Attitude dissimilation and persuasion

JOURNAL OF EXPERIMENTAL SOCIAL PSYCHOLOGY ARTICLE NO. 34, 182–201 (1998) JS971348 Attitude Dissimulation and Persuasion Gregory R. Maio University of Wales at Cardiff, UK and James M. Olson University of Western Ontario, Canada Received: March 12, 1997; revised: July 25, 1997; accepted: November 17, 1997 Because people’s true attitudes toward an object become more accessible immediately after indicating an attitude that they know to be false, we predicted that such attitude dissimulation might paradoxically cause the true attitude to have a stronger effect on subsequent judgments. In two experiments, participants were randomly assigned to express false, true, or no attitudes toward Albert Einstein. Next, in an ostensibly separate study, participants read a persuasive message describing Einstein’s dislike for a particular technology and then rated their attitude toward this technology. As expected, results indicated that participants who had previously indicated false or true attitudes toward Einstein indicated more dislike for the technology than participants who had not previously expressed attitudes toward him. A second experiment replicated the effect of attitude dissimulation using a manipulation that was similar to that employed in cognitive dissonance experiments. Possible effects of falsely indicating ‘‘politically correct’’ attitudes are discussed. r 1998 Academic Press Attitudes can be defined as associations in memory between objects and evaluations (Fazio, 1990; Olson & Zanna, 1993). For example, opera fans who like Luciano Pavarotti have formed an internal association between the evaluation ‘‘like’’ and the attitude object ‘‘Luciano Pavarotti.’’ Consequently, if asked about Pavarotti, fans can recall and express liking for him. The research reported in this paper was funded by a research grant to the second author and was conducted while the first author was supported by a doctoral fellowship, both from the Social Sciences and Humanities Research Council of Canada. We thank Russell Fazio, Richard Petty, Yaacov Trope, Mark Zanna, and several anonymous reviewers for their comments on a previous version of this paper. Address correspondence and reprint requests to Greg Maio, School of Psychology, Cardiff University of Wales, Cardiff, Wales, UK CF1 3YG. 182 0022-1031/98 $25.00 Copyright r 1998 by Academic Press All rights of reproduction in any form reserved. ATTITUDE DISSIMULATION 183 In everyday life, however, people sometimes express attitudes that they know are false. Research has examined numerous issues related to such dissimulation (e.g., DePaulo, Stone, & Lassiter, 1985; Olson, 1988; Olson & Ross, 1988; Zuckerman & Driver, 1985), but the aspect of attitude dissimulation that has received the most attention has been its effect on the deceivers’ own attitudes. Specifically, in past research on cognitive dissonance and self-perception processes (see Cooper & Fazio, 1984; Cooper & Scher, 1994), it has been found that people who are asked to express a false attitude toward an attitude object will sometimes change their attitude to make it more congruent with their lie. This attitude change occurs primarily when people freely express false attitudes in a manner that produces aversive consequences (Cooper & Fazio, 1984; Cooper & Scher, 1984; cf. Harmon-Jones, Brehm, Greenberg, Simon, & Nelson, 1996). Although attitude dissimulation that occurs without choice and without aversive consequences does not typically produce dissonance-motivated or selfperception-mediated attitude change, it might nonetheless have important psychological consequences that have not yet been recognized. Specifically, in the present set of studies, we tested the paradoxical prediction that such attitude dissimulation can cause people’s original attitudes to have stronger effects on subsequent attitude-relevant judgments. The Effect of Attitude Dissimulation on Attitude Accessibility Consistent with a model of attitude expression (Judd & Brauer, 1995), we suggest that a variety of pressures can induce people to express attitudes that are different from their true attitudes, while leaving the true attitudes unchanged. We propose that an indirect consequence of such dissimulation is increased accessibility of the true attitude from memory. For example, at a business luncheon, a person may be asked about his or her attitude toward the Republican party. In response, the business person might access and compute a negative attitude, but then recall that the question was asked by a staunch Republican. If the query came from a Republican who happened to be a valuable client, the respondent might quickly choose to indicate a neutral or positive attitude, while being aware of his or her true negative attitude. As a result of this dissimulation, the person might increase his or her chances of getting a valuable contract or promotion, assuaging any concerns about lying. Nevertheless, this dissimulation might remind the person of his or her true attitude, which could increase the attitude’s accessibility. We recently examined the effect of attitude dissimulation on attitude accessibility in two experiments (Maio & Olson, 1995). In these studies, we constructed a situation where attitude dissimulation was unlikely to cause attitude change. In particular, similar to the example above, we provided a strong external justification for attitude dissimulation, one which would override any potential dissonancemotivated attitude change. Specifically, participants were simply instructed to express false attitudes—no choice was made salient—and the dissimulation produced no aversive consequences. We also distinguished between attitude dissimulation, which involves indicat- 184 MAIO AND OLSON ing an attitude that one knows to be false, and uttering simple falsehoods: People can utter falsehoods without being aware of the truth, but dissimulation involves making a false statement while being aware of the truth. Thus, to elicit a laboratory analog of attitude dissimulation, we forced participants to access their true attitudes while expressing false attitudes. This was achieved by asking them to indicate the opposite of their true attitudes. We also included a condition that elicited attitude falsehoods, by asking participants to make specific statements that we knew were false (e.g., ‘‘horses are bad’’). In this condition, participants did not have to access their true attitudes in order to express false attitudes. Finally, we included a condition where participants indicated their true attitudes. After participants completed the experimental manipulation, we measured attitude accessibility by asking participants to express quickly and accurately their true attitudes toward (a) the objects for which they had expressed the opposite of their true attitudes, (b) the objects for which they had given specific false attitudes, (c) the objects that they had evaluated truthfully, and (d) other objects that they had not evaluated at all. Before and after the manipulation and attitude accessibility measure, participants rated their attitudes toward all of the attitude objects. As expected, results indicated that neither attitude dissimulation, nor indicating specific false attitudes, nor truthful expression caused attitude change from the pre- to posttest. Nonetheless, both attitude dissimulation and expression caused attitudes to become more accessible (i.e., more quickly recalled) than if specific false attitudes or no attitudes had been expressed. The Potential Effect of Attitude Dissimulation on Attitude-Relevant Judgments The effect of attitude dissimulation on attitude accessibility is interesting because, in general, highly accessible attitudes have larger effects on judgments and behavior than inaccessible attitudes (Fazio, 1990). Therefore, by making true attitudes more accessible, attitude dissimulation might paradoxically cause the true attitudes to have larger effects on subsequent judgments and behavior. Although this hypothesis is plausible, other perspectives suggest that attitude dissimulation might increase the impact on judgments and behavior of the false, rather than the true, attitude (even though the true attitude may become more accessible). For example, to the extent that expressing a false attitude is rewarded, individuals may be more likely to repeat such behavior in the future. Also, the mere expression of an assertion (such as a false attitude) is sometimes sufficient to increase the perceived plausibility of that assertion, especially when perceivers are too ‘‘cognitively busy’’ to monitor their judgments carefully (Arkes, Boehm, & Xu, 1991; Gilbert, Krull, & Malone, 1990). Finally, it is possible that attitude dissimulation makes both true and false attitudes more accessible, which might eliminate any effect of dissimulation on subsequent judgments and behavior. For example, recent conceptualizations of attitudes have suggested that positive and negative evaluations of an object can coexist and can even be simultaneously accessible (Cacioppo, Gardner, & ATTITUDE DISSIMULATION 185 Berntson, 1997; Maio, Bell, & Esses, 1996). Perhaps, then, attitude dissimulation might increase the accessibility of the true and false evaluations, creating attitudinal ambivalence and eliminating effects of dissimulation on judgments. Given these diverse possibilities, it is important to conduct an empirical test of the effects of attitude dissimulation on subsequent judgments. The present experiments provide such a test. Overview The present experiments build on our previous research and on an experiment by Roskos-Ewoldsen and Fazio (1992). Roskos-Ewoldsen and Fazio (1992) manipulated whether or not participants repeatedly expressed their attitudes toward Jacques Cousteau, who was a popular oceanographer. After completing this manipulation, participants were asked to rate their agreement with an essay ostensibly written by Cousteau. It was predicted that participants who repeatedly expressed attitudes toward Cousteau would agree more with his essay than participants who had not previously expressed attitudes toward him, because repeated expression should make more accessible participants’ positive attitudes toward Cousteau. Results confirmed this hypothesis. (In addition, manipulation checks showed that repeated attitude expression did not change participants’ attitudes toward Cousteau.) Using similar procedures, we explored whether people will also be more likely to accept persuasive messages from a likeable source when they have repeatedly expressed false attitudes toward him or her (i.e., indicated dislike toward the source), while being aware of their true attitudes. We also tested whether attitude dissimulation affected cognitive responses to the message. According to contemporary models of persuasion (e.g., Chaiken, Liberman, & Eagly, 1989; Petty & Cacioppo, 1986), variables can sometimes affect yielding to a message without affecting cognitive reactions to the message. For example, various heuristic cues (e.g., source likeability) can sometimes affect yielding without affecting people’s message-relevant thoughts (see Chaiken et al., 1989; Petty & Cacioppo, 1986). Roskos-Ewoldsen and Fazio (1992) did not examine the effect of attitude expression on cognitive responses in their experiments, but suggested that this issue should be examined. Consequently, we included a measure of participants’ message-relevant thoughts in our experiment (see, e.g., Chaiken, 1980; Millar & Millar, 1990; Petty, Schumann, Richman, & Strathman, 1993). EXPERIMENT 1 In our first experiment, attitude dissimulation was manipulated using a technique similar to the one employed in our previous research (Maio & Olson, 1995). In particular, we manipulated whether participants expressed the opposite of their true attitudes, true attitudes, or no attitudes toward Albert Einstein. Albert Einstein was chosen as the target because we expected that people strongly like him, making it unlikely that participants would change their attitudes toward him as a result of attitude dissimulation or expression. Participants were then presented 186 MAIO AND OLSON with an essay describing nanotechnology, which is a technology that might someday enable the creation of devices far smaller than a microchip. The essay indicated that Albert Einstein was opposed to research on nanotechnology. We then measured attitudes toward stopping nanotechnology research and cognitive reactions to the message. Finally, to ensure that attitudes toward Einstein were unaffected by the manipulation, we measured these attitudes and three characteristics of these attitudes: attitude certainty, importance, and intensity or strength. Attitude certainty is the extent to which people are confident that their attitude is correct; attitude importance refers to the degree to which people care deeply about their attitude; and attitude intensity is the strength of the emotional reaction elicited by an attitude object (Krosnick, Boninger, Chuang, Berent, & Carnot, 1993). Recent findings indicate that these three attitude characteristics produce effects similar to those of attitude accessibility (see Krosnick et al., 1993; cf. Bassili, 1996), so it seemed worthwhile to test whether these variables might mediate the effects of our manipulation. Method Participants Participants were 52 psychology undergraduates (7 men, 43 women, and 2 who did not indicate their sex) at the University of Western Ontario, who participated for course credit. Data from five additional participants were omitted because of failure to follow instructions (N 5 3), prior knowledge about nanotechnology (N 5 1), or suspicion (N 5 1). Overview One to four participants took part in each session. To ensure that the instructions were understood, we delivered them verbally as well as on paper. Thus, all participants in the same session were assigned to the same experimental condition. Participants were told that they would be participating in a number of different ‘‘surveys.’’ In this series of ‘‘surveys,’’ the second measured initial attitudes toward Albert Einstein; the seventh contained a manipulation of whether participants expressed true, false, or no attitudes toward Albert Einstein; and the eighth contained a persuasive message that (falsely) cited Albert Einstein as advocating a ban on nanotechnology. After reading this message, participants indicated (a) their attitudes toward banning nanotechnology, (b) their thoughts while reading the message, (c) their attitudes toward Einstein, and (d) the strength of their attitudes toward Einstein. The other ‘‘surveys’’ were included for unrelated research and served as fillers. After completing all of the materials, participants were probed for suspicion and debriefed. Premanipulation Attitudes toward the Source Premanipulation attitudes toward Einstein were measured within a ‘‘survey of students’ attitudes toward famous people from the past and present.’’ The experimenter stated that this survey was a pilot questionnaire that would enable him to choose some famous people as topics for a questionnaire to be used later in the year. Participants then rated their attitudes toward 26 famous people, using a 9-point, skewed, semantic-differential scale. This scale contained one negative value, 21 (unfavorable), a neutral point, 0, and seven positive values from 11 (slightly favorable) to 17 (extremely favorable). The seventeenth famous person rated in the survey was Albert Einstein. Experimental Manipulation Control condition. The experimenter stated that the purpose of the seventh ‘‘survey’’ was to examine perceptions of famous individuals from all walks of life. The experimenter stated that he was ATTITUDE DISSIMULATION 187 interested in discovering which famous individuals students liked the most, because he wanted to use those famous individuals as topics for a study later in the year. He then distributed a booklet containing the names of five different individuals who were well-known in Canada: Wayne Gretzky, Jean Chretien (the Canadian Prime Minister), Whitney Houston, Bill Clinton, and Donald Trump. Participants rated each individual using five 6-point (23 to 13) semantic-differential scales, which were anchored by awful/nice, worthless/valuable, unpleasant/pleasant, bad/good, and unfavorable/ favorable. (No neutral point was presented.) Each scale was presented twice for each famous individual. Thus, a total of ten scales (5 3 2) were presented for each famous individual. These scales were presented in random order, except that adjacent scales were never used to rate the same person. Attitude expression. The instructions in this condition were identical to those in the control condition, except that participants in this condition were asked to rate Albert Einstein instead of Jean Chretien. Participants rated Einstein favorably (M 5 1.78). Attitude dissimulation. The targets in this condition were identical to those in the attitude expression condition; thus, Albert Einstein was rated in both conditions. However, participants in the attitude dissimulation condition were asked to indicate the opposite of their true attitudes toward each famous individual. For example, if their true rating was 12, they were asked to indicate 22. To justify this request, the experimenter stated that this manner of responding was part of a new procedure for detecting attitudes. As instructed, participants rated Einstein unfavorably (M 5 22.18). Persuasive Message In the eighth ‘‘survey,’’ participants were told that we were interested in measuring their attitudes toward a wide variety of new social issues. The experimenter stated that he did not have time to ask each participant questions about all of the issues, so he was asking each person to draw the name of an issue from a box. He stated that they would be presented with a photocopy of an article describing the selected issue, because most people are unfamiliar with the issues contained in the box. Through appropriate contrivances, participants always drew the topic of ‘‘research on nanotechnology.’’ They were then given a booklet containing a (fictitious) article about research on nanotechnology. The article looked like a photocopy of an editorial page from a scientific magazine. The top of the page was entitled ‘‘Opinion,’’ and the bottom right showed the name and institutional affiliation of the (fictitious) author. Below the author’s name, a hydrogen atom was depicted above a caption indicating that the hydrogen atom is a basic building block of nanotechnology. The remaining two-thirds of the page displayed the text of the message. A portion of the text introduced and described nanotechnology (parts of the text were inaccurate): Despite the incredible achievements that were facilitated by the invention of the microchip, a new technology is emerging that will make the microchip look like a relic. It is expected that, in the next 30 years, this new technology will begin to create complicated ‘‘minifactories’’ that are one millionth the size of a microchip. This new technology is called nanotechnology, because its goal is the creation of ‘‘mini-factories’’ from individual atoms, which are approximately one nanometer in size (i.e., one millionth of a millimeter). In fact, these ‘‘atom factories’’ are nicknamed ‘‘nanites,’’ which is a name that is borrowed from a Star Trek episode that presented possible harmful consequences of these ‘‘mini-factories.’’ Because nanites would be built from atoms, nanites could take atoms from specific objects and recombine the atoms to form different objects. The objects that would be created would depend on the design of the nanite. For example, nanites could be designed to create and repair DNA molecules, which are the basic building blocks of all life. Nanites could even be programmed to create copies of themselves. After describing some benefits of nanotechnology, the text then (falsely) stated that Einstein foresaw its invention. Interestingly, Albert Einstein foresaw the invention of nanites. He was excited by the possible benefits of creating nanites, but he was also concerned that nanites could be 188 MAIO AND OLSON programmed to have destructive consequences. Specifically, Einstein was concerned that nanites could be designed to tear atoms from all objects in the environment, and not just specific objects, such as viruses. Such nanites could easily consume organic tissue and, consequently, kill people. Because of their destructive capabilities, nanites could be designed as weapons of war. For this reason and other reasons, Einstein felt that nanotechnology should not be explored. Despite Einstein’s concerns, research on nanotechnology has moved forward, because the scientific community now feels that nanotechnology can be researched safely. Nevertheless, in my opinion, it is only a matter of time before Einstein’s vision becomes reality. Eventually, we will be able to design nanites for use as deadly weapons. Thus, even though there are tremendous potential benefits of this new technology, the risks are too great. We should not explore nanotechnology. Dependent Measures Attitudes toward nanotechnology. After reading the persuasive message, participants rated their attitudes toward stopping the development of nanotechnology. Participants rated their attitudes in three different ways. First, they evaluated the idea of stopping research on nanotechnology using five 7-point (23 to 13) semantic-differential scales, which were anchored by bad/good, harmful/ beneficial, foolish/wise, negative/positive, and worthless/valuable. Responses to these scales were then averaged (a 5 .89). Next, using a 7-point scale from 23 (extremely unfavorable) to 13 (extremely favorable), participants rated their general favorability toward stopping research on nanotechnology. Finally, using a 9-point scale from 24 (disagree very strongly) to 14 (agree very strongly), they rated the extent to which they agreed with banning research on nanotechnology. Scores on the three attitude measures were significantly correlated (lowest r (45) 5 .76, p , .001). Thus, we created an aggregate index of these measures, weighting each measure equally. Because there were different response scales for the three measures, we first calculated z-scores for each measure, and then these z-scores were averaged to form an overall index of attitudes. Message-relevant thoughts. After indicating their attitudes toward stopping research on nanotechnology, participants were asked to write down any thoughts that occurred to them while they read the article. The instructions and coding procedures for the thought-listing task were adapted from Cacioppo and Petty (1981). Participants were given 18 lines to list their thoughts, and they were asked to number each thought. After listing their thoughts, participants were asked to go back and rate the extent to which each thought was favorable toward stopping the development of nanotechnology. They rated their thoughts using a 7-point scale from 23 (extremely unfavorable) to 13 (extremely favorable). The ratings for each thought were averaged to form an index of the extent to which the thoughts reflected favorability toward stopping the development of nanotechnology. Postmessage attitudes toward the source. After completing the thought-listing measure, participants were told that previous research has found that people’s opinions of an individual can influence their attitudes toward the person’s views. The experimenter stated that, in his analysis of attitudes toward nanotechnology, he wanted to statistically control for attitudes toward Albert Einstein. Participants were then asked to rate their attitudes toward Einstein. These ratings were performed using the same semantic-differential scale as was used to assess premanipulation attitudes toward Einstein. Confidence, importance, and strength of attitudes toward the source. Using 9-point scales from 0 (not at all) to 8 (extremely), participants rated their confidence about their attitude toward Einstein, the importance of their attitude, and the strength of their attitude. Results Preliminary Analyses Premanipulation and postmanipulation attitudes toward the source. Premanipulation and postmessage attitudes toward Einstein were submitted to a 3 3 2 189 ATTITUDE DISSIMULATION TABLE 1 EXPERIMENT 1: ATTITUDES TOWARD NANOTECHNOLOGY AMONG PARTICIPANTS WHO HAD EXPRESSED FALSE ATTITUDES, TRUE ATTITUDES, OR NO ATTITUDES TOWARD ALBERT EINSTEIN Attitude measure Attitude dissimulation Attitude expression Control Composite Differential scales Favorability Agreement 0.27 0.72 0.53 1.07 0.26 0.79 0.53 0.82 20.50 20.20 20.63 20.75 Note. The table presents the mean raw scores for each attitude measure, except for the composite measure, which produced z-scores. Higher scores indicate greater favorability toward nanotechnology. (condition 3 trial) mixed-model ANOVA, with trial (pre- vs postmessage) as the within-subjects factor. As expected, results indicated no significant effects. Thus, the manipulation did not affect attitudes toward Einstein. Overall, attitudes toward Einstein were positive (M 5 4.45; 5 5 very favorable), with no participants indicating a negative attitude toward him. Confidence, importance, and strength of attitudes toward the source. Threelevel one-way ANOVAs on participants’ ratings of attitude confidence, importance, and strength indicated no significant effects. Attitudes toward Einstein were characterized by high confidence (M 5 4.58; 4 5 moderately confident), mild importance (M 5 2.50; 2 5 mildly important) and mild to moderate strength (M 5 3.09; 4 5 moderately strong). Attitudes toward Nanotechnology A 3-level one-way ANOVA on participants’ attitudes toward nanotechnology indicated an effect of the experimental manipulation, F (2, 44) 5 3.90, p , .03. Examination of this effect confirmed our predictions. First, participants who were asked to repeatedly express false attitudes toward Albert Einstein were more favorable toward his recommendation that research on nanotechnology be stopped (M 5 0.27) than were participants who had not indicated true or false attitudes toward him (M 5 20.50), t (44) 5 2.41, p , .02. In addition, participants who were asked to repeatedly express true attitudes toward Albert Einstein were more favorable toward a ban on nanotechnology (M 5 0.26) than were participants who had not indicated true or false attitudes toward him, t (44) 5 2.42, p , .02. The attitudes of participants who had expressed false attitudes toward Einstein did not significantly differ from the attitudes of participants who had expressed true attitudes toward Einstein, t (44) 5 0.03, ns. (See Table 1.) Message-Relevant Thoughts A 3-level one-way ANOVA on thought favorability scores (average rating of the thoughts) indicated a significant effect of the experimental manipulation, F (2, 40) 5 3.20, p 5 .05. As expected, participants who were asked to repeatedly express false attitudes toward Albert Einstein indicated thoughts that were more 190 MAIO AND OLSON favorable toward stopping nanotechnology research (M 5 0.49) than did participants who had not indicated true or false attitudes toward him (M 5 20.44), t (40) 5 1.78, p , .05 (one-tailed). In addition, participants who were asked to repeatedly express true attitudes toward Albert Einstein indicated thoughts that were more favorable toward stopping nanotechnology (M 5 0.81) than did participants who had not indicated true or false attitudes toward him, t (40) 5 2.46, p , .02. Participants who expressed false attitudes toward Einstein did not indicate thoughts that were significantly different from the thoughts indicated by participants who expressed true attitudes toward him, t (40) 5 20.64, ns. We tested whether the effects of attitude dissimulation and attitude expression on thoughts mediated the effects on attitudes toward nanotechnology. To test whether thoughts played a mediating role, we conducted three different regression analyses for both the attitude expression and dissimulation conditions. First, experimental condition was used to predict attitudes toward stopping nanotechnology (control 5 0, dissimulation or expression 5 1). Second, experimental condition was used to predict the favorability of thoughts regarding stopping nanotechnology. Third, experimental condition and thought favorability were used to predict attitudes toward stopping nanotechnology. The analyses comparing the attitude dissimulation and control conditions were consistent with what would be expected if thoughts mediated the effect of dissimulation on attitudes. Specifically, the effects of attitude dissimulation on attitudes, B 5 .39, t (29) 5 2.25, p , .04, and on thought favorability, B 5 .32, t (25) 5 1.72, p , .05 (one-tailed) were significant in the first two regression analyses, and the effect of thought favorability on attitudes was significant in the third analysis, B 5 .66, t(24) 5 4.60, p , .001. The effect of attitude dissimulation was not significant in the third regression analysis, B 5 .19, t (24) 5 1.29, ns, where favorability of thoughts was simultaneously entered as a predictor. Similarly, the results in the attitude expression condition supported the idea that thoughts mediated the effect of repeated expression on attitudes. Specifically, the effects of attitude expression on attitudes, B 5 .40, t (30) 5 2.42, p , .03, and on thought favorability, B 5 .43, t (27) 5 2.49, p , .02, were significant in the first two regression analyses, and the effect of thought favorability on attitudes was significant in the third regression analysis, B 5 .73, t (25) 5 5.12, p , .001. The effect of attitude expression was not significant in the third regression analysis, B 5 .08, t (25) 5 0.55, ns, where favorability of thoughts was simultaneously entered as a predictor. Although these results are consistent with the hypothesis that thoughts about nanotechnology mediated the impact of attitude dissimulation and attitude expression on attitudes toward nanotechnology, their correlational nature means that they are not necessarily inconsistent with the possibility that attitude dissimulation and expression directly affected attitudes, which may, in turn, have affected thoughts about nanotechnology. That is, participants’ listed thoughts might have also served to rationalize their expressed attitudes. To examine this possibility, we used regression analyses to test whether the effects of attitude dissimulation and ATTITUDE DISSIMULATION 191 expression on thoughts were mediated by attitudes toward nanotechnology. Results indicated that the effect of attitude dissimulation on thoughts was not significant, B 5 .04, t (24) 5 0.27, ns, when attitudes were simultaneously entered as a predictor, B 5 .71, t (24) 5 4.60, p , .001. Similarly, the effect of attitude expression on thoughts was not significant, B 5 .16, t (25) 5 1.15, ns, when attitudes were simultaneously entered as a predictor, B 5 .70, t (25) 5 5.12, p , .001. These results indicate that attitudes toward nanotechnology may have mediated the effects of attitude dissimulation and expression on thoughts about nanotechnology. Overall, then, participants’ thoughts might have affected their attitudes and vice versa (see below).1 Discussion Expressing true or false evaluations of Albert Einstein did not alter attitudes toward him, but did affect attitudes toward his alleged point of view regarding nanotechnology. Specifically, participants were more favorable toward stopping nanotechnology when they had repeatedly expressed either true or false attitudes toward Einstein. Presumably, attitude expression and attitude dissimulation each made participants’ positive attitudes toward him more accessible, which caused them to agree more with his point of view. The effect of truthful attitude expression on yielding to the message replicated past findings (Roskos-Ewoldsen & Fazio, 1992), whereas the finding that lying about attitudes can increase the influence of true attitudes on yielding is novel. The possibility that dissimulation would increase the impact of the false attitude was not confirmed. Thus, the experiment showed that expression of an attitude does not necessarily cause subsequent attitude-relevant judgments to become more congruent with the expressed attitude (e.g., Houston & Fazio, 1989; Roskos-Ewoldsen & Fazio, 1992). That is, expression of dislike for something might not always cause subsequent judgments of it to be less favorable; if the expressed dislike was knowingly false, subsequent judgments could actually become more favorable. Also, by showing that attitude dissimulation caused true 1 For reasons of space, we have not included description of another study that was very similar to Experiment 1, with five notable differences. First, a premanipulation measure of attitudes toward Einstein was not included; only a postmanipulation measure was administered. Second, the experiment did not contain measures of the strength of attitudes toward Einstein. Third, participants in the control condition rated their attitudes toward David Suzuki, a popular Canadian scientist, rather than their attitudes toward Jean Chretien. Fourth, participants were asked to list fewer than six thoughts in their thought-listing. Fifth, the favorability of participants’ thoughts was rated by two trained coders and not by the participants themselves, possibly causing the scoring of favorability to be less accurate (see Bell, Esses, & Maio, 1996; Esses, Haddock, & Zanna, 1993; Maio et al., 1996). Despite these different procedures, the principal results were similar to the findings for Experiment 1. As expected, participants were significantly more favorable toward stopping nanotechnology when they had repeatedly expressed either true or false attitudes toward Einstein. Expressing true or false evaluations of Einstein did not alter attitudes toward him. Unlike the findings for Experiment 1, the experimental manipulation did not significantly affect responses to the thought-listing measure, perhaps because coders’ classifications of participants’ thoughts are less accurate than participants’ own ratings. 192 MAIO AND OLSON attitudes to have a larger effect on attitude-relevant judgments, the results indicated that the effects of attitude dissimulation on attitude accessibility (Maio & Olson, 1995) have practical significance. Mediational analyses suggested that the effects of attitude dissimulation and attitude expression on attitudes were mediated by message-relevant thoughts. However, mediational analyses also suggested that the effects of attitude dissimulation and attitude expression on thoughts were mediated by attitudes. These paradoxical findings might indicate that the effects of the manipulations on attitudes and thoughts were reciprocally interdependent. In other words, participants’ thoughts might have biased their attitudes, which, in turn, might have further biased their thoughts. To date, most persuasion researchers have not tested for such reciprocal interdependence of thoughts and attitudes (Eagly & Chaiken, 1993). Our data suggest that such reciprocal relations might occur and should be tested. Importantly, the results of Experiment 1 might also shed new light on previous research that has occurred under the rubric of cognitive dissonance theory (Festinger, 1957). A primary goal of past cognitive dissonance research has been the identification of conditions that are necessary for attitude dissimulation to elicit attitude change. Results have indicated that attitude dissimulation tends to elicit attitude change only when the dissimulator freely chooses to dissimulate in a context that produces aversive consequences for which the dissimulator feels personally responsible (Cooper & Fazio, 1984). Researchers did not appear to anticipate interesting consequences of attitude dissimulation when such conditions were absent. Experiment 1 demonstrates that there may be measurable effects of attitude dissimulation even when it does not elicit attitude change. Indeed, it is possible that, in past cognitive dissonance research, similar effects occurred undetected in the conditions where attitude dissimulation did not produce attitude change. Nevertheless, this implication would be strengthened if we found similar results using a different method to manipulate dissimulation. Our manipulation in Experiment 1 elicited attitude dissimulation in a heavy-handed fashion: Participants were asked to circle the opposite of their true attitudes. Participants might have felt that they were simply translating their responses into opposites, rather than dissimulating. A new procedure requiring more creative dissimulation would rectify this ambiguity. EXPERIMENT 2 Experiment 2 was similar to Experiment 1, except that Experiment 2 focused solely on the effects of attitude dissimulation, manipulating it using a context like that of the prototypical ‘‘no attitude change’’ condition in a cognitive dissonance experiment. As in many past dissonance studies (e.g., Scher & Cooper, 1989; Steele & Liu, 1983), attitude dissimulation was elicited by asking participants to write counterattitudinal essays. Specifically, we asked participants to write negative, untrue essays about Albert Einstein. Also, paralleling past cognitive ATTITUDE DISSIMULATION 193 dissonance studies, participants were told that their essays would be presented to others, whose attitudes might be affected by the essays. To minimize the likelihood that attitude dissimulation would produce attitude change, we gave participants no choice about engaging in attitude dissimulation and told them that we did not expect their stories about Albert Einstein to affect future participants’ perceptions of him (thus reducing possible aversive consequences). We should also note that, because we eliminated choice in our dissimulation condition, attitude change would not be expected from a self-perception perspective either (Bem, 1972; Olson, 1990, 1992; Olson & Roese, 1995). Method Participants Participants were 57 psychology undergraduates (12 men and 45 women) at the University of Western Ontario, who participated for course credit. Data from three additional participants were omitted because of failure to follow instructions (N 5 2), or suspicion (N 5 1). Two additional participants failed to complete the experimental manipulation. Overview Participation occurred individually. As in Experiment 1, participants were told that they would be completing several different ‘‘surveys.’’ The experimenter stated that all of the surveys were pretests that would help him develop questionnaires for studies later in the year. He also emphasized that responses were anonymous and confidential, stating that participants should not write their name on the booklets and that they should place their completed booklets inside a box in the room; a research assistant would collect the box at the end of the day and enter the data into a computer. Participants were then given the first ‘‘survey,’’ which measured their initial attitudes toward Albert Einstein; the third portion manipulated whether or not participants expressed false attitudes toward (i.e., wrote a negative story about) Albert Einstein; and the fourth portion contained the same persuasive message and dependent measures that were used in Experiment 1. The second ‘‘survey’’ was included for an unrelated study and served as a filler between the initial attitude measure and the experimental manipulation. After completing all of the ‘‘surveys,’’ participants were probed for suspicion and debriefed. Premanipulation Attitudes toward the Source Premanipulation attitudes toward Einstein were measured in the same manner as in Experiment 1. Experimental Manipulation Control condition. The experimenter stated that the focus of the third ‘‘survey’’ was tabloid newspapers and magazines (e.g., Globe, National Enquirer). Participants were told that many celebrities have been winning lawsuits against tabloid newspapers, because the tabloids frequently publish negative, untrue stories about celebrities. The experimenter stated that he was interested in this issue because celebrities allege that these stories negatively affect the public’s perceptions of them. He then claimed to be skeptical that the tabloids actually affect people’s perceptions of celebrities. He described a study that he claimed to have conducted in the previous summer, wherein he asked participants to write negative, untrue stories about two famous celebrities. He stated that, later in the year, other participants were asked to read the ‘‘tabloid-like’’ stories and then indicate their perceptions of the two celebrities. He said that there was no effect of these stories on the participants’ perceptions of the celebrities, supporting his theory. Nevertheless, the experimenter pointed out that he should have examined the effects of negative, 194 MAIO AND OLSON untrue stories about a variety of celebrities, rather than just two celebrities. He stated that he might discover effects for some celebrities but not for others, and, consequently, he was asking current participants to write stories about a variety of famous people. He told participants that he had different booklets containing the names of three different famous people and that he was asking each participant to write negative, untrue stories about the people named in one, randomly selected booklet. He then retrieved one of the booklets from another room. The booklet that he retrieved gave space for participants to write stories about John F. Kennedy, Harrison Ford, and Oprah Winfrey (in that order). For each celebrity, participants were given three-quarters of a page to write a negative, untrue story about the celebrity. In the bottom-quarter of each page, there were three six-point (23 to 13) semantic-differential scales, which were anchored by worthless/valuable, unpleasant/pleasant, and stupid/smart. The experimenter stated that he wanted future participants to believe that whoever wrote the story did not like the person described therein and that the story explained why the person was disliked. Participants were asked to indicate unfavorable attitudes using the semantic-differential scales. Attitude dissimulation. The procedure in the attitude dissimulation condition was identical to the procedure in the control condition, except that participants were given a booklet that contained Albert Einstein as the second celebrity, rather than Harrison Ford. As instructed, all participants indicated negative stories about Einstein (e.g., he was a fraud; he was a terrorist). They also rated him unfavorably (M 5 2.88). Dependent Measures Attitudes toward nanotechnology. Attitudes toward nanotechnology were measured using the same three attitude measures as in Experiment 1. As in Experiment 1, the three attitude measures were significantly correlated (lowest r (55) 5 .55, p , .001). Thus, we calculated z-scores for each measure, and these z-scores were averaged to form an overall index of attitudes. Message-relevant thoughts. After indicating their attitudes toward stopping research on nanotechnology, participants completed the same thought-listing task as in Experiment 1. Postmessage attitudes toward the source. Attitudes toward Einstein were measured using the same procedure and scale as in Experiment 1. Confidence, importance, and strength of attitudes toward the source. These measures were identical to those employed in Experiment 1. Results Preliminary Analyses Premanipulation and postmessage attitudes toward the source. Premanipulation and postmessage attitudes toward Einstein were submitted to a 2 3 2 (condition 3 trial) mixed-model ANOVA, with trial (pre- vs postmessage) as the within-subjects factor. As expected, results indicated no significant effects. Thus, the manipulation did not affect attitudes toward Einstein. Overall, attitudes toward Einstein were very positive (M 5 4.86; 5 5 very favorable), with no participants indicating a negative attitude toward him. Confidence, importance, and strength of attitudes toward the source. T-tests on participants’ ratings of attitude confidence, importance, and strength indicated no significant effects. As in Experiment 1, attitudes toward Einstein were characterized by high confidence (M 5 5.23; 4 5 moderately confident), mild to moderate importance (M 5 3.53; 4 5 moderately important), and mild to moderate strength (M 5 3.68; 4 5 moderately strong). ATTITUDE DISSIMULATION 195 TABLE 2 EXPERIMENT 2: ATTITUDES TOWARD NANOTECHNOLOGY AMONG PARTICIPANTS WHO HAD EXPRESSED FALSE ATTITUDES OR NO ATTITUDES TOWARD ALBERT EINSTEIN Attitude measure Attitude dissimulation Control Composite Differential scales Favorability Agreement 0.24 0.34 20.11 0.11 20.22 20.60 20.57 20.60 Note. The table presents the mean raw scores for each attitude measure, except for the composite measure, which produced z-scores. Higher scores indicate greater favorability toward nanotechnology. Attitudes toward Nanotechnology As expected, participants who a wrote negative, untrue story about Albert Einstein were subsequently more favorable toward his recommendation that research on nanotechnology be stopped (M 5 0.24) than were participants who did not write stories about him (M 5 20.22), t (55) 5 1.99, p 5 .05. This result replicated the principal finding in Experiment 1. Message-Relevant Thoughts Participants who wrote negative, untrue stories about Albert Einstein tended to indicate thoughts that were more favorable toward stopping nanotechnology research (M 5 0.09) than did participants who did not write stories about him (M 5 20.40), t (53) 5 1.63, p , .06 (one-tailed). As in Experiment 1, we conducted regression analyses that tested whether the favorability of participants’ thoughts about nanotechnology mediated the effect of attitude dissimulation on attitudes toward nanotechnology. Results indicated that the effects of attitude dissimulation on attitudes, B 5 .26, t (55) 5 1.99, p 5 .05, and on thought favorability, B 5 .22, t (53) 5 1.63, p , .06 (one-tailed) were significant. When attitude dissimulation and thought favorability were entered as simultaneous predictors of attitudes, the effect of thought favorability on attitudes was significant, B 5 .67, t (52) 5 6.60, p , .001, whereas the effect of attitude dissimulation was not significant, B 5 .09, t (52) 5 0.90, ns. Thus, the effect of attitude dissimulation on attitudes may have been mediated by message-relevant thoughts. Yet, when attitude dissimulation and attitudes were entered as simultaneous predictors of thoughts, the effect of attitudes was significant, B 5 .68, t (52) 5 6.60, p , .001, whereas the effect of attitude dissimulation was not significant, B 5 .06, t (52) 5 0.55, ns. Thus, the effect of attitude dissimulation on message-relevant thoughts may have been mediated by attitudes. Put simply, as in Experiment 1, participants’ thoughts appear to have affected their attitudes and vice versa. (See Table 2.) Discussion As in Experiment 1, expressing false evaluations of Albert Einstein did not affect participants’ attitudes toward him. In addition, despite substantial changes 196 MAIO AND OLSON in our procedure for eliciting attitude dissimulation, participants who expressed false attitudes toward Einstein again indicated attitudes and thoughts that were more favorable toward stopping nanotechnology than participants who did not express false attitudes toward him. Also, mediational analyses suggested that the effect of attitude dissimulation on attitudes was mediated by message-relevant thoughts and that the effect of attitude dissimulation on thoughts was mediated by attitudes. These findings replicated the mediational analyses in Experiment 1, lending further support to the hypothesis that the effects on attitudes and thoughts were reciprocally interdependent. Importantly, these results add to the construct validity of our conclusions because they revealed effects of attitude dissimulation using a procedure that was very different from our previous experiment. Specifically, we elicited false attitudes in a manner similar to low choice conditions in past cognitive dissonance experiments, wherein participants were asked to write counterattitudinal essays. These essays were written in a context that did not produce attitude change (low choice and no aversive consequences). Nonetheless, such attitude dissimulation caused participants’ true attitudes to have a larger effect on subsequent judgements than if no dissimulation occurred. Therefore, attitude dissimulation produced interesting consequences even though the circumstances that generate attitude change were absent. It is also important to note that, unlike the first experiment, participants in the attitude dissimulation condition believed that their characterization of Einstein would be taken as their true attitude by someone; therefore, this procedure created a closer analogue to real-world attitude dissimulation. GENERAL DISCUSSION The principal novel finding in both experiments was that falsely evaluating the source of a persuasive message can increase the influence of one’s true attitude on yielding to a message. These results provide evidence for the practical importance of our previous finding that lying about attitudes can make true attitudes more accessible from memory (Maio & Olson, 1995). In addition, it is interesting that attitude dissimulation affected message-relevant thoughts and attitudes simultaneously. These findings suggest that attitude dissimulation caused participants to process the message in a biased manner, which affected their message-relevant thoughts. Participants’ message-relevant thoughts and attitudes may then have reciprocally affected one another (see Smith, 1982, for evidence of similar reciprocal effects). We should also note that supplementary analyses revealed that only 5.39% of the total number of thoughts listed by participants referred to Albert Einstein. Thus, Einstein did not appear to be a salient aspect of the article on nanotechnology. In fact, during the debriefing, many participants indicated that they felt that their attitudes toward Einstein did not influence their agreement with his point of view. This belief was inconsistent with the finding that the expression of false or true attitudes toward Einstein affected attitudes toward banning nanotechnology. ATTITUDE DISSIMULATION 197 Thus, it appears that participants may have been unaware of the effect of their attitudes toward Einstein on their attitudes toward banning nanotechnology. This lack of awareness is consistent with what would be expected if attitudes toward Einstein were being spontaneously activated from memory during perusal of the persuasive message. After spontaneous activation from memory, attitudes can bias perception of attitude-relevant information without any conscious awareness of the attitude (Fazio, 1990). In contrast, if attitudes are deliberately accessed from memory, people should be more cognizant of their attitude. Given that participants’ thoughts rarely referred to Einstein, it seems unlikely that participants in our experiments deliberately accessed their attitudes toward Einstein as a means of calculating their attitudes toward banning nanotechnology. Instead, attitudes toward Einstein may have been activated spontaneously by the citation of him in the persuasive message, particularly among participants who expressed false (or true) attitudes toward him. To investigate this hypothesized process more closely, it would be necessary to test whether attitude dissimulation results in the spontaneous activation of true attitudes from memory and whether this spontaneous activation mediates the effects of attitude dissimulation. In the present experiments, it would have been ideal to measure the activation of attitudes toward Einstein prior to message exposure, but we did not do so because we were concerned that such a measure might indirectly activate participants’ attitudes toward Einstein prior to presentation of the persuasive message, causing these attitudes to be accessible in all of the experimental conditions. We hope in future research to include a direct assessment of the mediating role of attitude accessibility. In addition to raising such theoretical issues, the counterintuitive effect of attitude dissimulation has interesting implications for understanding some realworld behavior. These implications are particularly evident in situations where social norms put pressure on people to lie. In such situations, people might feel no choice but to lie and think that their dissimulation helps them to avoid unnecessary social friction. Thus, the prerequisites for attitude dissimulation to cause attitude change, namely freedom of choice and aversive consequences (see Cooper & Fazio, 1984), might not be present. An example of such a situation could be a party, where topics like politics are often discussed. During such conversations, one individual might express strong dislike for a particular political candidate. Politeness norms might then cause another individual, who privately admires the candidate, to echo a few negative things about him or her, while being aware of the numerous positive things that could be said. Consequently, the person who said a few negative things about the politician might become more aware of his or her liking for the politician. Our findings suggest that one consequence of this increased awareness could be increased agreement with the politician’s subsequent speeches (assuming that the increased accessibility of the original attitude endured for some period of time—an assumption not tested in our experiments). The above example represents a real-world instance where attitude dissimula- 198 MAIO AND OLSON tion might increase the impact of true attitudes. We expect that more serious consequences of attitude dissimulation could occur in other contexts. For example, so-called ‘‘political correctness’’ pressures may cause some people who strongly dislike particular minority groups to falsely express positive attitudes toward the groups. Such attitude dissimulation would be designed to minimize aversive consequences for the prejudiced individual, which means that it should not induce attitude change. Instead, such dissimulation might make the true, negative attitudes more accessible than if no attitude had been expressed. As a result, the prejudiced person might be more likely to discriminate against members of the minority groups, when they can do so anonymously or with impunity, relative to having not expressed positive attitudes. Interestingly, recent research has suggested similar paradoxical effects of stereotype suppression. That is, when people are asked to not think about particular negative stereotypes, these stereotypes tend to become more accessible from memory and exert stronger effects on subsequent judgments and behavior (e.g., Macrae, Bodenhausen, Milne, & Jetten, 1994). In general, trying to suppress a thought tends to make it easier to retrieve from memory (e.g., Wegner, 1994). It has been suggested that this effect occurs because of an ‘‘ironic’’ monitoring process, wherein people must be vigilant against appearances of the suppressed stereotype in consciousness. Paradoxically, such a process requires repeated awareness of the stereotype in order to suppress it. Thus, the stereotype is made more accessible. In contrast, we expect that attitude dissimulation does not involve a vigilant search process. Once it is decided that it is desirable to express a particular attitude, people might express that attitude without worrying that their true attitude will enter their awareness. That is, the true attitude might be primed only in the first moments of attitude dissimulation, where it is automatically activated from memory. On occasion, various psychological processes might extend the duration of this priming, but these processes might not involve a vigilant search process. For example, pressures to dissimulate might occasionally elicit psychological reactance (Brehm, 1966), which would cause recurrent contemplation of one’s attitude. Future research could examine the extent to which suppression and dissimulation have similar underlying mechanisms and effects. Caveats and Conclusions Our findings suggest that attitude dissimulation might produce the same effects as truthful attitude expression, but there are some important caveats to this conclusion. One important point is that attitude dissimulation can occur in a variety of contexts, including face-to-face interaction, telephone conversations, and written messages. Thus, the effects of dissimulation and expression need to be compared across these contexts. We included a truthful attitude expression condition only in Experiment 1 and not in Experiment 2, because our focus was on the effects of attitude dissimulation. Future research comparing the effects of attitude dissimulation and attitude expression across a variety of contexts might ATTITUDE DISSIMULATION 199 yield more information about the processes underlying the effects of these variables. In addition, we suspect that there are many conditions under which attitude dissimulation and attitude expression will produce dissimilar effects. First, dissimilar effects could occur when attitude dissimulation produces attitude change. Previous research has found that attitude change can result when people decide to tailor their attitude expressions to suit their audience (e.g., Higgins & Rholes, 1978; McCann, Higgins, & Fondacaro, 1991). Specifically, people’s true attitudes become similar in valence to the previously false attitudes. When this attitude change occurs, subsequent attitude-relevant judgments might become consistent with the changed attitudes, rather than with the previously held attitudes. (Such effects might be preempted if people initially hold strong, unambiguous attitudes, which are less susceptible to change.) Second, the frequency with which dissimulation has occurred might be an important factor. Perhaps false attitudes become strongly associated in memory with an attitude object after frequent dissimulations over an extended period of time. After a large number of dissimulations, the strength of the association between the attitude object and the false attitude might be strong enough to override the impact of the true attitude on judgments. Our experiments did not elicit repeated dissimulation over periods of days or weeks. This possibility could be examined in future research. In sum, our findings indicate that the effects of lying on subsequent judgments and behavior may be more complex than has been documented previously. Social psychologists have focused on the implications of lying for cognitive dissonance and self-perception processes. Our studies show that lying has diverse possible consequences. Given the ubiquity of deception in social behavior, it is important that these consequences be understood. REFERENCES Arkes, H. R., Boehm, L. E., & Xu, G. (1991). Determinants of judged validity. Journal of Experimental Social Psychology, 27, 576–605. Bassili, J. N. (1996). Meta-judgmental versus operative indices of psychological attributes: The case of measures of attitude strength. Journal of Personality and Social Psychology, 71, 637–653. Bell, D. W., Esses, V. M., & Maio, G. R. (1996). The utility of open-ended measures to assess intergroup ambivalence. Canadian Journal of Behavioural Sciences, 28, 12–18. Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 6, pp. 1–62). San Diego: Academic Press. Brehm, J. W. (1966). A theory of psychological reactance. San Diego: Academic Press. Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1997). Beyond bipolar conceptualizations and measures: The case of attitudes and evaluative space. Personality and Social Psychology Review, 1, 3–25. Cacioppo, J. T., & Petty, R. E. (1981). Social psychological procedures for cognitive response assessment: The thought-listing technique. In T. V. Merluzzi, C. R. Glass, & M. Genest, Cognitive assessment (pp. 309–342). New York: Guilford. Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39, 752–766. 200 MAIO AND OLSON Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and systematic processing within and beyond the persuasion context. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. 212–252). New York: Guilford. Cooper, J., & Fazio, R. H. (1984). A new look at dissonance theory. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 17, pp. 229–266). San Diego: Academic Press. Cooper, J., & Scher, S. J. (1994). When do our actions affect our attitudes? In S. Shavitt & T. Brock (Eds.), Persuasion: Psychological insights and perspectives (pp. 95–111). Boston: Allyn & Bacon. DePaulo, B. M., Stone, J. I., & Lassiter, G. D. (1985). Deceiving and detecting deceit. In B. R. Schlenker (Ed.), The self and social life (pp. 323–370). New York: McGraw-Hill. Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Hillsdale, NJ: Erlbaum. Esses, V. M., Haddock, G., & Zanna, M. P. (1993). Values, stereotypes, and emotions as determinants of intergroup attitudes. In D. M. Mackie & D. L. Hamilton (Eds.), Affect, cognition, and stereotyping: Interactive processes in group perception (pp. 137–166). San Diego: Academic Press. Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 23, pp. 75–109). New York: Academic Press. Festinger, L. (1957). A theory of cognitive dissonance. Evanston, Il: Row, Peterson. Gilbert, D. T., Krull, D. S., & Malone, P. S. (1990). Unbelieving the unbelievable: Some problems in the rejection of false information. Journal of Personality and Social Psychology, 59, 601–613. Harmon-Jones, E., Brehm, J. W., Greenberg, J., Simon, L., & Nelson, D. E. (1996). Evidence that the production of aversive consequences is not necessary to create cognitive dissonance. Journal of Personality and Social Psychology, 70, 5–16. Higgins, E. T., & Rholes, W. J. (1978). ‘‘Saying is believing’’: Effects of message modification on memory and liking for the person described. Journal of Experimental Social Psychology, 14, 363–378. Houston, D. A., & Fazio, R. H. (1989). Biased processing as a function of attitude accessibility: Making objective judgments subjectively. Social Cognition, 7, 51–66. Judd, C. M., & Brauer, M. (1995). Repetition and evaluative extremity. In R. E. Petty & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 43–71). Mahwah, NJ: Erlbaum. Krosnick, J. A., Boninger, D. S., Chuang, Y. C., Berent, M. K., & Carnot, C. G. (1993). Attitude strength: One construct or many related constructs? Journal of Personality and Social Psychology, 65, 1132–1151. Macrae, C. N., Bodenhausen, G. V., Milne, A. B., & Jetten, J. (1994). Out of mind but back in sight: Stereotypes on the rebound. Journal of Personality and Social Psychology, 67, 808–817. Maio, G. R., Bell, D. W., & Esses, V. M. (1996). Ambivalence and persuasion: The processing of persuasive messages about immigrant groups. Journal of Experimental Social Psychology, 32, 513–536. Maio, G. R., & Olson, J. M. (1995). The effect of attitude dissimulation on attitude accessibility. Social Cognition, 13, 127–144. McCann, C. D., Higgins, E. T., & Fondacaro, R. A. (1991). Primacy and recency in communication and self-persuasion: How successive audiences and multiple encodings influence subsequent evaluative judgments. Social Cognition, 9, 47–66. Millar, M. G., & Millar, K. U. (1990). Attitude change as a function of attitude type and argument type. Journal of Personality and Social Psychology, 59, 217–228. Olson, J. M. (1988). Misattribution, preparatory information, and speech anxiety. Journal of Personality and Social Psychology, 54, 758–767. Olson, J. M. (1990). Self-inference processes in emotion. In J. M. Olson & M. P. Zanna (Eds.), Self-inference processes: The Ontario symposium (Vol. 6, pp. 17–41). Mahwah, NJ: Erlbaum. Olson, J. M. (1992). Self-perception of humor. Evidence for discounting and augmentation effects. Journal of Personality and Social Psychology, 62, 369–377. Olson, J. M., & Roese, N. J. (1995). The perceived funniness of humorous stimuli. Personality and Social Psychology Bulletin, 21, 908–913. ATTITUDE DISSIMULATION 201 Olson, J. M., & Ross, M. (1988). False feedback about placebo effectiveness: Consequences for the misattribution of speech anxiety. Journal of Experimental Social Psychology, 24, 275–291. Olson, J. M., & Zanna, M. P. (1993). Attitudes and attitude change. Annual Review of Psychology, 44, 117–154. 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. 123–205). San Diego: Academic Press. Petty, R. E., Schumann, D. W., Richman, S. A., & Strathman, A. J. (1993). Positive mood and persuasion: Different roles for affect under high- and low-elaboration conditions. Journal of Personality and Social Psychology, 64, 5–20. Roskos-Ewoldsen, D. R., & Fazio, R. H. (1992). The accessibility of source likeability as a determinant of persuasion. Personality and Social Psychology Bulletin, 18(1), 19–25. Scher, S. J., & Cooper, J. (1989). Motivational basis of dissonance: The singular role of behavioral consequences. Journal of Personality and Social Psychology, 56, 899–906. Smith, E. R. (1982). Beliefs, attributions, and evaluations: Nonhierarchical models of mediation in social cognition. Journal of Personality and Social Psychology, 43, 248–259. Steele, C. M., & Liu, T. J. (1983). Dissonance processes as self-affirmation. Journal of Personality and Social Psychology, 45, 5–19. Wegner, D. M. (1994). Ironic processes of mental control. Psychological Review, 101, 34–52. Zuckerman, M., & Driver, R. E. (1985). Telling lies: Verbal and nonverbal correlates of deception. In A. W. Siegman & J. S. Feldstein (Eds.), Multichannel integrations of nonverbal behavior (pp. 129–147). Mahwah, NJ: Erlbaum.