Environmental Research Letters

LETTER • OPEN ACCESS

The impacts of political cues and practicalinformation on climate change decisionsTo cite this article: Gabrielle Wong-Parodi and Baruch Fischhoff 2015 Environ. Res. Lett. 10 034004

 

View the article online for updates and enhancements.

Related contentAssessing the environmental justiceconsequences of flood risk: a case studyin Miami, FloridaMarilyn C Montgomery and JayajitChakraborty

-

Environmental risk perception from visualcues: the psychophysics of tornado riskperceptionBarry Dewitt, Baruch Fischhoff, AlexanderDavis et al.

-

A multi-dimensional integrated approach toassess flood risks on a coastal city,induced by sea-level rise and storm tidesXu Lilai, He Yuanrong, Huang Wei et al.

-

Recent citationsA.H. Walker-

A decision science approach forintegrating social science in climate andenergy solutionsGabrielle Wong-Parodi et al

-

Resilience vs. Adaptation: Framing andactionGabrielle Wong-Parodi et al

-

This content was downloaded from IP address 128.237.216.150 on 19/10/2017 at 17:44

Environ. Res. Lett. 10 (2015) 034004 doi:10.1088/1748-9326/10/3/034004

LETTER

The impacts of political cues and practical information on climatechange decisions

GabrielleWong-Parodi1 andBaruch Fischhoff1,2

1 Department of Engineering and Public Policy, 129BakerHall, CarnegieMellonUniversity, Pittsburgh, PA 15213,USA2 Department of Social andDecision Sciences, 219 PorterHall, CarnegieMellonUniversity, Pittsburgh, PA 15213,USA

E-mail: gwongpar@cmu.edu

Keywords: decision-making, climate change, science communication, political communication

Supplementarymaterial for this article is available online

AbstractAdapting to climate changewill require people tomakemeasured decisions, informed by the sciencerelevant to those choices. Communicating that science is complicated by the politicization of the topic.In two studies, we ask howpolitical cues, designed to evoke individuals’ sense of identity as believers ornonbelievers in global warming, affect a hypothetical decision: buying a home vulnerable to coastalflooding exacerbated by global warming using the Zillow® real estate website. In both studies, wemanipulate participants’ frame of reference by focusing themon risks due to ‘elevation’, ‘globalwarming’, or both, ormentioning neither.We also examine how immersion in practical details affectsthe power of these cues bymanipulatingwhether participants have access to Risk Finder (http://sealevel.climatecentral.org), an interactive decision aid. Study 1 asks about global warming beliefsafter their decision; Study 2 asks beforehand. Bothfind that immersion in practical information, usingRisk Finder, overrode political identity cues.When framed in terms of both elevation and globalwarming andwithout explicit expression of global warming beliefs (Study 1), participants’ responsesreflected their beliefs. The results suggest that communications should acknowledge politicaldifferences and then focus on practical decisions and the science that can inform them.

1. Introduction

Evidence of a changing climate is all around us. It isseen in systemic changes, such as sea-level rise,meltingicecaps, and higher global mean temperatures [1], aswell as in episodic events [2, 3]. In the first fewmonthsof 2014 alone, the United States experienced unusualcold in the Midwest, heavy snowfall on the East Coast,and severe drought on the West Coast [4]. Suchextreme events are predicted to becomemore frequent[5–8], with implications for decisions made by busi-nesses (e.g., which crops to grow), governments (e.g.,what infrastructure to build), and individuals (e.g.,where to live). As with any topic, translating scientificknowledge into useful terms requires identifying theinformation most critical to decision makers and thencommunicating it in comprehensible terms [9, 10].

Communication about climate change faces theadditional challenge of overcoming the topic’s poli-tical polarization [11], with personal values,

ideological orientations, and social identity seeminglyacting as ‘perceptual screens’, leading people to inter-pret messages in ways that reinforce their existingviews and allegiances [12–15], particularly for thosewith strong beliefs about global warming [16, 17], ortriggering heuristics leading to biased inferences.Many psychological processes could contribute tosuch biased information processing. Some, such asconfirmation bias, occur with the mere existence ofbeliefs and influence reasoning without awareness[18]. Others arise from motivated reasoning, invokedmore or less consciously to defend desired beliefs andproduce behavior consistent with personal values [19–22]. For example, Costa and Kahn (2013) found thatproviding households with feedback about their ownand their neighbor’s electricity use was much moreeffective in reducing energy consumption among lib-erals (registered as Democrats, Greens, or Peace andFreedom) than among conservatives (registered asRepublicans) [23]. Indeed, for unknown reasons,

OPEN ACCESS

RECEIVED

25November 2014

REVISED

3 February 2015

ACCEPTED FOR PUBLICATION

4 February 2015

PUBLISHED

26 February 2015

Content from this workmay be used under theterms of theCreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

© 2015 IOPPublishing Ltd

conservatives who did not get their electricity fromrenewable resources and did not donate to environ-mental groups increased their energy consumption by1%, a result interpreted by the authors as their con-suming more to act in accordance with their politicalidentity [24]. Other results find that the choice ofterm, ‘global warming’ or ‘climate change’, can evokedifferent responses [25].

In two studies, we examine how two manipula-tions, designed to heighten the salience of politicalidentity related to global warming, affect responses toa realistic (although still hypothetical) decision: buy-ing a home in an area subject to storm surges, poten-tially affected by sea-level rise. Both manipulationswere designed to evoke the sense of identity that canaffect information processing [26, 27].

Our task asks participants to use Zillow®, a US realestate listing service, to choose a prospective home inSavannah, GA, and then describe it in terms of its loca-tion and price, and their reasons for selecting it. Theythen evaluate the home in terms of several risk-relatedconcerns, including ones relevant to climate change.

One manipulation of political identity varies howthe task is framed, by presenting either (i) no addi-tional information suggesting how to approach it (nocue); (ii) information about elevations above sea levelin the region (elevation), depicted on the Zillow® map;(iii) a paragraph on global warming; or (iv) both kindsof information (elevation+ global warming). ‘No cue’should evoke neither practical concerns nor politicalidentity, beyond what would occur to participantsnaturally. The ‘elevation’ information should evokepractical concern for all respondents. The ‘globalwarming’ information should evoke political identityfor both believers and nonbelievers in global warming;it should have practical value for believers, but not fornonbelievers. The combined information (‘eleva-tion + global warming’) should evoke both practicalconcerns and political identify for believers, hencehave additive impact, but evoke conflicting responsesfor nonbelievers.

The second manipulation varies whether partici-pants self-identify as believers or nonbelievers in glo-bal warming, prior to completing the task (in Study2) or afterward (in Study 1). The research literatureoffers conflicting predictions regarding the effects ofprior self-identification. On the one hand, partici-pants may feel pressure to make decisions consistentwith that expressed identify. On the other hand, self-affirmation theory [28] holds that allowing people toaffirm their identity makes them more receptive tomessages that might otherwise threaten it [29–32],thereby diminishing its role in their judgments. Wepredict that, with these practical decisions, made inprivate, the latter processes dominate. That is, parti-cipants express their identity, and then get on withthe tasks, reducing differences between believers andnonbelievers.

In addition to manipulating political identify, wealso manipulate how deeply participants can immersethemselves in the practical details of their task, pre-dicting that deeper immersion will reduce identityeffects. Specifically, we offer half of the participants theuse of Risk Finder, an interactive decision aid (http://sealevel.climatecentral.org) showing flood risk andassociated impacts for Savannah, GA. We expect thesepractical details to make coastal flooding risks morereal and relevant, reducing participants’ psychologicaldistance from the task. That engagement should leadto deeper processing, which reduces, in turn, the roleof political identity [33–37]. If involvement with apractical task can, indeed, overcome the effects of poli-tical identity, then the rancor of debates over the exis-tence of global warming might subside when peoplemake such decisions related to its potential impacts(unless, of course, those decisions become arenas forpolitical fights thatmake the details irrelevant).

Thus, in Study 1, participants self-identify asbelievers or nonbelievers in global warming after theycomplete tasks related to purchasing a home facingrisks related to global warming. We predict that theseparticipants’ political identity will express itself in thefollowing ways with the four frames: (1) with ‘no cue’and ‘elevation’, two conditionsmaking no reference toglobal warming, believers and nonbelievers willrespond similarly (with both seeing greater risk with‘elevation’). (2) With the ‘global warming’ frame,believers will see greater risk than will nonbelievers.(3) The combined frame (‘elevation + global warm-ing’) will elicit risk-related concerns for believers, andconflicting responses for nonbelievers. In Study 2, par-ticipants self-identify before completing these tasks.We predict that allowing these participants to expresstheir identity will free them to focus on the practicaldecisions, reducing differences between believers andnonbelievers. Finally, we predict that (in both studies)using Risk Finder will eliminate differences betweenbelievers and nonbelievers, by immersing them inpractical details of their task. Throughout, we seek tomake the task as realistic as possible, while still recog-nizing its hypothetical nature.

2. Study 1—global warming question aftertasks

2.1.Methods2.1.1. Experimental proceduresAfter a brief introduction, informed consent, andscreening for age (⩾18) and for not currently living inGeorgia, all participants were asked to imaginemovingwith their family to Savannah, GA. To enhancepersonal relevance [34], participants were told toimagine that their family wants to buy a ‘single-familyhome’ with ‘no plans to move ever again’. They thensaw a Zillow® map (figure 1) with current real estatelistings in Savannah, and chose a home that would be

2

Environ. Res. Lett. 10 (2015) 034004 GWong-Parodi and B Fischhoff

best for their family. After selecting it, they reported itsaddress, list price, and answered the open-endedquestions, ‘In a few sentences, tell us what you’d like toknow about Savannah that would help you decide ifyou want to buy this home’ and ‘What other expenseswould you consider in buying this home?’3 Next,participants were asked, ‘If you were, in fact, movingto Savannah and you had the money, would you buythis home?’ (1 =Yes, 2 =No).

Participants were then randomly assigned to oneof eight conditions, with Framing and Aid as crossedfactors. The four levels of Framingwere:

(i) no cue: participantswent directly to the responsemeasures (described below).

(ii) Elevation: participants were shown an elevationmap of Savannah, where they read, ‘Look whereyour home is on the Zillow® map. Now, find yourhome on the elevation map below. Places on themap in blue mean that the land is at low elevation.Take aminute to study the twomaps’. (figure 1).

(iii) Global Warming: participants read, ‘The bestscientific knowledge suggests that sea level is risingin Savannah. Sea level has been rising for as long asscientists have been able to measure it accurately.The more that sea level rises, the more coastal

floods increase. Global warming contributes to sealevel rise by melting glaciers and ice sheets into theoceans. It also heats ocean water, causing it toexpand’.

(iv) Elevation +Global Warming: participantsreceived both frames.

Note that, in addition to providing the political cueof global warming, frames (iii) and (iv) also had prac-tical value, providing information relevant for deci-sion making, and engaged participants furtherwith the task.

The two levels of Aidwere:

(v) without Risk Finder: participants searched theZillow®map on their own.

(vi)With Risk Finder: participants used a mock-upof the surging seas ‘Risk Finder’ tool (figure 2)(http://sealevel.climatecentral.org) to manipulateflood height between 0 and 10 + feet, allowing themto see predictions of the corresponding (a) chancesof such a flood ‘between today and 2050’, (b) mapof land under water, (c) percentage of land underwater, and (d) percentage of people, schools,homes, roadmiles, power plants, and sewage plantsaffected in Savannah, GA. They were theninstructed to ‘take a few minutes to learn aboutflooding risk for the area’.

Finally, participants completed the response mea-sures and answered demographic questions. Figure 3summarizes the study design.

Figure 1.The Zillow®map of homes for sale in Savannah, GA that participants used for Study 1 and Study 2 is on the left.Weembedded thismap in our studies with permission fromZillow® (www.zillow.com). The code that we used can be found insupplementalmaterial, available at stacks.iop.org/ERL/10/034004/mmedia. The elevationmap that participants used for both Study 1and Study 2 is on the right. Thismapwas created using theGeorgia FloodMappingAssessment&Planning (MAP.) tool fromGeorgia’s Department ofNatural Resources: Environmental ProtectionDivision (www.georgidfirm.com).

3A search revealed the following percentages of Study 1 participants

spontaneously mentioning each of these terms: ‘climate change’ or‘global warming’ (0%), ‘flood’ or ‘flooding’ (1.8%), ‘flood insur-ance’ (1.1%), ‘sea-level rise’ (0.1%). A search revealed the followingpercentages of Study 2 participants spontaneously mentioning eachof these terms: ‘climate change’ or ‘global warming’ (0.2%), ‘flood’or ‘flooding’ (2.2%), ‘flood insurance’ (1%), ‘sea-level rise’ (0%).

3

Environ. Res. Lett. 10 (2015) 034004 GWong-Parodi and B Fischhoff

2.1.2.Measures

2.1.2.1. Dependent variablesParticipants answered the open-ended question,‘Assume that you were going to buy the home that youchose earlier. If you had to guess, what do you thinkthat annual flood insurance is for the home?’ Theythen rated their agreement (1 = very unlikely, 7 = verylikely) with the following statements: ‘How likely is itthat youwould still buy the home if you had to pay thatmuch for flood insurance?’ (buying with insurance);‘How likely is it that you would still buy the home ifflood insurance were not available?’ (buying withoutinsurance); and ‘How likely is it that you would stillmove to Savannah if flood insurance were notavailable?’ (moving without insurance). We combinedthe three statements into an index (Cronbach’sα= 0.79), which was reverse coded to create thedependent variable (a) overall perceived risk. Wecomputed the difference between buying with insur-ance and buying without insurance in order to create asecond dependent variable, (b) sensitivity to insuranceavailability, with a positive score indicating greaterwillingness to buy the home when insurance was

available. As a manipulation check for how wellparticipants understood the task, we examinedwhether they reported beingmore likely to buy a homewhen insurance was available, compared to when itwas not.

2.1.2.2. Belief about global warmingWeused one question fromMaibach, Roser-Renouf &Leiserowitz’s (2009) Six Americas survey to assessbeliefs about global warming [38]. Participants read,‘Recently you may have noticed that global warminghas been getting some attention in the news. Globalwarming refers to the idea that the world’s averagetemperature has been increasing over the past 150years, may be increasing more in the future, and thatthe world’s climate may change as a result’. They werethen asked, ‘What do you think? Do you think thatglobal warming is happening?’ (1 =Yes, 2 =No, 3 = Idon’t know) and then, if they answered ‘Yes’ or ‘No’,to indicate their level of certainty on a scale anchoredat 1 =…and I’m extremely sure, 4 =…but I’m not atall sure. These questions appeared at the end of thestudy. Participants who answered ‘I don’t know’ wereexcluded from the study.

Figure 2.The interactive coastalflooding risk tool for Savannah, GA that participants used in both Study 1 and Study 2. The datawereadapted from those forKings County, NY, for which themost complete analyses were then available. The tool was developed forusability testing purposes for Climate Central’s Surging Seas Risk Finder (http://sealevel.climatecentral.org). It can be accessed athttp://wpweb2.tepper.cmu.edu/energy-behavior/c.c._studies/Study4a1_v3/c.c._4a1_2050_GA_8518750.html.

4

Environ. Res. Lett. 10 (2015) 034004 GWong-Parodi and B Fischhoff

2.1.3. ParticipantsWe recruited 1383 participants through an advertise-ment at Amazon’s Mechanical Turk (MTurk), anonline recruiting service [39]. Comparisons of beha-vioral experiments using MTurk and other recruit-ment methods have found few differences [40]. Apower analysis using G*Power [41] indicated a totalsample of 357 for a small effect size (η2 = 0.03) with80% power, for ANOVA (fixed effects, main effects,and interactions) with alpha at 0.05. We decided torecruit approximately four times that number ofparticipants in order to obtain enough nonbelievers,based on their (minority) prevalence in nationalpolls [42].

Participants’ self-reported mean age was 30.8(SD= 10.5), with 45.8% female, 76.5%White or Cau-casian, 49.0% having at least a bachelor’s degree and43.8% having household income of at least $51K. In

terms of party affiliation, 46.4% reported beingDemocrats, 34.1% Independents, 12.8% Republicansand 6.7% Other or Prefer not to answer. Nearly one-third reported having lived near the coast (31.1%),86.6% of whom reported being familiar with it. Mosthad vacationed along the coast (81.4%). More thanone-third had experienced a hurricane (38.4%), andmost (58.4%) had experienced a flood or knew some-one who had. Those who answered ‘yes’ to the ques-tion, ‘Do you think that global warming ishappening?’, were treated as ‘believers’ (n= 1212);those who answered ‘no’were treated as ‘nonbelievers’(n= 171). Nonbelievers were significantly more con-fident in their positions than were the believers,(M= 2.16 versus 1.97) p< 0.001, on the 1–4 con-fidence scale. Believers (85.4%) were significantlymore likely to be Democrats than were nonbelievers(23.4%), χ2(1,N= 818) = 189.18, p< 0.001.

Figure 3. Schematic of the study design. Study 2 directly primes political identity with respect to global warming prior to the task,whereas Study 1 elicits political identity afterward. Both studiesmanipulate that priming indirectly through task framing.

5

Environ. Res. Lett. 10 (2015) 034004 GWong-Parodi and B Fischhoff

2.2. Results2.2.1. Analysis planFor each dependent variable, we conducted separateAnalyses of Variance (ANOVA) examining the effectsof Framing (no cue, elevation, global warming, eleva-tion + global warming), Aid (with or without RiskFinder) and Belief (believer or nonbeliever). Thedependent variables were (a) overall perceived riskand (b) sensitivity to insurance availability. Partici-pants’ willingness to buy the home that they hadselected (if they were moving to Savannah and had themoney) and estimates of flood insurance cost (as aproxy for perceived elevation) did not differ byexperimental condition (H(7) = 8.19, p= 0.32; F(71 320) = .77, p= 0.62, respectively), consistent withsuccessful randomization.

2.2.2.Manipulation checkA paired-sample t-test found that participants weremuchmore willing to buy a homewhen insurance wasavailable (M= 4.33, SD= 1.70) than when it was not(M= 2.99, SD= 1.79), t(1383) = 26.30, p< 0.001, 95%CI of the difference [1.24, 1.44], indicating an under-standing of the task.

2.2.3. AnalysesAs shown in table 1, neither Belief nor Framing had amain effect for either dependent variable. However,Aid did for perceived overall risk: participants whoused Risk Finder saw greater risk, as expressed in beingless willing tomove to Savannah.

There was a significant interaction between Aidand Framing for both dependent variables. As seen infigure 4 (left-hand side), with the no cue and elevationframe conditions, participants with Risk Finder sawsignificantly greater risk than did those without it.However, having Risk Finder made no difference withthe global warming or elevation + global warmingframes, suggesting that those frames alone wereenough to evoke greater concern. We also found a sig-nificant interaction between Framing and Belief foroverall perceived risk. As seen in figure 5 (left-handside), believers saw greater risk (and nonbelievers lessrisk) with the elevation + global warming frame,whereas there was no difference in the other threeframes (figure 5). Tables S1 and S2 in supplementalmaterial present additional results.

2.3. SummaryStudy 1 manipulated the salience of political identitywith four alternative frames for considering a hypothe-tical decision about purchasing a home subject tocoastal flooding risks. For participants who used justthe Zillow® real estate interface to examine houses,believers and nonbelievers responded similarly forthree frames (no cue, elevation, global warming), butmuch differently for the fourth (elevation + globalwarming). When asked to consider both factors,believers saw greater risk, whereas nonbelievers sawless. That interaction (between Frame and Belief)disappeared for participants who had access to theRisk Finder decision aid and, with it, the opportunityfor greater immersion in practical details of thedecision. Having Risk Finder heightened perceivedrisk and sensitivity to the availability offlood insurancefor the no cue and elevation frames, but not the othertwo, suggesting that mentioning global warmingsufficed to evoke a greater sense of risk.

In Study 1, political identify was implicit, andmanipulated indirectly through task framing. Study 2makes it explicit, by having participants identify them-selves as believers or nonbelievers before performingthe tasks.

3. Study 2—global warming questionbefore tasks

3.1.Methods3.1.1. Experimental proceduresThe procedure was the same as in Study 1, except thatparticipants were asked about global warming beliefsat the beginning, rather than the end. They were also

Table 1.ANOVAswith factors of Belief (believer versus non-believer), Framing (no cue, elevation, global warming, or eleva-tion + global warming), andAid (without Risk Finder versuswithRisk Finder).

Dependent variables

Source (a)Overall risk

(b) Insurance

availability

No priming

(Study 1)

Main effects

Belief (B) F= 1.68, η2=0.02 F= 0.21, η2= 0.00

Frame (F) F= 2.09, η2=0.01 F= 0.75, η2= 0.00

Aid (A) F= 7.88, η2=0.01* F= 0.05, η2= 0.00

Interaction effects

BXF F= 2.71, η2=0.01* F= 1.69, η2= 0.00

BXA F= 1.62, η2=0.00 F= 2.32, η2= 0.00

FXA F= 3.55, η2=0.01* F= 2.99, η2= 0.01*

BXFXA F= 1.48, η2=0.00 F= 0.11, η2= 0.00

Priming (Study 2)

Main effects

Belief (B) F= 3.51, η2=0.01 F= 0.05, η2= 0.00

Frame (F) F= 1.45, η2=0.02 F= 0.91, η2= 0.01

Aid (A) F= 15.01,

η2= 0.05*

F= 5.06, η2= 0.02*

Interaction effects

BXF F= 0.16, η2=0.00 F= 1.04, η2= 0.01

BXA F= 0.46, η2=0.00 F= 0.22, η2= 0.00

FXA F= 2.69, η2=0.03* F= 2.17, η2= 0.01

BXFXA F= 0.74, η2=0.01 F= 1.24, η2= 0.01

Note. The ANOVAs for the dependent variables controlled for the

list price of the home (as found on Zillow®), participants’ estimates

for the cost of flood insurance, and whether participants said that

they would buy the chosen home. F-values with * are significant

at p<0.05.

6

Environ. Res. Lett. 10 (2015) 034004 GWong-Parodi and B Fischhoff

asked to explain why they hold the beliefs they doabout global warming.

3.1.2.MeasuresThe same measures were used as in Study 1. Cron-bach’s α for the dependent variable, overall perceivedrisk, was 0.85.

3.1.3. ParticipantsWe followed the same power analysis as in Study 1,and again used MTurk. In order to secure roughlyequal numbers of believers and nonbelievers, we firstscreened 1342 participants for their attitudes on avariety of policy topics, including global warming.Based on their responses, we recruited 194 globalwarming ‘believers’ and 134 ‘nonbelievers’ for Study 2.We treated the screening test as priming that aspect oftheir identity. As in Study 1, nonbelievers weresignificantly more confident in their positions thanwere the believers, (M= 2.25 versus 1.55) p< 0.001, onthe 1–4 confidence scale. Based on self-reports,participants’ average age was 31.7 (SD= 10.2), with36.7% female, 82.5%White or Caucasian, 42.9% withat least a bachelor’s degree, and 43.9%with householdincome of at least $51K. In terms of party affiliation,34.9% reported being Democrats, 35.5% Indepen-dents, 23.5% Republicans and 6.1% Other or Prefernot to answer. About one-quarter reported having

lived near the coast (25.2%), 83.3% of whom reportedbeing familiar with it. Most had vacationed along thecoast (76.9%).More than one-third had experienced ahurricane (39.4%) andmost (52.9%) had experienceda flood or knew someone who had. As before, therewas a significant association between Belief (believer,nonbeliever) and political affiliation (Democrat,Republican), χ2(1,N= 191) = 89.33, p< 0.001. Believ-ers (87.0%) were more likely to be Democrats thanwere nonbelievers (18.4%).

3.2. Results3.2.1. Analysis planThe analyses followed those of Study 1. As before,participants’willingness to buy the home they selectedand estimates of flood insurance cost did not differ bycondition (H(7) = 6.04, p= 0.54; F(7327) = 0.50,p= 0.83, respectively), consistent with successfulrandomization.

3.2.2.Manipulation checkA paired-sample t-test found that participants weremuchmore willing to buy a homewhen insurance wasavailable (M= 4.17, SD= 1.73) than when it was not(M= 2.96, SD= 1.83), t(327) = 13.50, p< 0.001, 95%CI of the difference [1.03, 1.39], indicating an under-standing of the task. As in Study 1, there was asignificant main effect for Aid, such that sensitivity to

Figure 4.The effects of Framing (no cue, elevation, global warming, elevation + global warming) andAid (with Risk Finder versuswithout Risk Finder) on overall perceived risk. ANOVAs found significantly higher risk perceptionswith Risk Finder for the no cueand elevation frames, but no difference for the global warming and elevation + global warming frames, for both Study 1 and Study 2.See table S1 in supplementalmaterial for additional results.

7

Environ. Res. Lett. 10 (2015) 034004 GWong-Parodi and B Fischhoff

insurance availability was greater with Risk Finderthanwithout, in each frame.

3.2.3. AnalysesAs seen in table 1 (right-hand side), there was, again,no main effect for Framing and one for Aid. As inStudy 1, participants who used Risk Finder expressedgreater reluctance to move to Savannah. There was,again, a significant Framing x Aid interaction foroverall risk perception, such that having Risk Finderincreased risk judgments for participants receiving noframe or the elevation frame, but not for thosereceiving either frame mentioning global warming(figure 4). As a measure of the relative impact of thetwo manipulations, with Risk Finder, participants inthe no cue condition saw risks as great as didparticipants without Risk Finder in the elevation +global warming frame. Table S1 in supplementalmaterial presents additional results.

As in Study 1, there was no main effect for Beliefon overall perceived risk or on sensitivity to the avail-ability of flood insurance. Unlike Study 1, there was nointeraction between Belief and Framing, suggesting

that the initial question evoked political identitystrongly enough that the frames added nothing. Inorder to see whether the difference in results reflectedthe smaller sample in Study 2, we generated three ran-dom subsamples of Study 1 participants with Study 2’ssample size equivalent and still found Study 1’sinteraction.

3.3. SummaryStudy 2 manipulated political identity directly byasking about it before participants began their task.Nonetheless, participants in Study 2 generallyresponded similarly to those in Study 1. As in Study 1,using Risk Finder heightened perceived risk when theframe made no mention of global warming (no cue,elevation), but not when it did, to the extent that RiskFinder eliminated any effect of the frame.Unlike Study1, though, believers and nonbelievers respondedsimilarly throughout. Thus, asking participants toexpress their political identity appeared to allow themto focus on the decision-making task, in ways thatrevealed similar sensitivity to the varying detailsrelated to coastalflooding risks.

Figure 5.The effects of Framing (no cue, elevation, global warming, elevation + global warming) andBelief (believer versusnonbeliever) on judgments of overall perceived risk. ANOVAs found significantly higher risk perceptions for believers for theelevation + global warming frame, but no difference for other three frames, for Study 1 but not Study 2. See table S2 in supplementalmaterial for additional results.

8

Environ. Res. Lett. 10 (2015) 034004 GWong-Parodi and B Fischhoff

4.General discussion

Sound communications provide information relevantto recipients’ decisions. For people thinking aboutmoving to an area subject to coastal flooding, thatinformation includes the risk of flooding, capturedhere by the elevation of the home that they havechosen, and their financial exposure, captured here bythe availability of insurance. Global warming influ-ences those risks, as well as the value and cost ofinsurance.

In both of the present studies, participants respon-ded to such information in orderly ways. They sawgreater risk, as expressed in their reduced willingnessto move to a flood-prone area, when flood insurancewas unavailable, when global warming was men-tioned, and when they could learn more about flood-ing risks using the Risk Finder decision aid. Generallyspeaking, participants who identified themselves asbelievers and nonbelievers in global warming respon-ded similarly. The one exception was that, in Study 1,where belief was elicited at the end, nonbelieversreported less risk with the combined frame (eleva-tion + global warming). That difference effectively dis-appeared in Study 2, where participants expressedtheir beliefs about global warming before doingthe task.

Thus, it appears that even the simplest version ofour task (with no cues andwithout Risk Finder), parti-cipants were sufficiently engaged in the details of thedecision to make political identity largely irrelevant.Any remaining differences disappeared when partici-pants used Risk Finder, a decision aid designed [43] tomake decisions more personal, local, immediate andreal, thereby addressing the four dimensions of Con-strual Level Theory [35, 36]. These results contrastwith the polarized discussion regarding general issues(e.g., is global warming happening? are people respon-sible? shouldwe take political action?)

Drawing on self-affirmation [32] and reactancetheory [44], we speculate that the one differencebetween believers and nonbelievers (with the eleva-tion + global warming frame in Study 1) arose becausenonbelievers, without the immersion of Risk Finder,found it heavy handed—as though they were beingmanipulated into endorsing global warming as a pricefor acknowledging the importance of elevation. Thatinteraction vanished in Study 1, with the additionaldetail provided by Risk Finder, and in Study 2, wherethe legitimacy of participants’ identity was acknowl-edged by having them state their beliefs prior to per-forming the task. If so, then, rather than accentuatingmotivated reasoning, the opportunity to affirm theiridentity may have allowed participants to focus on thetask [29–32].

A topic for future study is hownonbelievers recon-ciled their general beliefs about global warming withthese specific decisions, enough to respond like believ-ers. We offer several possible explanations. One is that

the decision allowed nonbelievers to accept a risk ofglobal warming without having to acknowledge itsanthropogenic origin. A second is that politicaldebates, focusing on public policies (e.g., carbon tax,cap-and-trade), have only weak implications for per-sonal decisions. A third is that nonbelievers see theprobability of global warming as high enough to affecthome purchases, even if it is not high enough to affectpublic policies (which might also be opposed on othergrounds). A fourth is that any transient emotionsevoked by references to global warming dissipated asparticipants engaged in the concrete tasks. A fifth isthat some nonbelievers may actually believe the sci-ence, but not want to admit it publicly, in order toshow solidarity with political allies and avoid implicitendorsement of policies that they oppose (e.g., gov-ernment intervention) [45, 46]. They are free of thoseconstraints with personal decisions, especially whenmade in the privacy of an experiment. Fuller under-standing of these processes awaits additional studies,in which there may be special value for using realistic,engaging practical tasks such as those made possiblewith Zillow® and Risk Finder, both developed by sub-stantive experts with resources beyond those availablefor most studies of social and political identity [39–44, 47–49]. Looking at self-reports of actual decisions,Maibach et al (2009) found similar, greater energy-conservation behavior among Alarmed Americansand Dismissive Americans, with members of bothgroups paying more attention to these issues than domembers of the other four Americas (Concerned,Cautious, Disengaged, Doubtful) [38]. For commu-nicators interested in informed decision making, therecommendation emerging from these results is tofocus on facts that people need, while avoiding termsthat divide them.

Acknowledgments

This work was supported by a research grant from theRockefeller Foundation and by the Center for Climateand Energy Decision Making (NSF/SES-0949710).The authors thank Jack Wang and Dan Rizza for theirsupport with the experiment and Nichole Argo, SarahHailey and Ben Strauss for their feedback. Finally, theauthors thank Zillow® for permission to use their realestate search feature.

References

[1] National Climate Assessment 2014 2014National ClimateAssessment (Washington,DC:USGlobal Change ResearchProgram)

[2] CoumouDandRahmstorf S 2012A decade ofweatherextremesNat. Clim. Change 2 491–6

[3] Min SK, ZhangX, Zwiers FWandHegerl GC 2011Humancontribution tomore-intense precipitation extremesNature470 378–81

[4] Gillis J 2014 Freezing out the Bigger Picture (TheNewYorkTimes) February 11 2014D3

9

Environ. Res. Lett. 10 (2015) 034004 GWong-Parodi and B Fischhoff

[5] Grinsted A,Moore J C and Jevrejeva S 2013 ProjectedAtlantichurricane surge threat from rising temperatures Proc. NatlAcad. Sci. USA 110 5369–73

[6] HollandG andBruyère C L 2014Recent intense hurricaneresponse to global climate changeClim.Dyn. 42 617–27

[7] KimBM, Son SW,Min SK, Jeong JH, KimS J, ZhangX andYoon JH2014Weakening of the stratospheric polar vortex byArctic sea-ice lossNat. Commun. 5 4646

[8] Trenberth KE,Dai A, van der Schrier G, Jones PD,Barichivich J, Briffa KR and Sheffield J 2014Global warmingand changes in droughtNat. Clim. Change 4 17–22

[9] Fischhoff B 2013The sciences of science communication Proc.Natl Acad. Sci. USA 110 14033–9

[10] PidgeonNand Fischhoff B 2011The role of social and decisionsciences in communicating uncertain climate risksNat. Clim.Change 1 35–41

[11] OreskesN andConway EM2010Merchants of Doubt (NewYork: Bloomsbury)

[12] Hardisty D J, Johnson E J andWeber EU 2010Adirtyword ora dirty world? attribute framing, political affiliation, and querytheory Psychol. Sci. 21 86–92

[13] Hart P S andNisbet EC 2012 Boomerang effects in sciencecommunication howmotivated reasoning and identity cuesamplify opinion polarization about climatemitigation policiesCommun. Res. 39 701–23

[14] NisbetMC2009Communicating climate change: why framesmatter for public engagement Environ.: Sci. Policy SustainableDev. 51 12–23

[15] KahanDM,Peters E,WittlinM, Slovic P,Ouellette L L,BramanD andMandel G 2012The polarizing impact ofscience literacy and numeracy on perceived climate changerisksNat. Clim. Change 2 732–5

[16] Myers TA,Maibach EW,Roser-Renouf C, Akerlof K andLeiserowitz AA 2012The relationship between personalexperience and belief in the reality of global warmingNat.Clim. Change 3 343–7

[17] Weber E 2013 Seeing is believingNat. Clim. Change 3 312–3[18] NickersonR S 1998Confirmation bias: a ubiquitous phenom-

enon inmany guisesRev. Gen. Psychol. 2 175[19] Fiske S T andNeuberg S L 1990A continuumof impression

formation, from category-based to individuating processes:influences of information andmotivation on attention andinterpretationAdv. Exp. Soc. Psychol. 23 1–74

[20] Kunda Z 1990The case formotivated reasoning Psychol. Bull.108 480–98

[21] Lundgren S R and Prislin R 1998Motivated cognitiveprocessing and attitude changePersonality Soc. Psychol. Bull.24 715–26

[22] Taber C S and LodgeM2006Motivated skepticism in theevaluation of political beliefsAm. J. Political Sci. 50 755–69

[23] CostaD L andKahnME2013 Energy conservation ‘nudges’and environmentalist ideology: evidence from a randomizedresidential electricity field experiment J. Eur. Econ. Assoc. 11680–702

[24] CostaD L andKahnME2010 Energy conservation nudgesand environmentalist ideology: evidence from a randomizedresidential electricity field experimentWorking paper No.15930 (Cambridge,MA:National Bureau of EconomicResearch) (April)

[25] Schuldt J P, Konrath SH and SchwarzN 2011 ‘Globalwarming’ or ‘climate change’? whether the planet is warmingdepends on questionwording Public Opin. Q. 75 115–214

[26] Tajfel H 1978Differentiation Between Social Groups (London,UK: Academic)

[27] Stets J E andBurke P J 2000 Identity theory and social identitytheory Soc. Psychol. Q. 63 224–37

[28] Steele CM1988The psychology of self-affirmation: sustainingthe integrity of the selfAdv. Exp. Soc. Psychol. 21 261–302

[29] CohenGL, Aronson J and Steele CM2000When beliefs yieldto evidence: reducing biased evaluation by affirming the selfPersonality Soc. Psychol. Bull. 26 1151–64

[30] Crocker J, Niiya Y andMischkowski D 2008Why does writingabout important values reduce defensiveness? self-affirmationand the role of positive other-directed feelings Psychol. Sci. 19740–7

[31] NyhanB andReifler J Blank slates or closedminds? the role ofinformation deficits and identity in the prevalence ofmis-perceptionsAm. J. Political Sci. in press

[32] ReedMBandAspinwall LG 1998 Self-affirmation reducesbiased processing of health-risk informationMotivation Emo-tion 22 99–132

[33] Fiske S T, LinMandNeuberg S L 1999The continuummodelDual-Process Theories in Social Psychology ed SChaiken andYTrope (NewYork: Guilford Press) pp 321–254

[34] Petty RE andCacioppo J T 1984 Source factors and theelaboration likelihoodmodel of persuasionAdv. Consum. Res.11 668–72

[35] Trope Y and LibermanN2003Temporal construalPsychol.Rev. 110 403–21

[36] Trope Y and LibermanN2010Construal-level theory ofpsychological distance Psychol. Rev. 117 440–63

[37] Förster J, FriedmanR S and LibermanN2004Temporalconstrual effects on abstract and concrete thinking: conse-quences for insight and creative cognition J. Personality Soc.Psychol. 87 177

[38] Maibach E, Roser-Renouf C and Leiserowitz A 2009GlobalWarming’s Six Americas 2009: anAudience SegmentationAnalysis (NewHaven, CT: YaleUniversity)

[39] BuhrmesterM,KwangT andGosling SD 2011Amazon’smechanical turk a new source of inexpensive, yet high-quality,data? Perspect. Psychol. Sci. 6 3–5

[40] CrumpM J,McDonnell J V andGureckis TM2013 Evaluatingamazon’smechanical turk as a tool for experimental beha-vioral researchPloSOne 8 e57410

[41] Faul F, Erdelder E, Buchner A and LangA-G 2009 Statisticalpower analyses usingG*Power 3.1: tests for correlation andregression analysesBehav. Res.Methods 41 1149–60

[42] Leiserowitz A,Maibach E, Roser-Renouf C and SmithN2011GlobalWarming’s Six Americas,May 2011 (YaleUniversity andGeorgeMasonUniversity)

[43] Wong-Parodi G, Fischhoff B and Strauss B 2014Amethod toevaluate the usability of interactive climate change impactdecision aidsClim. Change 126 485–93

[44] Brehm JW1966ATheory of Psychological Reactance (NewYork: Academic)

[45] Swim JK (ed) 2011 Special issue: psychology and globalclimate changeAm. Psychol. 86

[46] HulmeM2010WhyWeDisagree About Climate Change(Cambridge: CambridgeUniversity Press)BenjaminD J, Choi J J and StricklandA J 2007 Social identityand preferencesAm. Econ. Rev. 100 1913–28

[47] ForehandMR,Deshpandé R andReedA II 2002 Identitysalience and the influence of differential activation of the socialself-schema on advertising response J. Appl. Psychol. 87 1086

[48] HalloranM J andKashima E S 2004 Social identity andworldview validation: the effects of in-group identity primesandmortality salience on value endorsement Personality Soc.Psychol. Bull. 30 915–25

[49] Hertel G andKerrN L 2001 Priming in-group favoritism: theimpact of normative scripts in theminimal group paradigmJ. Exp. Soc. Psychol. 37 316–24

10

Environ. Res. Lett. 10 (2015) 034004 GWong-Parodi and B Fischhoff