Improving Climate Change Acceptance Among U.S. Conservatives Through Value-Based Message Targeting

Abstract

Although prior research has identified communication strategies for reducing climate change skepticism, recent research suggests such approaches can backfire. To explore this issue, we report on a preliminary study investigating two prominent messaging styles: consensus and targeted messages. While consensus messaging did not produce significant effects, targeted messages emphasizing free market solutions to climate change were effective at improving conservatives’ climate change acceptance. Furthermore, the inclusion of consensus information did not improve or diminish the effects of the targeted messages. These findings underscore the utility of targeted messaging in improving climate change acceptance among political conservatives.

Keywords

climate change message targeting consensus messaging climate skepticism

Introduction

Climate change is considered a major global threat, with organizations such as the Intergovernmental Panel on Climate Change (IPCC) and American Association for the Advancement of Science (AAAS) calling for immediate action (AAAS, 2006; IPCC, 2014). Findings from the IPCC’s 2014 report confirm the scientific consensus that the earth’s atmosphere and oceans have warmed since 1950 and that human activity is a main contributor. However, many Americans, in particular conservatives, are skeptical of human-caused climate change. Recent polling data from Pew Research, for example, show that belief about climate change varies considerably by political ideology, with only 29% of U.S. conservatives believing climate change is human-caused (Funk & Rainie, 2015). In fact, most conservatives (e.g., 39%) believe climate change is not happening, whereas a majority of moderates (e.g., 56%) and liberals (e.g., 76%) believe climate change is caused by humans. Such widespread skepticism threatens the adoption of policy initiatives and societal changes that are necessary for addressing the risks associated with climate change (McCright & Dunlap, 2003; Nisbet, 2009). Thus, there is a critical need to address climate change skepticism among those who identify themselves as conservatives on an ideology scale.

Communication plays a key role in cultivating greater public understanding of climate change. Indeed, decades of communication research has provided important guidance on how to address climate change skepticism (Nisbet, 2009). However, recent findings on common messaging practices, such as framing and consensus messaging, show that persuasive messages might backfire for key audiences, thus increasing climate change skepticism (Cook & Lewandowsky, 2016; Hart & Nisbet, 2012; Zhou, 2016). In addressing these concerns, we focus on two communication strategies that have garnered recent interest among scholars—consensus messaging and message targeting—and explore their effectiveness at improving climate change acceptance among conservative audiences.

Consensus messaging emphasizes the importance of conveying scientists’ beliefs about issues such as climate change to the public. The idea is that learning about the scientific consensus relative to important scientific and environmental issues could shift people’s beliefs to be more in line with those held by scientists. Indeed, a number of recent studies have shown that providing people information relative to the scientific consensus on climate change not only raise consensus estimates (Myers, Maibach, Peters, & Leiserowitz, 2015; van der Linden, Leiserowitz, Feinberg, & Maibach, 2014) but can also indirectly affect beliefs, risk perception, and mitigation support (van der Linden, Leiserowitz, Feinberg, & Maibach, 2015). Moreover, consensus messaging has been found to neutralize the effect of worldviews, such that it increases climate change acceptance among conservatives and free market supporters (Cook, 2016; Kotcher, Meyers, Maibach, & Leiserowitz, 2014; Lewandowsky, Gignac, & Vaughan, 2013)

Other scholars have noted the importance of taking into account the existing views of the public and creating messages that resonate with their existing beliefs. Research has noted that persuasive messages tied to controversial topics, such as climate change, sometimes result in boomerang effects whereby an effect occurs opposite to the strategic intent of the message (Byrne & Hart, 2009). As a result, messaging strategies intended to reduce climate change skepticism may actually increase it for key audiences (Cook & Lewandowsky, 2016; Hart & Nisbet, 2012; Zhou, 2016). These effects occur because people process new information through the lens of their political, cultural, and value predispositions. Thus, information inconsistent with a person’s prior views is typically processed in a manner that reinforces rather than refutes their prior views, resulting in motivated reasoning (Lodge & Taber, 2013). For example, recent research suggests messages encouraging support of governmental intervention or personal engagement against climate change increase Republicans’ opposition to government and personal climate change mitigation (Zhou, 2016). Similarly, conservatives exposed to message frames focusing on socially distant victims of climate change reported lower support for climate change mitigation policies when compared to conservatives receiving no message at all (Hart & Nisbet, 2012). Last, there is even new evidence suggesting that consensus messaging directed at American audiences result in more skeptical climate change views for individuals with strong free market beliefs (Cook & Lewandowsky, 2016).

With concerns tied to biased processing of information and boomerang effects, scholars have noted the importance of considering the underlying values of the skeptical audience to increase the effectiveness of climate change messages. As a result, messages targeted to resonate with conservatives’ values could eliminate boomerang effects, and more important, reduce climate change skepticism. Targeting involves shaping messages to reflect the common characteristics, such as political ideology, of a target subgroup (Schmid, Rivers, Latimer, & Salovey, 2008). Doing so can make a message effective where nontargeted interventions fail. Indeed, we recognize and draw on recent research that has found messages framed in terms of interpersonal warmth, purity, and societal development to be effective at shifting conservatives’ environmental attitudes and behaviors (Bain, Hornsey, Bongiorno, & Jeffries, 2012; Feinberg & Willer, 2013). Moreover, messaging that highlights the free market benefits to climate change action also appears effective at reducing skepticism among conservatives (Campbell & Kay, 2014).

With inconsistent findings occurring with climate change messaging, we seek to extend as well as replicate previous research. First, in replicating past findings (e.g., Campbell & Kay, 2014), we examine whether targeted messages designed to resonate with conservative Americans shift their scientific beliefs on climate change. Specifically, we use two conservative values that have been identified as factors in climate change acceptance (Campbell & Kay, 2014; Heath & Gifford, 2006; McCright & Dunlap, 2011; Nisbet, 2009)—religion and free market ideology—and investigate whether statements targeted around these ideas can persuade conservatives that climate change is happening and human caused. In extending this previous work, we examine how targeted messaging strategies compare with consensus messaging strategies in terms of their effects on conservative audiences and also consider whether the inclusion of a consensus statement enhances the effectiveness of a targeted message for political conservatives. Thus, our study represents an important test of two prominent climate change messaging strategies that can provide further insight on climate change communication best practices.

Method

Participant Sample

Data for this study were collected via the Internet from August 18th to August 20th, 2016. Survey Sampling International (SSI), an online survey panel provider, sent our Qualtrics experiment link to members of their online participant pool. Panel participants are generally recruited via e-mail and receive monetary compensation for participating in studies (see http://www.surveysampling.com/ for more information on SSI panels). A total of 660 respondents completed the experiment. Though SSI does not obtain a random sample of respondents, it is heterogeneous and tends to be representative of the general population in terms of basic demographics such as age, race, and gender. Our sample is slightly older than the U.S. population (age M = 45.37 years, SD = 15.42), slightly less White (61.8% White), and includes more men than the general population (57% male). Quota samples have become more popular in recent years as the public shifts from relying on landlines to cell phones, which has reduced researchers’ ability to collect data using random digit dialing to reach a probability sample (Ansolabehere & Schaffner, 2014). Although some evidence suggests that using quota groups provide higher levels of error compared with traditional probability methods (Yeager et al., 2011), there is evidence that using these types of samples makes little difference in terms of response quality (Ansolabehere & Schaffner, 2014). Moreover, this article is testing a theoretical process, which means it is more interested in making process inferences rather than population inferences (Hayes, 2005).

Manipulation

For our study, we conducted an experiment that varied whether respondents (N = 660) saw a consensus statement, a testimonial from a climate change scientist (targeted-religious, targeted-free market, nontargeted, or no testimonial), or the combination these two factors. Thus, our experiment involved a between-groups 2 (consensus statement vs. no statement) × 4 (targeted-religious, targeted-free market, nontargeted, or no testimonial) factorial design. There was generally equal distribution of respondents across our eight conditions (ranged from 81 to 84 respondents per condition). Before reading our messages, respondents first filled out a series of pretest questions that included demographics (e.g., age, race, gender) and a measure of political ideology.

After filling out the pretest battery of questions, respondents were placed into one of our eight experimental conditions.¹ The consensus message simply stated, “Did you know? 97% of climate scientists have concluded that human-caused climate change is happening.” This is nearly identical to the statements that have been used in previous work examining the effects of consensus messages (van der Linden et al., 2015). We then included (or excluded) this statement in our climate change testimonials. Our fictional climate scientist, Dr. James Hoffmann from the University of Arizona, presented the testimonials. We included a head shot of Dr. Hoffmann as part of the testimonial. The generic/nontargeted testimonial simply focused on how future generations may find innovative solutions to address the issue of climate change. The testimonials aimed at conservatives focused on religious and free market solutions to climate change. The religious testimonial emphasized Dr. Hoffman’s support for a “Christ-centered, evangelical” solution to the issue of climate change. To create this testimonial, we used information about religious solutions for climate change from the organization Christians and Climate (www.christiansandclimate.org). The testimonial provided people with an example of how religious organizations have worked with local businesses and government to develop solutions to this problem. Finally, the free market testimonial emphasized Dr. Hoffmann’s support for a free market solution to climate change. To create this testimonial, we included content from speeches given to conservative audiences by conservative climate scientist, Dr. Katharine Hayhoe. For instance, the testimonial provided an example of how deregulation of the energy market has led to an increased supply of renewable energy in Arizona. The testimonials were similar in length ranging from 130 words for the nontargeted testimonial to 148 for the free market testimonial. After reading the message, respondents then filled out a battery of posttest questions. These questions included our measure of climate change beliefs.

Manipulation Check

To ensure that respondents understood the emphasis of our targeted messages, we included items asking respondents the extent to which they agreed that their message included religious and free market-based solutions to climate change. Results confirm a successful manipulation of our targeted messages and confirm that our nontargeted messages were not perceived to include religious² or free market-based solutions.³

Pretest Measure Variable

Political Ideology

Ideology was assessed by asking participants to respond to the statement, “I would describe my political views as . . .” using a scale of “very conservative” (1) to “very liberal” (7), M = 3.85, SD = 1.79.

Posttest Measure Variable

Climate Change Beliefs

Climate change beliefs were measured with six items adapted from previous research (Cook & Lewandowsky, 2016; Deryugina & Shurchkov, 2016). Specifically, respondents were asked using a 6-point Likert-type scale (1= strongly disagree, 6 = strongly agree) to what extent they agreed/disagreed with the following statements: (1) The climate is always changing and what we are currently observing is just natural fluctuation (reverse coded); (2) The burning of fossil fuels over the past 50 years has caused serious damage to the planet’s climate; (3) Human CO₂ emissions cause climate change; (4) Humans are too insignificant to have an appreciable impact on global temperature (reverse coded); (5) Climate change is a process that is already underway; (6) Climate change is not happening (reverse coded). We achieved acceptable reliability, M = 4.20, SD = 0.97, α = .73, and averaged the six items together to create a single measure of climate change beliefs.

Analysis

Our interests were to examine how our various message types (testimonials vs. consensus messaging) affected political conservatives. Thus, our main results involve examination of whether the effects of these different message conditions vary by political ideology. In doing so, we examined the effects of the message conditions on climate change beliefs with political ideology situated as a moderator using the RLM macro developed by Darlington and Hayes (2017). This macro allows scholars to use heteroscedasticity-consistent standard error estimators to reduce the effects of heteroscedasticity on inference for regression analysis (Darlington & Hayes, 2017; Hayes & Cai, 2007). The message testimonial groups were dummy coded with the “no testimonial group” assigned the reference category. The consensus message factor was coded (1 = consensus, 0 = no consensus). We then ran hierarchical multiple regression models exploring the interaction effects of political ideology and the message factors.

Results

We first examine the main effects and interaction effects of our message factors without consideration of political ideology. The main effects for consensus messages, the testimonials, and their interaction were nonsignificant (see Table 1, Model 1 and Model 2). However, our main interests involved whether the effects of these different message conditions vary by political ideology. For the consensus messages, there was a nonsignificant interaction with political ideology, b = .01, p = .75 (see Table 1, Model 3). Although the consensus messages were ineffective at influencing beliefs about climate change, it is notable that a backfiring effect among conservatives was not observed (see Table 2). On the other hand, our results found a significant free market messages by political ideology interaction, b = −.13, p = .02. In contrast, we did not find that the religious messages, b = −.08, p = .22, or nontargeted messages varied by political ideology, b = −.07, p = .24 (see Table 1, Model 3). Probing the free market messages by political ideology interaction showed that this targeted message was only effective for conservatives (i.e., those identifying themselves as “conservative” and “very conservative”; see Table 3 and Figure 1).⁴

Table 1.

Regression With Climate Change Beliefs as Dependent Variable.

	Model 1	Model 2	Model 3	Model 4
	B (SE)	B (SE)	B (SE)	B (SE)
Constant	4.09 (.09)***	4.1 (.11)***	3.28 (.16)***	3.11 (.16)***
Consensus (1 = yes; 0 = no)	.11 (.08)	.1 (.15)	−.05 (.21)	.3 (.32)
Nontargeted (1 = yes; 0 = no)	.11 (.11)	.11 (.14)	.36 (.23)	.68 (.26)*
Free market (1 = yes; 0 = no)	.04 (.11)	.11 (.14)	.57 (.24)*	.55 (.29)
Religious (1 = yes; 0 = no)	.06 (.11)	−.02 (.16)	.21 (.29)	.59 (.38)
Consensus * nontargeted		.002 (.21)	.04 (.2)	−.63 (.47)
Consensus * free market		−.13 (.21)	−.05 (.2)	−.02 (.47)
Consensus * religious		.17 (.22)	.3 (.21)	−.49 (.53)
Political ideology			.22 (.04)***	.27 (.04)***
Consensus * political ideology			.01 (.04)	−.08 (.08)
Nontargeted * political ideology			−.07 (.06)	−.15 (.07)*
Free market * political ideology			−.13 (.06)*	−.13 (.07)
Religious * political ideology			−.08 (.06)	−.18 (.09)*
Consensus * nontargeted * political ideology				.17 (.11)
Consensus * free market * political ideology				−.01 (.11)
Consensus * religious * political ideology				.2 (.12)
R ²	.005	.008	.107	.115
F	F(4,653) = 0.73	F(7,650) = 0.65	F(12,645) = 6.23***	F(15,642) = 6.32***

Note. Testimonial factors were dummy coded with “no testimonial” as the reference category. Heteroscedasticity consistent standard errors (H3) enabled. Unstandardized coefficients reported.

p < .05. **p < .01. ***p < .001.

Table 2.

Effect of Consensus Messages by Political Ideology (With Standard Errors).

Moderator value	Effect on climate change beliefs	p
Very conservative	−.04 (.18)	.84
Conservative	−.02 (.16)	.88
Slightly conservative	−.01 (.14)	.95
Moderate	.004 (.14)	.98
Slightly liberal	.02 (.15)	.91
Liberal	.03 (.17)	.86
Very liberal	.05 (.19)	.82

Note. Heteroscedasticity consistent standard errors enabled (H3) in all models. Probed from Model 3 in Table 1 with PROCESS macro (Hayes, 2013).

Table 3.

Effect of Free Market Messages by Political Ideology (With Standard Errors).

Moderator value	Effect on climate change beliefs	p
Very conservative	.44 (.19)	.02
Conservative	.31 (.15)	.047
Slightly conservative	.18 (.13)	.18
Moderate	.05 (.13)	.72
Slightly liberal	−.08 (.16)	.59
Liberal	−.21 (.19)	.27
Very liberal	−.34 (.24)	.15

Note. Heteroscedasticity consistent standard errors enabled (H3) in all models. Probed from Model 3 in Table 1 with PROCESS macro (Hayes, 2013).

Figure 1.

Estimated means of climate change beliefs by testimonial factor and political ideology. Probed from Model 3 in Table 1 using PROCESS macro (Hayes, 2013).

We also investigated whether the effectiveness of our targeted testimonials to reach conservatives depended on the presence of a consensus statement. Overall, our regression model reports null effects across all our three-way interaction terms (see Table 1, Model 4). These results show that the presence of a consensus message in a free market testimonial does not appear to make a difference in its effect on climate change beliefs.

Based on our findings, highlighting free market solutions to climate change, with or without a consensus statement, appears effective in shifting beliefs about climate change among conservatives (see Figures 2 and 3).

Figure 2.

Estimated means of climate change beliefs for free market and no testimonial conditions containing a consensus statement. Probed from Model 3 in Table 1 using PROCESS macro (Hayes, 2013).

Figure 3.

Estimated means of climate change beliefs for free market and no testimonial conditions without a consensus statement. Probed from Model 3 in Table 1 using PROCESS macro (Hayes, 2013).

Discussion

Our findings provide direction for addressing climate change skepticism among U.S. conservatives. First, we did not find that consensus statements affected people’s beliefs about climate change; nor were there any effects of consensus messages across political ideologies. In addition, we did not find that adding a consensus statement enhanced the effects of our targeted messages. One reason for the lack of findings relative to consensus messages is the dependent variable of interest. Indeed, many of the consensus messaging studies have shown effects flowing through consensus estimates (e.g., van der Linden et al., 2015), which we do not examine in this particular article. Because of that, including a consensus statement in a targeted message might matter more when the goal is to raise consensus estimates. We encourage further research in this area.

Second, we provide an important replication of previous work that adds greater credence to the utility of targeted climate change messaging. Consistent with the findings from Campbell and Kay (2014), emphasizing free market solutions to climate change might be a particularly effective messaging strategy for generating more favorable beliefs among conservative Americans. However, for conservatives it appears that only free market targeting, rather than religious targeting, affects beliefs. It could be that religious targeting did not adequately address factors that predict conservative skepticism toward climate change. For instance, conservative politicization of climate change has occurred largely because of fears that climate mitigation policies would hamper free market commerce due to greater government oversight and regulations (McCright & Dunlap, 2011). Indeed, cross-sectional survey research suggests free market ideology is strongly correlated with climate change skepticism (Heath & Gifford, 2006). Therefore, emphasizing free market solutions to climate change might alleviate core concerns among conservatives, which could make a climate change message more effective among these individuals.

We also recognize that more positive changes in climate change beliefs typically associate with other outcomes, including policy support, and risk perception (see van der Linden et al., 2015). Therefore, we encourage future research to explore how free market climate change messaging impacts additional outcomes, and whether changes in beliefs may only translate into support for policy that adheres to a free market perspective rather than support for a wider range of policies.

Given its ubiquity and capability for targeted messaging, social media sites such as Facebook could be a useful platform to ensure that targeted messages reach only their target audiences. However, we recognize that the general public views targeted political messaging with discomfort, and scholars fear that long-term reliance on targeting could erode public trust in political candidates (Turow, Delli Carpini, Draper & Howard-Williams, 2012). By the same token, long-term reliance on targeted climate change messaging could negatively impact trust on scientific institutions and authorities involved in strategic messaging. Therefore, it is important to consider that targeting cannot and should not be the sole method for addressing conservative climate change skepticism. Efforts made to build trust in science and its institutions via public science engagement should also be used in concert with strategic messaging.

In sum, this research demonstrates that improving climate change acceptance should involve careful consideration of the target audience’s values. Specifically, messages that reflect on the core reasons behind conservative climate change skepticism (i.e., free market ideology) might overcome the persuasion barriers documented in previous climate change communication research. We hope that our findings can spark further research with additional attention to innovative message dissemination strategies.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Graham Dixon (PhD, Cornell University) is an assistant professor in the School of Communication at The Ohio State University. His research interests include science and risk communication.

Jay Hmielowski (PhD, The Ohio State University) is an assistant professor in the Edward R. Murrow College of Communication at Washington State University. His research interests include political and environmental communication.

Yanni Ma is a PhD student in the Edward R. Murrow College of Communication at Washington State University. Her interests include science, environmental, and risk communication.

References

American Association for the Advancement of Science. (2006). AAAS board statement on climate change. Retrieved from https://www.aaas.org/sites/default/files/migrate/uploads/aaas_climate_statement2.pdf

Ansolabehere

Schaffner

B. F.

(2014). Does survey mode still matter? Findings from a 2010 multi-mode comparison. Political Analysis, 22, 285-303.

Bain

Hornsey

Bongiorno

Jeffries

(2012). Promoting pro-environmental action in climate change deniers. Nature Climate Change, 2, 600-603.

Byrne

Hart

P. S.

(2009). The “boomerang” effect: A synthesis of findings and a preliminary theoretical framework. Communication Yearbook, 33, 3-37

Campbell

Kay

(2014). Solution aversion: On the relation between ideology and motivated disbelief. Journal of Personality and Social Psychology, 107, 809-824.

Cook

(2016). Countering climate science denial and communicating scientific consensus. In Nisbet

(Ed.), Oxford encyclopedia of climate change communication. London, England: Oxford University Press. doi:10.1093/acrefore/9780190228620.013.314

Cook

Lewandowsky

(2016). Rational irrationality: Modeling climate change belief polarization using Bayesian networks. Topics in Cognitive Science, 8, 160-179.

Darlington

R. B.

Hayes

A. F

. (2017). Regression analysis and linear models: Concepts, application, and implementation. New York, NY: Guilford Press.

Deryugina

Shurchkov

(2016). The effect of information provision on public consensus about climate change. Plos One, 11, e0151469.

10.

Feinberg

Willer

(2013). The moral roots of environmental attitudes. Psychological Science, 24, 56-62.

11.

Funk

Rainie

(2015). Chapter 2: Climate change and energy issues. Americans, Politics and Scientific Issues. Retrieved from http://www.pewinternet.org/2015/07/01/chapter-2-climate-change-and-energy-issues/

12.

Hart

P. S.

Nisbet

E. C.

(2012). Boomerang effects in science communication: How motivated reasoning and identity cues amplify opinion polarization about climate mitigation policies. Communication Research, 39, 701-723.

13.

Hayes

A. F.

(2005). Statistical methods for communication science. New York, NY: Routledge.

14.

Hayes

A. F.

(2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guilford Press.

15.

Hayes

A. F.

Cai

(2007). Using heteroskedasticity-consistent standard error estimators in OLS regression: An introduction and software implementation. Behavior Research Methods, 39, 709-722.

16.

Heath

Gifford

(2006). Free-market ideology and environmental degradation the case of belief in global climate change. Environment and Behavior, 38, 48-71.

17.

Intergovernmental Panel on Climate Change. (2014). Climate change 2014 synthesis report. Retrieved from https://www.ipcc.ch/pdf/assessment-report/ar5/syr/SYR_AR5_FINAL_full_wcover.pdf

18.

Kotcher

Meyers

Maibach

Leiserowitz

(2014, May). Correcting misperceptions about the scientific consensus on climate change: Exploring the role of providing an explanation for the erroneous belief. Paper presented at 2014 annual conference of the International Communication Association, Seattle, WA.

19.

Lewandowsky

Gignac

G. E.

Vaughan

(2013). The pivotal role of perceived scientific consensus in acceptance of science. Nature Climate Change, 3, 399-404.

20.

Lodge

Taber

C. S.

(2013). The rationalizing voter. Cambridge, England: Cambridge University Press.

21.

McCright

Dunlap

(2003). Defeating Kyoto: The conservative movement’s impact on U.S. climate change policy. Social Problems, 50, 348-373.

22.

McCright

Dunlap

(2011). The politicization of climate change and polarization in the American public’s views of global warming, 2001-2010. Sociological Quarterly, 52, 155-194.

23.

Myers

T. A.

Maibach

Peters

Leiserowitz

(2015). Simple messages help set the record straight about scientific agreement on human-caused climate change: The results of two experiments. PloS One, 10, e0120985.

24.

Nisbet

M. C.

(2009). Communicating climate change: Why frames matter for public engagement. Environment: Science and Policy for Sustainable Development, 51, 12-23.

25.

Schmid

K. L.

Rivers

S. E.

Latimer

A. E.

Salovey

(2008). Targeting or tailoring? Maximizing resources to create effective health communications. Marketing Health Services, 28, 32-37.

26.

Turow

Delli Carpini

Draper

Howard-Williams

(2012). Americans roundly reject targeted political advertising (Annenberg School for Communication Department Papers). Retrieved from http://repository.upenn.edu/cgi/viewcontent.cgi?article=1414&context=asc_papers

27.

van der Linden

S. L.

Leiserowitz

A. A.

Feinberg

G. D.

Maibach

E. W

. (2014). How to communicate the scientific consensus on climate change: Plain facts, pie charts or metaphors? Climatic Change, 126, 255-262.

28.

van der Linden

S. L.

Leiserowitz

A. A.

Feinberg

G. D.

Maibach

E. W

. (2015). The scientific consensus on climate change as a gateway belief: Experimental evidence. Plos One, 10, e0118489.

29.

Yeager

D. S.

Krosnick

J. A.

Chang

Javitz

Levendusky

M. S.

Simpser

Wang

(2011). Comparing the accuracy of RDD telephone surveys and internet surveys conducted with probability and non-probability samples. Public Opinion Quarterly, 75, 709-747.

30.

Zhou

(2016). Boomerangs versus javelins: How polarization constrains communication on climate change. Environmental Politics, 25, 788-811.