that the order of naming attitudes versus exemplars did not affect the relationship between naming the same exemplar and retaining one’s initial attitude. The results, of course, were only correlational, and did not, by themselves, rule out the possibility that students assessed their attitudes first-perhaps relying only on a spontaneous affective response to the category’s name-and then named an exemplar that they thought would justify such an attitude. This “attitude first” claim, however, is called into doubt by the results of the activation study described earlier (Sia, et al., in press), in which students were merely asked to assess and report their attitudes, and yet they still selectively activated exemplars. Automatic acti- vation of an affective response takes less than 500 ms (Logan, 1989; Neely, 1977), after which students could have pressed a key to go on to the next question. Would they have waited 1500 additional milliseconds before going on (see response times in Figure 3) if they were only activating exemplars after-the-fact to justify their attitude rather than activating exemplars be- cause they were relevant to assessing an attitude? Also, as we shall see, empirical evidence supports a causal relationship such that manipulating exemplar accessibility affects the stability of evaluative responses to what is objectively the same stimulus.

C. CONSISTENT ACTIVATION AND RESISTANCE TO ATTITUDE CHANGE

If consistent exemplar activation promotes the stability of evaluative responses, then it might also produce resistance to attitude change. Many previous studies have addressed the factors that impart resistance to attitude change, for instance, in the face of disconfirming evidence (e.g., Lord, Ross, & Lepper, 1979; Lydon & Zanna, 1990; McGuire, 1985; Petty & Cacioppo, 1986; Watts & Holt, 1979). The gist of the empirical results is that attitudes resist change better when they consist of strong, durable, or interconnected beliefs about the attitude object.

People who always activate the same exemplar when they assess a particu- lar attitude are likely to have, according to ART, a consistent and interre- lated set of assumptions regarding the attitude object. People routinely attribute cross-temporally and cross-situationally consistent characteristics to others (Ross, 1977). Someone who always activates the same exemplar for a social category (the same assumed target identity), therefore, is very likely to activate the same set of assumed characteristics to represent the category. Also, someone who activates the same exemplar for a social category might have a history of previous responses associated with the exemplar (the same associated responses). An individual who always acti-

Page 21: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 285

vates Bill Clinton to represent “politicians,” for instance, is very likely to activate the same set of Clinton’s characteristics every time, and the same associated responses, such as voting for or against him. If the exemplar remains unchanged from one attitude-relevant response to the next, inter- ventions that do not change the activated exemplar might not have much effect on the underlying attitude.

A person who activates Clinton at one time, Dole the next, and Gingrich the time after that, in contrast, is likely to activate different sets of individu- ating characteristics and a different set of previous evaluative responses (perhaps knowing that Gingrich used to be a professor, or being bored by one of Dole’s televised speeches) on each occasion. If the exemplar changes from one attitude-relevant response to the next, even interventions that do not address the activated exemplar might have a good chance of altering the underlying attitude, because that attitude has a shaky foundation. A persuasive message that said nothing about any exemplars, for instance, might introduce a new set of principles or associations that-without an “anchoring” exemplar to counteract them-might more easily undermine an existing attitude.

From just such reasoning about the implications of ART, Sia, Thomas, Lord, and Lepper (1998) devised the first longitudinal study of exemplar activation. At Time 1, they asked college students to assess their attitudes toward several social categories, one of which was politicians, and to name the first exemplar that came to mind to represent each category. Once a week for the next five weeks (Times 2-5), the students completed the same questionnaire. Finally, at Time 6, all students participated in an “unrelated” study, conducted by a different experimenter from the one who had been giving them attitude questionnaires for the previous five weeks, in which they were asked to “help us prepare some stimulus materials.” To avoid possible demand characteristics had participants thought the persuasive messages were aimed at them, the experimenter asked the students to read several persuasive essays and, for each message, to summarize the topic, the author’s position, and the main arguments, as well as to rate the essay’s clarity and objectivity.

Unknown to the students, the essay about politicians had been prepared to argue a position opposite to their own initial attitude at Time 1. Each of the two essays (one in favor of politicians and the other against) was also prepared to take a strong position without ever mentioning a specific politician exemplar. Instead, each essay concentrated on logical, principled arguments. The “pro” essay, for instance, argued that politicians served a valuable purpose in our society and that elected office was a more difficult task than in previous years. The “con” essay argued that the campaign

Page 22: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

286 CHARLES G. LORD AND MARK R. LEPPER

finance laws allowed large contributors to exert undue influence on the political process and that reform of the political system is badly needed.

After they had summarized and critiqued these persuasive essays, the students were thanked for “helping us prepare these stimulus materials,” and dismissed. Immediately thereafter, the experimenter who had been giving them attitude questionnaires for five consecutive weeks entered the room and distributed the attitude questionnaires for the sixth time (Time 6). Figure 5 shows how much student’s attitudes toward politicians changed from Time 5 (before the persuasive message) to Time 6 (after the persua- sive message).

As the figure shows, students who retained the same politician exemplar every week for five consecutive weeks changed their attitudes only slightly, and nonsignificantly, in the direction of the persuasive message (i.e., adopted slightly less extreme attitudes at Time 6 than at Time 5). Students who had changed exemplars at least once during the initial five weeks. in

0 Pre-message Post-message

Retained Changed Exemplar generated prior to message

Fig. 5. Mean attitude extremity before and after reading a persuasive message, by partici- pants who named the same versus different politician exemplars across five weekly attitude assessments. (Data from Sia, Thomas, Lord, & Lepper, 1998.)

Page 23: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 287

contrast, changed their attitudes significantly in the direction of the persua- sive m e ~ s a g e . ~ This study thus established that, even beyond the previously demonstrated impact of exemplar stability on attitude stability, exemplar stability lends resistance to persuasive messages that are intended to change attitudes.

D. MAKING ASPECTS OF A REPRESENTATION ACCESSIBLE

As described, people who name the same first exemplar of a social category twice, one month apart, are more likely to report an unchanged attitude than are people who name different exemplars (Sia et al., 1997). One interesting implication of these findings is that the latter group might have changed their attitudes, even without the introduction of new informa- tion about the attitude object, merely because they activated different aspects of the attitude representation at Time 2 than at Time 1. This finding thus suggests that temporarily accessible constructs can alter attitude- relevant responses from one time to the next, even in the absence of new information about the stimulus, as Eagly and Chaiken (1998, p. 217) have suggested. Reminding people of a very likable social category exemplar, for instance, might cause them to favor the category and its interests more than they had in the past.

In one particularly relevant study, students first estimated the heights of several celebrities (Bodenhausen, Schwarz, Bless, & Wanke, 1995). For some of the students, the last name on the list was a well-liked successful African-American (either Michael Jordan or Oprah Winfrey). For others it was just another White person, like all the others whose heights they had estimated. Immediately after these ratings, in an “unrelated” experi- ment, the same students indicated how much they believed that “racial discrimination against Blacks is no longer a significant problem in the U.S.” (Bodenhausen et al., 1995, p. 53). As the authors predicted, students who had been reminded of Michael or Oprah saw present-day racial discrimina- tion as more of a problem than did students who had not been so reminded, presumably because the height-estimation task made a very likable exem- plar temporarily more accessible than it would otherwise have been. The accessible exemplar presumably ameliorated attitudes toward African- Americans, which in turn altered beliefs about the problem of discrimi- nation. ’ Although students who retained the same politician exemplar had more stable attitudes

across the first five weeks than did students who changed exemplars, the relationship between exemplar stability and attitude change was independent of attitude variance during the first five weeks.

Page 24: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

288 CHARLES G. LORD AND MARK R. LEPPER

In a related study that measured attitude change more directly, Sia and her colleagues (1997, Experiment 3) had students report their attitudes toward, and exemplars for, several social categories, one of which was politicians. From a separate experimental session, the researchers had al- ready learned how much each student like each of the exemplars that he or she had generated. Approximately one month after the initial attitude questionnaire, the same students rated the heights of several celebrities, the last of whom was either the same politician exemplar they had named earlier, a politician the individual liked more, or a politician the individual liked less. Immediately after the height-estimation task, in a supposedly unrelated additional experiment, the students completed the same attitude questionnaire they had completed one month earlier. Even when queried directly, no student realized the connection between the height-estimation task and the attitude questionnaires.

Figure 6 shows the Time 1 and Time 2 attitudes of students in each condition. As can be seen, the manipulation of exemplar accessibility had a significant impact. Students who estimated the height of the same politi- cian they had named earlier tended to retain their initial attitudes. Students who estimated the height of an exemplar they liked more than their earlier first politician exemplar changed their attitudes toward politicians in a positive direction, whereas students who estimated the height of an exem- plar they liked less changed their attitudes in a negative direction. Consistent with many other studies of temporarily accessible constructs (Schwarz & Bless, 1992; Smith, 1992; Tesser, 1978; Wilson & Hodges, 1992), manipula- tions that alter the probability of activating different aspects of a representa- tion can change attitudes without the presentation of additional infor- mation.

E. ART AND ALLPORT’S “CONTACT HYPOTHESIS”

The ART model shown in Figure 1 also has implications for attitude change that occurs in response to new information. According to the ART perspective, such change can occur either when new information alters the weights attached to various aspects of a representation or when it changes the underlying representation itself. One instance of the latter case that is of particular social and theoretical interest occurs when a new, previously unknown exemplar becomes part of the way an individual cognitively repre- sents an attitude object. Just as a temporarily accessible but previously known exemplar such as Bill Clinton might suggest different characteristics than would be suggested by a different politician who was more accessible the last time an attitude was measured, so a new, previously unknown

Page 25: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 289

0 . 5

0

cn C 3 -0 .5 .- .- U

Ti E - 1

Y !! 9) U 3 - 1 . 5 .- s

- 2

- 2 . 5

0 Pre-height estimation 0 Post-height estimation

Same liked exemplar

More liked

Politican whose height was estimated

Mean pre- and postmanipulation attitudes toward politicians, for students who had recently estimated the height of a politician they liked less, the same politician, or a politician they liked more than the politician they had named when their attitudes were assessed 1 month earlier. (Data from Sia, Lord, Blessurn, Ratcliff, & Lepper, 1997. Experi- ment 3.)

Fig. 6.

exemplar might become more accessible than those previously associated with the attitude, changing the default assumptions associated with that category. Attitudes might change because the assumed exemplar at Time 2 does not match the assumed exemplar from Time 1.

This reasoning about the potential impact of new exemplars is part of the rationale behind Allport’s (1 954) famous contact hypothesis. People who avoid interacting with members of a negatively stigmatized group will never get to know any of them as potential exemplars of the category, so they will have little reason to change their negative attitudes. When we arrange situations so that people meet and interact positively with one or more members of a previously despised group, we increase the probability that a new, more positive exemplar will supplant whatever negative exem- plar had previously symbolized or represented the category (Hewstone & Brown, 1986; Rothbart & John, 1985; Sherif, Harvey, White, Hood, &

Page 26: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

290 CHARLES G. LORD AND MARK R. LEPPER

Sherif, 1961). When the well-liked basketball player Magic Johnson an- nounced that he had AIDS, for instance, surveys showed widespread ame- lioration of negative attitudes toward people with AIDS (Graham, Weiner, Giuliano, & Williams, 1993).

1. Abstract versus Specific Representations

Such an analysis of the contact hypothesis suggests an additional aspect of ART that we have not yet discussed. As Smith (1998) demonstrates, an exemplar representation can range from a specific, personally encountered instance to a fuzzy abstraction garnered from media and other second- hand accounts. One of these representations, however, ought to be easier for a new, concrete, and immediate exemplar to supplant. One goal of the contact hypothesis is to present the individual with a specific exemplar who undeniably disconfirms their assumptions (Blaney, Stephan, Rosenfield, Aronson, & Sikes, 1977; Brewer & Miller, 1984; Brown & Wade, 1987; Cook, 1984; Norvell & Worchel, 1981; Slavin, 1979). Contact thus pits a new, more positive exemplar against the old exemplar to determine which will represent the stimulus in the future. The new exemplar presumably has a better chance of winning such a contest if the old exemplar is a vague abstraction than if the old exemplar is a specific, concrete individual such as a person with whom one has had personal experience.

Werth and Lord (1992), therefore, asked students to assess their attitudes toward persons with AIDS and to report whether, when they thought of a typical person with AIDS, they had in mind a specific person (whose initials they were asked to provide) or an abstraction. Subsequently, as part of a regularly scheduled small-enrollment class, some of these students had an actual AIDS victim-a very personable man who was an excellent public speaker-visit their classroom, give a brief presentation, and then answer students’ questions for the rest of the class period. During the next class period, all students completed the attitude questionnaire again.

Figure 7 shows the mean attitudes of students who had the two types of pre-contact representations. Students who initially represented the category with an abstract exemplar changed their attitudes toward people with AIDS more following contact with an actual AIDS sufferer than did students who initially represented the category with a specific, namable exemplar. At Time 1, before the classroom visit, approximately 45% of students said they represented the category with a specific exemplar. At Time 2, approximately the same percentage of students in classes that had not been visited said they had in mind a specific, namable person. In the classes he visited, the percentage of students who said they had a specific exemplar rose to 68%

Page 27: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY

4.2 -

4 - rn

9 3 3.0 -

r 8 3.6 - u) C 0

; g - 3.4 - 0) TI 3 U - ’ 3.2 -

291

3 1

0 Pre-contact 0 Post-contact

1

Specific Abstract

Pre-contact exemplar representation

Mean attitudes toward persons with AIDS by students whose exemplar for this category was a specific person or an abstraction. before and after they had positive contact with an actual AIDS victim. (Data from Werth & Lord, 1992.)

Fig. 7.

after the visit.6 One of the ways in which the contact hypothesis works, then, is by altering the underlying attitude representation and supplanting old exemplars with new, more positive ones-a process that is easier when the old exemplars are nothing more than vague abstractions.

2. Representativeness and the Problem of Generalization

ART also offers some insight into the vexing problem of generalization in Allport’s (1954) contact hypothesis. Allport believed that contact would, when it entailed equal status and mutual goals, change attitudes not only toward the specific group member or members with whom the contact

Naturally, students who had a specific exemplar in mind at Time 1 reported having met a larger number of persons with AIDS in person than did students who had an abstraction in mind. This effect, however. was statistically independent of, and did not qualify, the effect of exemplar specificity on attitude change.

Page 28: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

292 CHARLES G . LORD AND MARK R. LEPPER

occurred, but also toward the larger group as a whole-that the beneficial effects of contact would generalize. Instead, most of the empirical evidence suggests that even when contact is very successful in changing specific attitudes, it often leaves negative attitudes toward the group unchanged (Hewstone & Brown, 1986).

The ART explanation is that contact with a positive exemplar will be unlikely to change general attitudes unless it alters the underlying represen- tation of the category by adding one or more new, more positive exemplars or by altering the default assumptions that go with existing exemplars. The important construct, as recognized by many theorists, is representativeness (Hewstone & Lord, 1998; Pettigrew, 1979; Rothbart & John, 1985). The crucial element is “ensuring that in some way the participants in the contact encounters see each other as representatives of their groups and not merely as ‘exceptions to the rule’ ” (Hewstone & Brown, 1986, p. 18). An individual who is an exception, no matter how positively he or she may be regarded after contact, cannot possibly alter the underlying attitude representation. That is why positive contact with a typically dressed member of a disliked group changes general attitudes toward the group more than does positive contact with an atypically dressed member (Wilder, 1984) and why positive contact with a self-disclosing group member is less effective in producing generalization than positive contact with someone who reveals very little individuating information (Scarberry, Ratcliff, Lord, Lanicek & Des- forges, 1997).

In a relevant study by Desforges et al. (1991), for instance, students assessed their attitudes toward several social groups and categories, includ- ing former mental patients. The students also provided a personality trait profile of “the typical former mental patient.” Approximately one month later, the same students participated in an “unrelated” study of cooperative learning. They studied cooperatively for 50 minutes with another student, about whom they had seen background information that led them to believe she had been successfully treated for mental illness. Actually, the other student was a confederate who was blind to the hypothesis, but had been trained to be pleasant and friendly.

Before contact, when all they knew was that their partner-to-be was a former mental patient, the students described the personality traits that they expected her to have. This trait profile matched very well the trait profile the students had earlier generated as their cognitive representation of the typical former mental patient, as it would if they were reporting assumed characteristics from an underlying representation. After contact, the students described their partner again, now that they had met her. These after-contact descriptions were uniformly more positive than the expectancies that had been based only on categorical assumptions. Subse- quently, in a “separate” experiment conducted by a different experimenter,

Page 29: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 293

the students completed the initial survey of global attitudes yet again. Some students described the typical former mental patient more positively than they had before contact, but others did not. As predicted, changes in the profile of the typical former mental patient significantly predicted changes in general attitudes. Changes in attitudes toward the group as a whole-the generalization part of Allport’s (1954) contact hypothesis-were signifi- cantly related to changes in the underlying representation.

In a follow-up study, Desforges, Lord, Pugh, Scarberry, and Ratcliff (1997, Experiment 1) manipulated perceived representativeness by telling students that their interaction partner belonged to not one, but two, groups that they disliked equally (for instance, a former mental patient who was also homosexual or a drug-abuser). Some students were told that the other student had received an affirmative action preference, being selected over other applicants to participate in this study for money, because he belonged to one (or the other) of the two groups and was representing that group. The group that he represented had, naturally, been selected at random.

Students in the two conditions liked their partner equally, but, as Figure 8 shows, the proportion of that specific liking that generalized to more

1.2

1

S g 0.8 .- g -

0.6 0 U a

U 0.4

0.2

0

Nontarget category E Target category

1

Neither

group Target

group

Partner was chosen to represent

Fig. 8. Mean attitude change toward the nontarget and target social categories following positive contact with a partner who was said to have been chosen as a representative of the target category. (Data from Desforges et al., 1997, Experiment 1.)

Page 30: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

294 CHARLES G . LORD AND MARK R. LEPPER

positive attitudes toward each group depended on his being thought of as a group representative. When the confederate was merely said to belong to two groups (one of which was randomly selected for analysis as the “target group”) but represented neither, students translated a statistically equal proportion of their liking for him to each of the groups. When he represented one of those groups, even though membership in that group got him what some might regard as an unfair advantage, students transferred a significantly higher proportion of their liking for him to the group that he represented than the other group to which he belonged. The ART interpretation is that newly encountered group members, no matter how much they are liked, are not likely to be adopted as exemplars, and their characteristics are not likely to become part of the underlying representa- tion that informs general attitudes, unless they are viewed as “legitimate” representatives of that group.’

F. SUMMARY OF RESEARCH RELEVANT TO THE REPRESENTATION POSTULATE

To review, the representation postulate of Attitude Representation The- ory holds that a person’s response to any attitude-relevant stimulus depends not only on the perceived properties of that stimulus and the situation surrounding it, but also on the subjective representation of that stimulus by the person. Supporting the ART model shown in Figure 1, studies have shown that people do often activate such subjective representations when they assess their attitudes and that they base their responses at least in part on the evaluative implications of whichever aspects of the subjective representation were activated (Sia et al., in press). When they activate the same representation at two separate times, they respond consistently (Sia et al., 1997) and are less likely to alter their attitudes in response to persuasive arguments (Sia, Thomas, & Lord, 1998). When they activate a different representation that is more likable or less likable than the first, they change their attitudes in a predictable direction (Sia et al., 1997). These studies support the contention that attitude-relevant responses involve activating aspects of the attitude objects’s subjective representation.

’ Related studies (e.g., Smith, Fazio, & Cejka, 1996) showed that people who belong to two social categories (e.g., Mike Tyson as a boxer or as a rapist) are identified more closely with the category toward which the individual has more frequently assessed his or her attitude. One category or the other might also be rendered more important for a particular judgment by the norms that exist within a particular culture (Smith & Zarate, 1992).

Page 31: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 295

If attitude-relevant responses involve activating aspects of the attitude object’s representation, then it follows that one might change attitudes by changing aspects of the representation. Consistent with this reasoning, other studies have shown that positive contact with one member of a social category is more likely to ameliorate negative attitudes toward the group as a whole when the person’s initial attitude involves an abstract exemplar rather than a specific, namable exemplar (Werth & Lord, 1992). Likewise, positive contact with one group member ameliorates negative attitudes toward the group to the extent that such contact alters the way the group is represented. Furthermore, positive contact with a person who belongs to two equally disliked groups is also more likely to ameliorate negative attitudes toward whichever group the person more clearly represents (Des- forges et al., 1991).

All of these studies-of exemplar activation, response stability, resistance to persuasion, exemplar accessibility, exemplar specificity, and attitude change via equal-status contact-thus provide results consistent with deri- vations of the representation postulate. We turn next to studies that have addressed derivations of our second, matching postulate.

IV. Research Relevant to the Matching Postulate

The matching postulate of ART holds that the closer the match between the representations and perceptions to which a person is responding in different situations, the more consistency there will be in that person’s responses to attitude-relevant stimuli. The ART model in Figure 1 shows four types of representation-based assumptions: target identity, target char- acteristics, associated responses, and surrounding context. Relevant studies have provided evidence concerning the matching postulate for each of these four types of assumptions. In one set of studies we tested whether consistent exemplar activation promotes consistent attitude-relevant responding (of- ten labeled “attitude-behavior consistency”). In a second set of studies we tested whether a match between assumptions and perceptions of an attitude target’s characteristics promotes attitude-behavior consistency. Other stud- ies tested whether the matching principle can also be applied to matches between assumptions and perceptions with regard to actions and to contexts. Finally, additional studies investigated the impact on the matching princi- ple of factors that increase or decrease the relevance and importance of representation-based assumptions.

Page 32: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

296 CHARLES G. LORD AND MARK R. LEPPER

A. CONSISTENT EXEMPLAR ACTIVATION AND RESPONSE GENERALITY

Technically, studies that are described in the literature as concerning “attitude-behavior consistency” do not measure the relationship between “a behavior’’ and “an attitude.” According to one widely accepted defini- tion, an attitude is “a psychological tendency that is expressed by evaluating a specific entity with favor or disfavor” (Eagly & Chaiken, 1993, p. 1). A tendency is a bias, inclination, or propensity to do something. It is not a sure thing. When we say that “Susan tends to vote for conservative political candidates,” we do not mean that she always votes for the more conservative candidate in every election. Any single sample of Susan’s voting behavior, therefore, is just that-only a sample. Similarly, any single attitude-relevant response is merely a sample.

An individual’s underlying attitude is a hypothetical construct-a hypoth- esized thread of consistency that weaves together many of a person’s thoughts, feelings, and actions. The attitude is something that we infer from one or more attitude-relevant responses. When LaPiere’s (1934) proprietors answered “no” on the questionnaire about serving Chinese people, the answer was not an “attitude.” It was an attitude-relevant evaluative re- sponse. When they served the actual Chinese couple, that was also an attitude-relevant evaluative response.

One could infer an underlying attitude from either of these responses. It is customary in the literature to call a written or oral response to only the name of the attitude object an “attitude” or “attitude measure” and to call a response to any actual or hypothetical behavioral situation a “behavior” or “behavioral intention,” but technically either could be con- sidered an expression of the underlying attitude. Studies of “attitude- behavior consistency,” then, are actually studies of consistency between two or more evaluative responses to the same stimulus, or studies of re- sponse generality.

From a practical standpoint, however, a response to the mere name of the attitude object, as in “What is your attitude toward Chinese people” or “What is your attitude toward capital punishment?” ought usually to be the most direct and comprehensive expression of the underlying attitude, because it theoretically includes all possible instantiations of the target, all possible evaluative responses, and all possible contexts. It is also the sort of question that has been most influential in the great success of the attitude concept. Hence, although individual construals of targets, responses, and contexts lie at the heart of ART, in describing studies of “attitude-behavior consistency,” we will follow the tradition and label one of the measures an “attitude” and the other a “behavior” or “behavioral intention.” The

Page 33: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 297

reader should remember, however, that an answer to a single attitude question is merely an expression of an underlying tendency and should not be confused with the attitude itself.

B. MATCHING AND ATTITUDE-BEHAVIOR CONSISTENCY

That said, let us examine the implications of ART for attitude-behavior consistency. Consider one common case: Suppose that the attitude-relevant response required at Time 2 in Figure 1 is not a second attitude assessment, but is instead an assessment of behavior or behavioral intentions toward the category as a whole, rather than toward specific category members. ART would argue that when people make an attitude-relevant response to a general stimulus rather than to a specific one, they will use the default values that they assume from their underlying representation of the stimu- lus. People who activate Bill Clinton as their exemplar of politicians when they assess their attitudes, for instance, might also activate Bill Clinton as their politician exemplar when they decide whether to support an otherwise unspecified politician or politicians (e.g., to sign a petition to encourage university administrators to bring more “politicians” to campus as guest lecturers). They have no information, that is, that would allow them to replace Bill Clinton with a different exemplar.

With no behavioral target other than the group as a whole, attitude- behavior consistency will be greater when the assumed target identity at Time 1 matches rather than mismatches the assumed target identity at Time 2. Put another way, people who are thinking about Bill Clinton at both times-when they assess their attitudes and when they choose a behavior or behavioral intention toward otherwise unspecified “politicians”-will display greater consistency or generality across evaluative responses than will people who are thinking about Bill Clinton on only one of the two occasions and about a different politician on the other occasion. Further- more, it should be possible to predict not only the occurrence of attitude- behavior inconsistency, but also the direction of attitude-behavior inconsis- tency.

To test these hypotheses, Sia and her colleagues (1997, Experiment 2) obtained students’ attitudes and first exemplars for politicians and several other social categories in an initial questionnaire. Two weeks later, the same students participated in an “unrelated” study in which they reported their liking for 150 well-known objects and people (e.g., Rice Krispies or Madonna), a few of which were the politician exemplars the individual student had named earlier. Two weeks after that, the students again com- pleted the initial questionnaire, in which they provided politician and other

Page 34: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

298 CHARLES G. LORD AND MARK R. LEPPER

0.5 - 2 0.4 - 0 zz m - 0.3 - E 8 8 0.2 - ‘E B O * ’ - 2 CI 0 -

I al

E a -0 .1 -

-0.2 - -0.3

category exemplars for a second time. This questionnaire session, however, was “unexpectedly” interrupted by a representative of the student council who had supposedly received university permission to distribute petitions for bringing an increased number of homosexuals, environmentalists, and politicians to speak on campus. For each of these three (otherwise unspeci- fied) groups, the student was asked to sign the petition, and to agree to write letters, make phone calls, work in groups, and commit to other supportive activities.

Figure 9 shows how well students’ initial attitudes toward politicians predicted subsequent behavior (signing the petition) and behavioral inten- tions (expressing a willingness to engage in the other supportive behaviors). For both measures, attitude-behavior consistency was higher for students who named the same politician to represent the category both times (match- ing exemplars) than for students who named different politicians (mis- matching exemplars). Further analyses revealed a strong correlation, for

--

o.6 1 0 Mismatching exemplar 0 Matching exemplar

I 1

Signing the Other supportive petition activities

Behavioral response

Fig. 9. Attitude-behavior correlations for behavior (signing a petition) and behavioral intentions (willingness to engage in other supportive activities) toward otherwise unspecified “politicians,” by participants who named the same or a different politician exemplar 1 month apart. (Data from Sia, Lord, Blessum, Ratcliff, & Lepper, 1997, Experiment 2.)

Page 35: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 299

those who named a different exemplar, between the difference in liking for the two exemplars and the extent to which students behaved either more positively or more negatively than would have been expected from their initial attitudes. Those who named a politician exemplar they liked better than the one they had named a month earlier made more positive behavioral decisions than their earlier attitudes had suggested. Those who named a politician exemplar they liked less made more negative behavioral decisions than their earlier attitudes had suggested. ART thus allowed researchers to predict not only the occurrence of attitude-behavior incon- sistency, as had been predicted in dozens of earlier studies, but also its specific direction.

C. MATCHING TARGET IDENTITIES

The study just described was designed to test the impact of consistent exemplars on attitude-behavior consistency when the behavior is directed toward the group as a whole. Another frequent scenario occurs when behavior is directed toward one or more individuals who are identified as a subtype within the larger group. Many categories have a hierarchical structure, with distinctions at various levels of the hierarchy that play an important role in perceptions and judgments concerning both natural (Rips, Shoben, & Smith, 1973; Rosch, 1978) and social objects (Brewer, Dull, & Lui, 1981; Brewer, Ho, Lee, & Miller, 1987; Hewstone & Brown, 1986; Stangor, Lynch, Duan & Glas, 1992; Weber & Crocker, 1983). Politicians, for instance, include subtypes such as Democrats and Republicans, plus finer distinctions such as southern Democrats and northern Democrats. From this perspective, an individual exemplar, such as Bill Clinton or the Chinese man who accompanied LaPiere (1934), represents the most specific level in the representational hierarchy for social categories.

ART, then, makes the same prediction for matching subtypes and atti- tude-behavior consistency as it does for matching exemplars and attitude- behavior consistency. Attitudes will predict behavior and behavioral inten- tions better when assumed and perceived target subtypes match rather than mismatch. If LaPiere‘s (1934) proprietors had in mind the subtype “Chinese laborers” and not “Chinese academics” when they completed the attitude questionnaire, then their attitudes toward members of the Chinese race might have predicted their behavior had a coolie laborer requested service, but not when an academic couple requested service.

To test this derivation, Ramsey, Lord, Wallace, and Pugh (1994, Experi- ment 2) asked undergraduate students to report their attitudes toward former substance abusers and to write a paragraph describing the typical

Page 36: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

300 CHARLES G . LORD AND MARK R. LEPPER

former substance abuser. The researchers classified these descriptions ac- cording to the subtype of substance abuser the student had in mind. Most of the descriptions involved one and only one type of substance abuser, such as an alcohol, cocaine, heroin, or marijuana abuser.

Approximately three weeks later, the 80% of the students who had ex- pressed negative attitudes toward “substance abusers” participated in an “unrelated” study in which they were told that students from their own university were working in pairs with partners from another university to conduct a joint “survey of campus life.” To facilitate the survey, the students were allowed to read descriptions of two potential partners and to indicate, for each of 23 possible activities that varied in psychological closeness, from “just be introduced” to “host the student for the weekend,” how willing they would be to interact in that way with each of the potential partners. Unknown to the students, the two descriptions had been carefully prepared so that pretest participants had rated them as equally likable, but one of these potential partners was said to have been treated successfully for abusing the substance the student had described, whereas the other was said to have been treated successfully for abusing one of the other sub- stances (and was thus identified as belonging to a different or mismatch- ing subtype).

As predicted by ART, students displayed greater attitude-behavior con- sistency toward a substance abuser who was identified as belonging to the subtype they had described when they had earlier assessed their attitudes than toward a substance abuser who was identified as belonging to a differ- ent subtype. Figure 10 shows attitude-behavior correlations, both for the number of activities the students were willing to undertake with each of the potential partners and for the psychological closeness of the average activity to which they agreed. Clearly, attitudes predicted these supportive activities and their psychological closeness better when the target was identi- fied as from a matching than from a mismatching subtype.

The subtypes study also yielded two other results that supported the matching postulate of ART. First, one week later, all students returned, expecting to meet their partner. The experimenter randomly told students that their partners were from either the matching or a mismatching subtype. The partner had apparently left the room momentarily, but had left some books where he or she was sitting. The students, all of whom had negative attitudes, chose to sit farther from a substance abuser of the subtype that matched rather than mismatched the subtype they had earlier described as representing the typical category member.

Second, an additional group of students were told only that their potential partner was a former substance abuser, but got no descriptions that would help them to identify the subtype. According to ART, these additional

Page 37: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 301

0.6

0 . 5

16 C 0 0.4 P m V 5 0.3

-

.- 6 z 0.2

s 4

0.1 c .- *: 4 0

-0 .1

Mismatching subtype Matching subtype

Supportive Psychological activities closeness

Behavioral response

Attitude-behavior correlations for supportive activities and psychological close- ness of those activities toward a substance abuser who was said to belong to a subtype that matched or mismatched the substance abuser subtype participants described when they assessed their attitudes. (Data from Ramsey, Lord, Wallace, & Pugh. 1994, Experiment 2.)

Fig. 10.

students, with the subtype unspecified, should have retained their assump- tions about the target’s identity and assumed that the partner belonged to the subtype they had in mind when they assessed their attitudes. Their behavioral responses should have been equivalent to those of students who got exactly the subtype they expected. Consistent with this reasoning, the behavioral responses of the additional students were almost identical to those of students who got the “confirming” description of someone from the subtype they associated with their attitude, and significantly different from those of students who got a “disconfirming” description of someone from a different subtype.

D. MATCHING TARGET CHARACTERISTICS

Sometimes a behavioral target is unspecified, as it was in the study of bringing “politicians” to campus, or only partly specified, as in the subtype

Page 38: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

302 CHARLES G. LORD AND MARK R. LEPPER

study. More frequently, the behavioral response is directed toward a specific target with known characteristics. LaPiere’s (1934) proprietors, for instance, did not have to decide whether to serve a tour bus full of unspecified “Chinese people,” but instead whether to serve a specific Chinese couple, whose characteristics may have been different from those that the propri- etors had in mind when they had assessed and reported their attitudes. At such times, the ART prediction is clear: Response consistency depends on the match between the assumed and perceived target characteristics. Attitudes will predict behavior better when assumed and perceived target characteristics match rather than mismatch.

To test this hypothesis, which concerns target characteristics rather than the identity or subtype of exemplars, Lord, Lepper, and Mackie (1984, Experiment 1) asked Princeton undergraduates to express their attitudes toward various campus eating clubs (similar to fraternities and sororities) and to list six trait characteristics that they attributed to “the typical mem- ber” of each club. One student, for instance, described the typical member of Cap & Gown as ambitious, aggressive, playful, approval-seeking, domi- neering, and persistent. Approximately three months later, in an “unre- lated” study, the same students learned that they would be working closely on a task with one of two potential partners.

To help the students express a preference for working with one or the other of these two potential partners, the experimenter let them read two “thumbnail sketches” that had supposedly been written by an experienced clinical psychologist as part of an earlier interview study. According to the thumbnail sketches, both potential partners belonged to the same eating club-one of those on the earlier attitude questionnaire. Although the student did not realize it, the two sketches had also been meticulously prepared and pretested so that, although the two potential partners were equally likable, one had 100% of the trait characteristics the individual student had earlier attributed to the typical member of the club, whereas the other potential partner had only 50% of those characteristics and was portrayed as exactly opposite to the student’s reported attributions on the other 50%.

Attitudes toward the eating club predicted commitment to work with the 100% typical club member ( r = .69) significantly better than they predicted commitment to work with the 50% typical club member ( r = .32). An ART analysis is that for six of the 100% typical member’s characteristics, assumed and perceived target characteristics were identical-a match found on only three of the 50% typical member’s characteristics. For the other three characteristics, perceived characteristics were opposite to assumed characteristics.

Page 39: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 303

The experimenters also asked participants, at the end of the experiment, to describe the “typical” member of each eating club again. LaPiere’s (1934) proprietors reported no less negative attitudes when they had pre- viously served an obviously cultured Chinese couple than when they had not, so it seemed unlikely that a one-paragraph written description, no matter that it was 50% disconfirming, would have much impact on the underlying representation itself (see Lord, Ross, & Lepper, 1979). Consis- tent with this reasoning, the students changed their portraits of the “typical” club member very little, and changed their portraits no more on the trait dimensions that the potential partner’s description had explicitly discon- firmed than on the trait dimensions that a potential partner’s description had explicitly confirmed. In short, the underlying attitude representation had escaped unscathed, ready to be activated the next time an evaluative response was required.

The ART principle of matching target characteristics applies to all types of representations.* In associated systems theory (Carlston, 1994), for in- stance, cognitive representations can include not only verbal propositions such as personality traits, but also visual images. In addition, cognitive representations need not correspond with reality. An individual might get a visual image of the typical gay male as dressed in drag, for instance, or the typical mental patient as having very wide, frantic eyes and wild, bushy hair. It does not matter whether these mental images are fanciful or realistic. According to ART, fanciful mental images should affect attitude-behavior consistency just as much as realistic ones, as long as they are part of the attitude representation.

To explore the role of mental images in attitude representations and attitude-behavior consistency, Lord, Desforges, Ramsey, Trezza, and Lep- per (1991, Study 1) created attitudes toward a social group from scratch, using a traditional concept formation task. Students viewed yearbook pho- tographs of women with different hairstyles. For each photograph, they guessed whether the woman belonged to an unnamed “organization” (pre- sumably a sorority) at another university, after which the experimenter told them the “correct” answer. In reality, the experimenter told students that the woman belonged to the organization each time she had any two of three key physical characteristics-short hair, dark hair, and bangs-and that she did not belong when she had only one or none of the three. Although students had no idea how they formed the concept, all of them could guess at better than chance, by the end of the training session, which women belonged to the group and which did not. To manipulate attitudes,

As Zajonc (1998) suggests, a “representation” need not be purely cognitive. but can include affective and behavioral aspects as well.

Page 40: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

304 CHARLES G. LORD AND MARK R. LEPPER

the experimenter also told students each photographed woman’s opinions on controversial issues. As expected (Byrne, 1971), students who were told that most group members agreed with them reported a positive attitude toward the fictitious group, whereas students who were told that most group members disagreed with them reported negative attitudes toward the group.

Subsequently, all students attended a class taught by a woman who was said to have been a group member at the other university. The woman wore either her normal hair style, which was medium length, dirty blond, half off the forehead, or she wore a short, dark wig with bangs. The students were asked to evaluate her lecture for a professor in her graduate speech communications course, believing that their evaluations would influence the woman’s grade in the course.

Although the students had never seen a member of the organization who had all three of these physicai characteristics during the training phase, they were expected to have formed a visual image of the prototypic group member and to use that image as part of their cognitive representation for the group. Match to the cognitive representation, in turn, was predicted to influence attitude-behavior consistency, even though the visual image had not been the basis for adopting a positive or negative attitude. Attitudes predicted class evaluations better for students who saw the woman lecture in the wig than with her normal hair, presumably because when she wore the wig, she matched a visual image the students had constructed of the typical group member.

Similarly, in a follow-up study (Lord et al., 1991, Study 2), students responded more in line with their attitudes toward former mental patients when they had an opportunity to interact socially with a former mental patient who had wide eyes and bushy hair than with a former mental patient who had normal eyes and hair, even though the two men’s looks had been rated as equally likable by pretest students, none of whom thought that the man with wide eyes looked anything other than normal. Whatever characteristics comprise an individual’s attitude representation, whether they be personality traits or physical features, can affect attitude-behavior consistency when perceived target characteristics do not match assumed target characteristics. General attitude will better predict responses to cate- gory members who fit one’s stereotype and “look the part” than to category members who do not.

E. MATCHING ASSOCIATED RESPONSES

As shown in Figure 1, ART holds that an attitude representation can involve not only default assumptions concerning a target’s identity and

Page 41: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 305

characteristics, but also default assumptions concerning the types of re- sponses typically associated with that target. Ajzen and Fishbein (1977) were the first to observe that attitudes predict behavior better when the attitude and the behavior correspond in terms of both target and action. Lapiere’s (1934) restaurant and hotel proprietors, for instance, might have responded negatively to Chinese people in many ways, including voting for stricter immigration laws, refusing to donate money to help the Chinese poor, and avoiding public transportation because it might put them in close contact. LaPiere (1934) might have happened to measure the one type of evaluative response (accepting Chinese people as paying customers) in which the proprietors’ deeds would not match their words-even though both responses in LaPiere’s (1934) study did concern the same action.

We know from previous research that people’s actions toward an attitude object can alter their attitudes (Bem, 1972; Festinger, 1957). Indeed, actions toward the object have long been regarded as one of the three components that form the basis for attitudes (Breckler, 1984; Esses, Haddock, & Zanna, 1994), and actions are widely regarded as one of three ways in which attitudes are expressed (Eagly & Chaiken, 1993). It seems likely, then, that one aspect of the representation for a stimulus might be concerned with the types of responses that the stimulus typically either has elicited or might elicit from the individual. In assessing their attitudes, that is, people might activate both assumptions about the target and assumptions about attitude- relevant actions. If they do, then the consistency or generality of evaluative responses toward that stimulus might be influenced as much by a match between the assumed and perceived actions as by a match between the assumed and perceived target identity or characteristics. Women who say they have a positive attitude toward Bill Clinton might mean they would vote for him to continue as president, not necessarily that they would trust him in a long-term romantic relationship.

Thomas, Lord, and Lepper (1998) noted that studies of stereotyping had established the types of personality traits, physical features, and exemplars that people associate with attitudes toward negatively stigmatized or un- fairly treated minority groups (Stangor & Lange, 1994), but that no previous studies had identified the types of actions that might be associated with such attitudes. They sought to create an action typology that might be used to determine whether a match between assumed and perceived actions would promote response consistency. In terms of the ART model shown in Figure 1, Thomas and her colleagues (1998) wanted a typology they could use to test whether attitudes predict behavior better when assumed and perceived behavioral responses match than when they do not match.

In an initial study, the researchers asked 100 college students to assess their attitudes toward gay males, politicians, and persons with AIDS. For

Page 42: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

306 CHARLES G. LORD AND MARK R. LEPPER

each social category, the researchers also asked the students to list up to five actions that they either had performed or might perform toward one or more members of the category, plus up to five actions that people with the opposite attitude to theirs had performed or might perform. After eliminating nonactions such as characteristics and cfeelings, the researchers were left with 1703 different actions. Many of these specific actions were redundant, such as “hug them” and “give them a hug.” Eliminating all such redundancies left 301 specific actions. The researchers classified these 301 specific actions into eight positive and eight negative action types. Naive raters validated the researchers’ placement of the 301 specific actions into the 16 action types.

The researchers then asked additional students to rate the similarity of each pair of the eight positive action types and the similarity of each pair of the eight negative action types. Separate ADDTREE clustering solutions produced the tree diagrams shown in Figure 11, from which the researchers concluded that the 16 action types formed eight bipolar action dimensions. Thomas and her colleagues (1998, Experiment 1) further discovered that

r

I L PURSUE

SEEK CONTACT ’t IN-rERAcr

HELP

SUPPORT GROUP 46 AGREE W T H

I-,RESPECT

OPPOSE GRO

DISAGREE WITH

Fig. 11. ADDTREE solution derived from similarity ratings of eight types of positive and eight types of negative actions found in the specific actions that students associate with their attitudes toward various social categories. (Data from Thomas. Lord, & Lepper, 1998, Experiment 1.)

Page 43: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 307

the eight bipolar action dimensions differed in how frequently the initial participants had said they associated each dimension with their attitudes. Of the 1703 specific actions that students had listed, 369 were on the respect versus disrespect dimension; 255 were on the support versus oppose dimension; 229 were on the help versus harm dimension; 202 were on the seek information about versus ignore dimension; 197 were on the pursue versus be aloof dimension; 175 were on the interact versus shun dimension; 161 were on the agree with versus disagree with dimension; and 115 were on the contact versus no contact dimension.

These differing frequencies suggested that, at least for these three social categories, some action dimensions are more likely to be part of people’s representations than others. Just as participants in the exemplar study described earlier were more likely to associate Bill Clinton with their atti- tudes toward “politicians” than other exemplars, so participants in the action types study were more likely to associate actions such as being courteous, rude, accepting or scornful (respect versus disrespect) with their attitudes toward the three categories than actions such as accepting or rejecting their ideas (agree with versus disagree with), or seeking physical closeness or distance (contact versus no contact).

Thomas and her associates (1998) believed that these action dimensions and their relative frequencies were not specific to the three social categories, but rather represented general action tendencies that people cognitively associate with social category attitudes. To test this possibility, they re- viewed the published literature on attitude-behavior consistency toward social groups or categories. Examining only studies published between 1934 (the publication date of Lapiere’s study) and 1994 that measured attitudes as an individual difference (not as a stereotype), the researchers identified 30 publications that had reported 42 effect sizes for predicting from attitudes to a wide range of behaviors and behavioral intentions. Then they classified each behavioral measure into one of the eight bipolar action dimensions and validated with naive raters that their classifications were reliable.

Using meta-analytic techniques (Cooper & Hedges, 1994), the research- ers found a significant relationship between the frequency with which the 100 students had spontaneously associated an action dimension with their attitudes and the magnitude of the attitude-behavior correlation for pub- lished studies that used such a dimension. Figure 12 shows the mean effect sizes for studies within each of three sets of action dimensions, arranged according to how frequently the 100 students generated each attitude action dimension (Thomas et al., 1998, Experiment 2). The linear arrangement of these mean effect sizes is especially compelling because most of the published studies investigated attitudes toward social categories different from the ones used to develop the attitude action typology. Approximately

Page 44: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

308

0.6 -

0.5 -

0.4 -

CHARLES G . LORD AND MARK R. LEPPER

u)

u)

.- 3

8 - 0.3

a= w 0.2

0.1

0 Least frequent Medium frequent Most frequent

actions actions actions

Studies that used each type of action as “behavior“

Mean effect sizes in published studies of attitude-behavior consistency that used as the “behavior” action dimensions that varied in how frequently students associated those actions with their attitudes toward social categories. (Data from Thomas, Lord, & Lepper, 1998, Experiment 2.)

Fig. 12.

half of the published studies, for instance, concerned attitudes toward Blacks, which was not one of the three categories used in developing the typology.

F. MATCHING CONTEXTS

Finally, according to ART, an attitude representation can contain not just assumed target characteristics and types of evaluative response, but also assumptions about the social context that might accompany an evaluative response. When LaPiere’s proprietors assessed their attitudes toward “ac- cepting members of the Chinese race,” for instance, they might have meant not only a particular type of Chinese person and a particular way of serving that person, but also a particular type of context. Ross and his colleagues, for example, have demonstrated the ubiquity with which people answering questions about their own or others’ behaviors, or making a decision about

Page 45: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 309

how to behave, typically go far beyond the information immediately avail- able to make inferences about “indiscernible or unspecified” details of the situations in which those behaviors will take, or have taken, place (Griffin & Ross, 1991; Griffin, Dunning, & Ross, 1990; Ross, 1977; Ross & Ward, 1995). This process of construal, which appears to operate in a relatively automatic and unrecognized fashion, may in turn determine the responses or actions of the individual, just as Asch (1940) might have suggested. LaPiere’s (1934) proprietors might have imagined, for instance, that other customers would become irate and openly voice displeasure had they tried to serve a Chinese person-an imaginary circumstance that, at least accord- ing to LaPiere’s (1934) account, never arose. The proprietors might also have imagined that it would be very easy for them to eject Chinese custom- ers, whereas the real situation imposed formidable obstacles to doing so.

The studies described earlier suggest the benefits for predicting behavior if we can learn more than just an individual’s attitude. In the study of subtypes of former substance abusers (Ramsey et al., 1994), for instance, knowing the assumed target subtype allowed the researchers to predict more precisely than did knowing only the attitude. Of two participants who had equally negative attitudes toward former substance abusers, that is, the researchers could predict that the one who meant by substance abusers “people who used marijuana” would sit farther from an alleged former marijuana abuser. In general, the more we know about the individual’s assumptions regarding target identities, target characteristics, and the type of evaluative response involved, the better we can predict what the individ- ual will do in a specific behavioral situation.

Two well-established social psychology theories hold that the predictive power of attitudes is also enhanced by knowing about an individual’s subjec- tive norms and perceived behavioral control. According to the theory of reasoned action (Ajzen & Fishbein, 1980), attitudes guide behavior through a reasoning process that takes into account the subjective norms surround- ing a behavior. In one relevant study, Manstead, Proffitt, and Smart (1983) tried to predict from expectant first-time mothers’ attitudes toward breast- feeding their babies to their actual behavior after the baby was born. The researchers realized, however, that attitudes by themselves might not be sufficient, because once they became mothers, many of the women would be subjected to intense social pressure. Before the birth, many of the women might not have discussed their private decision about breast-feeding versus bottle-feeding with others, assuming that everyone else would just quietly go along with their own personal preference. Once the baby was born, in contrast, their mothers, spouses, friends, and relatives might have exerted subtle and not-so-subtle influence on the behavioral situation.

Page 46: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

310 CHARLES G. LORD AND MARK R. LEPPER

Because Manstead and his colleagues (1983) realized that people do not act in a social vacuum, they did not merely measure the women’s attitudes toward breast-feeding. Instead, they also asked the women about the subjec- tive norms that they associated with breast-feeding, posing questions such as “How much does your husband want you to breast-feed and how much do you care what he wants?” and “How much does your mother want you to breast feed and how much do you care what she wants?” They then calculated the positive or negative impact of subjective norms by multiplying each person’s assumed desires by how important they were to the expectant mother, and then summing across persons.

Attitudes by themselves predicted whether the women would breast-feed, but knowing about the women’s subjective norms allowed the researchers to predict subsequent breast-feeding significantly better than merely knowing about the attitudes by themselves. Information about subjective norms has also added to the predictive power of attitudes in studies of behaviors as diverse as getting exercise, recycling, dating, the use of contraceptives, and AIDS risk prevention (Bentler & Speckart, 1979; Cochran, Mays, Ciarletta, Caruso, & Mallon, 1992; Doll & Orth, 1993; Gallois et al., 1992; Page1 & Davidson, 1984; Vining & Ebreo, 1992; Winslow, Franzini, & Hwang, 1992). The ART interpretation is that, if we are trying to predict from attitudes to behavior, it helps to know what type of facilitating or inhibiting context people assume at the time they assess and report their attitudes and to know whether those assumptions match perceptions of a specific behav- ioral situation.

Subjective norms might be one part of the assumed context; perceived control might be another. In the theory of planned behavior (Ajzen, 1985), people differ in how much control they think they have over whether to behave in a particular way. Perceived behavioral control is sometimes unnecessarily limited by a misperceived lack of time, money, skills, and other resources. People sometimes do not realize how much they can accom- plish if they would only try (Bandura, 1977). Assumptions about perceived control may be more salient in the representation of some attitude objects than others. Especially for actions that involve self-discipline such as dieting, however, people often overestimate how much control they can exert.

According to the theory of planned behavior, then, we can increase the predictive power of attitudes by knowing how capable people think they are of engaging in a particular behavior. The theory has been well supported by studies of attitude objects as diverse as eating less, getting better grades, and engaging in health-protective behaviors such as breast self-exams (Aj- zen & Madden, 1986; Madden, Ellen, & Ajzen, 1992: McCaul, Sandgren, O’Neill, & Hinsz, 1993; Schifter & Ajzen, 1985). In each case, merely knowing that someone has a positive attitude toward engaging in the behav-

Page 47: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

AlTITUDE REPRESENTATION THEORY 31 1

ior only moderately predicts whether he or she will actually do it. Informa- tion about the individual’s perceptions of control allows researchers to predict behavior in the real situation significantly better than merely know- ing about attitudes, or even knowing about both attitudes and subjective norms (Ajzen, 1987). According to ART, both subjective norms and per- ceived control can constitute assumptions within an individual’s representa- tion of a stimulus when assessing his or her attitude. By learning these assumptions about contextual factors, researchers can better predict when behavior in attitude-relevant situations will match versus mismatch gen- eral attitudes.

G. WHEN MATCHING REPRESENTATIONS BECOME IRRELEVANT

One of the key concepts in the ART model, as suggested by Figure 1, is that the assumptions that inform responses to a general attitude question can be contradicted by the more direct perceptions that define a specific behavioral situation. Assumptions about barefoot coolies in LaPiere’s (1934) study, for instance, might have been rendered relatively unimportant when a smiling, cultured, well-dressed couple requested service, because perceptions of concrete, specific target characteristics were more immediate and salient. More generally, ART predicts that discrepancies between as- sumptions and perceptions will have less impact on attitude-behavior con- sistency when either the assumptions or the perceptions seem vague or unimportant, and thus receive small or zero weights. One such instance arises when particular assumptions play a negligible role in an individual’s attitude representation.

Lord, Desforges, Fein, Pugh, and Lepper (1994) applied this reasoning about the importance or relevance of target characteristics to attitudes concerning social policy issues such as capital punishment and welfare. In one of their studies, Lord and his colleagues (1994, Study 4) had students draw “attitude concept maps” about welfare, by starting from a central node labeled “welfare” and drawing whatever nodes and pathways came to mind-a procedure used in previous studies to investigate people’s asso- ciative networks (Armbruster & Anderson, 1984; Holley & Dansereau, 1984; Reder & Anderson, 1982). In this study, these graphic representations contained three reliably recognizable types of nodes: person nodes (e.g., minorities, promiscuous women, people lacking in ambition), principle nodes (e.g., social safety net, wasteful use of tax dollars, fosters depen- dency), and factual nodes (e.g., begun in the 1930s).

Page 48: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

312 CHARLES G. LORD AND MARK R. LEPPER

Two weeks after drawing the concept maps, the same students partici- pated in an “unrelated” study in which they pretended to be members of a welfare review board charged with deciding the fate of welfare recipients who had been accused of cheating by having hidden sources of income. In each instance, the caseworker had discovered unusual assets, such as a new projection television set, which the welfare recipient explained by bargain shopping or claimed were merely on loan from someone else. The facts of the case were deliberately ambiguous, so that positive or negative attitudes might bias the student’s decisions about whether to cut off each recipi- ent’s funds.

Each student read and made decisions about two case histories. The recipient in one of the case histories matched perfectly the prevailing stereo- type of people on welfare. She was said to be intellectually limited and somewhat overwhelmed by her responsibilities, lacking in self-confidence, lacking in drive, and resigned to her impoverished lifestyle. The recipient in the other case history mismatched the prevailing stereotype. She was said to be shrewd and exceedingly ambitious, proud, self-assured, competi- tive, and wasting an excellent education. These two sets of target character- istics, of course, had previously been rated by other students as equally (un)- likable.

The researchers counted the number of person nodes and the number of principle nodes in each student’s attitude concept map. Some students included more person nodes than principle nodes. Others included more principle nodes than person nodes. The ART prediction was that students who included relatively many person nodes in their attitude representation of welfare would show less attitude-behavior consistency when faced with mismatching target characteristics than would students who included rela- tively few person nodes, because target characteristics would matter more to the former than to the latter. Students who included few person nodes in their attitude concept maps would presumably use lower weights for assumed target characteristics, so a mismatch between assumptions and perceptions would be less important for them than for others.

Figure 13 shows that this prediction was confirmed. Students who in- cluded many person nodes behaved in a more attitude-consistent way (those with positive attitudes deciding to keep this person on the welfare roles and those with negative attitudes deciding to cut her off) toward the matching or stereotypical than the mismatching or nonstereotypical recipient. Students who included few person nodes in their welfare concept maps, however, made no such distinction. When they were asked, after making the decision, how typical each case history seemed of welfare recipients, both “person- oriented students” and “principle-oriented students” equally recognized that one recipient was far more typical than the other. Indeed, when asked

Page 49: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 313

0.6

0 .5

C

a o 0.4 .- C -

0.3 s L 0 .s 0.2

5 s g o

0.1 - .-

-0.1

- 0 . 2

1 0 Mismatching recipient

Matching recipient

Few person nodes

Many person nodes

Nodes in welfare attitude concept map

Attitude-behavior correlations for a recipient who matched versus mismatched stereotypical characteristics, for students who included either few or many person nodes in their welfare attitude concept maps. (Data from Lord. Desforges, Fein, Pugh, & Lepper, 1994.)

Fig. 13.

directly which of the two was more typical, every student answered cor- rectly. To some of them, however, the information about target characteris- tics was less relevant for making a behavioral decision.

The study of attitude concept maps and welfare thus supported the ART prediction that the more heavily weighted one particular feature is within an individual’s attitude representation, the more important matches or mismatches on that feature will be to consistency between two attitude- relevant responses. The other side of the coin-the lessened impact of information in the immediate stimulus when that information is made less salient or important-was studied by Blessum, Lord, and Sia (1998), in an investigation of the effects of cognitive load and positive mood on attitude-behavior consistency.

According to ART, individuals who assess their general attitudes rely on default assumptions from an attitude representation (e.g., that gay men are sensitive, nonathletic, have feminine characteristics, and display a good sense of humor). When they encounter a man they know to be gay, these representation-based assumptions are integrated with perceived individuat-

Page 50: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

314 CHARLES G. LORD AND MARK R. LEPPER

ing characteristics (e.g., that this specific gay man lifts weights and is very serious). To the extent that the representation-based assumptions dominate the impression that the individual forms of a specific gay man, attitudes will predict behavior, because the representation that informs the attitude and the representation that informs the behavior will both be based on assumed default values. When immediate perceptions of individuating char- acteristics dominate the impression, in contrast, attitudes might not predict behavior, because the perception that informs the behavior might be differ- ent from the representation that informs the attitude. Assumptions about target characteristics, as we have seen in the previously described studies, might not match perceptions of a specific category member.

Many studies of stereotype impact have identified two factors that reliably increase reliance on representation-based assumptions over more directly perceived characteristics in forming an impression: cognitive load and posi- tive mood (Forgas, 1995; Gilbert, 1989; Mackie & Worth, 1989; Schwarz, Bless, & Bohner, 1991). In one illustrative study of cognitive load, for instance, students who performed a complicated letter-matching task while listening to a tape of a woman they thought was Asian concluded that she was more timid, intelligent, and calm (the stereotypic representation) than did students who listened to the same tape without performing a concurrent task (Gilbert & Hixon, 1991, Experiment 2). Similarly, in one representative study of positive mood, students who read the case history of a man named Juan Garcia found him more guilty of assault (thus relying more on the stereotypic representation of Hispanic men as violent) than did students who read the same case in a neutral mood (Bodenhausen, Kramer, & Susser, 1994).

Cognitive load increases reliance on representation-based assumptions over individuating characteristics because it decreases the cognitive capacity available to integrate the assumptions with perceptions of the immediate stimulus. Positive mood, in contrast, increases reliance on representation- based assumptions by reassuring people that all is well, so that in-depth thinking about the target is not required (Bless, et al. 1996; Macrae, Milne, & Bodenhausen, 1994; Schwarz & Clore, 1996). In both circumstances, how- ever, people rely more on their default assumptions and less on perceptions of a person’s individuating characteristics than they might otherwise.

Blessum and his colleagues (1998) asked students to report their attitudes toward several social categories, one of which was gay men. Two to three weeks later, in an “unrelated” experiment, the same students were asked to form impressions of two men from written personality descriptions. Before they read the descriptions, some of the students were placed under cognitive load by being given an eight-digit number to rehearse-a manipu-

Page 51: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 315

lation used previously by Gilbert and Osborne (1989). Other students were placed in an especially positive mood by writing vivid accounts of recent events that made them very happy-a manipulation used previously by Strack, Schwarz, and Gschneidinger (1985). A third group of students merely read the personality descriptions without any preamble.

Both of the men were said to be gay, but one had characteristics that fit the student’s stereotype perfectly, whereas the other gay man had several opposite characteristics. The two descriptions, though, had previously been deemed equally likable by another set of students. After the students had read both descriptions, the experimenter revealed that both men would be visiting the campus soon, as new transfer students from other universities. The student participants were asked to volunteer for as many or as few as they wanted of 15 activities that would help a potential transfer student get to know the new university in advance, such as showing him around campus, introducing him to friends, and hosting him for a weekend visit.

Figure 14 shows how well attitudes toward gays predicted students’ will- ingness to help each of the gay transfer students. In the control condition, where neither cognitive load nor positive mood were involved, attitudes predicted willingness to interact with and help the transfer student who had stereotypical characteristics better than they predicted willingness to interact with and help the transfer student with nonstereotypical character- istics, presumably because his assumed and perceived characteristics matched when he was typical, but not when he was atypical. No such difference occurred in the other two conditions.

Although students in the three conditions equally realized that one of the gay men was more typical than the other, students who were under cognitive load when they formed an impression of the target behaved as much in line with their general attitudes whether he had typical or atypical characteristics, presumably because their limited cognitive resources made them rely more on assumed characteristics. Similarly, students who were in a positive mood when they formed an impression of the target behaved as much in line with their attitudes whether he had typical or atypical characteristics, presumably because their positive mood reassured them that they did not need to think deeply and instead could rely on what has been termed “heuristic processing” (Chaiken, 1987; Schwarz & Clore, 1996). With either cognitive load or a positive mood, students seemed to be saying “he’s one of them, and that’s good enough for me to know what he’s like and how I should behave toward him.” The message from this study and from the welfare study is that the match between assumptions and perceptions matters only to the extent that the perceptions are salient, important, and thus heavily weighted in the overall evaluation.

Page 52: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

316

0.6 - 0.55 -

0.5 - llD C 0

0.45 - 8

2

s s a 0.3 -

L .p 0.4 -

? 0.35 - s

0

0.25 -

0.2

CHARLES G. LORD AND MARK R. LEPPER

-

0 Mismatching characteristics

Ci Matching characteristics

1

1

1

Cognitive load Positive mood No manipulation

Experimental condition

Attitude-behavior correlations for participants who formed an impression of a gay male, who matched versus mismatched participants’ stereotypes, while participants were under cognitive load, in a positive mood, or received no experimental manipulation. (Data from Blessum, Lord, & Sia, 1998.)

Fig. 14.

H. SUMMARY OF RESEARCH RELEVANT TO THE MATCHING POSTULATE

To review, the matching postulate of Attitude Representation Theory holds that the closer the match between the representations and perceptions to which a person is responding in different situations, the more consistency there will be in that person’s responses to various attitude-relevant stimuli. Supporting the ART model in Figure 1, several studies have shown that people who consistently identify the same exemplar to represent a social category also display greater response generalization (attitude-behavior consistency) than do people who identify different exemplars (Sia et al., 1997). Also people who identify a social category with a specific category subtype behave more predictably toward category members who belong

Page 53: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 317

to that subtype than to a different subtype (Ramsey et al., 1994). These studies support the contention that attitude-relevant responses are more consistent when they involve matching rather than mismatching target iden- tities.

Other studies have shown that attitude-relevant responses are more con- sistent when they involve matching than mismatching target characteristics. Thus people behave more predictably toward social group members whose characteristics match than mismatch what they assumed the typical group member to be like, whether those characteristics were personality traits or physical features (Lord et al., 1984, 1991). Still other studies have shown that attitude-relevant responses are more consistent when they involve matching rather than mismatching types of associated responses. Specifi- cally, attitude-relevant actions toward members of a social category may be divided into eight bipolar attitude dimensions, some of which are more likely to come to mind as ways to express social attitudes (Thomas et al., 1998, Experiment 1). Furthermore, published studies of attitude-behavior consistency, collapsed across target categories, have reported larger effect sizes when they used behavioral measures that matched rather than mis- matched the action dimensions that naive participants most frequently mentioned when they assessed their social category attitudes (Thomas e t al., 1998, Experiment 2).

Other studies have shown that attitude-relevant responses are more con- sistent when they involve matching than mismatching contexts. Thus one can predict better from people’s general attitudes to their behavioral re- sponses when the social norms that surround those behavioral responses occur in contexts that match rather than mismatch participants’ assumptions regarding subjective norms (Ajzen & Fishbein, 1980; Manstead et al., 1983). One can also predict better from people’s general attitudes to their behav- ioral responses when the difficulty of enacting the behavioral response matches rather than mismatches participants’ assumptions about the extent to which they can exert control (Ajzen, 1987; Ajzen & Madden, 1986).

Finally, studies derived from ART correctly predicted times when match- ing might not matter. One such case occurs when an individual’s representa- tion for an attitude object contains few assumptions regarding a social policy’s targets (Lord et al., 1994). Another such time occurs when the individual is under cognitive load or in an especially good mood, because at such times people rely almost exclusively on representation-based as- sumptions at the expense of immediate perceptions (Blessum et al., 1998). Like any effective process model, then, ART predicts times when the postulated processes will and will not occur. In the final section of the chapter, we will discuss not only limitations, but also several other issues

Page 54: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

318 CHARLES G. LORD AND MARK R. LEPPER

that are raised by examining the similarities and differences between ART and other attitude theories.

V. Issues Related to the Attitude Representation Theory Model

Numerous studies, then, have supported both the representation postu- late and the matching postulate of ART. The model built around these studies, though, raises several related issues.

A. A CONTINUUM OF ATTITUDE-RELEVANT SITUATIONS

The first step in the ART model shown in Figure 1 is that an individual identifies what we call an “attitude-relevant situation.” We define an attitude-relevant situation as any situation in which the individual has an opportunity to respond evaluatively to the attitude object (Eagly & Chaiken, 1993). Attitude-relevant situations, according to ART, differ in how much immediate perceptual information they afford. At one end of the continuum is a situation in which the individual responds evaluatively to the mere name of the attitude object, with no other information available. In such a relatively impoverished situation the individual is being asked, in theory, to respond to some “average” of all possible identities, character- istics, associated responses, and contexts that involve the attitude object. Thus, this particular evaluative response is often equated with, or at least taken as the most general and direct indicator of, the individual’s underlying attitude-i.e., the underlying psychological tendency to evaluate the atti- tude object with favor or disfavor (Eagly & Chaiken, 1993). At the other end of the continuum is first-hand contact with some actual, specific attitude object. The word snake, for instance does not afford the same direct percep- tions as an actual snake (Breckler, 1984). The general attitude question presumably involves many assumptions that are derived from a representa- tion of the attitude object, whereas first-hand contact might involve fewer assumptions about the attitude object and more immediate perceptions of a specific stimulus, response type, and context.

Most attitude-relevant situations probably fall somewhere on the contin- uum between a general attitude question and a first-hand encounter. Near the general attitude side of the continuum are situations in which the individual has a little more information available than merely the name of the attitude object. Instead of assessing their attitudes toward “Chinese people,” for instance, people might assess their attitudes toward “Chinese

Page 55: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 319

professors” (target identity specified), “polite Chinese” (target characteris- tic specified), “serving Chinese people” (type of response specified), or “serving Chinese people when other customers do not object” (type of response and context specified). Thus the valence of the evaluative response might be altered by the addition of even a small amount of extra informa- tion-by knowing, for instance, that the attitude object is Chinese academ- ics rather than Chinese people in general, or that the evaluative response in question involves serving them rather than other ways (cognitive, affec- tive, and/or behavioral) of responding evaluatively.

Similarly, as we approach the other, first-hand-encounter end of the continuum, the importance of assumptions about the general class might decrease and the importance of immediate perceptions might increase. Thus the proprietor who is asked by a Caucasian professor about “serving my polite Chinese friends” has available additional information that is almost, but not quite, as compelling as would be available from a first- hand encounter. Depending on the source’s perceived reliability, individuals might regard second-hand information of this type as almost as important as first-hand perceptions, thus decreasing the importance of representation- based assumptions. Such second-hand information about the target’s iden- tity and characteristics, the type of response involved, and the context is often the basis for behavioral intentions rather than for actual behaviors (Ajzen & Fishbein, 1980).

Even when the attitude-relevant situation conveys considerable percep- tual information about the stimulus, however, it is still possible for represen- tation-based assumptions to influence evaluative responses. “Educated” Chinese people could be interpreted as “having attended elementary school before dropping out.” “Accepting” Chinese people as guests in a restaurant might be interpreted to mean “giving them take-out food from a window at the back of the restaurant.” We know from numerous studies of stereo- typing that reality itself can be misinterpreted when assumptions and expec- tations bias our perceptions (Bruner, 1957; Olson, Roese, & Zanna, 1996; Trope, 1986). People who read the same evidence from a court case, for instance, perceive exactly the same defendant, who engaged in exactly the same criminal activities, as more guilty if his surname triggers assumptions about the propensity of his race or nationality to commit certain types of crimes (Bodenhausen, 1988; Gordon, 1993). Similarly, personality trait- relevant behavior that appears to be inconsistent often varies because indi- viduals differ in how they construe situations that might appear to be objectively similar (Bern & Funder, 1979; Lord, 1982; Mischel, 1968).

Finally, when concrete situations appear to afford all the information one would need to respond evaluatively, seemingly obviating the need for representation-based assumptions, many studies have shown that people

Page 56: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

320 CHARLES G. LORD AND MARK R. LEPPER

may still differ in their construals of the situation (Ross, 1987,1989). People who are distracted or not motivated to think in depth about the “real” situation, in particular, may often fall back on default assumptions from their representation of a stimulus and rely on heuristic as opposed to systematic processing (Chaiken, 1987; Schwarz & Clore, 1996). It is all too easy for assumptions to influence evaluative responses at all points along the continuum of attitude-relevant situations (Ajzen, 1996).

B. NUMBER AND TYPE OF REPRESENTATION- BASED ASSUMPTIONS

The classical explanation of attitude change had been that people altered their evaluations of the same attitude object-for instance, that they changed their opinions of politicians. When it became apparent that such changes were too frequent and too facile to fit what we knew about ourselves and other people, Asch (1940) turned the classical explanation on its head, by speculating that such changes could also occur because people had altered the specific instance that they had in mind. Asch’s (1940) ideas applied only to changes in the specific category exemplar that the individual assumed would represent or symbolize a category. Attitude Representation Theory, in contrast, holds that people assume not only exemplars, but also characteristics of the attitude object, likely responses associated with that object, and the context in which such responses might occur.

We chose to include these four aspects of an attitude representation because we believe that they can all play a part in determining an individu- al’s evaluative response to a stimulus. As we explain in the next section, we do not believe that all four aspects of the object’s representation are always activated every time an individual makes an evaluative response (a positive or negative thought, feeling, or action) toward an attitude object. We do, however, believe that at least one of them is activated for every evaluative response, whether the individual is aware of it or not (Tesser & Martin, 1996). We would rather err on the side of including too many possible aspects of an attitude representation than too few. Some authors, in fact, have suggested an even larger number factors that determine evalua- tive responses than in ART, including, for instance, the impact of a particu- lar action on the individual’s self-identity (Eagly & Chaiken, 1993).

We included the target’s identity among representation-based assump- tions for obvious reasons. Asch’s (1940) initial insight about changes in the object of judgment involved the assumed identity of politicians. Smith and Zarate’s (1992) extensive program of research established that exemplars are activated for and influence many types of social judgments, and that

Page 57: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 321

exemplars are merely instances and need not involve people who are mem- bers of a social group or category. “Electrocution” might be an exemplar of “capital punishment.” “Recycling” might be an exemplar of “helping the environment.” Exemplars can also be based on direct experience or only imagined (Smith, 1998). An exemplar of “Chinese people” might be a coolie that a 1930s proprietor actually saw working in the fields or a fictional hired assassin from a lurid novel. In fact, it is sometimes difficult to separate in memory the real exemplars from the imagined ones (Johnson, Hastroudi, & Lindsay, 1993). Exemplars can also occur in subsets (Zarate & Smith, 1990), as has been shown by numerous studies of subtyping (e.g., Hewstone, Hopkins, & Routh, 1992) including the study of substance abuser representations that we described earlier (Ramsey et al., 1994). Finally, ART argues that an attitude object can be identified not merely as a name (e.g., former mental patients), but also as a mental image that has physical characteristics, just as people can have both verbal propositions and mental images about objects as seemingly abstract as “nuclear war” (Fiske, Pratto, & Pavelchak, 1983). Considerable modern research suggests that the target’s identity constitutes an important part of how we represent not only social categories, but also attitude objects of all types.

Particular characteristics or attributes of the attitude object have likewise played an important role in an array of attitude theories. The most fre- quently included characteristic in early theories of evaluative responses was the object’s instrumental value. In these theories, one could predict an individual’s overall evaluation of an attitude object by asking about the object’s desirable or undesirable attributes, and the probability that the object would have or produce those attributes (Fishbein, 1963, 1967; Peak, 1955; Rosenberg, 1956). Thus people’s attitudes toward engaging in a health-protective behavior could be predicted from their assumptions about whether the procedure would do any good (Rogers, 1983), and people’s attitudes toward racial integration could be predicted by knowing what the individual assumed would be its consequences (Carlson, 1956). ART acknowledges that instrumental beliefs are important assumptions about the characteristics of an attitude object, but also includes characteristics that are usually referred to as the elements within a stereotype (i.e., rude and uneducated for Chinese people in the 1930s, or shy and polite for Asian people today). Some of these characteristics may be personality traits and some may be demographic, as in the erroneous assumption that the typical murderer has a dark skin (Lord et al., 1994). Some may also be attached to specific exemplars. “Jogs,” for instance, could be activated as an assumed characteristic of politicians, but only for individuals who have Bill Clinton as their primary politician exemplar.

Page 58: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

322 CHARLES G. LORD AND MARK R. LEPPER

A potentially controversial facet of ART is the inclusion of “associated responses” among representation-based assumptions about an attitude ob- ject. “It seems redundant,” one might argue, “to say that people base their evaluative responses on other, associated responses to the same attitude object. It is like saying that we base responses on responses.” We argue, however, that modern research suggests just such a possibility. Recall that evaluative responses include thoughts, feelings, and actions-the three ways of expressing an underlying attitude (Eagly & Chaiken, 1993). The first time an individual evaluates an attitude object, the individual necessarily has to take into account whatever information seems relevant. For subsequent evaluations, however, many modern theories hold that an additional piece of information has now been added-namely, the outcome of the previous evaluative response (Eagly & Chaiken, 1993; Judd & Brauer, 1995). Consid- erable evidence suggests, in fact, that people sometimes remember a previ- ous evaluative response without being able to recall the specific pieces of information on which the response was based, for instance, when we say that “I remember that I loved that book, but I can’t for the life of me remember what it was about” (Chaiken & Stangor, 1987; Dreben, Fiske, & Hastie, 1979; Hastie & Park, 1986; Lingle & Ostrom, 1981; Zajonc, 1980). Self-generated thoughts and feelings from a previous real or imagined encounter with the attitude object clearly influence future evaluative re- sponses to the object (Greenwald, 1968; Petty & Cacioppo, 1986).

Another type of evaluative response that deserves to be included in the attitude representation is a real or imagined action toward the attitude object. According to self-perception theory (Bern, 1972), people often base their evaluative responses to an attitude object on their own previous actions. Many models of attitude processes have included past behaviors as influences on current evaluative responses (Bentler & Speckart, 1979; Liska, 1984; Triandis, 1977,1980). Moreover, as part of their mental repre- sentations, people can have scripts about their own actions or people’s usual actions toward an attitude object (Abelson, 1976, 1981; Schank & Abelson, 1977). In addition, our own study of effect sizes in published studies of attitude-behavior consistency indicated that the actions lay- people frequently associate with their attitudes have produced higher atti- tude-behavior correlations than the actions that laypeople infrequently associate with their attitudes toward social categories (Thomas et al., 1998). It would be interesting in this regard to investigate the details of an individu- al’s assumptions about an attitude object, including the level of specificity at which the individual identifies an attitude-relevant action (Vallacher & Wegner, 1987). Finally, by including previous evaluative responses as part of an attitude representation, ART allows for the influence of numerous factors of which the individual is not aware, such as classical conditioning

Page 59: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 323

(Cacioppo, Marshall-Goodell, Tassinary, & Petty, 1992; Krosnick, Betz, Jussim, & Lynn, 1992; Staats, 1983: Zanna, Kiesler, & Pilkonis, 1970), mere exposure (Zajonc, 1968), and physical movements (Cacioppo, Priester, & Berntson, 1993; Strack, Martin, & Stepper, 1988).

If we agree that attitude assessment can involve assumptions about associ- ated evaluative responses, then it is obvious from theories of attribution that the meaning of such responses can be dramatically altered by the surrounding context (Gilbert, 1998: Jones & Davis, 1965; Kelley, 1967). An action performed in a facilitative context has a different meaning than the same action performed in an inhibiting context. Thus assumptions about contextual factors such as subjective norms and perceived control are likely aspects of an attitude object’s representation (Miniard & Cohen, 1981; Smetana & Adler, 1980). Like the other assumptions about the target’s identity and characteristics, assumptions about contextual factors can be at odds with perceptions of an actual situation. People often fail to realize, for example, the extent to which social norms of tolerance and politeness constrain actual behaviors (Campbell, 1963).

When LaPiere’s (1934) proprietors projected that they would refuse service to members of the Chinese race, they might well have assumed that they could rely on the staff and patrons to cooperate rather than protest. Similarly, people often underestimate the skills and resources required to perform an action (Fishbein & Ajzen, 1975). Perceived control over a behavior often falls short of actual control (Buehler, Griffin, & Ross, 1994; Kahneman & Tversky, 1979; Langer, 1975), which is one reason why stu- dents’ perceived control over getting an A near the end of a course corres- ponds better with their actual grades than their perceived control earlier in the course (Ajzen & Madden, 1986). Including contextual factors such as subjective norms and perceived control in the ART model signals yet another way in which the matching principle can be violated-another way in which assumptions can mismatch perceptions.

C. COMBINING ASSUMPTIONS WITH PERCEPTIONS

In the ART model shown in Figure 1, representation-based assumptions and more direct perceptions are combined to form an overall evaluation of the attitude object. Although individuals might not be aware of the combination process that they use to form an overall evaluation (Tesser & Martin, 1996), many attitude theories and studies have provided important insights. Previous research suggests, for instance, that people base their evaluative responses on a very small subset of all the information they have

Page 60: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

324 CHARLES G. LORD AND MARK R. LEPPER

available about a stimulus (Ajzen, 1996; Ajzen & Fishbein, 1980; Eagly & Chaiken, 1998; Higgins, 1996; Wilson & Hodges, 1992).

In some instances, people may consider just one salient dimension of the attitude representation or perceived situation to evaluate an attitude object ( Jaccard, 1981; McGuire, 1985; Tversky, 1972), especially if the evaluative task seems simple or trivial (Guerin & Innes, 1989; Zajonc, 1960). When more than one dimension matters, though, the weights assigned to specific assumptions and perceptions during the combination process exert considerable impact on an overall evaluation (Abelson & Levi, 1985; Ander- son, 1991; Atkinson, 1958; Feather, 1982; Kaplan & Anderson, 1973; Lewin, 1938). Political conservatives and political liberals, for instance, often agree on the characteristics that they attribute to a social policy, but differ in the weights that they assign to each characteristic, thus arriving at different overall evaluations (Kerlinger, 1984; Tourangeau, Rasinski, & D’Andrade, 1991). Smokers and nonsmokers can agree on the characteristics of smoking, but differ in how heavily they weight those characteristics (van der Pligt & Eiser, 1984), just as Blacks and Whites can agree on the Characteristics of affirmative action, but differ in the weight they assign to each characteristic (Kluegel & Smith, 1986). Especially if an individual knows of both positive and negative characteristics of an attitude object, the individual’s overall attitude can change from one time to the next, merely because of assigning different weights at different times to the same representation-derived assumptions (Bargh et al., 1992; Hass, Katz, Rizzo, Bailey, & Einstadt, 1991; Katz, 1981).

Previous research also suggests some factors that can increase or decrease the weights assigned to specific aspects of an attitude representation and/ or to the corresponding attitude-relevant perceptions. First, the scope of the attitude object logically limits the weights assigned to specific assumptions. Frequently, as described earlier, people are asked to assess their attitudes toward a stimulus that is identified only by its name, as in “Chinese people” or “capital punishment.” At such times, according to ART, they are free to assume “default values” for the target’s identity and characteristics, the responses it has elicited or might elicit, and the context surrounding those responses. At other times, people may be asked to report their attitudes toward a stimulus that specifies a particular type of evaluative response, as in “serving Chinese people.” In such cases, they can be more confident that they know the nature of the evaluative response involved, and assump- tions about the type of response (e.g., its likely consequences) will merit greater importance than would have been the case had a specific type of evaluative response not been specified.

Second, some factors are likely to increase the weights given to immediate perceptions rather than to representation-based assumptions. We have al-

Page 61: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 325

ready discussed the simple availability of more specific information. An- other factor might be a reluctance to use stereotypical assumptions, o r at least to weight them heavily, in arriving at an overall evaluation. Research suggests that prejudiced and nonprejudiced people are equally aware of the characteristics that comprise the stereotype of various racial and ethnic minority groups (Devine, 1989). They differ, however, in their willingness to use those assumptions when forming an overall evaluation (Devine & Monteith, 1993; Yzerbyt, Schadron, Leyens, & Rocher, 1994). In terms of the ART model, prejudiced and nonprejudiced people might activate the same representation of the social group or category, but nonprejudiced people might assign much lower weights to the assumed target characteris- tics than do prejudiced people, because they feel guilty about relying on representation-based assumptions that might be politically incorrect, or because the social structure discourages stereotyping and prejudice (De- vine & Monteith, 1993; Fiske, 1998; Monteith, Devine, & Zuwerink, 1993). Low weights on assumptions might reflect inhibition of a stereotype’s initial activation, inhibition of its use, or both (Bodenhausen & Macrae, 1998; Dijksterhuis & van Knippenberg, 1998; Stangor, Thompson, & Ford, 1998; Stroessner, 1998).

Other factors may work in the opposite direction and increase the likely weights accorded to representation-based assumptions and decrease the weights accorded to immediate perceptions. An overall evaluative response to the mere name of an attitude object, which involves no immediate stimulus, for instance, may predict evaluative responses in subsequent be- havioral situations when people are not willing or able to pay much attention to the perceptions that the concrete behavioral situation affords (Fazio, 1990; Schuette & Fazio, 1995). At such times, people become “cognitive misers” who fall back on (i.e., give greater weight to) their representation- based assumptions than to specific situational perceptions (Chaiken, 1987; Fiske & Neuberg, 1990; Taylor & Fiske, 1978).

Experts on a topic might also be more likely than novices to attach large weights to their representation-based assumptions. In the study we described earlier of responses to the women whose hairstyles matched versus mismatched the mental image we had created for members of a fictitious organization, the average participant displayed greater attitude- behavior consistency toward a matching than a mismatching exemplar (Lord et al., 1991, Experiment 1). Some participants, however, were rela- tively more expert (i.e., more accurate) at discriminating members from nonmembers, whereas others could tell the difference at just barely better than chance. The novices’ attitudes toward the group predicted behavior toward a woman with the hairstyle that matched their representation, but not their behavior toward a woman with a mismatching hairstyle-as

Page 62: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

326 CHARLES G. LORD AND MARK R. LEPPER

though novices took into account both their representation-based assump- tions and their more direct perceptions.

Experts’ attitudes, in contrast, predicted behavior toward both women, regardless of their hairstyle, as though the experts relied almost entirely on their representation-based assumptions about the group and gave very little weight to their immediate perceptions. In a follow-up study, partici- pants were divided into “experts,” who knew many, and “novices,” who knew few, gay men (Lord et al., 1991, Experiment 2). Again, novices’ attitudes predicted their behavior toward a typical but not an atypical gay man, whereas experts’ attitudes predicted their behavior toward both targets, as though experts relied on heuristic or assumption-dominated processing in the behavioral situation. Expertise, then, may alter the relative weights that people assign to representation-based assumptions versus more direct perceptions in determining their overall evaluation and evaluative re- sponse.

D. EFFECTS OF EXPERIENCE

“Experience” can occur either in reality or in the imagination. A number of studies have shown that direct experience with a real attitude object increases the consistency of evaluative responses toward that object (Fazio & Zanna, 1981). Attitudes toward getting flu shots predicted actually getting the shots better, for instance, for people who had had flu shots previously than for people who had not (Davidson, Yantis, Norwood, & Montano, 1985). Similarly, attitudes toward smoking predicted behavior better for people who had previously smoked than for people who had never smoked (Sherman et al., 1982). These effects of direct personal experience fit well with the ART model shown in Figure 1, if we assume that direct experience will, on average, increase the correspondence between representations and perceptions of attitude objects.

Direct experience may also make the representation of an attitude object more accessible (Fazio, Chen, McDonel, & Sherman, 1982). People base their evaluative reactions on whatever information is most readily accessible at the time (Bodenhausen & Wyer, 1985; Fishbein & Ajzen, 1975; Fishbein, Ajzen, & MacArdle, 1980; Mitchell & Olson, 1981; Olson, Toy, & Dover, 1982; Zanna & Rempel, 1988). Thus people are more apt to base evaluations of their lives on marital satisfaction if they have recently thought about their marriages (Schwarz, Strack, & Mai, 1991), and people are more apt to evaluate the president’s performance on the state of the economy if they have recently thought about their incomes (Iyengar & Ottati, 1994). Also, people who have recently recalled their own actions toward an atti-

Page 63: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 327

tude object resist attitude change better than do people who have not re- cently recalled their actions (Ross, McFarland, Conway, & Zanna, 1983). In weighting assumptions and perceptions to form an overall evaluation, then, direct experience with the attitude object might lend weight to representation-based assumptions by making them more readily accessible for future evaluative responses (Doll & Ajzen, 1992).

Even imagined experience, involving nothing more than thinking about the attitude object in depth, can increase the accessibility of representation- based assumptions. Repeated expression of an attitude using the same type of evaluative response over and over may make the process of combining assumptions with perceptions more efficient ( Judd & Brauer, 1995). With each response, the process becomes more proceduralized (Anderson, 1983). All other aspects of the representation except for the one most recent evaluative response may come to have near-zero weights (Judd & Brauer, 1995). Thus people who are asked about their attitude toward the same stimulus six times in a row can answer very rapidly, merely by repeating the previous overall evaluation (Fazio, 1990). They will also display consid- erable attitude-behavior consistency (Houston & Fazio, 1989).

Repeated evaluative responses to an attitude object might not only make representation-based assumptions more accessible, but might also create a sense of fluency that imparts confidence in those assumptions (Mandler, Nakamura, & Van Zandt, 1987). We get two types of information when representations are activated: the content of the representation itself, and the subjective ease of bringing that content to mind (Schwarz, 1998). Partici- pants in one relevant study reported more favorable attitudes toward BMWs after generating one reason for buying a BMW than after generating ten reasons for buying one, because the ones who had to generate only one reason found the task so easy that such a purchase must be wise, whereas the ones who had to generate ten reasons found the task so difficult that they drew the opposite conclusion (Wanke, Bohner, & Jurkowitz, 1997). Repeated evaluative responses to the same stimulus might thus increase the importance assigned to representation-based assumptions by making them seem to come to mind very easily (Downing, Judd, & Brauer, 1992; Roese & Olson, 1994).

Finally, thinking in depth about an attitude object could also decrease the consistency of evaluative responses, according to the matching princi- ple-if it makes accessible a different aspect of the representation than is activated as the basis for a subsequent behavioral response. In several studies, for instance, Wilson and his colleagues have shown that people who think in depth about the characteristics of an attitude object (specifi- cally those characteristics that might justify the attitude) sometimes report more extreme attitudes and sometimes report more moderate attitudes

Page 64: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

328 CHARLES G. LORD AND MARK R. LEPPER

(Wilson, Dunn, Bybee, Hyman, & Rotondo, 1984; Wilson, Dunn, Kraft, & Lisle, 1989; Wilson & Kraft, 1993).

The reason is that such thinking increases the weights given to the attitude object’s characteristics at the expense of other aspects of the representation such as previous evaluative reactions, often altering the overall evaluation (Wilson et al., 1989). When the individual subsequently enters an attitude- relevant situation that involves more direct perceptions, in contrast, target characteristics might not carry as much weight as previous evaluative re- sponses, so the attitude will not predict the subsequent behavior as well. Having people think about their previous affective responses to an attitude object, on the other hand, may increase attitude-behavior consistency for subsequent evaluative responses that are primarily concerned with feelings, rather than primarily concerned with instrumental calculations (Millar & Tesser, 1992). Thus the type of direct experience that we get from actual encounters with an attitude object is likely to make evaluative responses more consistent, but imagined encounters with the attitude object can make subsequent evaluative responses either more or less consistent, depending on the types of perceptions that are most useful for the specific response.

We will close with one final speculation.’ Throughout this chapter we have argued, and presented evidence, that people will often think of broad categories and even abstract concepts in terms of specific exemplars. Indeed, several theorists have suggested that people seldom reason with abstractions (Griggs & Cox, 1982; Kolodner, 1992; Reich & Ruth, 1982). Instead, their reasoning is much more concrete, based on knowledge about specific situa- tions. They make judgments about abstract concepts by using specific re- membered instances (Brooks, 1987; Estes, 1993; Hintzman, 1986; Nosofsky, 1986; Reed, 1972; Shin & Nosofsky, 1992). They find it easier to reason with concrete examples than with abstractions (Bracewell & Hidi, 1974; Gilhooly & Falconer, 1974; Johnson-Laird, Legrenzi, & Sonino-Legrenzi, 1972), especially when they are familiar with a particular judgment task or decision (Griggs & Cox, 1982), as people presumably are with assessing their own attitudes.

Beyond that, however, why might people prefer to think of abstract attitude objects in concrete terms? One possible perspective on this question is an evolutionary one. Perhaps, early in our ancestral history, before the ability for abstract, conceptual thinking had become highly developed, there may have been adaptive advantages to the development of a process that automatically evaluated the concrete stimuli that the organism encountered in terms of associated positive or negative affect (e.g., Zajonc, 1980, 1998).

’ We owe this idea to Daniel Gilbert, who could no doubt have expressed it much more ele- gantly.

Page 65: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 329

If such affective reactions were indeed an evolutionarily older development than the higher cognitive processes involved in conceptual abstraction, our tendency to think of specific exemplars even when assessing our attitudes to complex categories or abstract social policies may be, in part, the residual of earlier adaptations. We may be “programmed,” as it were, to reduce the general to the particular in making affective evaluations.

VI. Concluding Remarks

The concept of general social attitudes became ubiquitous in social psy- chology because of their implicit promise of stability across time and consis- tency across situations in people’s responses to various attitude-relevant stimuli. Empirical and theoretical challenges to these assumptions offered by Asch’s (1940) analysis of the “attitude-object problem” and LaPiere’s analysis of the “attitude-behavior problem,” however, strongly questioned the utility of both of these premises. Like many previous efforts, ART involves an attempt to specify at least some of the sources of these problems and to identify at least some of the conditions under which attitudes will show cross-temporal stability versus instability and cross-situational consis- tency versus inconsistency.

In particular, the ART framework, as outlined in Figure 1, involves two basic postulates. The first, representation postulate holds that a person’s responses to any attitude-relevant object will be a joint product of that person’s current subjective representation of that object or category and that person’s more direct perceptions of any specific attitude object that is immediately present. The second, matching postulate then holds that the consistency or stability of different attitude-relevant responses will depend critically on the degree of match between both the subjective representa- tions and the immediate perceptions of the attitude object at those different times or in those different situations.

If the central “message,” in a sentence, of the attitude concept could perhaps be summarized as that to know how a person will act, one must first know the person’s attitudes, then the central “message” of Asch’s and LaPiere’s criticisms of the attitude concept could be summarized as that to know a person’s attitudes, one must first know how that person thinks. Together, as described by the ART model, they point to the critical role of recollection, construal, and other subjective cognitive processes in deter- mining how people will behave in various attitude-relevant situations. As Lord Halifax put it, quite succinctly, over three hundred years ago:

Page 66: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

330 CHARLES G. LORD AND MARK R. LEPPER

Could we know what men are most apt to remember, we might know what they are most apt to do.

Acknowledgments

We thank Alice Eagly, Daniel Gilbert, Eliot Smith, and Mark Zanna for their helpful comments on an earlier draft. Preparation of this chapter was supported, in part, by National Institute of Mental Health Grant MH-44321 to the second author.

References

Abelson, R. P. (1976). Script processing in attitude formation and decision making. In J. S. Carroll & J. W. Payne (Eds.), Cognition and social behavior (pp. 33-46). Hillsdale, NJ: Erlbaum.

Abelson, R. P. (1981). The psychological status of the script concept. American Psychologist, 36, 715-729.

Abelson, R. P., & Levi, A. (1985). Decision making and decision theory. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed., Vol. 1, pp. 231-309). New York: Random House.

Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., & Sanford, R. N. (1950). Theauthoritar- ian personality. New York: Harper & Row.

Ajzen, 1. (1985). From intentions to actions: A theory of planned behavior. In J. Kuhl & J. Beckman (Eds.), Action-control: From cognition to behavior (pp. 11-39). Heidelberg: Springer.

Ajzen, I. (1987). Attitudes, traits, and actions: Dispositional prediction of behavior in personal- ity and social psychology. In L. Berkowitz (Ed.), Advances in experimental social psychol- ogy (Vol. 20, pp. 1-63). San Diego, CA: Academic Press.

Ajzen, I. (1996). The social psychology of decision making. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 297-325). New York: Guil- ford Press.

Ajzen, I., & Fishbein, M. (1977). Attitude-behavior relations: A theoretical analysis and review of empirical research. Psychological Bulletin, 84, 888-918.

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall.

Ajzen, I., & Madden, T. J. (1986). Prediction of goal-directed behavior: Attitudes, intentions, and perceived behavioral control. Journal of Experimental Social Psychology, 22,453-474.

Allport, G. W. (1935). Attitudes. In C. Murchison (Ed.), A handbook of social psychology (pp. 798-844). Worcester, MA: Clark University Press.

Allport, G. W. (1954). The nature of prejudice. Reading, MA: Addison-Wesley. Anderson, J. R. (1983). The architecture of cognition. Cambridge, MA: Harvard University

Anderson, N. H. (Ed.) (1991). Contributions to information integration theory (Vols. 1 ,2 and Press.

3). Hillsdale, NJ: Erlbaum.

Page 67: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 33 1

Anderson, R. C., & Pichert, J. W. (1978). Recall of previously unrecallable information following a shift in perspective. Jourtial of Verbal Learning and Verbal Behavior, 17, 1-12.

Armbruster, B. B., & Anderson, T. H. (1984). Mapping: Representing informative text dia- grammatically. In C. H. Holley & D. F. Dansereau (Eds.). Spatial learning straregies: Techniques, applications, and related issues (pp. 189-209). San Diego, CA: Academic Press.

Asch, S. E. (1940). Studies in the principles of judgments and attitudes: 11. Determination of judgments by group and by ego standards. The Journal of Social Psychology, S.P.S.S.1. Bulletin, 12, 433-465.

Atkinson. J. W. (Ed.) (1958). Motives in fantasy, action, and society: A method of assessment and siudy. Princeton, NJ: Van Nostrand.

Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191-215.

Bargh, J. A. (1992). The ecology of automaticity: Toward establishing the conditions needed to produce automatic processing effects. American Journal of Psychology, 105, 181-199.

Bargh, J. A. (1996). Automaticity in social psychology. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 169-183). New York: Guil- ford Press.

Bargh, J. A,, Chaiken, S., Govender, R., & Pratto, F. (1992). The generality of the automatic attitude activation effect. Journal of Personaliry and Social Psychology, 62, 893-912.

Bellezza, F. S. V. (1984). Reliability of retrieval from semantic memory: Common categories. Bulletin of the Psychonomic Society, 22, 324-326.

Bern, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experirnentrd social psychology (Vol. 6, pp. 1-62). San Diego. CA: Academic Press

Bem, D. J., & Funder, D. C. (1979). Predicting more of the people more of the time: Assessing the personality of situations. Psychological Review, 85, 485-501.

Bentler, P. M., & Speckart, G. (1979). Models of attitude-behavior relations. Psychological Bulletin, 86, 452-464.

Blaney, N. T., Stephan. S., Rosenfield, D., Aronson, E., & Sikes, J. (1977). Interdependence in the classroom: A field study. Journal of Educational Psychology, 69, 121-128.

Bless, H.. Clore, G. L., Schwarz, N., Golisano, V., Rabe, C., & Wolk. M. (1996). Mood and the use of scripts: Does a happy mood really lead to mindlessness? Journal of Personality and Social Psychology, 71, 665-679.

Blessum, K. A,, Lord, C. G., & Sia, T. L. (1998). Cognitive load and positive mood reduce typicality effects in attitude-behavior consistency. Personality and Social Psychology Bulerin, 24, 496-504.

Bodenhausen, G. V. (1988). Stereotyopic hiases in social decision making and memory: Testing process models for stereotype use. Journal of Personality and Social Psychology, 55, 726-737.

Bodenhausen, G. V., Kramer. G. P., & Siisser K. (1994). Happiness and stereotypic thinking in social judgment. Journal of Personality and Social Psychology, 66, 621-632.

Bodenhausen. G. V., & Macrae, C. N. (1998). Stereotype activation and inhibition. In R. Wyer (Ed.), Advances in social cugnifion (Vol. 1 1 , pp. 1-52). Mahwah, NJ: Erlbaum.

Bodenhausen, G. V., Schwarz, N., Bless, H., & Wanke, M. (1995). Effects of atypical exemplars on racial beliefs: Enlightened racism or generalized appraisals? Journal of Experitnentul Sociul Psychology, 31, 48-63.

Bodenhausen, G. V., & Wyer. R. S. (1985). Effects of stereotypes on decision making and information-processing strategies. Journal of Personality and Social Psychology, 48,

Bracewell, R. J. , & Hidi, S. E. (1974). The solution of an inferential problem as a function of stimulus materials. Quarterly Journal of Experimenial Psychology, 26, 480-488.

267-282.

Page 68: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

332 CHARLES G. LORD AND MARK R. LEPPER

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.

Brewer, M. B., Dull, V., & Lui, L. (1981). Perceptions of the elderly: Stereotypes as prototypes. Journal of Personality and Social Psychology, 41, 656-670.

Brewer, M. B., Ho, H. K., Lee, J. Y., & Miller, N. (1987). Social identity and social distance among Hong Kong school children. Personality and Social Psychology Bulletin, 13,

Brewer, M. B., & Miller, N. (1984). Beyond the contact hypothesis: Theoretical perspectives on desegregation. In N. Miller & M. B. Brewer (Eds.), Groups in contact: The psychology of desegregation (pp. 281-302). San Diego, CA: Academic Press.

Brooks, L. (1987). Decentralized control of categorization: The role of prior processing epi- sodes. In U. Neisser (Ed.), Concepts and conceptual development (pp. 141-174). Cam- bridge: Cambridge University Press.

Brown, R., & Wade, G. (1987). Superordinate goals and intergroup behavior: The effect of role ambiguity and status on intergroup attitudes and task performance. European Journal of Social Psychology, 17, 131-142.

156-165.

Bruner, J. S. (1957). On perceptual readiness. Psychological Review, 64, 123-152. Buehler, R., Griffin, D., & Ross, M. (1994). Exploring the “planning fallacy”: Why people

underestimate their task completion times. Journal of Personality and Social Psychology, 67, 366-381.

Byrne, D. (1971). The attraction paradigm. New York: Academic Press. Cacioppo, J. T., Marshall-Goodell, B. S., Tassinary, L. G., & Petty, R. E. (1992). Rudimentary

determinants of attitudes: Classical conditioning is more effective when prior knowledge about the attitude stimulus is low than high. Journal of Experimental Social Psychology,

Cacioppo, J. T., Pnester, J. R., & Berntson, G. G. (1993). Rudimentary determinants of attitudes: 11. Arm flexion and extension have differential effects on attitudes. Journal of Personality and Social Psychology, 65, 5117.

Campbell, D. T. (1963). Social attitudes and other acquired behavioral dispositions. In S . Koch (Ed.), Psychology: A study of a science (Vol. 6, pp. 94-172). New York: Mcgraw-Hill.

Carlson, E. R. (1956). Attitude change through modification of attitude structure. Journal of Abnormal and Social Psychology, 52, 256-261.

Carlston, D. E. (1994). Associated systems theory: A systematic approach to cognitive repre- sentations of persons. In R. S . Wyer (Ed.), Advances in social cognition (Vol. 7, pp. 1-78). Hillsdale, NJ: Erlbaum.

Carlston, D. E., & Skowronski, J. J. (1994). Savings in the relearning of trait information as evidence for spontaneous inference generation. Journal of Personality and Social

Chaiken, S. (1987). The heuristic model of persuasion. In M. P. Zanna, J. M. Olson, & C. P. Herman (Eds.), Social influence: The Ontario Symposium (Vol. 5, pp. 3-34). Hillsdale, NJ: Erlbaum.

Chaiken, S., Pomerantz, E. M., & Giner-Sorolla, R. (1995). Structural consistency and attitude strength. In R. E. Petty & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 387-412). Mahwah, NJ: Erlbaum.

Chaiken, S., & Stangor, C. (1987). Attitudes and attitude change. Annual Review of Psychology, 38, 575-630.

Cochran, S. D., Mays, V. M., Ciarletta, J., Caruso, C., & Mallon, D. (1992). Efficacy of the theory of reasoned action in predicting AIDS-related sexual risk reduction among gay men. Journal of Applied Social Psychology, 22, 1481-1501.

28, 207-233.

PSyCholOgy, 66, 840-856.

Page 69: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 333

Collins, A,, & Quillian, M. R. (1969). Retrieval time from semantic memory. Journal of Verbal Learning and Verbal Behavior, 8, 240-247.

Converse, P. E. (1970). Attitudes and non-attitudes: Continuation of a dialogue. In E. R. Tufte (Ed.), The quantitative analysis of social problems (pp. 168-189). Reading, MA: Addison-Wesley.

Cook, S. W. (1984). Cooperative interaction in multiethnic contexts. In N. Miller & M. B. Brewer (Eds.), Groups in contact: Thepsychology of desegregation (pp. 155-1 85) . Orlando, FL: Academic Press.

Cooper, H., & Hedges. L. V. (Eds.) (1994). The handbook of research synthesis. New York: Russell Sage Foundation.

Davidson, A. R., Yantis. S.. Norwood, M., & Montano, D. E. (1985). Amount of information about the attitude object and attitude-behavior consistency. Journal of Personality and Social Psychology, 49, 1184-1198.

Desforges. D. M., Lord, C. G., Pugh, M. A,, Scarberry, N. C., & Ratcliff, C. D. (1997). Role of group representativeness in the generalization part of the contact hypothesis. Basic and Applied Social Psychology, 19, 183-204.

Desforges, D. M.. Lord, C. G.. Ramsey, S. L., Mason. J. A., Van Leeuwen, M. D., West, S. C., & Lepper, M. R. (1991). Effects of structured cooperative contact on changing negative attitudes toward stigmatized social groups. Journal of Personality and Social Psychology, 60, 531-544.

Deutscher, I. (1973). What we say/whaf we do: Sentiments and acts. Glenview, IL: Scott, Foreman.

Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal of Personality and Social Psychology, 56, 5-1 8.

Devine, P. G., & Monteith, M. J. (1993). The role of discrepancy-associated affect in prejudice reduction. In D. M. Mackie & D. L. Hamilton (Eds.). Affect, cognition, and stereotyping: fnteractive processes in intergroup perception (pp. 317-344). San Diego, CA: Academic Press.

Dijksterhuis, A., & van Knippenberg, A. (1998). Inhibition, Aberdeen, and other cloudy subjects. In R. Wyer (Ed.), Advances in social cognition (Vol. 11, pp. 83-96). Mahwah, NJ: Erlbaum.

Doll, J., & Ajzen, I. (1992). Accessibility and stability of predictors in the theory of planned behavior. Journal of Personalily and Social Psychology, 63, 754-765.

Doll, J., & Orth. B. (1993). The Fishbein and Ajzen theory of reasoned action applied to contraceptive behavior: Model variants and meaningfulness. Journal of Applied Social

Dovidio, J. F., Evans, N., & Tyler, R. B. (1986). Racial stereotypes: The contents of their cognitive representations. Journal of Experimental Social Psychology, 22, 22-37.

Downing, J. W., Judd, C. M., & Brauer, M. (1992). Effects of repeated expressions on attitude extremity. Journal of Personality and Social Psychology. 63, 17-29.

Dreben, E, K., Fiske, S. T., & Hastie, R. (1979). The independence of evaluative and item information: Impression and recall order effects in behavior-based impression formation. Journal of Personality and Social Psychology, 37, 1758-1768.

Eagly, A. H., & Chaiken. S. (1993). The psychology of anirudes. New York: Harcourt Brace Jovanovich.

Eagly, A. H., & Chaiken, S. (1998). Attitude structure and function. In D. T. Gilbert, S. T. Fiske. & G. Lindzey (Eds.), Handbook of socialpsychology, (4th ed., Vol. 1, pp. 269-322). New York: McGraw-Hill.

Edwards, K. (1990). The interplay of affect and cognition in attitude formation and change. Journal of Personality and Social Psychology, 59, 212-216.

Psychology, 23, 39.5-415.

Page 70: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

334 CHARLES G. LORD AND MARK R. LEPPER

Esses, V. M., Haddock, G., & Zanna, M. P. (1994). The role of mood in the expression of intergroup stereotypes. In M. P. Zanna & J. M. Olson (Eds.), The psychology ofprejudice: The Ontario symposium (Vol. 7, pp. 77-101). Hillsdale, NJ: Erlbaum.

Estes, W. (1993). Concepts, categories, and psychological science. Psychological Science, 4, 143-153.

Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior: The MODE model as an integrative framework. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 75-109). New York: Academic Press.

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

Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). O n the automatic activation of attitudes. Journal of Personality and Social Psychology, 50, 229-238.

Fazio, R. H., & Zanna, M. P. (1981). Direct experience and attitude-behavior consistency. In L. Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 14, pp. 161-202). New York: Academic Press.

Feather, N. T. (Ed.) (1982). Expectations and actions: Expectancy-value models in psychology. Hillsdale, NJ: Erlbaum.

Festinger, L. (1957). A theory ofcognitive dissonance. Stanford, CA: Stanford University Press. Festinger, L. (Ed.) (1964). Conflict, decision, and dissonance. Stanford, CA: Stanford Univer-

sity Press. Fishbein, M. (1963). An investigation of the relationships between beliefs about an object

and the attitude toward that object. Human Relations, 16, 233-240. Fishbein, M. (1967). Attitude and the prediction of behavior. In M. Fishbein (Ed.), Readings

in attitude theory and measurement (pp. 477-492). New York: Wiley. Fishbein, M., & Ajzen, I. (1975). Belief; altitude, intention, and behavior: An introduction to

theory and research. Reading, MA: Addison-Wesley. Fishbein, M., Ajzen, I., & McArdle, J. (1980). Changing the behavior of alcoholics: Effects

of persuasive communication. In I. Ajzen & M. Fishbein (Eds.), Understanding attitudes and predicting social behavior (pp. 217-242). Englewood Cliffs, NJ: Prentice-Hall.

Fiske, S. T. (1998). Stereotyping, prejudice, and discrimination. In D. T. Gilbert, S. J. Fiske, & G. Lindzey (Eds.), Handbook of social psychology (4th ed., Vol. 2, pp. 357-411). New York: McGraw-Hill

Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression-formation, from category- based to individuating processes: Influences of information and motivation on attention and interpretation. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 1-74). New York: Academic Press.

Fiske, S. T., Pratto, F., & Pavelchak, M. A. (1983). Citizens’ images of nuclear war: Content and consequences. Journal of Social Issues, 39, 41-65.

Forgas, J. P. (1995). The affect infusion model (AIM): Review and an integrative theory of mood effects on judgments. Psychological Bulletin, I 1 7, 39-66.

Gallois, C., Kashima, Y., Terry, D., McCamish, M., Timmins, P., & Chauvin, A. (1992). Safe and unsaf.: sexual intentions and behavior: The effects of norms and attitudes. Journal of Applied Social Psychology, 19, 1521-1545.

Gilbert, D. T. (1989). Thinking lightly about others: Automatic components of the social inference process. In J. S. Uleman & J . A. Bargh (Eds.), Unintended thought (pp. 189-21 I). New York, NY: Guilford Press.

Gilbert, D. T. (1998). Ordinary personology. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handbook of Social Psychology (4th Ed., Vol. 2, pp. 89-150). New York: McGraw-Hill.

Page 71: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 335

Gilbert, D. T.. & Hixon, J. G. (1991). The trouble of thinking: Activation and application of stereotypic beliefs. Journal of Personality and Social Psychology, 60, 509-51 7.

Gilbert, D. T., & Osborne, R. E. (1989). Thinking backward: Some curable and incurable consequences of cognitive busyness. Journal of Personality and Social Psychology, 57, 940-949.

Gilhooly, K. J., & Falconer, W. (1974). Concrete and abstract terms and relations in testing a rule. Quarterly Journal of Experimental Psychology, 26, 355-359.

Gilovich. T. (1981). Seeing the past in the present: The effect of associations to familiar events on judgments and decisions. Joirrnal of Personalicy and Social Psychology, 40, 797-807.

Gordon, R. A. (1993). The effect of strong versus weak evidence on the assessment of race stereotypic and race nonstereotypic crimes. Journal of Applied Social Psychology, 23, 734-749.

Graham, S., Weiner, B., Giuliano, T., & Williams. E. (1993). An attributional analysis of reactions to Magic Johnson. Journal of Applied Social Psychology, 23, 996- 1010.

Greenwald, A. G. (1968). Cognitive learning, cognitive response to persuasion, and attitude change. In A. G. Greenwald, T. C. Brock, & T. M. Ostrom (Eds.), Psychologicalfounda- tions of attitudes (pp. 147-170). New York Academic Press.

Greenwald, A. G. (1990). What cognitive representations underlie social attitudes? Bulletin of the Psychononiic Society, 28, 254-260.

Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27.

Griffin, D. W.. Dunning, D., & Ross, L. (1990). The role of construal processes in overconfident predictions about the self and others. Journal of Personality and Social Psychology,

Griffin, D., & Ross, L. (1991). Subjective construal, social inference, and human understanding. In M. P. Zanna (Ed.), Advances in experimental socialpsychology (Vol. 24, pp. 319-359). New York: Academic Press.

Griggs, R., & Cox. J. R. (1982). The elusive thematic-materials effect in Wason’s selection task. British Journal of Psychology, 73, 407-420.

Guerin, B., & Innes, J. M. (1989). Cognitive tuning sets: Anticipating the consequences of communication. Current Psychology: Research & Reviews, 8, 234-239.

Hass, R. G., Katz, I., Rizzo, N., Bailey, J., & Einstadt, D. (1991). Cross-racial appraisal as related to attitude ambivalence and cognitive complexity. Personality and Social Psychol- ogy Bulletin, 17, 83-92.

Hastie, R., & Park, B. (1986). The relationship between memory and judgment depends on whether the judgment task is memory-based or on-line. Psychological Review, 93,258-268.

Hewstone, M., & Brown, R. (1986). Contact is not enough: An intergroup perspective on the ‘contact hypothesis.’ In M. Hewstone & R. Brown (Eds.). Contact and conflict in infergroup encounters (pp. 1-44). New York: Basil Blackwell Ltd.

Hewstone, M., Hopkins, N., & Routh, D. A. (1992). Cognitive models of stereotype change: Generalization and subtyping in young people’s views of the police. European Journal of Social Psychology, 22, 219-234.

Hewstone, M., & Lord, C. G. (1998). Changing intergroup cognitions and intergroup behavior: The rolc of typicality. In C. Sedikides, J. Schopler. & C. Insko (Eds.), lnrergroup cognition and intergroup behavior (pp. 367-392). Hillsdale, NJ: Erlbaum.

Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability, and salience. In E. T. Higgins & A. W. Kruglanski (Eds.), Socialpsychobgyc Handbook ofbasic principles (pp. 131-168). New York: Guilford Press.

Higgins, E. T.. & King, G. (1981). Accessibility of social constructs: Information processing consequences of individual and contextual variability. In N. Cantor & J. F. Kihlstrom (Eds.), Personality, cognition, and social interaction (pp. 69-121). Hillsdale. NJ: Eribaum.

59, 1128-1 139.

Page 72: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

336 CHARLES G. LORD AND MARK R. LEPPER

Higgins, E. T., Rholes, W. S., & Jones, C. R. (1977). Category accessibility and impression formation. Journal of Experimental Social Psychology, 13, 141-154.

Hintzman, D. L. (1986). “Schema abstraction” in a multiple-trace memory model. Psychologi- cal Review, 93, 411-428.

Holley, C. D., & Dansereau, D. F. (Eds.) (1984). Spatial learning strategies: Techniques, applications, and related issues. San Diego, CA: Academic Press.

Houston, D. A,, & Fazio, R. H. (1989). Biased processing as a function of attitude accessibility: Making objective judgments subjectively. Social Cognition, 7, 51-56.

Hovland, C. L., Janis, I. L., & Kelley, H. H. (1953). Communication andpersuasion: Psychologi- cal studies of opinion change. New Haven, CT Yale University Press.

Iyengar, S., & Ottati, V. (1994). Cognitive perspect ve in political psychology. In R. S. Wyer, Jr. & T. K. Srull (Eds.), Handbook of social cognition (Vol. 2, pp. 143-187). Hillsdale, NJ: Erlbaum.

Jaccard, J. J. (1981). Attitudes and behavior: Implications of attitudes toward behavioral alternatives. Journal of Experimental Social Psychology, 17, 286-307.

Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114, 3-28.

Johnson-Laird, P., Legrenzi, P., & Sonino-Legrenzi, M. (1972). Reasoning and a sense of reality. British Journal of Psychology, 63, 395-400.

Jones, E. E., & Davis, K. E. (1965). From acts to dispositions. The attribution process in person perception. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 2, pp. 220-266). New York: Academic Press.

Judd, C. M., & Brauer, M. (1995). Repetition and evaluative extremity. In R. E. Petty & J. A. Krosnick (Ed.), Attitude strength: Antecedents and consequences (pp. 43-72). Mah- wah, NJ: Erlbaum.

Kahneman, D., & Tversky, A. (1979). Intuitive prediction: Biases and corrective procedures. TIMS Studies in Management Science, 12, 313-327.

Kaplan, M. F., & Anderson, N. H. (1973). Information integration theory and reinforcement theory as approaches to interpersonal attraction. Journal of Personality and Social Psychol- ogy, 28, 301-312.

Katz, I . (1981). Stigmas: A social psychological analysis. Hillsdale, NJ: Erlbaum. Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levine (Ed.), Nebraska

Symposium on Motivation, 14, 192-241. Kerlinger, F. N. (1984). Liberalism and conservatism: The nature and structure of social attitudes.

Hillsdale, NJ: Erlbaum. Kiesler, C. A., Collins, B. E., & Miller, N. (1969). Attitude change: A critical analysis of

theoretical approaches. New York: Wiley. Klein, S. B., & Loftus, J. (1993). The mental representation of trait and autobiographical

knowledge about the self. In T. S. Srull & R. S. Wyer, Jr. (Eds.), Advances in social cognition (Vol. 5, pp. 1-49). Hillsdale, NJ: Erlbaum.

Klein, S. B., Loftus, J., & Burton, H. A. (1989). Two self-reference effects: The importance of distinguishing between self-descriptiveness judgments and autobiographical retrieval in self-referent encoding. Journal of Personality and Social Psychology, 56, 853-865.

Kluegel, J. R., & Smith, E. R. (1986). Beliefs about inequality: Americans’ views of what is and what ought to be. New York: Aldine de Gruyter.

Kolodner, J. (1992). An introduction to case-based reasoning. Artificial Intelligence Review, 6, 3-34.

Krosnick, J. A., Betz, A. L., Jussim, L. J., & Lynn, A. R. (1992). Subliminal conditioning of attitudes. Personality and Social Psychology Bulletin, 18, 152-162.

Page 73: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 337

Kutner, H., Wilkins, C., & Yarrow, P. R. (1952). Verbal attitudes and overt behavior involving

Langer, E. J. (1975). The illusion of control. Journal of Personality and Social Psychology,

LaPiere, R. T. (1934). Attitudes versus actions. Social Forces, 13, 230-237. Lewicki, P. (1986). Nonconscious social information processing. San Diego, CA: Academic

Press. Lewin. K. (1938). The conceptual representation and the measurement of psychological forces.

Durham, N C Duke University Press. Lingle, J. H., & Ostrom, T. M. (1981). Principles of memory and cognition in attitude formation.

In R. E. Petty, T. M. Ostrom, & T. C. Brock (Eds.), Cognitive responses in persuasion (pp. 399-420). Hillsdale, NJ: Erlbaum.

Liska, A. E. (1984). A critical examination of the causal structure of the FishbeinIAjzen attitude-behavior model. Social Psychology Quarterly, 47, 61-74.

Logan, G. D. (1989). Automaticity and cognitive control. In J. S. Uleman & J. A. Bargh (Eds.). Unintended thought (pp. 52-74). New York: Guilford Press.

Lord, C. G . (1982). Predicting behavioral consistency from an individual’s perception of situational similarities. Journal of Personality and Social Psychology, 42, 1076-1088.

Lord, C. G., Desforges, D. M., Chacon. S., Pere, G. , & Clubb, R. (1992). Reflections on reputation in the process of self-evaluation. Social Cognition, 10, 2-29.

Lord. C. G., Desforges, D. M., Fein, S.. Pugh. M. A,, & Lepper, M. R. (1994). Typicality effects in attitudes toward social policies: A concept-mapping approach. Journalof Person- ality and Social Psychology, 66, 658-673.

Lord, C. G., Desforges, D. M., Ramsey, S. L.. Trezza, G . R., & Lepper, M. R. (1991). Typicality effects in attitude-behavior consistency: Effects of category discrimination and category knowledge. Journal of Experimental Social Psychology, 27, 550-575.

Lord, C. G. , Lepper, M. R., & Mackie, D. (1984). Attitude prototypes as determinants of attitude-behavior consistency. Journal of Personality and Social Psychology, 46, 1254- 1266.

Lord, C. G., Ross, L., & Lepper, M. R. (1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of Personaliry and Social Psychology, 37, 2098-2109.

Lydon, J. E., & Zanna, M. P. (1990). Commitment in the face of adversity: A value-affirmation approach. Journal of Personality and Social Psychology, 58, 1040-1057.

Macht, M. L.. & O’Brien. E. J. (1980). Familiarity-based responding in item recognition: Evidence for the role of spreading activation. Journal of Experimental Psychology: Human Memory and Learning, 6, 301-318.

Macht, M. L.. & Spear, N. E. (1977). Priming effects in episodic memory. Journalof Experimen- tal Psychology: Human Memory and Learning, 3, 733-741.

Mackie, D. M., & Worth, L. T. (1989). Cognitive deficits and the mediation of positive affect in persuasion. Journal of Personality and Social Psychology, 57, 27-40.

Macrae, C. N., Milne, A. B., & Bodenhausen, G. V. (1994). Stereotypes as energy-saving devices: A peek inside the cognitive toolbox. Journal of Personality and Social Psychology,

Madden, T. J., Ellen, P. S., & Ajzen, I. (1992). A comparison of the theory of planned behavior and the theory of reasoned action. Personality and Social Psychology Bulletin, 18, 3-9.

Mandler, G. , Nakamura, Y., & Van Zandy, B. J. S. (1987). Nonspecific effects of exposure on stimuli that cannot be recognized. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 646-648.

racial prejudice. Journal of Abnormal and Social Psychology, 47, 649-652.

32, 311-328.

66, 37-47.

Page 74: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

338 CHARLES G. LORD AND MARK R. LEPPER

Manstead, A. S. R., Proffitt, C., & Smart, J. L. (1983). Predicting and understanding mothers’ infant-feeding intentions and behavior: Testing the theory of reasoned action. Journal of Personality and Social Psychology, 44, 657-671.

McArthur, L. Z., & Baron, R. M. (1983). Toward an ecological theory of social perception. Psychological Review, 90, 215-238.

McCaul, K. D., Sandgren, A. K., O’Neill, H., & Hinsz, V. B. (1993). The value of the theory of planned behavior, perceived control, and self-efficacy for predicting health-protective behaviors. Basic and Applied Social Psychology, 14, 231-252.

McGuire, W. J. (1969). The nature of attitudes and attitude change. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology (2nd ed., Vol. 3, pp. 136-315). Reading, MA: Addison-Wesley.

McGuire, W. J. (1976). Some internal psychological factors influencing consumer choice. Journal of Consumer Research, 2, 302-319.

McGuire, W. J. (1985). Attitudes and attitude change. In G. Lindzey & E. Aronson (Eds.), Handbook ofsocialpsychology (3rd ed., Vol. 2, pp. 233-346). New York: Random House.

Millar, M. G., & Tesser, A. (1992). The role of beliefs and feelings in guiding behavior: The mismatch model. In L. L. Martin & A. Tesser (Eds.), The construction of social judgments (pp. 277-300). Hillsdale, NJ: Erlbaum.

Miniard, P. W., & Cohen, J. B. (1981). An examination of the Fishbein-Ajzen behavioral- intentions model’s concepts and measures. Journal of Experimental Social Psychology, 17, 309-339.

Mischel, W. (1968). Personality and assessment. New York: Wiley. Mitchell, A. A,, & Olson, J. C. (1981). Are product attribute beliefs the only mediator of

advertising effects on brand attitude? Journal of Marketing Research, 18, 318-332. Monteith, M. J., Devine, P. G.. & Zuwerink, J. R. (1993). Self-directed versus other-directed

affect as a consequence of prejudice-related discrepancies. Journal of Personality and Social Psychology, 64, 198-210.

Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Roles of inhibitionless spreading activation and limited-capacity attention. Journal of Experimental Psychology: General, 106, 225-254.

Neisser, U. (1967). Cognitive Psychology. EnglewoodCliffs, NJ: Prentice Hall. Newell, A., Rosenbloom, P. S., & Laird, J. E. (1989). Symbolic architectures for cognition.

In M. Posner (Ed.), Foundations in cognitive science (pp. 93-131). Cambridge, MA: MIT Press.

Niedenthal, P. M. (1992). Implicit perception for affective information. Journal of Experimental Social Psychology, 26, 505-527.

Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 231-259.

Norvell, N., & Worchell, S. A. (1981). A reexamination of the relation between equal status contact and intergroup attraction. Journal of Personality and Social Psychology, 41, 902-908.

Nosofsky, R. (1986). Attention, similarity, and the identification-categorization relationship. Journal of Experimental Psychology: General, 115, 39-57.

Olson, J. M., Roese, N. J., & Zanna, M. P. (1996). Expectancies. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 211-238). New York: Guilford.

Olson, J. C., Toy, D. R., & Dover, P. A. (1982). Do cognitive responses mediate the effects of advertising content on cognitive structure? Journal of Consumer Research, 9, 345-362.

Olson, J. M., & Zanna, M. P. (1993). Attitudes and attitude change. Annual Review of Psychology, 44, 117-154.

Page 75: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 339

Ottati, V. C. (1997). When the survey question directs retrieval: Implications for assessing the cognitive and affective predictors of global evaluation. European Journal of Social

Pagel, M. D., & Davidson, A. R. (1984). A comparison of three social-psychological models of attitude and behavior plan: Prediction of contraceptive behavior. Journal of Personality and Social Psychology, 47, 517-533.

Peak, H. (1955). Attitude and motivation. In M. R. Jones (Ed.), Nebraska Symposium on Motivation (Vol. 3, pp. 149-188). Lincoln: University of Nebraska Press.

Perdue, C. W.. & Gurtman, M. B. (1990). Evidence for the automaticity of ageism. Journal of Experimental Social Psychology, 26, 199-216,

Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport’s cognitive analysis of prejudice. Per.sonality and Social Psychology Bulletin, 5, 461 -476.

Petty, R. E., & Cacioppo. J. T. (1981). Attitudes and persuasion: Classic and conretnporary approaches. Dubuque, IA: Brown.

Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19. pp. 124-205). New York: Academic Press.

Rajaram, S., & Roediger. H. L. (1993). Direct comparison of four implicit memory tests. Journal of Experimental Psychology: Learning, Memory and Cognition, 19, 765-776.

Ramsey, S. L., Lord, C. G., Wallace, D. S., & Pugh. M. A. (1994). The role of subtypes in attitudes towards superordinate social categories. British Journal of Social Psychology, 33, 387-403.

Reder, L. M., & Anderson, J. R. (1982). Effects of spacing and embellishment on memory for the main points of a text. Memory and Cognition, 10, 97-102.

Reed. S. (1972). Pattern recognition and categorization. Cognitive Psychology, 3, 383-407. Reich, S. S., & Ruth, P. (1982). Wason’s selection task: Verification. falsification, and matching.

Rips, L. J.. Shoben, E. J., & Smith, E. E. (1973). Semantic distance and the verification of

Rock, 1. (1984). Perceptioti. New York: W. H. Freeman. Roediger, H. L. (1990). Implicit memory: Retention without remembering. American Psycholo-

gist, 45, 1043-1056. Roese, N. J., & Olson, J. M. (1994). Attitude importance as a function of repeated attitude

expression. Joitrnal of Experimental Social Psychology, 30, 39-51. Rogers, R. W. (1983). Cognitive and physiological processes in fear appeals and attitude

change: A revised theory of protection motivation. In J. T. Cacioppo & R. E. Petty (Eds.), Social psychophysiology: A sourcebook (pp. 153-176). New York: Guilford Press.

Rosch. E. (1978). Principles of categorization. In E. Rosch & B. Lloyd (Eds.), Cognition and categorization (pp. 28-48). Hillsdale, NJ: Erlbaum.

Rosenherg. M. J. (1956). Cognitive structure and attitudinal affect. Journal of Abnormal and Social P,yychology, 53, 367-372.

Ross, L. (1977). The intuitive psychologist and his shortcomings: Distortions in the attribution process. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 10. pp. 173-220). San Diego. CA: Academic Press.

Ross. L. (1987). The problem of construal in social inference and social psychology. In N. E. Grunberg, R. E. Nisbett, J. Rodin, & J. E. Singer (Eds.), A distinctive approach to psychological research: The influence of Stanley Schachter (pp. 77-96). Hillsdale, NJ: Erlbaum.

Ross, L. (1989). Recognizing construal processes. I n I. Rock (Ed.), The legacy of Solomon Asch: Essays in cognition and social psychology. Hillsdale. NJ: Erlbaum.

P.yychology, 27, 1-21.

Brirish Journal of Psychology, 73, 395-405.

semantic relations. Journal of Verbal Learning and Verbal Behavior, 12, 1-20.

Page 76: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

340 CHARLES G. LORD AND MARK R. LEPPER

Ross, L., & Ward, A. (1995). Psychological barriers to dispute resolution. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 21, pp. 255-304). San Diego, CA: Academic Press.

Ross, M., McFarland, C., Conway, M., & Zanna, M. P. (1983). Reciprocal relation between attitudes and behavior recall: Committing people to newly formed attitudes. Journal of Personality and Social Psychology, 45, 257-267.

Rothbart, M., & John, 0. P. (1985). Social categorization and behavioral episodes: A cognitive analysis of the effects of intergroup contact. Journal of Social Issues, 41, 81-104.

Scarberry, N. C., Ratcliff, C. D., Lord, C. G., Lanicek, D. L., & Desforges, D. M. (1997). Effects of individuating information on the generalization part of Allport’s contact hypothesis. Personality and Social Psychology Bulletin, 23, 1291-1299.

Schank, R. C., & Abelson, R. P. (1971). Scripts, plans, goals, and understanding: An inquiry into human knowledge structures. Hillsdale, NJ: Erlbaum.

Schifter, D. B., & Ajzen, I. (1985). Intention, perceived control, and weight loss: An application of the theory of planned behavior. Journal of Personality and Social Psychology, 49,

Schuette, R. A., & Fazio, R. H. (1995). Attitude accessibility and motivation as determinants of biased processing: A test of the MODE model. Personality and Social Psychology Bulletin, 21, 104-110.

Schwarz, N. (1994). Judgments in a social context: Biases, shortcomings, and the logic of conversation. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 26, pp. 123-162). San Diego, CA: Academic Press.

Schwarz, N. (1998). Accessible content and accessibility experiences: The interplay of declara- tive and experiential information in judgment. Personality and Social Psychology Review, 2, 87-99.

Schwarz, N., & Bless, H. (1992). Scandals and the public’s trust in politicians: Assimilation and contrast effects. Personality and Social Psychology Bulletin, 18, 514-579.

Schwarz, N., Bless, H., & Bohner, G. (1991). Mood and persuasion: Affective states influence the processing of persuasive communications. In M. P. Zanna (Ed.) Advances in experi- mental social psychology (Vol. 24, pp. 161-199). San Diego, CA: Academic Press.

Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judgments of well-being: Informative and directive functions of affective states. Journal of Personality and Social Psychology, 45, 513-523.

Schwarz, N., & Clore, G. L. (1996). Feelings and phenomenal experiences. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 433-465). New York: Guilford Press.

Schwarz, N., Strack, F., & Mai, H. P. (1991). Assimilation and contrast effects in part-whole question sequences: A conversational logic analysis. Public Opinion Quarterly, 55, 3-23.

Sedikides, C. (1990). Effects of fortuitously activated constructs versus activated communica- tion goals on person impressions. Journal of Personality and Social Psychology, 58,

Sherif, M., Harvey, 0. J., White, B. J,, Hood, W. R., & Sherif, C. W. (1961). Intergroup conflict and cooperation: The Robber’s Cave experiment. Norman, OK. University of Oklahoma Press.

Sherman, S. J., Presson, C. C., Chassin, L., Bensenberg, M., Corty, E., & Olshavsky, R. W. (1982). Smoking intentions in adolescents: Direct experience and predictability. Personal- ity and Social Psychology Bulletin, 8, 376-383.

Shin, H., & Nosofsky, R. (1992). Similarity-scaling studies of dot-pattern classification and recognition. Journal of Experimental Psychology: General, 121, 278-304.

843-851.

397-408.

Page 77: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 34 1

Sia, T. L., Lord, C. G., Blessum, K. A., Thomas, J. C., & Lepper. M. R. (in press). Activation of exemplars in the process of assessing social-category attitudes. Journal of Personality and Social Psychology.

Sia, T. L., Lord, C. G., Blessum, K., Ratcliff. C. D., & Lepper, M. R. (1997). Is a rose always a rose? The role of social category exemplar-change in attitude stability and attitude- behavior consistency. Journal of Personality and Social Psychology, 72, 501-514.

Sia, T. L., Thomas J. C., Lord, C. G., & Lepper, M. R. (1998). A longitudinal investigation of social category exemplar activation: Consistency imparts resistance to attitude change. Unpublished manuscript.

Slavin, R. E. (1979). Effects of biracial learning teams on cross-racial friendships. Journal of Educational Psychology, 71, 38 1-387.

Smetana, J. G., & Adler, N. E. (1980). Fishbein’s Value X Expectancy Model: An examination of some assumptions. Personality and Social Psychology Bulletin, 6 , 89-96.

Smith, E. R. (1992). The role of exemplars in social judgment. In L. L. Martin & A. Tesser (Eds.), The construction of social judgments (pp. 37-65). Hillsdale, NJ: Erlbaum.

Smith, E. R. (1998). Mental representation and memory. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handbook ofsocialpsychology (4th ed., Vol. 1, pp. 391-445). New York: McGraw-Hill.

Smith. E. R., Fazio, R. H., & Cejka, M. A. (1996). Accessible attitudes influence categorization of multiply categorizable objects. Journal of Personality and Social Psychology, 71,

Smith, E. R., Stewart, T. L., & Buttram, R. T. (1992). Inferring a trait from a behavior has long- term, highly specific effects. Journal of Personality arid Social Psychology, 62, 753-759.

Smith. E. R.. & Zarate, M. A. (1992). Exemplar-based model of social judgment. Psychological Review, 99, 3-21.

Srull. T. K., & Wyer, R. S. (1979). The role of category accessibility in the interpretation of information about persons: Some determinants and implications. Journal of Personality and Social Psychology, 37, 1660-1672.

Staats, A. W. (1983). Paradigmatic behaviorism: Unified theory for social-personality psychol- ogy. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 16, pp. 125-179). San Diego, CA: Academic Press.

Stangor, C. (1990). Arousal, accessibility of trait constructs, and person perception. Journal of Experimental Social Psychology, 26, 305-321.

Stangor, C., & Lange, J. E. (1994). Mental representations of social groups: Advances in understanding stereotypes and stereotyping. In M. P. Zanna (Ed.), Advances in experimen- tal social psychology (Vol. 26, pp. 357-416). San Diego, CA: Academic Press.

Stangor, C., Lynch, L., Duan, C., & Glass, B. (1992). Categorization of individuals on the basis of multiple social features. Journal of Personality and Social Psychology, 62,207-218.

Stangor, C., Thompson, E. P., & Ford, T. E. (1998). An inhibited model of stereotype inhibition. In R. Wyer (Ed.), Advances in social cognition (Vol. 11. pp. 193-210). Mahwah, NJ: Erlbaum.

Strack, F. (1992). The different routes to social judgments: Experimental versus informational strategies. In L. L. Martin & A. Tesser (Eds.), The construction of social judgments (pp. 37-65). Hillsdale, NJ: Erlbaum.

Strack, F., Martin, L. L.. & Stepper, S. (1988). Inhibiting and facilitating conditions of the human smile: A non-obtrusive test of the facial feedback hypothesis. Journalof Personality and Social Psychology, 53, 168-777.

Strack, F., Schwarz, N., & Gschneidinger, E. (1985). Happiness and reminiscing: The role of time perspective, mood, and mode of thinking. Journal of Personality and Social Psychol- ogy, 49, 1460-1469.

888-898.

Page 78: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

342 CHARLES G. LORD AND MARK R. LEPPER

Stroessner, S. J. (1998). Varieties of inhibition in social stereotyping. In R. Wyer (Ed.), Advances in social cognition (Vol. 11, pp. 211-226). Mahwah, NJ: Erlbaum.

Taylor, S. E., & Fiske, S. T. (1978). Salience, attention, and attribution: Top of the head phenomena. In L. Berkowitz (Ed.), Advances in experimentalsocial psychology (Vol. 11, pp. 249-288). San Diego, CA: Academic Press.

Tesser, A. (1978). Self-generated attitude change. In L. Berkowitz (Ed.), Advances in experi- mental social psychology (Vol. 11, pp. 289-338). San Diego, CA: Academic Press.

Tesser, A,, & Martin, L. (1996). The psychology of evaluation. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of bnsic principles (pp. 400-432). New York: Guilford.

Thomas, J. C., Lord, C . G. , & Lepper, M. R. (1998). Which behaviors do attitudes predict? A typology of attitude actions. Manuscript in preparation.

Thorndike, E. L. (1920). A constant error in psychological ratings. Journal of Applied Psychol-

Tourangeau, R., & Rasinski, K. A. (1988). Cognitive processes underlying context effects in attitude measurement. Psychological Bulletin, 103, 299-314.

Tourangeau, R., Rasinski, K. A,, & D’Andrade, R. (1991). Attitude structure and belief accessibility. Journal of Experimental Social Psychology, 27, 48-75.

Triandis, H. C. (1977). Interpersonal behavior. Monterey, CA: Brooks/Cole. Triandis, H. C. (1980). Values, attitudes, and interpersonal behavior. In M. E. Howe, Jr. &

M. M. Page (Eds.), Nebrasku Symposium on Motivation, 1979 (Vol. 27, pp. 195259). Lincoln: University of Nebraska Press.

Trope, Y. (1986). Identification and inferential processes in dispositional attribution. Psycho- logical Review, 9.7, 239-257.

Tulving, E., & Schacter, D. L. (1990). Priming and human memory systems. Science, 247, 301 -305.

Tulving, E., Schacter, D. L., & Stark, H. (1982). Priming effects in word-fragment completion are independent of recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 8, 336-342.

Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 7Y,

Vallacher, R. R., & Wegner, D. M. (1987). What do people think they’re doing? Action identification and human behavior. Psychological Review, 94, 3-15.

van der Pligt, J., & Eiser, J. R. (1984). Dimensional salience, judgment, and attitudes. In J. R. Eiser (Ed.), Attitudinal judgment (pp. 161 -177). New York: Springer-Verlag.

Vining, J., & Ebreo, A. (1992). Predicting recycling behavior from global and specific environ- mental attitudes and changes in recycling opportunities. Journal of Applied Social Psychol-

Wanke, M.. Bohner, G., & Jurkowitz, A. (1997). There are many reasons to drive a BMW: Does imagined ease of argument generation influence attitudes? Journal of Consumer Research.

Watts, W. A., & Holt, L. E. (1979). Persistence of opinion change induced under conditions of forewarning and distraction. Journal of Personality and Social Psychology, 37,778-789.

Weber, R., & Crocker, J. (1983). Cognitive processes in the revision of stereotypic beliefs. Journal of Personality and Social Psychology, 45, 961 -977.

Werth, J. L., & Lord, C. G. (1992). Previous conceptions of the typical group member and the contact hypothesis. Basic and Applied Social Psychology, 13, 351-369.

Wicker, A. W. (1969). Attitude versus actions: The relationship of verbal and overt behavioral responses to attitude objects. Journal of Social Issues, 25, 41-78.

Wilder, D. A. (1984). Intergroup contact: The typical member and the exception to the rule. Journal of Experimental Social Psychology, 20, 177-194.

ogy, 4, 25-29.

281-299.

ogy, 22, 1580-1607.

Page 79: [Advances in Experimental Social Psychology] Advances in Experimental Social Psychology Volume 31 Volume 31 || Attitude Representation Theory

ATTITUDE REPRESENTATION THEORY 343

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, 5-16.

Wilson, T. D., Dunn, D. S.. Kraft, D., & Lisle. D. J. (1989). 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 e.wperimental social psychology (Vol. 22, pp. 287-338). New York: Academic Press.

Wilson, T. D.. & Hodges. S. D. (1992). Attitudes as temporary constructions. In L. L. Martin & A. Tesser (Eds.), The construction ofsocialjudgmenfs (pp. 37-65). Hillsdale, NJ: Erlbaum.

Wilson, T. D., & Kraft, D. (1993). “Why do I love thee? Effects of repeated introspections about a dating relationship on attitudes toward the relationship. Personality and Social Psychology Bulletin, 19, 409-418.

Winslow, R. W., Franzini, L. R., & Hwang. J. (1992). Perceived peer norms. casual sex, and AIDS risk reduction. Journal of Applied Social Psychology, 22, 1809-1827.

Wyer, R. S., & Srull, T. K. (1986). Human cognition in its social context. Psychological Review.

Yzerbyt, V. Y., Schadron, G., Leyens, J., & Rocher, S. (1994). Social judgeability: The impact of meta-informational cues on the use of stereotypes. Journal of Personality and Social Psychology, 66, 48-55.

Zajonc, R. B. (1960). The process of cognitive tuning in communication. Journalof Abnormal arid Social Psychology, 61, 159-167.

Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personaliry and Social Psychology [Monograph]. 9, 1-27.

Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psycholo-

Zajonc, R. B. (1998). Emotions. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handbook of social psychology, (4th ed., Vol. 1, pp. 591-632). New York: McGraw-Hill.

Zanna. M. P., Kiesler, C. A.. & Pilkonis. P. A. (1970). Positive and negative attitudinal affect established by classical conditioning. Journal of Personality and Social Psychology,

Zanna, M. P.. & Rempel, J. K. (1988). Attitudes: A new look at an old concept. In D. Bar- Tal & A. W. Kruglanski (Eds.), The social psychology of knowledge (pp. 315-334). Cambridge, England: Cambridge University Press.

Zirate, M. A,, &Smith, E. R. (1990). Person categorization and stereotyping. Social Cognition, 8. 161-185.

93, 322-359.

gist, 35, 151-175.

14, 321-328.