Transcription
822697Catapano et al.Perspective Taking and Self-PersuasionASSOCIATION FORPSYCHOLOGICAL SCIENCEResearch ArticlePerspective Taking and Self-Persuasion:Why “Putting Yourself in Their Shoes”Reduces Openness to Attitude ChangePsychological Science2019, Vol. 30(3) 424 –435 The Author(s) 2019Article reuse guidelines:sagepub.com/journals-permissionsDOI: 56797618822697www.psychologicalscience.org/PSRhia Catapano1 , Zakary L. Tormala1, andDerek D. Rucker21Stanford University, Graduate School of Business, and 2Northwestern University, KelloggSchool of ManagementAbstractCounterattitudinal-argument generation is a powerful tool for opening people up to alternative views. On the basisof decades of research, it should be especially effective when people adopt the perspective of individuals who holdalternative views. In the current research, however, we found the opposite: In three preregistered experiments (total N 2,734), we found that taking the perspective of someone who endorses a counterattitudinal view lowers receptivenessto that view and reduces attitude change following a counterattitudinal-argument-generation task. This ironic effectcan be understood through value congruence: Individuals who take the opposition’s perspective generate argumentsthat are incongruent with their own values, which diminishes receptiveness and attitude change. Thus, trying to “putyourself in their shoes” can ultimately undermine self-persuasion. Consistent with a value-congruence account, thisbackfire effect is attenuated when people take the perspective of someone who holds the counterattitudinal view yethas similar overall values.Keywordsattitude change, persuasion, perspective taking, receptiveness, resistance, open data, open materials, preregisteredReceived 11/20/17; Revision accepted 10/18/18People are more divided, and less open to opposingviews, than they have been in years. For example,Democrats and Republicans are further apart ideologically than at any point in the past two decades (PewResearch Center, 2017). This problem is exacerbated byselective exposure—individuals predominantly attendto information that supports their existing attitudes,beliefs, and values (Eagly, Kulesa, Chen, & Chaiken,2001; Hart et al., 2009; Iyengar & Hahn, 2009; Smith,Fabrigar, & Norris, 2008)—and by ideological silos, inwhich individuals gravitate toward friends with attitudesthat match their own (Pew Research Center, 2017). People increasingly live in echo chambers that reinforce,rather than question, their views (Quattrociocchi, Scala,& Sunstein, 2016).How can we change attitudes on controversial issuesor open people up to alternative viewpoints? Oneapproach is to leverage insights from research on selfpersuasion. Broadly speaking, self-persuasion occurswhen people reflect on a topic and change their attitudes without outside input (Petty, Wheeler, & Tormala,2003). Perhaps the most studied self-persuasion methodinvolves asking people to consider or generate arguments that contradict their own attitudes and beliefs.An extensive literature suggests that generating counterattitudinal arguments can shift people’s attitudes inthe counterattitudinal direction (e.g., Briñol, McCaslin,& Petty, 2012; Carlsmith, Collins, & Helmreich, 1966;Cialdini, 1971; Greenwald & Albert, 1968). For example,role-playing the opposition—even merely brainstormingarguments in anticipation of role playing—can causeattitude change in the direction of the opposing view(Cialdini & Petty, 1981; Janis & King, 1954). Likewise,Corresponding Author:Rhia Catapano, Stanford University, Graduate School of Business, 655Knight Way, Stanford, CA 94305E-mail: [email protected]
Perspective Taking and Self-Persuasionencouraging people to “consider the opposite” or “consider alternatives” can reduce bias in social and cognitive judgments (e.g., Adame, 2016; Hirt & Markman,1995; Lord, Lepper, & Preston, 1984; Mussweiler, Strack,& Pfeiffer, 2000). Thus, inducing people to generatearguments against their views can soften those viewsand trigger shifts toward more moderate positions.Perspective Taking as a Catalyst forSelf-PersuasionGiven extant research, it is reasonable to expect theeffect of generating counterattitudinal arguments to beeven greater when accompanied by perspective taking—that is, when people are instructed to imagine thethoughts and feelings of individuals who would makethose arguments. After all, if people put themselves inothers’ shoes, they might generate more persuasivearguments, or relate more to opposing positions, orboth. Consistent with this proposition, previous studieshave shown perspective taking to have positive consequences for interpersonal outcomes. For instance, perspective taking promotes self-other merging andcloseness with others (Davis, Conklin, Smith, & Luce,1996), generous attributions for others’ negative outcomes (Gould & Sigall, 1977), and helpful behaviortoward individuals in distress (Toi & Batson, 1982).Perspective taking can also decrease stereotypic biasesand enhance out-group evaluations (Galinsky &Moskowitz, 2000). If perspective taking enhances perceptions of others—including those with divergentviews—it could open people up to opposing opinionsand strengthen the effect of counterattitudinal-argumentgeneration on attitude change.Indeed, initial work suggests that perspective takingmight moderate people’s opinions on controversialissues. Tuller, Bryan, Heyman, and Christenfeld (2015)had participants generate proattitudinal arguments fromtheir own perspective or counterattitudinal argumentsfrom the perspective of someone from a disagreeinggroup. Participants showed more attitude change whenthey wrote from the perspective of the other person.However, because perspective takers were the onlyparticipants in the study to generate counterattitudinalarguments, it is unclear whether perspective taking orthe counterattitudinal-argument task itself was responsible for the effect. Nevertheless, Tuller and colleaguesuncovered potential support for the idea that perspective taking facilitates attitude change in a self-persuasionparadigm. This possibility also has face validity. Weasked 350 people on Amazon’s Mechanical Turk whichwould be a more effective way to change someone’sattitude: to have them generate arguments for theopposing view or take the perspective of someone on425the opposing side and generate arguments from theirviewpoint. Sixty-seven percent indicated that perspective taking would be the better strategy.Perspective Taking as an Inhibitor ofSelf-PersuasionDiverging from the notion that perspective takingboosts openness to change in a self-persuasion paradigm, we submit that it ironically inhibits the effect ofcounterattitudinal-argument generation and undermines feelings of receptiveness and openness tochange. Consistent with this hypothesis, an emergingliterature cautions that perspective taking can backfire,causing selfishness (Epley, Caruso, & Bazerman, 2006)and negative intergroup perceptions (Galinsky, Wang,& Ku, 2008; Ku, Wang, & Galinsky, 2015; Skorinko &Sinclair, 2013; Tarrant, Calitri, & Weston, 2012; Vorauer& Sasaki, 2009). Such effects are particularly likelywhen groups are perceived as especially divided or incompetition. For example, recent work suggests that incompetitive contexts, perspective taking intensifiesrather than attenuates competitive feelings and actions.Specifically, Pierce, Kilduff, Galinsky, and Sivanathan(2013) found that perspective taking fosters unethicalbehavior in negotiations by promoting feelings of competitive threat. Likewise, when groups are divided byseemingly unbridgeable gaps, perspective takingappears to push people apart rather than bring themtogether (Sassenrath, Hodges, & Pfattheicher, 2016). Inshort, although considerable research attests to thepositive consequences of perspective taking, there isreason to suspect that it can reduce, rather thanincrease, openness to opposing views.The current research examined whether perspectivetaking inhibits openness to attitude change in the classic self-persuasion paradigm. We hypothesized thatwhen people take the perspective of someone whoendorses a counterattitudinal view and generate arguments for that view, they become less receptive andshow reduced attitude change compared with whenthey generate counterattitudinal arguments without perspective taking. This effect, we argue, stems from valuecongruence—the extent to which arguments generatedreflect values that align with, versus differ from, one’sown. In essence, because perspective taking leads people to put themselves in the opposition’s shoes, it mightlead them to generate arguments that are less alignedwith their own values and more aligned with the valuesof the opposition, which undermines persuasion (e.g.,Feinberg & Willer, 2015). This logic resonates with thegeneral proposition of Pierce and colleagues (2013; seealso Sassenrath et al., 2016), but in this case, perspectivetaking fosters value-incongruent thinking. Ironically,
Catapano et al.426then, rather than facilitating self-persuasion, perspectivetaking may inhibit it. We tested this hypothesis, and thevalue-congruence mechanism, in three preregisteredexperiments.Experiment 1In Experiment 1, participants generated arguments fora counterattitudinal view. Some were instructed to takethe perspective of an individual who holds this view;others simply generated arguments. As noted, pastresearch suggests that generating counterattitudinalarguments can shift attitudes in the counterattitudinaldirection. Our interest was in whether perspective taking enhances or undermines this effect. Thus, wetreated counterattitudinal-argument generation (withoutperspective taking) as a control condition. On the basisof our value-congruence hypothesis, we expected participants who generated arguments from the counterattitudinal perspective to show less receptiveness to theopposing view and less attitude change than controlparticipants would. Moreover, we expected these effectsto be mediated by value congruence: Perspective takerswould generate arguments that were less congruentwith their own values, thus undermining their receptiveness and rendering an attitude change less likely.Our hypotheses and analytic strategy were preregistered on the Open Science Framework (https://osf.io/juah5/).Experiment 1 tested these hypotheses using an issuethat people cared about and a context in which peopletook the perspective of an ostensibly real individualwhose attitude differed from their own. Specifically,participants read about a current political issue—universal health care—and were led to believe they woulddiscuss this issue with another individual. To ensurethat our sample would be interested in universal healthcare, we focused recruitment on politically orientedindividuals from the website reddit.com. Reddit is anonline discussion forum where users interact with specific communities around particular topics, includingpolitics, specific political issues, and politicians. Consequently, Reddit enabled us to recruit people whocared about politics and were already active in discussing politics online.Participants were informed that they would discussuniversal health care with another Reddit user, and theythen received background information about this user.The goal was to give participants the impression of areal, multidimensional target person by presenting theperson’s political ideology and attitude toward universalhealth care in the context of other personal information.To increase the psychological realism of the experiment, we wanted participants to believe they wouldactually interact with the target person. In reality, allparticipants received the same information about thetarget (except for political ideology and attitude, whichwere always opposite to participants’), and the interaction never occurred.MethodOn the basis of a pilot test indicating a Cohen’s d of0.175, we set a target sample size of 1,000 participants,yielding 75% power to detect a small effect size afterallowing for 100 exclusions (10%). However, becauseof the unpredictable nature of recruitment on Reddit,we also preregistered an end date at which data collection would be stopped even if the target were notmet. On the end date, 568 Reddit users had completedthe survey. Following our preregistration plan, weexcluded 84 participants (14.8%) who failed the attention check or did not follow instructions (final N 484,44.4% female; mean age 30.57).At the outset, participants were told that we wereconducting a preliminary survey for an event duringwhich they would interact with another Reddit user ona political topic. After providing demographic information, including age, gender, ethnicity, home state, andpolitical ideology, participants received a descriptionof the target issue (universal health care) and reportedtheir attitude toward it on a scale from 0 (stronglyagainst) to 100 (strongly support). 1 Participants thenreceived information about their interaction partner.Participants were given their partner’s Reddit username, age (22), gender (male), ethnicity (white), homestate (Ohio), political ideology (liberal or conservative),and attitude toward universal health care. All participants were told that the target was of the oppositepolitical ideology and held the opposite opinion ofuniversal health care from theirs. For instance, if a participant were liberal and in favor of universal healthcare, he or she was informed that the interaction partner was conservative and against universal health care.As an attention check, participants were asked questions about their partner’s demographics.Next, participants were told they would complete ashort task to prepare for the event. Here, each participant was assigned to one of two conditions. In theperspective-taking condition, participants were instructedto take the perspective of their interaction partner:Before you write an argument, please take amoment to consider the perspective of thisindividual, reflecting on their intentions andinterests. Try to visualize clearly and vividly whatthis individual’s life and experiences may be like,and how they may feel.
Perspective Taking and Self-PersuasionThese instructions were adapted from past research(Clore & Jeffery, 1972; Galinsky & Moskowitz, 2000).Participants then generated an argument that the interaction partner might give for his position. In the controlcondition, participants generated an argument for thecounterattitudinal position, without perspective-takinginstructions. To reduce self-presentation concerns, wetold participants in both conditions that the activity wasfor their own preparation and would not be shown tothe other participant. After generating the argument, allparticipants indicated their attitude using the same scaleas before.Finally, we measured receptiveness toward the counterattitudinal position and value congruence. Receptiveness was measured using four items assessingparticipants’ receptiveness and openness toward thespecific argument they wrote down and the counterattitudinal position in general. Responses, provided onscales ranging from 1 (not at all) to 7 (extremely), wereaveraged (α .88). Value congruence was measuredwith three items assessing participants’ perceptions thatthe argument they wrote down was congruent withtheir own values, fit their worldview, and reflected theirpersonal morals (α .92). Responses were provided onscales ranging from 1 (not at all) to 7 (extremely).Finally, to gauge whether we had selected an issue thatReddit users cared about, we asked participants to indicate the importance of universal health care (1 extremely unimportant, 7 extremely important). Seethe Open Science Framework for complete stimuli(https://osf.io/heun6).ResultsFirst, participants indicated that universal health carewas an important issue (M 6.00, 95% confidence interval, or CI [5.90, 6.11]). Moreover, initial attitudes didnot significantly differ between the control (M 81.48,95% CI [77.93, 85.03]) and perspective-taking (M 81.61, 95% CI [77.82, 85.38]) conditions, t(482) 0.05,p .96. Attitude change was computed as the differencebetween Time 1 (preargument) and Time 2 (postargument) attitudes. Change was coded in the direction ofthe generated argument; positive numbers indicatedmovement toward the counterattitudinal position, andnegative numbers indicated movement away from thecounterattitudinal position.We found a significant effect of the manipulationon receptiveness and attitude change. Participantsreported less receptiveness to the counterattitudinalview in the perspective-taking (M 3.27, 95% CI [3.11, 3.43]) relative to the control (M 3.53, 95%CI [3.36, 3.69]) condition, t(482) 2.15, p .032,d 0.20. Participants also showed less attitude change427in the perspective-taking (M 0.39, 95% CI [ 0.17,0.94]) relative to the control (M 1.52, 95% CI [0.55,2.50]) condition, t(482) 1.98, p .048, d 0.18.Moreover, as indicated by the 95% CIs, attitude changewas greater than zero in the control but not the perspective-taking condition.To determine whether these effects were driven byvalue congruence, we conducted mediation analysesusing bootstrapping (Preacher & Hayes, 2008). Theseanalyses indicated that condition affected value congruence, β 0.21, t(482) 2.30, p .02, and valuecongruence predicted both receptiveness, β 0.49,t(482) 12.06, p .001, and attitude change, β 0.18,t(482) 4.09, p .001. As hypothesized, the indirecteffects (the ab paths) of condition on receptiveness andattitude change were mediated by value congruence—receptiveness: indirect effect 0.10, 95% CI [ 0.19, 0.02], p .02; attitude change: indirect effect 0.04,95% CI [ 0.09, 0.00], p .02. See the Open ScienceFramework (https://osf.io/heun6/) for discussion ofalternative mediation models.DiscussionExperiment 1 revealed that perspective taking reducedopenness to opposing views in a self-persuasion paradigm and provided initial support for a value-congruence account. That we observed this effect in a sampleof politically oriented Reddit users, who generatedarguments on a personally important topic and believedthat they would interact with the person whose perspective they took, suggests that this finding has ecological validity. However, conducting an experimentwith Reddit users also had a downside: We were unableto recruit the target number of participants (N 1,000)within the prespecified time frame. Thus, we present aconceptual replication of Experiment 1 (N 998 afterexclusions) using participants from a Qualtrics panel.This experiment, detailed in the Supplemental Materialavailable online, fully replicated the results of Experiment 1.Experiment 2In Experiment 2, we sought to replicate the findings ofExperiment 1 and provide further evidence for thevalue-congruence account with a moderation approach.Specifically, if the backfire effect of perspective takingis driven by the perception that the arguments generated are incongruent with one’s personal values, thiseffect should be attenuated when people take the perspective of someone who disagrees on the target issueyet has similar values. To investigate this possibility, weincluded two perspective-taking conditions in
Catapano et al.428Experiment 2, varying whether participants took theperspective of someone with a similar or different political ideology. Unlike in Experiment 1, participants inthese conditions received no information about theperspective-taking target beyond his or her attitude andpolitical ideology. Our hypotheses and analytic strategywere preregistered on the Open Science Framework(https://osf.io/p5fta).MethodAll participants began by completing a demographicquestionnaire that included a political-ideology measure (1 extremely liberal, 7 extremely conservative).Because our manipulation required that participantsself-identified as either liberal or conservative, participants who responded using the midpoint of the scale(neither) were disqualified. Sample size was determinedon the basis of the same Cohen’s d as in Experiment 1(0.175), suggesting approximately 1,350 participantsacross three conditions for 75% power. To allow forexclusions, we collected data from 1,411 participantson Amazon’s Mechanical Turk. Following our preregistration plan, we excluded 159 participants (11.3%)who failed the attention check, completed the surveymultiple times, or did not follow instructions (final N 1,252; 58.1% female; mean age 35.3).Participants in Experiment 2 were introduced to theissue of universal basic income and reported their attitudes toward it on a scale from 0 (strongly disagree/against universal basic income) to 100 (strongly agree/in favor of universal basic income). Each participantwas then randomly assigned to one of three conditions.In the perspective-taking/different-ideology condition,participants generated an argument that a person withthe opposite political ideology would give for the counterattitudinal viewpoint. For example, liberals withfavorable attitudes were instructed as follows: “Manyconservatives in America are against a universal basicincome. What is one argument that a conservative whois against a universal basic income system in Americamight give (i.e., against universal basic income)?” Participants in this condition also received the perspective-takinginstructions from Experiment 1. In the perspective-taking/same-ideology condition, participants generated anargument that someone who shared their political ideology would give for the counterattitudinal viewpoint.Here, liberals in favor of universal basic income generated arguments from the perspective of a liberal whoopposes universal basic income. The goal was to haveparticipants generate a counterattitudinal argumentfrom the perspective of someone who disagreed with themon the issue but shared their overall values. Participantsin this condition also received the perspective-takinginstructions. Participants in the control condition generated a counterattitudinal argument without perspectivetaking.Finally, participants reported attitudes on the samescale as before and answered the receptiveness (α .86) and value-congruence (α .93) questions fromExperiment 1. For exploratory analysis, we also askedparticipants to complete two items assessing the extentto which they believed liberals and conservatives havesimilar values to theirs (1 not at all, 7 very much).Primary resultsThere were no differences in initial attitudes, F(2, 1249) 1.05, p .35, d 0.08 (see Table 1). We submittedreceptiveness, attitude change, and value congruenceto one-way analyses of variance (ANOVAs) with thethree-level manipulation as a between-participants factor, and we observed significant effects on each (seeTable 1). First, the manipulation affected receptiveness,F(2, 1249) 18.36, p .001, d 0.35. Participants inthe control and perspective-taking/same-ideology condition were more receptive than participants in theperspective-taking/different-ideology condition, F(1,1249) 33.78, p .001, d 0.35. Participants weremarginally more receptive in the perspective-taking/same-ideology condition than in the control condition,F(1, 1249) 3.27, p .07, d 0.13. Likewise, we founda significant effect on attitude change, F(2, 1249) 27.47, p .001, d 0.42. Participants in the control andperspective-taking/same-ideology conditions changedmore than participants in the perspective-taking/different-ideology condition, F(1, 1249) 41.47, p .001, d 0.38. In addition, change was greater in theperspective-taking/same-ideology condition than in thecontrol condition, F(1, 1249) 14.21, p .001, d 0.24.Viewed differently, relative to the control condition,generating counterattitudinal arguments from the perspective of someone with different values underminedreceptiveness, t(1249) 4.17, p .001, d 0.28, andattitude change, t(1249) 3.73, p .001, d 0.28,whereas generating counterattitudinal arguments fromthe perspective of someone with similar valuesincreased receptiveness, t(1249) 1.81, p .07, d 0.13, and attitude change, t(1249) 3.78, p .001, d 0.24.These results suggest that value congruence drivesthe inhibitory effect of perspective taking on receptiveness and attitude change in the self-persuasion paradigm. Providing additional evidence, results showedthat the manipulation affected the value-congruenceindex, F(2, 1249) 34.78, p .001, d 0.47. Participantsin the control and perspective-taking/same-ideologyconditions perceived their arguments as more value
Perspective Taking and Self-Persuasion429Table 1. Mean Scores on Each of the Dependent Measures in Experiment 2ConditionVariableInitial attitude(100-point scale)Receptiveness(7-point scale)Attitude change(100-point scale)Value congruence(7-point scale)Control(n 431)Perspective-taking/same-ideology(n 407)Perspective-taking/different-ideology(n 414)53.26a[49.97, 56.56]3.93a†[3.80, 4.07]5.51a[4.06, 6.96]3.83a[3.66, 3.99]56.73a[53.32, 60.15]4.11a†[3.98, 4.24]9.37b[7.69, 11.04]3.87a[3.70, 4.02]55.42a[51.99, 58.84]3.53b[3.39, 3.67]1.72c[0.63, 2.80]3.00b[2.84, 3.16]Note: Values in brackets are 95% confidence intervals. Within a row, means with differentsubscripts differ significantly (p .05); means with the same subscript and a dagger differmarginally significantly (p .10).congruent than did participants in the perspective-taking/different-ideology condition, F(1, 1249) 69.56,p .001, d 0.50. In this case, the control and perspective-taking/same-ideology conditions did not differfrom each other, F(1, 1249) 0.10, p .75, d 0.02,suggesting that other factors might account for theunexpected boost in receptiveness and attitude changein the perspective-taking/same-ideology condition.Mediation analysis followed recommendations fortesting process with multicategorical variables (Hayes& Preacher, 2014). Value congruence mediateddecreased receptiveness in the perspective-taking/different-ideology condition relative to the other conditions (indirect effect 0.06, 95% CI [0.04, 0.08], p .001), but not increased receptiveness in the perspective-taking/same-ideology condition relative to the control condition (indirect effect 0.004, 95% CI [ 0.02,0.03], p .73). Similarly, value congruence mediateddecreased attitude change in the perspective-taking/different-ideology condition relative to the other conditions (indirect effect 0.03, 95% CI [0.01, 0.04], p .001), but not increased attitude change in the perspective-taking/same-ideology condition relative to the control condition (indirect effect 0.002, 95% CI [ 0.009,0.01], p .74).Exploratory analysesOur value-congruence account suggests that the receptiveness and attitude-change results might be especiallylikely to emerge when participants view themselves asmore similar to the in-group and different from theout-group. Thus, we conducted exploratory analysesusing participants’ self-reported similarity to liberalsand conservatives. Only participants who completedboth items were included in this analysis.Not surprisingly, liberals perceived other liberals asmore similar to themselves (M 5.30) than conservatives did (M 2.84), t(1246) 33.55, p .001, d 1.99,and conservatives perceived other conservatives asmore similar to themselves (M 5.28) than liberals did(M 2.58), t(1246) 36.15, p .001, d 2.14. Wecomputed the distance (i.e., difference) between participants’ perceptions of similarity to liberals and conservatives and coded scores such that higher valuesindicated greater similarity to the in-group and greaterdissimilarity to the out-group. For example, liberalswho viewed themselves as highly similar to other liberals and dissimilar to conservatives would have high(positive) distance scores, whereas liberals who viewedthemselves as similar to both groups (or as more similarto conservatives) would have low (or negative) distancescores.We surmised that the effect of condition on receptiveness and attitude change would be amplified whenparticipants had high distance scores (i.e., when theyperceived themselves as more similar to their in-groupand less similar to their out-group) and attenuatedwhen they had low distance scores (i.e., when theyperceived themselves as less similar to their in-groupand more similar to their out-group). Indeed, we foundsignificant interactions between condition (control andperspective-taking/same-ideology conditions versusperspective-taking/different-ideology condition) andin-group/out-group distance on both attitude change,β 0.04, t(1246) 2.28, p .02, and receptiveness,β 0.10, t(1246) 4.99, p .001. In each case the inhibitory effect of the perspective-taking/different-ideologycondition relative to the other conditions increased asdistance grew—that is, as participants saw themselvesas increasingly similar to their in-group, different fromtheir out-group, or both (see Figs. 1 and 2).
Catapano et al.430Control ConditionPerspective-Taking/Different-Ideology ConditionPerspective-Taking/Same-Ideology ConditionReceptiveness543–6–4Similar to Out-Group/Dissimilar to In-Group0–2246Similar to In-Group/Dissimilar to Out-GroupDistanceMagnitude of Receptiveness Backfire(Difference Between Regression ilar to Out-Group/Dissimilar to In-Group–20246Similar to In-Group/Dissimilar to Out-GroupDistanceFig. 1. Receptiveness results from Experiment 2. The top panel shows receptiveness as afunction of in-group/out-group distance and argument-generation condition. The dotted linesindicate the Johnson-Neyman numbers for the contrast comparing the perspective-taking/different-ideology condition with the perspective-taking/same-ideology condition and thecontrol condition. Shaded regions indicat
tered on the Open Science Framework (https://osf.io/ juah5/). Experiment 1 tested these hypotheses using an issue that people cared about and a context in which people took the perspective of an ostensibly real individual whose attitude differed from their own. Specificall
