lable at ScienceDirect

Journal of Environmental Psychology 45 (2016) 192e200

Contents lists avai

Journal of Environmental Psychology

journal homepage: www.elsevier .com/locate/ jep

Compassion for climate change victims and support for mitigationpolicy

Hang Lu*, Jonathon P. SchuldtDepartment of Communication, Cornell University, USA

a r t i c l e i n f o

Article history:Received 23 November 2015Received in revised form11 January 2016Accepted 13 January 2016Available online 15 January 2016

Keywords:CompassionClimate changePolicy supportCommunicationEmotional appeals

* Corresponding author.E-mail addresses: hl896@cornell.edu (H. Lu), jps56

http://dx.doi.org/10.1016/j.jenvp.2016.01.0070272-4944/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

The lack of broad public support for climate change mitigation policy hampers efforts to adopt timelyapproaches to the climate crisis. Echoing prior calls for examining the role of emotions in climate changecommunication, this study explores effects of compassion on support for government actions to addressclimate change. A diverse sample of U.S. participants (N ¼ 400) was randomly assigned to differentmessage treatments as part of a 2 (compassion: high or low) � 2 (climate change cue: present or absent)between-subjects factorial design. Results showed that the high-compassion condition elicited greaterself-reported compassion and stronger belief that a climate-related humanitarian crisis was caused byhuman activities, both of which, in turn, mediated increased policy supportdparticularly among politicalconservatives and moderates (compared to liberals). Overall, these findings add to the nascent literatureexamining emotions in climate change public opinion and help inform compassion appeals in climatechange communication.

© 2016 Elsevier Ltd. All rights reserved.

Despite increasing evidence that climate change is primarilycaused by human activities and is already producing severe nega-tive impacts, the American public's concern about climate changehas remained relatively stable in recent years (Leiserowitz,Maibach, Roser-Renouf, Feinberg, & Rosenthal, 2015) and climatechange continues to emerge near the bottom of the public's policypriorities for the President and Congress (Pew, 2015a). Because thepreferences and demands of citizens play a crucial and foundationalrole in the government policy agenda, timely and effective actionsto mitigate and/or adapt to climate change will likely require broadpublic support for building political will (Bliuc et al., 2015). How-ever, the persistent partisan divide on the climate issue (e.g.,McCright & Dunlap, 2011; Pew, 2015b) threatens to underminepolicy initiatives that the world's leading scientists believe will becritical if humanity is to avert the most severe warming projectionsand their negative impacts (Drews & van den Bergh, 2015).

Due in part to the disconnect between the beliefs of scientistsand the public on this issue, understanding how message featuresinfluence public support for climate policy has emerged as a majorfocus for research in environmental communication (Drews & van

@cornell.edu (J.P. Schuldt).

den Bergh, 2015). Some have suggested that the lack of widespreadpublic support on climate policy may be due, in part, to ineffectivecommunication efforts that fail to engage and resonate with in-dividuals of diverse backgrounds (e.g., political ideologies, culturalworldviews; Newell & Pitman, 2010; Kahan, Jenkins-Smith, &Braman, 2011; Pearson & Schuldt, 2014). Efforts to enhanceclimate change communication usually concentrate on cognitiveframing aspects related to the development of a persuasive mes-sage (e.g., Nisbet, Hart, Myers, & Ellithorpe, 2013; Schuldt, Konrath,& Schwarz, 2011; Wiest, Raymond, & Clawson, 2015). Recently,research has begun to explore the role of emotions, a powerfulmotivator for human behaviors, in influencing how individualsprocess and react to climate change information (e.g., Chadwick,2015; Lu & Schuldt, 2015). Building on this emerging literature aswell as research on political ideology and climate change commu-nication, the present study investigates how compassionda posi-tive emotion that is frequently featured in narratives about thenegative impacts of climate change for human beingsd influencesindividuals' support for government actions to address climatechange and how it may interact with individuals' political ideology.

1. Emotions and climate change communication

Barriers to establishing effective climate change communication

H. Lu, J.P. Schuldt / Journal of Environmental Psychology 45 (2016) 192e200 193

have been well documented (Markowitz & Shariff, 2012; Moser,2010; Van der Linden, Maibach, & Leiserowitz, 2015). Amongthem are the perception that climate change is a distant issue andthe fact that different segments of the public have divergentopinions regarding its existence and origins (Bliuc et al., 2015;Spence, Poortinga, & Pidgeon, 2012). Researchers have sought toaddress these communication barriers by examining how differentways of framing the causes and impacts of climate change (e.g.,attribution of responsibility, social distance, etc.) can shape in-dividuals' attitudes and behaviors (e.g., Hart & Nisbet, 2012; Jang,2013). These efforts have led to promising but also mixed results.

While emotional appeals are often conceptualized as a partic-ular type of framing and frequently promoted as an efficaciousstrategy for persuasion (Dillard & Nabi, 2006; Nabi, 2003), fewstudies have directly explored their potential in communicatingclimate change; however, accumulating evidence suggests thatemotions may play an unappreciated role in this domain. First, inrelation to the climate-change-as-distant issue, literature incognitive psychology has shown that increased emotional intensityof an event/object can reduce its perceived distance (Cole, Balcetis,& Dunning, 2013; Van Boven, Kane, McGraw,& Dale, 2010). Second,with regards to the deep-seated political polarization on climatechange, previous psychology and communication research suggeststhat emotions can affect policy preferences that are typicallystrongly linked to political ideology (Kühne& Schemer, 2015; Small& Lerner, 2008). Although many of these studies investigated ef-fects of emotions rather than emotional appeals per se, we exam-ined whether emotions induced through an emotional appeal mayhave similar effects.

Most relevant to this research are a small but growingnumber of studies linking emotions to climate change policysupport. In an early study investigating factors accounting forthe American public's support for climate policy, affective re-sponses to climate change emerged as a strong predictor(Leiserowitz, 2006). A more recent study extended this work andfound that discrete emotions such as worry, interest, and hopeexplained more variance in climate policy preferences than did awide range of socio-demographic variables, including politicalideology and cultural worldviews (Smith & Leiserowitz, 2014).Apart from these cross-sectional survey studies, emotions eli-cited in randomized controlled experiments have also beenshown to influence support for climate policy, suggesting acausal role of emotions in the processing of climate information.For instance, Lu and Schuldt (2015) explored how incidentalemotions might influence the way that audiences reacted toclimate change information. Results revealed that climate policypreferences were contingent on the particular emotion (namely,guilt or anger) that people were induced to feel before makingthe decisiondin addition to other factors including politicalpartisanship and the framed social distance of climate changeimpacts. Taken together, while relatively little is knownregarding the role played by discrete emotionsdand by exten-sion, persuasive messaging that utilizes emotional appealsdinclimate policy support, the current literature hints at thepossible strengths of this particular framing strategy.

1.1. Compassion and motivation

Existing climate change communication campaigns have pre-dominantly featured information that is likely to induce negativeemotions, such as fear, guilt, and shame (Markowitz & Shariff,2012). However, research has shown that overly dire messagesabout the consequences of climate change can potentially

backfire, by increasing skepticism concerning the existence ofclimate change (Feinberg & Willer, 2011). This finding is consis-tent with research in other domains indicating that negativeemotional appeals can sometimes elicit unintended conse-quences (Nabi, 2015; Williams, 2011). Thus, scholars areincreasingly advocating more attention to positively valencedemotions (e.g., Markowitz & Shariff, 2012), and in this vein,Chadwick (2015) found evidence that hope appeals can increasethe perceived effectiveness (e.g., whether the message isconvincing) of climate change messages.

While it has received less attention than other emotions(notably, fear) as an influence tool in the persuasion literature, thepositive emotion of compassion is especially relevant to the contextof climate change communication. In this paper, we follow discreteemotion theories and adopt the view that compassion is anemotion elicited from witnessing another's suffering that subse-quently generates a desire within the perceiver to help alleviatethat suffering (DeSteno, 2015; Goetz, Keltner, & Simon-Thomas,2010). Because of its approach tendency, compassion has the po-tential to enhance social engagement with others and reducepsychological distance (Oveis, Horberg, & Keltner, 2010), the latterof which may be essential for reducing the public's tendency toconstrue climate change in psychologically distant terms (Weber,2010). More importantly, compassion is a prosocial emotion thatis linked to increased care and concern for others, decreasedattention to one's own needs, and a motivation to aid anotherperson for their own sake (Goetz et al., 2010). Numerous studieshave indeed shown that the experience of compassion stimulateshelping behaviors even at a personal cost to the helper (e.g.,Condon & DeSteno, 2011; Leiberg, Klimecki, & Singer, 2011).

This prosocial feature of compassion is readily applicable toclimate change communication because in order to mitigatenegative effects of climate change, individuals need to bewilling tomake personal sacrifices for the sake of others (e.g., other species,people in other countries, future generations, etc.). These “others”are oftentimes depicted as victims of climate change, which willlikely induce compassion among the audiences who encountersuch information. While empirical studies of climate changemessage effects abound, compassion has rarely been the primaryfocus in this line of research; it is therefore unknown whethercompassion can exert beneficial effects in climate changecommunication. In particular, because we are interested in theimplications of compassion appeals for climate change commu-nication, our study does not examine measures that can help thesuffering victims directlydfor instance, intentions to donate.Instead, we focus on measures that can help prevent the futureoccurrence of similar tragedies, namely, support for governmentactions aimed at mitigating climate change. In so doing, we alsoexplore whether effects of compassion can extend beyond theprovision of immediate help to shape support for societal-levelinterventions that, ultimately, would reduce the likelihood ofsimilar tragedies in the future.

Importantly, beyond examining the potential for main effects ofcompassion on climate-related policy preferences, we also seek toexamine the underlying mechanism for the hypothesized effect.We suggest three factors that may serve as mediators of compas-sion effects in the present context. The first one is the subjectivefeeling of compassion. If claims are to be made about the effects ofcompassion appeals on policy support, subjective feelings ofcompassion should serve as the mediator between the appeal andthe outcome such that any effect on policy support is contingent onthe extent to which the compassion appeal evoked the targetemotion among audience members. The second factor is perceived

H. Lu, J.P. Schuldt / Journal of Environmental Psychology 45 (2016) 192e200194

similarity with the suffering other. Research has shown thatincreased compassion leads to increased perceived self-othersimilarity (Oveis et al., 2010). If compassion appeals are to raisepublic support for climate policy, this effect may be more likelyamong audience members who feel more connected or similar tothe suffering other; this abridged social distance may, in turn,motivate people to support government actions that can helpalleviate the suffering. The last factor we highlight is the perceptionthat the suffering other can, in fact, be helped (i.e., action efficacy).The feeling of compassion motivates moral judgments and is sen-sitive to harm-related concerns such as responsibility and vulner-ability (Goetz et al., 2010)dthus, the stronger the belief that thereare ways to aid the suffering other, the more likely it is that peoplewho feel compassionate will offer help. In the context of climatechange, if a tragedy is believed to be caused by human activities (ascompared to nature), support for government actions to preventsimilar tragedies may be expected to increase.

2. Motivated reasoning and climate change

A number of recent studies have investigated how politicalideology influences public opinion about climate change (e.g.,Bohr, 2014; Hamilton, Hartter, Lemcke-Stampone, Moore, & Saf-ford, 2015; Leiserowitz, Maibach, Roser-Renouf, Feinberg, &Rosenthal, 2014; Schuldt, Roh, & Schwarz, 2015). The generalpattern found from this line of research is a large divide betweenDemocrats and Republicans with regards to basic climate beliefsand support for climate mitigation policies. That is, most Demo-crats and liberals believe that anthropogenic forces have causedand continue to exacerbate climate change, and furthermore, ex-press greater support for government actions to address the issue;by comparison, Republicans and conservatives are consistentlyless likely to espouse these beliefs and opinions. This disagree-ment largely reflects the political and ideological divisions be-tween the two parties that have persisted over the past twodecades (Krosnick, Holbrook, & Visser, 2000; McCright & Dunlap,2011).

Increasingly, scholars have interpreted such findings as reflectingmotivated reasoning on the part of individuals who are predisposedto process climate-related information in ways that reinforce theirpreexisting views (Kunda, 1990). In practice, the tendency towardbiased information processing may be so pronounced that exposureto the same information can engenderopposite effects across climatebelievers andclimate skeptics. Accumulatingevidencehasconfirmedthat cognitive framing effects can be readily eclipsed by ideologicalcommitments amongthoseatopposite endsof thepolitical spectrum(e.g., Hart & Nisbet, 2012). Some scholars have therefore suggestedthat attempts to persuade strong partisansmay be an ineffective andinefficient strategy for heightening support for mitigation policiesamong the general public (Carrico, Truelove, Vandenbergh, & Dana,2015). However, a growing body of work has shown that situa-tional cues (e.g., weather, incidental emotions) can influence strongpartisans' beliefs about climate change (Egan & Mullin, 2012;Hamilton & Stampone, 2013; Lewandowski, Ciarocco, & Gately,2012; Li, Johnson, & Zaval, 2011; Lu & Schuldt, 2015; Schuldt &Roh, 2014). Since the subjective feeling of compassion elicited bycompassion appeals may serve as a situational cue, our study at-tempts to examine if such appeals are able to overcome stalwartresistance from conservatives, who may hold beliefs against climatechange that are highly crystallized. Additionally, because someresearch indicates that making climate change cues salient inmessaging (e.g., by explicitly linking a natural disaster to climate

change) can further polarize attitudes and behavioral intentionsbetween climatebelievers andskeptics (Chapman& Lickel, 2016),wealso explorewhether increased compassion can override the salienceof climate change cues in our study.

3. Methods

3.1. Participants

Participants were 400 adults (186 females, 214 males; mean age(years)¼ 35.03, SD¼ 12.20; 312 self-identifiedWhites, 20 Blacks, 29Asians, and 23 Hispanics) recruited online from around the UnitedStates via Amazon's Mechanical Turk (see Paolacci & Chandler,2014). Of particular interest, the sample's political ideology leanedliberal, on average (M ¼ 2.52, SD ¼ 1.16; 5-point scale from 1 ¼ veryliberal to 5 ¼ very conservative). All participants received modestmonetary compensation. Each participantwas randomly assigned toone of four experimental conditions as part of a 2 (compassion: highor low) � 2 (climate change cue: present or absent) between-subjects factorial design.

3.2. Procedure and materials

On the first webpage they encountered, participants learnedthat they were going to participate in a study focusing on howpeople evaluate stories in the media. After providing informedconsent, they were randomly assigned to one of two compassionconditions. In the high-compassion condition, participants read acompassion-promoting instruction detailing the manner in whichthey should read the news article on the following page that wasmodeled on previous studies (cf. Batson, Chang, Orr, & Rowland,2002; Pfattheicher, Sassenrath, & Schindler, 2015), specifically,“While you are reading the article, try to imagine how the child inthe picture feels about what has happened and how it has affectedher life. Try to feel the full impact of what she has been through andhow she feels as a result. You can let yourself be guided by yourfeelings.” By contrast, in the low-compassion condition, the in-struction stated, “While you are reading the article, try to take anobjective perspective toward what is described. Try not to getcaught up in how the child in the picture feels. Just remain objectiveand detached.”

After these instructions, participants proceeded to the next pagewhich displayed a fictitious news article ostensibly obtained from amainstream news website. This 180-word article described “theworst drought in half a century” to afflict East Africa and gave ageneral description of its impacts on tens of thousands of families inthe region (Appendix A). Particularly, to avoid psychic numbing(Slovic, 2007), this article featured a two-year-old girl sufferingfrom severe malnutrition and showed a picture of her laying in hermother's arm and looking directly at the readers. There were twoversions of the article. One version made explicit connections be-tween the drought and climate change, and the other did not. Forinstance, the title of the climate-change-cue-present article was“Worst drought in 60 years grips Africa, linked to climate change,”whereas the title of the climate-change-cue-absent article wassimply “Worst drought in 60 years grips Africa.”

Because the manipulation for inducing high versus lowcompassion was adapted from one originally used for audio stories(e.g., Batson et al., 1997), and that such adaptations are not alwayssuccessful in generating different levels of compassion for text-based stories (e.g., Swim & Bloodhart, 2015), we added a writingtask after participants had read the news article for purposes of

Fig. 1. Parallel multiple mediator model with self-reported compassion and belief in anthropogenic drought as parallel mediators. Note. *p < .05. **p < .01. ***p < .001.

H. Lu, J.P. Schuldt / Journal of Environmental Psychology 45 (2016) 192e200 195

maintaining the salience of the manipulation. In the high-compassion condition, participants were told to “briefly summa-rize what happened in the news article, paying particular attentiontowhat the childmay be feeling.” In the low-compassion condition,participants were asked to “briefly summarize what happened inthe news article.” After the writing task, participants filled out aquestionnaire assessing various variables detailed in the followingsection.

3.3. Measures1

3.3.1. EmotionsParticipants completed a measure of emotions that required

them to indicate, on 7-point scales, the extent to which they werefeeling each of a listed set of emotions while reading the newsarticle (1 ¼ none to 7 ¼ a lot). This randomized list of emotionsincluded a compassion scale (Batson, 1987) and a short Positive andNegative Affect Schedule (PANAS) scale (Mackinnon et al., 1999).The compassion scale included five items: compassionate, sympa-thetic, softhearted, tender, and moved (a ¼ .93; M ¼ 4.67, SD ¼ 1.67).The short PANAS scale contained ten items: inspired, alert, excited,enthusiastic, determined, afraid, upset, nervous, scared, and distressed(Positive Affect scale: a ¼ .76, M ¼ 2.66, SD ¼ 1.17; Negative Affectscale: a ¼ .89, M ¼ 2.96, SD ¼ 1.52).

3.3.2. Support for government actionsParticipants indicated on an 11-point scale to what extent they

thought the government should take actions to address climatechange (0 ¼ do nothing at all to 10 ¼ do everything they can;M ¼ 7.70, SD ¼ 2.73; adapted from Arbuckle, Morton, & Hobbs,2015).

3.3.3. Perceived similarityParticipants completed the Inclusion of Other in Self (IOS) scale

(Aron, Aron, & Smollan, 1992). Specifically, they chose the picturethat best described their felt relationship to the child in the articlefrom seven pictures comprised of two circlese one symbolizing theparticipant and one symbolizing the child e representing differ-ential degrees of overlap (1 ¼ most distant to 7 ¼ closest; M ¼ 2.78,SD ¼ 1.66).

3.3.4. Belief in anthropogenic droughtParticipants indicated on a 7-point scale to what extent they

thought the drought mentioned in the article was caused by human

1 We also measured pro-environmental behavioral intentions. Participants ratedon 7-point scales how willing they were to adopt 9 behaviors, such as switching toflorescent light bulbs and participating in recycling, in the next six months (a ¼ .86,M ¼ 2.96, SD ¼ 1.52). However, no main or interaction effects of the experimentalconditions were found for this outcome variable. We speculate that, as compared togovernment actions, individual-level behavioral changes might be perceived as lesseffective for helping climate change victims in the future.

actions (1 ¼ not at all to 7 ¼ very much so; M ¼ 4.29, SD ¼ 1.86).

3.3.5. Manipulation check for climate change cuesParticipants indicated on a 7-point scale to what extent they

thought the drought mentioned in the article was linked to climatechange (1 ¼ not at all to 7 ¼ very much so; M ¼ 4.96, SD ¼ 1.73).

4. Results

4.1. Manipulation checks

We performed two-way ANOVAs featuring terms for thecompassion condition (high or low), climate change cue condition(present or absent), and their interaction term as the independentvariables, and self-reported compassion, positive affect, negativeaffect, and belief in the connection between the drought andclimate change, as the dependent variables, respectively.

Our analyses suggested that the compassion induction wassuccessful, such that the high-compassion condition (M ¼ 5.04,SD ¼ 1.61) elicited more compassion than the low-compassioncondition (M ¼ 4.34, SD ¼ 1.65), F(1, 396) ¼ 17.98, p < .001,h2 ¼ .04. In contrast, there was no significant difference betweenthe high-compassion (M ¼ 2.77, SD ¼ 1.22) and low-compassion(M ¼ 2.57, SD ¼ 1.13) conditions in positive affect, F(1,396) ¼ 3.03, ns, nor was there any difference in negative affectbetween conditions (Mhigh ¼ 3.04, SDhigh ¼ 1.58; Mlow ¼ 2.88,SDlow ¼ 1.46), F(1, 396) ¼ 1.09, ns.

Turning to our manipulation of the climate change cue (i.e.,present or absent), we found that the climate-change-cue-presentcondition led to significantly stronger belief that the drought waslinked to climate change (M ¼ 5.13, SD ¼ 1.66) compared to theclimate-change-cue-absent condition (M ¼ 4.77, SD ¼ 1.78), F(1,396) ¼ 4.38, p < .05, h2 ¼ .01. Therefore, the manipulation of theclimate change cue was also successful.

4.2. Tests of main dependent variables

We performed a series of two-way ANOVAs featuring terms forcompassion condition (high or low), climate change cue condition(present or absent), and their interaction term as the independentvariables, and support for government actions, perceived similarity,and belief in anthropogenic drought, as the dependent variables,respectively.

Consistent with the main hypothesis, results revealed a maineffect of compassion condition on support for government actionsto address climate change, F(1, 396) ¼ 6.30, p < .05, h2 ¼ .02, suchthat the high-compassion condition led to more support for gov-ernment actions (M ¼ 8.06, SD ¼ 2.53) than did the low-compassion condition (M ¼ 7.38, SD ¼ 2.86). No other main orinteraction effects emerged for support for government actions,and moreover, we observed no significant main or interaction

H. Lu, J.P. Schuldt / Journal of Environmental Psychology 45 (2016) 192e200196

effects on the perceived similarity measure. Turning to belief inanthropogenic drought, a main effect of compassion condition wasobserved, F(1, 396) ¼ 4.96, p < .05, h2 ¼ .01, such that the high-compassion condition (M ¼ 4.51, SD ¼ 1.85) resulted in strongerbelief in anthropogenic drought than did the low-compassioncondition (M ¼ 4.10, SD ¼ 1.85). No other main or interaction ef-fects were observed.

Fig. 2. Graph depicting the effect of compassion condition on support for governmentactions to address climate change, by political ideology (conservatives (Mþ1SD),moderates (M), liberals (M�1SD)). Error bars represent standard errors of the mean.

4.3. Tests of mediation effects

Because we observed a main effect of compassion conditionon support for government actions, self-reported compassion,and belief in anthropogenic drought, we conducted follow-upanalyses testing whether self-reported (felt) compassion andbelief in anthropogenic drought might mediate the effect ofcompassion condition on support for government actions. Weused PROCESS Model 4 from Hayes (2013), with the independentvariable being the compassion condition (high ¼ .5, low ¼ �.5),with the climate change cue condition (present ¼ .5,absent ¼ �.5) and their interaction term as the covariates; self-reported compassion and belief in anthropogenic drought beingthe simultaneous mediators; and support for government actionsbeing the dependent variable (5000 bootstrap resamples, 95%bias corrected confidence interval)2. Results revealed that bothself-reported compassion and belief in anthropogenic droughtemerged as significant mediatorsdthat is, the 95% confidenceintervals for the indirect effects of the compassion condition didnot include zero (Bcompassion ¼ .16, 95% CI:.05 to .33; Bbelief ¼ .27,95% CI: .05 to .52) and the main effect of compassion conditionwas no longer significant after controlling for self-reportedcompassion and belief in anthropogenic drought (B ¼ .26,t(394) ¼ 1.04, ns; Fig. 1). Thus, it appears that the high-compassion (vs. low-compassion) condition increased self-reported compassion and belief in anthropogenic drought, bothof which, in turn, positively influenced support for governmentactions to address climate change.

4.4. Tests of moderating effects of political ideology

Recall that we were also interested in the potential moderatingrole of political ideology on support for government actions. To testthis possibility, we used a multiple regression model in whichsupport for government actions was regressed on compassioncondition, climate change cue condition, and political ideology(mean-centered), and all interaction terms. While no significantthree-way interaction emerged, there was a significant two-wayinteraction between compassion condition and political ideology,B¼ .42, t(392)¼ 2.06, p < .05, h2¼ .01 (Fig. 2). More specifically, thehigh-compassion condition was more effective than the low-compassion condition in increasing support for government actionsamong moderates (M), (Mhigh ¼ 8.08, Mlow ¼ 7.35, B ¼ .72,t(392) ¼ 3.12, p < .01) and conservatives (Mþ1SD), (Mhigh ¼ 6.90,Mlow¼ 5.70, B¼ 1.20, t(392)¼ 3.67, p < .001). In contrast, there was

2 This was essentially a parallel multiple mediator model (Hayes, 2013), whichwe chose because of its several advantages over separate simple mediation models(see Ding, Ng, & Li, 2014; Preacher & Hayes, 2008a; 2008b). We tested self-reportedcompassion and belief in anthropogenic drought in separate simple mediationmodels and found that both variables served as full mediators between the maineffect of compassion conditions and support for government actions. Additionally,we also tested whether the two variables would fit in a serial multiple mediatormodel and found that the partial correlation between self-reported compassion andbelief in anthropogenic drought was not significant after controlling for condition(Hayes, 2013), suggesting that the parallel mediation model we report is the moreappropriate model.

no difference between high- and low-compassion conditions forliberals (M�1SD), (Mhigh ¼ 9.26, Mlow ¼ 9.01, B ¼ .24, t(392) ¼ .74,ns)3.

5. Discussion

The lack of broad public support for climate mitigation policyhampers efforts to address climate change efficiently and effec-tively (Shwom, Bidwell, Dan, & Dietz, 2010). This study aligns withrecent efforts in various disciplines, such as psychology andcommunication, that attempt to increase public support throughframing the issue in ways that are engaging and accessible for thepublic (Markowitz & Shariff, 2012). Importantly, this study echoesprior calls for emphasizing the role of emotions in climate changedecision-making research (Roeser, 2012), and specifically, in-vestigates the effects of compassion as an emotional frame inincreasing support for government actions among the generalpublic and across the spectrum of the political ideology, a topic thathas received very limited attention in the climate changecommunication literature, to date.

Our experiment provides initial evidence for the efficacy ofcompassion appeals, as we observed stronger support for govern-ment actions under conditions of high versus low compassion. Onthe one hand, this result was not surprising given that a number ofstudies have uncovered the prosocial role that compassion plays inmotivating individuals to alleviate others' suffering (Goetz et al.,2010). On the other hand, the fact that increased compassionpositively influenced our primary outcome variable, support forgovernment actions to address climate change, is notable. Incontrast to many previous studies on the effects of compassion, weopted not to use ameasure to assess direct aiding behaviors, such asdonation. While it is possible that such measures would show

3 We also examined whether the mediating roles of self-reported compassionand belief in anthropogenic drought were themselves contingent on political ide-ology (i.e., moderated mediation; Muller, Judd, & Yzerbyt, 2005; Hayes, 2015).Significant results emerged from PROCESS Model 15 (10,000 bootstrap resamples,95% bias corrected confidence interval) from Hayes (2013), which tests formoderation on mediator / outcome path, such that the indirect effect throughself-reported compassion was significant for moderates and conservatives (boot-strap CIs: .03 to .27, and .10 to .56, respectively) whereas the indirect effect throughbelief in anthropogenic drought was significant for liberals, moderates, and con-servatives alike (bootstrap CIs: .01 to .23, .03 to .37, and .04 to .58, respectively).

H. Lu, J.P. Schuldt / Journal of Environmental Psychology 45 (2016) 192e200 197

stronger effects of compassion, our primary focus on support forclimate change mitigation policy directly led us to employ mea-sures that might not provide immediate help to victims but thatcould reasonably be expected to prevent similar tragedies fromhappening in the future. Thus, our findings suggest that effects ofcompassion are not limited to increasing support for actions thatprovide immediate help to alleviate another person's suffering.Future work should continue to explore other outcome measuresand boundary conditions of compassion effects in climate changecommunication.

Moreover, we considered three possible routes through whichcompassion might influence support for government actions, andfound that the present effect could be fully accounted for bygreater self-reported compassion and belief in the anthropogeniccause of the disaster elicited by the high compassion appeal.While the former may be self-evident, the latter may benefitfrom more explanation. Following theorists who contend thatemotions are a primary driver of human actions (Niedenthal,Krauth-Gruber, & Ric, 2006), we assume that increased compas-sion evoked a desire to help prevent similar situations in thefuture. The more likely individuals are to think that the droughtwas caused by human activities, the more preventable they maythink similar droughts will be, which may engender greatersupport for government actions on climate change as an effectiveprevention approachdperhaps one of many. Future research isneeded to systematically examine this presumed cognitivepathway and the role that different emotions may play in shapingthese judgments.

A primary objective of this study is to investigate if increasedcompassion can produce similar impacts on climate change policysupport among individuals with disparate political ideologies. Ourresults suggest that increased compassion is effective for height-ening support among moderates and conservatives, but less soamong liberals. This finding is consistent with other workreporting that moderates' and conservatives' beliefs about climatechange are more likely to be affected by situational and contextualfactors, including other types of framing, because liberals mayhave developed strong pro-climate change beliefs that may bedifficult to bolster (e.g., Hamilton & Stampone, 2013; Schuldt et al.2011; 2015). As the divide between liberals and conservatives onclimate change issues widens, compassion, as illustrated by ourstudy, can potentially help to bridge the well-known partisandivide in this domain.

While this study adds to the climate change communicationliterature by presenting evidence for a causal link betweencompassion and support for government actions to address climatechange, some limitations should be noted. First, whereas partici-pants in the present experiment were recruited from an onlineparticipant pool, psychological studies involving induced emotionsare more typically carried out in laboratory settings, thus affordinggreater control. On the other hand, it should be noted that thisonline pool provided a larger and more diverse sample than thosetraditionally recruited in lab experiments (Buhrmester, Kwang, &Gosling, 2011). In addition, studies have shown that the quality ofdata obtained from Amazon Mechanical Turk workers is compa-rable to data from other sources (e.g., Bartneck, Duenser,Moltchanova, & Zawieska, 2015; Crump, McDonnell, & Gureckis,2013), and the between-subjects experimental design of this on-line study may also help limit effects of environmental factors infavor of the focal independent variables (Paolacci & Chandler,2014).

Second, while we employed a common method for inducingcompassion (Batson et al., 2002), it is possible that the differencewe found between the high-compassion and low-compassionconditions might also be explained by differential attentionacross conditions, with the high-compassion manipulation pro-moting increased attention to information contained in thefictitious news article. Although our current design cannot fullyrule out this alternative explanation, we would note that furtheranalyses revealed no difference between compassion conditionsin the time spent reading the news article, nor did the conditionsgenerate differential word lengths on the writing tasks. Moreimportantly, participants in the high-compassion condition wereno more likely to notice the climate change cue in the conditionwhere it was present (as indicated by participants' reportedbelief that the drought was caused by climate change). Thus, webelieve it is unlikely that differential attention across conditionsprovides a satisfactory alternative explanation for the observedeffects.

Third, we note some measurement limitations and relatedconsiderations for future work. Although the hypothesized effect ofcompassion condition on perceived similarity to the climate changevictim did not emergedthus challenging a prominent assumptionin the emotion literature on compassiondthis lack of an effect maybe partly attributable to some idiosyncratic features of the presentdesign. For instance, although we employed the IOS scale (Aronet al., 1992) for its ease of comprehension and high reliability, it isarguably a better measure of subjective relationship closeness thanof perceived similarity, per se (see G€achter, Starmer, & Tufano,2015). Future work may employ other, more direct measures ofperceived similarity (e.g., the single-itemmeasure from Oveis et al.,2010), to examine whether the expected effect would emerge withalternative measures. Moreover, the particular characteristics of theclimate victim portrayed in our fictional news article (an impov-erished child, stricken by malnutrition, living in East Africa) mayhave elicited a great deal of psychological distance (e.g., social andspatial distance; Trope & Liberman, 2010) among our predomi-nately adult, White U.S. sampledperhaps too much dissimilarity toovercome with this compassion induction. Moreover, although weexamined the mediating role of belief in anthropogenic drought asa proxy for perceived efficacy of government actions, weacknowledge that this is a topically relevant but rather indirectmeans for doing so. Future work may thus benefit from includingmeasures that more directly assess people's beliefs about whichgovernment actions are most likely to prevent similar humanitar-ian disasters from occurring in the future, and to what extent.Finally, whilewe found significant effects of the compassion appeal,these findings should be interpreted with caution as they representrelatively small effects.

To conclude, by demonstrating that the influence of compassionextends beyond increasing the motivation to act in ways thatalleviate immediate suffering, the present work underscores itspreviously overlooked role in contributing to citizens' policy sup-port regarding climate change mitigation. Notably, since sufferingvictims are often featured in persuasive messaging about climatechange impacts, the present study invites climate change com-municators to more carefully consider whether and how to usecompassionda positive emotion or “emotional carrot,” rather thannegative emotions or “emotional sticks”dto connect and engagetraditionally skeptical audiences.

H. Lu, J.P. Schuldt / Journal of Environmental Psychology 45 (2016) 192e200198

Appendix A

Climate change cue present message

H. Lu, J.P. Schuldt / Journal of Environmental Psychology 45 (2016) 192e200 199

Climate change cue absent message

References

Arbuckle, J. G., Morton, L. W., & Hobbs, J. (2015). Understanding farmer perspectiveson climate change adaptation and mitigation: the roles of trust in sources ofclimate information, climate change beliefs, and perceived risk. Environmentand Behavior, 47, 205e234.

Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of other in the self scale and thestructure of interpersonal closeness. Journal of Personality and Social Psychology,63, 596e612.

Bartneck, C., Duenser, A., Moltchanova, E., & Zawieska, K. (2015). Comparing thesimilarity of responses received from studies in Amazon's mechanical turk tostudies conducted online and with direct recruitment. PLoS One, 10, e0121595.

Batson, C. D. (1987). Prosocial motivation: is it ever truly altruistic?. In L. Berkowitz

(Ed.), Advances in experimental social psychology (Vol. 20, pp. 65e122) New York,NY: Academic Press.

Batson, C. D., Chang, J., Orr, R., & Rowland, J. (2002). Empathy, attitudes, and action:can feeling for a member of a stigmatized group motivate one to help thegroup? Personality and Social Psychology Bulletin, 28, 1656e1666.

Batson, C. D., Sager, K., Garst, E., Kang, M., Rubchinsky, K., & Dawson, K. (1997). Isempathy-induced helping due to selfeother merging? Journal of Personality andSocial Psychology, 73, 495e509.

Bliuc, A. M., McGarty, C., Thomas, E. F., Lala, G., Berndsen, M., & Misajon, R. (2015).Public division about climate change rooted in conflicting socio-politicalidentities. Nature Climate Change, 5, 226e229.

Bohr, J. (2014). Public views on the dangers and importance of climate change:predicting climate change beliefs in the United States through income

H. Lu, J.P. Schuldt / Journal of Environmental Psychology 45 (2016) 192e200200

moderated by party identification. Climatic Change, 126, 217e227.Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon's mechanical turk: a new

source of inexpensive, yet high-quality, data? Perspectives on Psychological Sci-ence, 6, 3e5.

Carrico, A. R., Truelove, H. B., Vandenbergh, M. P., & Dana, D. (2015). Does learningabout climate change adaptation change support for mitigation? Journal ofEnvironmental Psychology, 41, 19e29.

Chadwick, A. E. (2015). Toward a theory of persuasive hope: effects of cognitiveappraisals, hope appeals, and hope in the context of climate change. HealthCommunication, 30(6), 598e611.

Chapman, D. A., & Lickel, B. (2016). Climate change and disasters: how framingaffects justifications for giving or withholding aid to disaster victims. SocialPsychological and Personality Science, 7, 13e20.

Cole, S., Balcetis, E., & Dunning, D. (2013). Affective signals of threat increaseperceived proximity. Psychological Science, 24, 34e40.

Condon, P., & DeSteno, D. (2011). Compassion for one reduces punishment foranother. Journal of Experimental Social Psychology, 47, 698e701.

Crump, M. J. C., McDonnell, J. V., & Gureckis, T. M. (2013). Evaluating Amazon's me-chanical turk as a tool for experimental behavioral research. PLoS One, 8, e57410.

DeSteno, D. (2015). Compassion and altruism: how our minds determine who isworthy of help. Current Opinion in Behavioral Sciences, 3, 80e83.

Dillard, J. P., & Nabi, R. L. (2006). The persuasive influence of emotion in cancerprevention and detection messages. Journal of Communication, 56, S123eS139.

Ding, Z., Ng, F., & Li, J. (2014). A parallel multiple mediator model of knowledgesharing in architectural design project teams. International Journal of ProjectManagement, 32, 54e65.

Drews, S., & van den Bergh, J. C. (2015). What explains public support for climatepolicies? A review of empirical and experimental studies. Climate Policy. http://dx.doi.org/10.1080/14693062.2015.1058240. Advanced online publication.

Egan, P. J., & Mullin, M. (2012). Turning personal experience into political attitudes:the effect of local weather on Americans' perceptions about global warming.The Journal of Politics, 74, 796e809.

Feinberg,M.,&Willer, R. (2011). Apocalypse soon?Diremessages reduce belief inglobalwarming by contradicting just-world beliefs. Psychological Science, 22, 34e38.

G€achter, S., Starmer, C., & Tufano, F. (2015). Measuring the closeness of relation-ships: a comprehensive evaluation of the 'Inclusion of the Other in the Self'scale. PLoS One, 10(6), e0129478.

Goetz, J. L., Keltner, D., & Simon-Thomas, E. (2010). Compassion: an evolutionaryanalysis and empirical review. Psychological Bulletin, 136, 351e374.

Hamilton, L. C., Hartter, J., Lemcke-Stampone, M., Moore, D. W., & Safford, T. G.(2015). Tracking public beliefs about anthropogenic climate change. PLoS One,10(9), e0138208. http://dx.doi.org/10.1371/journal.pone.0138208.

Hamilton, L. C., & Stampone, M. D. (2013). Blowin’ in the wind: short-term weatherand belief in anthropogenic climate change. Weather, Climate and Society, 5,112e119.

Hart, P. S., & Nisbet, E. C. (2012). Boomerang effects in science communication: howmotivated reasoning and identity cues amplify opinion polarization aboutclimate mitigation policies. Communication Research, 39, 701e723.

Hayes, A. F. (2013). An introduction to mediation, moderation, and conditional processanalysis. New York: The Guilford Press.

Hayes, A. F. (2015). An index and test of linear moderated mediation. MultivariateBehavioral Research, 50, 1e22.

Jang, S. M. (2013). Framing responsibility in climate change discourse: ethnocentricattribution bias, perceived causes, and policy attitudes. Journal of EnvironmentalPsychology, 36, 27e36.

Kahan, D. M., Jenkins-Smith, H., & Braman, D. (2011). Cultural cognition of scientificconsensus. Journal of Risk Research, 14, 147e174.

Krosnick, J. A., Holbrook, A. L., & Visser, P. S. (2000). The impact of the fall 1997debate about global warming on American public opinion. Public Understandingof Science, 9, 239e260.

Kühne, R., & Schemer, C. (2015). The emotional effects of news frames on informationprocessing and opinion formation. Communication Research, 42, 387e407.

Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108,480e498.

Leiberg, S., Klimecki, O., & Singer, T. (2011). Short-term compassion training increasesprosocial behavior in a newly developed prosocial game. PLoS One, 6, e17798.

Leiserowitz, A. (2006). Climate change risk perception and policy preferences: therole of affect, imagery, and values. Climatic Change, 77, 45e72.

Leiserowitz, A., Maibach, E., Roser-Renouf, C., Feinberg, G., & Rosenthal, S. (2014).Politics & global warming, Spring 2014. New Haven, CT: Yale Project on ClimateChange Communication.

Leiserowitz, A., Maibach, E., Roser-Renouf, C., Feinberg, G., & Rosenthal, S. (2015).Climate change in the American mind: March, 2015. Yale University and GeorgeMason University. New Haven, CT: Yale Project on Climate ChangeCommunication.

Lewandowski, G. W., Jr., Ciarocco, N. J., & Gately, E. L. (2012). The effect of embodiedtemperature on perceptions of global warming. Current Psychology, 31, 318e324.

Li, Y., Johnson, E. J., & Zaval, L. (2011). Local warming daily temperature changeinfluences belief in global warming. Psychological Science, 22, 454e459.

Lu, H., & Schuldt, J. P. (2015). Exploring the role of incidental emotions in support forclimate change policy. Climatic Change, 131, 719e726.

Mackinnon, A., Jorm, A. F., Christensen, H., Korten, A. E., Jacomb, P. A., & Rodgers, B.(1999). A short form of the positive and negative affect schedule: evaluation offactorial validity and invariance across demographic variables in a communitysample. Personality and Individual Differences, 27, 405e416.

Markowitz, E. M., & Shariff, A. F. (2012). Climate change and moral judgement.Nature Climate Change, 2, 243e247.

McCright, A. M., & Dunlap, R. E. (2011). The politicization of climate change andpolarization in the American public's views of global warming, 2001e2010. TheSociological Quarterly, 52, 155e194.

Moser, S. C. (2010). Communicating climate change: history, challenges, process andfuture directions. Wiley Interdisciplinary Reviews: Climate Change, 1, 31e53.

Muller, D., Judd, C. M., & Yzerbyt, V. Y. (2005). When moderation is mediated andmediation is moderated. Journal of Personality and Social Psychology, 89, 852e863.

Nabi, R. L. (2003). Exploring the framing effects of emotion do discrete emotionsdifferentially influence information accessibility, information seeking, andpolicy preference? Communication Research, 30, 224e247.

Nabi, R. (2015). Emotional flow in persuasive health messages. Health Communi-cation, 30, 114e124.

Newell, Ben, & Pitman, Andrew (2010). The psychology of global warming:improving the fit between the science and the message. Bulletin of the AmericanMeteorological Society, 91, 1003e1014.

Niedenthal, P. M., Krauth-Gruber, S., & Ric, F. (2006). Psychology of emotion: Inter-personal, experiential, and cognitive approaches. New York, NY: Psychology Press.

Nisbet, E. C., Hart, P. S., Myers, T., & Ellithorpe, M. (2013). Attitude change incompetitive framing environments? Open-/closed-mindedness, framing effects,and climate change. Journal of Communication, 63, 766e785.

Oveis, C., Horberg, E. J., & Keltner, D. (2010). Compassion, pride, and social intuitionsof self-other similarity. Journal of Personality and Social Psychology, 98, 618e630.

Paolacci, G., & Chandler, J. (2014). Inside the turk: understanding mechanical turk asa participant pool. Current Directions in Psychological Science, 23, 184e188.

Pearson, A. R., & Schuldt, J. P. (2014). Facing the diversity crisis in climate science.Nature Climate Change, 4, 1039e1042.

Pew Research Center for the People and the Press. (2015a). Public's policy prioritiesreflect changing conditions at home and abroad. Retrieved from http://www.people-press.org/files/2015/01/01-15-15-Policy-Priorities-Release.pdf.

Pew Research Center for the People and the Press. (2015b). Ideological divide overglobalwarming aswideasever. Retrieved fromhttp://www.pewresearch.org/fact-tank/2015/06/16/ideological-divide-ove-global-warming-as-wide-as-ever/.

Pfattheicher, S., Sassenrath, C., & Schindler, S. (2015). Feelings for the suffering ofothers and the environment compassion fosters proenvironmental tendencies.Environment and Behavior. http://dx.doi.org/10.1177/0013916515574549.Advanced online publication.

Preacher, K. J., & Hayes, A. F. (2008a). Asymptotic and resampling strategies forassessing and comparing indirect effects in multiple mediator models. BehaviorResearch Methods, 40, 879e891.

Preacher, K. J., & Hayes, A. F. (2008b). Contemporary approaches to assessingmediation in communication research. In A. F. Hayes, M. D. Slater, & L. B. Snyder(Eds.), The sage sourcebook of advanced data analysis methods for communicationresearch (pp. 13e54). Thousand Oaks, CA: Sage Publications.

Roeser, S. (2012). Risk communication, public engagement, and climate change: arole for emotions. Risk Analysis, 32, 1033e1040.

Schuldt, J. P., Konrath, S. H., & Schwarz, N. (2011). “Global warming” or “climatechange”?: whether the planet is warming depends on question wording. PublicOpinion Quarterly, 75, 115e124.

Schuldt, J. P., & Roh, S. (2014). Of accessibility and applicability: how heat-relatedcues affect belief in “global warming” versus “climate change.” Social Cogni-tion, 32, 217e238.

Schuldt, J. P., Roh, S., & Schwarz, N. (2015). Questionnaire design effects in climatechange surveys: implications for the partisan divide. Annals of the AmericanAcademy of Political and Social Science, 658, 67e85.

Shwom, R., Bidwell, D., Dan, A., & Dietz, T. (2010). Understanding US public supportfor domestic climate change policies. Global Environmental Change, 20,472e482.

Slovic, P. (2007). “If I look at the mass I will never act”: psychic numbing andgenocide. Judgment and Decision Making, 2, 79e95.

Small, D. A., & Lerner, J. S. (2008). Emotional policy: personal sadness and angershape judgments about a welfare case. Political Psychology, 29, 149e168.

Smith, N., & Leiserowitz, A. (2014). The role of emotion in global warming policysupport and opposition. Risk Analysis, 34, 937e948.

Spence, A., Poortinga, W., & Pidgeon, N. (2012). The psychological distance ofclimate change. Risk Analysis, 32, 957e972.

Swim, J. K., & Bloodhart, B. (2015). portraying the perils to polar bears: the role ofempathic and objective perspective-taking toward animals in climate changecommunication. Environmental Communication, 9, 446e468.

Trope, Y., & Liberman, N. (2010). Construal-level theory of psychological distance.Psychological Review, 117, 440e463.

Van Boven, L., Kane, J., McGraw, A. P., & Dale, J. (2010). Feeling close: emotionalintensity reduces perceived psychological distance. Journal of Personality andSocial Psychology, 98, 872e885.

Weber, E. U. (2010). What shapes perceptions of climate change? Wiley Interdisci-plinary Reviews: Climate Change, 1, 332e342.

Van der Linden, S., Maibach, E., & Leiserowitz, A. (2015). How to improve publicengagement with climate change: five “best practice” insights from psycho-logical science. Perspectives on Psychological Science, 10, 758e763.

Wiest, S. L., Raymond, L., & Clawson, R. A. (2015). Framing, partisan predispositions,and public opinion on climate change. Global Environmental Change, 31,187e198.

Williams, K. C. (2011). Improving fear appeal ethics. Journal of Academic and BusinessEthics, 1e24.