The Partisan Brain

Abstract

There has been deepening concern about political polarization in public attitudes toward the scientific community. The “intrinsic thesis” attributes this polarization to psychological deficiencies among conservatives as compared to liberals. The “contextual thesis” makes no such claims about inherent psychological differences between conservatives and liberals, but rather points to interacting institutional and psychological factors as the forces driving polarization. We evaluate the evidence for both theses in the context of developing and testing a theoretical model of audience response to dissonant science communication. Conducting a national online experiment (N = 1,500), we examined audience reactions to both conservative-dissonant and liberal-dissonant science messages and consequences for trust in the scientific community. Our results suggest liberals and conservatives alike react negatively to dissonant science communication, resulting in diminished trust of the scientific community. We discuss how our findings link to the larger debate about political polarization of science and implications for science communicators.

Keywords

political polarization science communication motivated reasoning institutional trust affect persuasion

The scientific community remains one of the most trusted institutions in the United States (National Science Board 2014), yet over the past 30 years public confidence in science has become increasingly politically polarized as liberal and conservative attitudes toward science have diverged (Gauchat 2012; Mooney 2005, 2012). Political polarization in trust threatens the scientific community’s ability to provide evidence as the basis for policy consensus, as well as its longstanding cultural legitimacy; but explanations for this phenomenon are highly contested (Gauchat 2012; Nelkin 1975, 1992).

Two competing sets of explanations have emerged for the political polarization in attitudes about science. One set of explanations, what we term the intrinsic thesis, argues that political polarization about science is due to fundamental psychological differences between political conservatives and liberals, asserting that conservatives are consistently more likely to reject such information and/or to be distrustful of the scientific community (Mooney 2005, 2012; see also discussion by Kahan 2013). In contrast, the contextual thesis asserts that ideological differences in the public’s scientific denialism and distrust are a consequence of which science policy issues are most salient in political and public discourse, often driven by institutional actors and political entrepreneurs (Hiltgartner and Bosk 1988; M. Nisbet and Huge 2006). The contextual thesis asserts that liberals are as likely as conservatives to engage in motivated reasoning and biased processing when they are exposed to ideologically “dissonant” scientific evidence (Braman et al. 2012; Kahan 2013).

Dissonant evidence is information that challenges one’s ideological worldview or set of cultural values (Kahan and Braman 2006; Kahan et al. 2012; M. Nisbet 2005; M. Nisbet and Markowitz 2014). Claims about the benefits of embryonic stem cell research, for example, would be dissonant for social conservatives because such claims challenge beliefs about life beginning at inception. The contextual thesis posits that evidence about the advantages of nuclear power over fossil fuels for climate change mitigation, for example, would be dissonant for liberals because those claims challenge their beliefs about environmental protection and would prompt them to engage in motivated reasoning

To advance the debate over science communication and political ideology, we explicate the psychological processes by which science communication about controversial topics influences trust in the scientific community. Drawing from the fields of communication and persuasion, we evaluate how conservatives and liberals react when exposed to ideologically dissonant science communication and what this means for institutional trust in the scientific community. Results show that both political conservatives and liberals exhibit negative affective and cognitive reactions to ideologically dissonant evidence from the scientific community. As a consequence, although political conservatives currently exhibit a lower base-level of trust in the scientific community than do liberals (Gauchat 2012), the trust of both liberals and conservatives varies significantly when presented with ideologically dissonant science messages. We argue that these findings are consistent with the contextual thesis and that they provide insight into the psychological basis for the deepening divide in trust of the scientific community.

Motivated Reasoning and Political Polarization about Science

There are two competing explanations for political polarization about science: both propositions draw on the concept of motivated reasoning but differ in their interpretation and application of this concept (Kunda 1990; Lodge and Taber 2000; Lord, Ross, and Lepper 1979; Taber, Cann, and Kucsova 2009). Motivated reasoning is the desire to reach conclusions consistent with previously held beliefs and it leads to biased information processing (Kunda 1990). People do not approach evidence and arguments about controversial issues in a purely rational, even-handed manner (Lodge and Taber 2000). Instead, an individual’s prior beliefs and political ideology strongly bias how he or she responds to these arguments through selective exposure, attention, comprehension, and recall (Kunda 1990; Lodge and Taber 2000; Lord, Ross, and Lepper 1979; Taber and Lodge 2006; Taber, Cann, and Kucsova 2009). For example, those with greater issue involvement or with strongly held opinions are less likely to modify their beliefs when confronted with new information and so will frequently ignore and misinterpret ideologically incongruent arguments (Johnson and Eagly 1989). Others may respond to counterattitudinal or ideologically dissonant messages with source derogation, counterarguing, reactance, or negative affect (Byrne and Hart 2009; Jacks and Cameron 2003; Lewandowsky et al. 2012).

Therefore, equally informed citizens may employ scientific evidence in very different ways depending on their ideological orientation. This type of motivated processing is most likely to occur among those with greater cognitive sophistication (Braman et al. 2012; Hart and E. Nisbet 2012; Kahan 2013; Scheufele 2006). For example, ideological orientations may bias the interpretation of information about climate change, even possibly resulting in opinion or belief outcomes that were unintended by the message creators (a boomerang effect, see Hart and E. Nisbet 2012; Nyhan and Reifler 2010). As a result, those with differing ideologies become more polarized in their beliefs upon being exposed to the same scientific evidence, creating a differentiation in belief accuracy between ideological groups, or a belief gap (Hart and E. Nisbet 2012; Hindman 2009; Kahan 2013; E. Nisbet, Cooper, and Ellithorpe, forthcoming; Taber, Cann, and Kucsova 2009).

Where the intrinsic and contextual theses differ is on how each applies the concept of motivated reasoning to the question of public polarization in beliefs and attitudes toward science. The intrinsic thesis argues that the polarization of institutional trust in the scientific community is due to a fundamental antiscience bias among political conservatives (Mooney 2012). The intrinsic thesis bases its arguments on two sets of scholarship. The first stems from analysis of politically charged scientific controversies over the past 30 years and identifies political conservatism and institutions, as compared to political liberalism and institutions, as more often in the position of denying scientific evidence (e.g., Mooney 2012; Oreskes and Conway 2010). Supporters of the intrinsic thesis also point to Gauchat’s (2012) analysis of the General Social Survey, which demonstrates that confidence in the scientific community has declined substantially for conservatives since the early 1990s while remaining steady for liberals.

The second set of scholarship underlying the intrinsic thesis focuses on psychological factors and argues that psychological biases among conservatives increase the likelihood of distrust in the scientific community via the use of motivated reasoning to shape value-inconsistent information so that the information conforms to preexisting beliefs. It is important to note that liberals are considered substantially less likely to engage in this form of biased processing when presented with ideologically dissonant scientific claims (Oreskes and Conway 2010; Nam, Jost, and Van Bavel 2013; Mooney 2005, 2012). The intrinsic thesis relies on research asserting that political conservatives have unique psychological characteristics relative to liberals, such as dogmatism and a higher need for closure, that create stark differences in how members of these groups form attitudes and process information about science (Jost et al. 2003; Kruglanski 2004). For instance, Nam and colleagues (2013) demonstrated that conservatives are more strongly motivated to avoid dissonant messages and dissonant-arousing tasks.

In recent years, scholarship has identified some nuanced cognitive differences between liberals and conservatives (e.g. Carney et al. 2008; Dodd et al. 2012; Garrett and Stroud, forthcoming; Jost and Amodio 2012; Shook and Fazio 2009). However, it is a mistake to overstate these subtle dissimilarities between liberals and conservatives and to use them to make broad claims about biased information processing, especially in the area of science, as the intrinsic thesis does. For example, other scholarship has demonstrated that conservatives are no more likely to reject de-biasing information or engage in motivated reasoning relative to liberals when considering counterattitudinal political information (Gaines et al. 2007; Nyhan and Reifler 2010; Weeks and Garrett, forthcoming).

Neither is there any evidence of innate ideological differences when people consider scientific arguments and evidence. For instance, individuals from disparate value groups were found to make similarly biased risk assessments about ideologically incongruent science issues (Braman et al. 2012). Individuals in both groups employed motivated reasoning when forming their risk judgments about dissonant science topics, but polarization was greatest among those most likely to engage in systematic processing (i.e., those with greater cognitive sophistication; Braman et al. 2012). Consequently, ideological polarization was not a result of disproportionate reliance on heuristic processing or of scientific incomprehension by either liberals or conservatives, but rather resulted from the decision to process scientific information in a manner consistent with ideological and group affiliation.

Kahan (2013) has countered the idea of intrinsic ideological asymmetries by administering the cognitive reflection test to liberals and conservatives; he found no significant difference between the two ideological groups in their likelihood to engage in systematic processing of information. Consistent with Braman and his colleagues’ (2012) findings, individuals who scored highest on the cognitive reflection test were the most likely to engage in ideologically motivated reasoning. These findings support the idea that ideological polarization around science issues is not a consequence of either an information deficit between groups or of an overreliance on heuristic processing by one group compared to another (Kahan 2013). Instead, Kahan asserts that both liberals and conservatives engage in the “the reliable employment of more effortful, conscious information processing” that magnifies “the polarizing effects of identity-protective cognition” (2013, 410).

McCright and colleagues (2013) examined ideological differences in trust of scientists engaged in policymaking. They argued that conservatives are not distrustful of all scientists, but instead react negatively to those scientists who are engaged in what they call reflexive science (i.e., science that assesses the environmental and health impacts of modern society). The researchers manipulated the type of science to which participants were exposed and evaluated how much they consequently trusted scientists to advise on policy. Some respondents were asked how much they trust scientists such as climate scientists, epidemiologists, etc. (i.e., impact scientists), while others were asked about food scientists, industrial chemists, etc. (i.e., production scientists). Their results were consistent with the contextual thesis: liberals were more trusting and conservatives less trusting of impact scientists, whereas liberals were less trusting and conservatives more trusting of production scientists.

Political Dynamics of the Contextual Thesis

Collectively, this evidence from the fields of political psychology and risk and decision-making challenges the central assertion of the intrinsic hypothesis that conservatives are fundamentally less capable of rationally processing scientific evidence. To the contrary, the evidence is consistent with the contextual thesis, which posits that distrust and denialism of scientific evidence is greatest in those science policy contexts in which individuals feel their values, identity, or interests are threatened, regardless of their ideology (Braman et al. 2012; Kahan 2013; McCright et al. 2013; M. Nisbet 2005, 2011). Liberals and conservatives are equally likely to engage in motivated reasoning and exhibit bias when scientific evidence is incongruent with their values or prior beliefs (e.g., nuclear power for liberals, climate change for conservatives; Braman et al. 2012; Kahan 2013).

The contextual thesis asserts that the use of science for political ends by both Democrats and Republicans has contributed to political polarization about science issues as part of a larger, ongoing competition between American political actors attempting to differentiate themselves and mobilize base constituencies within an increasingly polarized political environment (Dunlap and McCright 2008, 2011; Leeper and Slothuus 2014; Sarewitz 2013; M. Nisbet 2011). For instance, increased media attention to the 1997 Kyoto Treaty, organized largely by the Clinton administration, substantially increased political polarization between liberals and conservatives about the topic of climate change (Krosnick, Holbrook, and Visser 2000). Likewise, the documentary An Inconvenient Truth, featuring a former Democratic presidential candidate from one of the most polarizing presidential elections in recent American history, along with the corresponding media attention to the topic driven by the film, had a similarly polarizing impact between liberals and conservatives on public opinion about climate change (M. Nisbet 2011).

This does not exempt the well-organized and heavily funded efforts by a range of conservative and industry actors, often termed the “denialist movement,” from being the primary cause of the present political polarization about climate change and climate mitigation policies in the United States (Dunlap and McCright 2008, 2011; M. Nisbet 2011). This institutional form of denialism has systematically attacked climate science and climate mitigation policies through extensive public communication campaigns that misrepresent science and scientific consensus, funding and targeting of political candidates, and intimidation of climate scientists. But here again, communication about the issue is polarizing due to processes of motivated reasoning: liberals tend to discount the claims made by these groups, whereas conservatives view them as raising legitimate questions about science.

Within this framework, the current low levels of institutional trust in the scientific community and rejection of scientific evidence (about some topics) among conservatives is an outcome of a reinforcing dynamic of institutional mobilization and psychological reactance among politically conservative segments of society. Conservative political, social, and economic actors engage in agenda-building activities to increase the salience of science-policy issues that are inconsistent with their values and interests, pushing the science topic from mostly administrative arenas into more public areas of discourse and contestation (Cobb and Elder 1983; Hiltgartner and Bosk 1988; Jasanoff 1987; M. Nisbet and Huge 2006).

As a consequence of these highly salient and polarized science policy debates, conservative audiences are exposed to cues from both the Left and the Right in the media and political discourse about how their ideology and values are largely incongruent with scientific evidence (Krosnick, Holbrook, and Visser 2000; Slothuus and de Vreese 2010). These cues provide information to conservative audiences about what they should believe and whom they should trust (Hmielowski et al. 2013; Slothuus and de Vreese 2010). Accordingly, conservative audiences engage in motivated reasoning and biased interpretations of scientific information as a means of identity maintenance and protection when such beliefs or opinions are challenged (Cohen 2003; Gaines et al. 2007; Kahan 2013).

In sum, although both the intrinsic and contextual theses explain public polarization about and distrust toward the scientific community in terms of motivated reasoning, these theses presume different sources of bias. The intrinsic thesis cites innate psychological differences as the basis for polarization in beliefs and attitudes toward science. In contrast, the contextual thesis focuses on how differences in the amount and nature of political mobilization around ideologically dissonant science-policy issues interact with ideological and value predispositions of the public. Thus, it is the dynamic interplay between institutional and psychological factors that result in different outcomes for conservatives and liberals rather than intrinsic difference between the two.

A Model of Audience Response to Dissonant Science Messages

Though motivated reasoning provides a useful psychological framework for understanding how exposure to science messages may lead to differences in beliefs and opinions about science between liberals and conservatives, we wish to further unpack the psychological processes underlying biased responses to science communication. We focus on two sets of intervening, or mediating, variables that may determine how exposure to ideologically dissonant science messages influences institutional trust in the scientific community. Since motivated reasoning and ideological dissonance have a strong affective component, and because affect is one of the key factors in message resistance, the first intervening variable that we consider is the affective (metacognitive) experience of exposure to a dissonant communication (Jacks and Cameron 2003; Schwarz 2011). Metacognitive refers to the individual’s experience of cognition (not cognition per say), including the emotional reactions that accompany thought generation, with the valence of the emotional reactions being a key factor in determining the perceived value of the presented information (Schwarz et al. 2007). This metacognitive experience provides information that audiences use to make judgments about the value of the message and source that may either amplify or dampen persuasiveness and trust toward the source depending on whether the net valence of the affective experience is positive or negative (Lewandowsky et al. 2012; Schwarz 2011; Schwarz et al. 2007).

The other mediating variable we consider is what we term motivated resistance to persuasion (Brehm and Brehm 1981; Jacks and Cameron 2003; Knowles and Linn 2004; McGuire 1964). In the context of exposure to science messages presenting evidence that is dissonant with an individual’s ideology, we view audience resistance as a “motivational state: the motivation to oppose and counter pressures to change” (Knowles and Linn 2004, 5). We consider two highly intercorrelated components of motivated resistance to persuasion—counterarguing and reactance. Counterarguing refers to the generation of cognitions that undermine and rebut a message’s informational content, thereby reducing the persuasiveness and credibility of both message and source (Jacks and Cameron 2003; McGuire 1964; Petty and Cacioppo 1986; Slater and Rouner 2002). Psychological reactance, which includes both affective and cognitive elements, refers to an oppositional response to perceived pressure for change that occurs when a person believes that a message threatens his or her agency or freedom (Brehm and Brehm 1981; Dillard and Shen 2005; Rains 2013). This opposition can lead to message rejection or source derogation (Byrne and Hart 2009; Knowles and Linn 2004; Rains 2013; Smith 1977). In the case of dissonant science communication presenting factual evidence that is ideologically inconsistent, the likelihood of perceiving a threat to the freedom to believe what one wants is high, which induces reactance. Although consonant messages may induce negative affect or motivated resistance, message dissonance (in terms of ideology or belief) increases the likelihood of these outcomes (Festinger 1957; Kunda 1990; Lewandowsky et al. 2012; Schwarz et al. 2007).

In sum, we propose a moderated-mediation model (Hayes 2013) of audience response that explicates some of the psychological processes that underlie resistance to science communication. First, we posit that an individual’s affective experience (H1a) and degree of motivated resistance (H1b) to the message from the scientific community will mediate the relationship between exposure to the message and institutional trust in the scientific community. Second, we hypothesize that more strongly held ideological orientations (H2a) and belief (in)accuracy (H2b) will moderate (amplify) the negative affective experience and/or motivated resistance involved in an individual’s response to a dissonant science message.

Methodology

Experimental design

For our study, we recruited a demographically diverse sample of 1,518 adult participants via a national paid opt-in online survey panel through Qualtrics Panels.¹ Participants were randomly assigned to one of three conditions: (1) conservative-dissonant science (climate change or human evolution), (2) liberal-dissonant science (hydraulic fracking of natural gas or nuclear power), or (3) ideologically neutral science (astronomy or geology). We used stimulus sampling (inclusion of multiple science topics within each condition to which participants were randomly assigned) to increase the external validity of our findings without reducing the internal validity of the experimental design (Wells and Windschitl 1999).

The issues for each condition were selected based on recent national polling data indicating sharp ideological divides within each of the four “dissonant” science policy contexts of climate change, human evolution, fracking, and nuclear power (Pew Research Center 2013a, 2013b, 2013c). Furthermore, we conducted a manipulation check to examine how political ideology was associated with the accuracy of prior beliefs about each set of topics within the study. With sociodemographic controls entered as covariates, political ideology did not significantly predict accuracy of beliefs about geology and astronomy, but did so for both of the ideologically dissonant science conditions (i.e., liberals were more accurate in their beliefs about climate change and human evolution whereas conservatives were more accurate about fracking and nuclear power).

Participants were asked four items assessing the accuracy of their beliefs about the science topic to which they were randomly assigned. The participants then viewed what they were told was a sample entry from an educational website for college students and adults about contemporary science topics (called ScienceWise). They were asked to carefully review the material and told that several questions about the material would follow. The stimulus provided information that would have allowed participants to correctly answer all of the previously asked knowledge questions. The full text of questions and examples of the stimuli are provided in the appendix. After viewing the stimulus, each participant responded to survey items measuring affective response and motivated resistance (counterarguing and reactance) to the message, as well as institutional trust in the scientific community. Participants were then debriefed.

Independent variables

We collected sociodemographics, which were used as control variables in subsequent analyses, including age (M = 45.18, SD = 15.45), gender (51 percent female), race (80 percent white), and educational attainment (1 = none at all, 8 = postgraduate training or professional schooling after college; M = 5.78, SD = 1.37). Analyses also controlled for biblical literalism (M = 1.94, SD = 0.70) and whether participants considered themselves evangelical Christians (yes coded high; M = 0.27, SD = 0.45).

General scientific literacy was measured using a slightly modified version of the index employed biannually by the National Science Foundation (2012). Participants were asked four items on a 5-point true/false scale ranging from definitely true to definitely false, with accuracy coded high and averaged into one overall measure of biophysical scientific knowledge (M = 3.80, SD = 0.83, see appendix for item wording). Science and technology news attention/interest was measured by averaging three items asking respondents how much attention they paid to environmental issues, advancements in science and technology, and energy (1 = no attention at all, 7 = a great deal of attention) along with two items asking respondents how much interest they had in the topics of “nature and wildlife” and “science and technology” (1 = no interest at all, 7 = a great deal of interest; M = 5.06, SD = 1.20, α = .87). Political news attention/interest was measured by combining two items asking participants how much interest they had in politics and how much attention they paid to political news (M = 4.37, SD = 1.76, r = .89).

Political ideology and prior belief accuracy were included as moderator variables to test our hypotheses regarding conditional indirect effects. Political ideology was measured by averaging two items asking participants how liberal or conservative they were on economic issues and social issues, respectively. Participants rated themselves on a 7-point scale for each item ranging from very liberal (0) to very conservative (6; M = 2.96, SD = 1.49, r = .79).

Prior belief accuracy was measured by four items about the topic to which the participant was assigned, scored on a 5-point true/false scale ranging from definitely true to definitely false (accuracy coded high). See the appendix for full text of the belief items for each of the six topics in the study. Participant responses to the four items were averaged in an overall measure of belief accuracy for each topic on a 1–5 scale: climate change (M = 3.6, SD = .71), human evolution (M = 3.0, SD = 1.00), hydraulic fracking (M = 2.9, SD = .68), nuclear power (M = 2.6, SD = .79), astronomy (M = 3.5, SD = .74), or geology (M = 2.6, SD = .73).

Dummy variables were coded to indicate whether the participant was assigned to the conservative-dissonant condition (33.1 percent), the liberal-dissonant condition (33.5 percent), or the ideologically neutral reference condition (33.4 percent). Dummy variables were coded for climate change (16.6 percent), fracking (16.8 percent), and astronomy (16.7 percent) to indicate which topic the participant was exposed to within each condition; they were also included as controls.

Affective response to the message was measured using the following item adapted from Watson, Clark, and Tellegen (1988): “When you think about the information you just read on the ScienceWise website, how do you feel? Do you feel …” followed by the emotions confident, encouraged, angry, and annoyed. Each item was scored on a 1–7 scale with the two positive and two negative emotions averaged, and the positive affect score was subtracted from the negative affect score (–6 = strong positive affective response, 6 = strong negative affective response; M = −1.65, SD = 2.54). Resistance to persuasion was measured using a composite scale of eleven items (see appendix) adapted (see Moyer-Gusé and Nabi 2010; Nabi, Moyer-Gusé, and Byrne 2007) to assess both counterarguing and reactance to the science communication featured on the ScienceWise website (1 = strongly disagree, 7 = strongly agree; M = 3.23, SD = 1.31, Cronbach’s α = .93).

Dependent variable

Institutional trust in the scientific community was measured by averaging five items (see appendix). Sample items included, “I am suspicious of the scientific community” (reverse coded) and “Information from the scientific community is trustworthy” (1 = strongly disagree, 7 = strongly agree; M = 4.70, SD = 1.18, Cronbach’s α = .87).

Data analysis

We tested our hypothetical moderated-mediation model using Hayes (2013) SPSS PROCESS macro (hereafter PROCESS) to explore conditional indirect effects. Moderated-mediation analyses allow the simultaneous examination of the moderating effects of ideology and belief accuracy on the mediated relationships between exposure to dissonant science communication and institutional trust in the scientific community through metacognitive affective response and motivated resistance to persuasion. Bootstrapped 99 percent and 99.9 percent confidence intervals employing 5,000 samples were calculated to identify significant conditional indirect effects on institutional trust via the two mediators.

Results

Results from our moderated-mediation analysis and hypothesis testing employing PROCESS are provided in three different tables. Table 1 shows the results of the first step of the moderated-mediation analysis and reports the effects of exposure to liberal- and conservative-dissonant science communications, as compared to the neutral reference condition, on metacognitive affective experience and motivated resistance. It also shows whether prior belief accuracy and political ideology moderated these relationships. Table 2 reports the results of the equation predicting how exposure to dissonant science communications, again contrasted to the ideologically neutral reference condition, influenced institutional trust in the scientific community with mediators entered into the model. Tables A1 and A2 in the appendix report the total conditional indirect effects—via the mediating variables of metacognitive affective response and resistance to persuasion—of exposure to each treatment condition.

Table 1

Moderated Mediation Analysis Predicting Metacognitive Affective Experience and Motivated Resistance to Persuasion

	Metacognitive Affective Experience (Negative)		Motivated Resistance to Persuasion
	Model 1	Model 2	Model 3	Model 4
Variable	b (SE)	b (SE)	b (SE)	b (SE)
Constant	−.69 (.50)	−1.69 (.54)**	3.44 (.06)***	2.94 (.28)***
Ideology (conservatism coded high)	.02 (.04)	.11 (.07)	.06 (.02)**	.08(.03)*
Prior belief accuracy	−.86 (.08)***	−.24 (.13)	−.62 (.04)***	−.24 (.07)***
Conservative-dissonant condition^a	1.84 (.21)***	2.23 (.50)***	1.03(.11)***	1.50 (.26)***
Liberal-dissonant condition^a	.45 (.21)*	3.13 (.43)***	.93(.11)***	2.13 (.23)***
Conservative dissonant × ideology	–	.20 (.09)*	–	.15 (.05)**
Conservative dissonant × belief accuracy	–	−.64 (.17)****	–	−.55 (.09)**
Liberal dissonant × ideology	–	−.49 (.09)***	–	−.25 (.05)***
Liberal dissonant × belief accuracy	–	−.77 (.18)***	–	−.30 (.10)***
% Variance explained (Total R²)	26.4	30.3	24.0	29.4

NOTE: Unstandardized coefficients and standard error reported. Results are controlling for age, sex, educational attainment, race, evangelical Christian, biblical literalism, attention to science and tech news, attention to/interest in politics, general scientific literacy; dummy codes for social indicator and political cues and stimulus sampling topics (climate change, fracking, astronomy) entered as covariates.

Reference condition is ideologically neutral condition (geography/astronomy).

p ≤ .05, **p ≤ .01, ***p ≤ .001.

Table 2

OLS Equation Predicting Institutional Trust in the Scientific Community

	Model 1	Model 2	Model 3
Variable	b (se)	b (SE)	b (SE)
Constant	4.00 (.25)	4.40 (.27)***	5.40 (.25)***
Ideology (conservatism coded high)	−.16 (.02)***	−.19 (.03)***	−.15 (.03)***
Prior belief accuracy	.32 (.04)***	.09 (.07)	−.01 (.01)
Conservative-dissonant condition^a	−.49 (.10)***	−1.16 (.25)***	−.44 (.22)*
Liberal-dissonant condition^a	−.22 (.10)***	−.85 (.22)***	.19 (.20)
Conservative dissonant × ideology	–	.01 (.02)	.08 (.04)*
Conservative dissonant × belief accuracy	–	.37 (.09)***	.12 (.08)
Liberal dissonant × ideology	–	.12 (.05)**	−.02 (.04)
Liberal dissonant × belief accuracy	–	.18 (.09)*	.02 (.08)
Metacognitive affective experience	–	–	−.07***
Motivated resistance to persuasion	–	–	−.38***
% Variance explained (Total R²)	19.2	20.3	38.5

NOTE: Unstandardized coefficients and standard error reported. Results are controlling for age, sex, educational attainment, race, evangelical Christian, biblical literalism, attention to science & tech news, attention/interest to politics, general scientific literacy, dummy codes for social indicator and political cues and stimulus sampling topics (climate change, fracking, astronomy) entered as covariates.

Reference condition is ideologically neutral condition (geography/astronomy).

p ≤ .05, **p ≤ .01, ***p ≤ .001.

Models 1 and 3 of Table 1 show how the two ideologically dissonant science conditions, as compared to the neutral condition, influence metacognitive affective responses and motivated resistance to persuasion after accounting for the covariates described above.² Participants in both the conservative-dissonant (climate change/evolution) condition and liberal-dissonant (fracking/nuclear) condition experienced a more negative metacognitive affective response while reading the ScienceWise stimulus compared with the ideologically neutral condition (astronomy/geology), though the size of the treatment effect for the conservative-dissonant condition was four times greater than that for the liberal-dissonant condition. Similarly, participants in both the conservative-dissonant condition and liberal-dissonant condition reported greater motivated resistance to persuasion as compared with the referent condition, though in contrast to the metacognitive affective response the strength of the resistance to persuasion in each condition was almost equal.

Models 2 and 4 in Table 1 test our hypotheses (H2a, H2b) that ideology and belief accuracy moderate the relationship between dissonant science communications and the two mediators. Within each condition, the regression results show that the more accurate one’s prior beliefs were about a topic, the less negative affect and resistance to persuasion one experienced in response to a corrective science message in the dissonant conditions as compared to those in the ideologically neutral condition. Greater accuracy of prior beliefs about climate change or human evolution reduced the effect size of exposure to the conservative-dissonant message on metacognitive affective and resistance to persuasion. Likewise, greater accuracy of prior beliefs about fracking of natural gas and nuclear power diminished the effect size of the liberal-dissonant message on metacognitive affective experience and resistance to persuasion. The moderating influence of ideology (with conservatism coded high) had a different relationship within each dissonant condition as compared to the referent condition. Conservatives had heightened negative affect experiences and greater resistance to persuasion in the conservative-dissonant condition as compared to the referent condition, whereas the converse was true in the liberal-dissonant condition with liberals experiencing greater affect and resistance to persuasion than conservatives.

Using PROCESS, we graphed these patterns and calculated the marginal mean values of the metacognitive affective and resistance to persuasion measures for liberal (1 standard deviation below the mean ideology score), moderate (mean ideology score), and conservative (1 standard deviation above the mean ideology score) participants in each condition. The results are depicted in Figures 2 and 3. Within the ideologically neutral condition, all three ideological groups appear to have an equally positive metacognitive experience while reading the science message about geology or astronomy with no significant differences. In contrast, the valence of the metacognitive affective experience for all three groups was significantly more negative in both the conservative-dissonant and liberal-dissonant conditions, though with some variance within each condition. The emotional valence of conservatives varied between the conservative and liberal-dissonant conditions, with conservatives experiencing net negative emotions in the conservative-dissonant condition and net positive emotions in the liberal-dissonant condition. Liberals experienced the mirror image of conservatives with a net negative emotional response in the liberal-dissonant condition and a net positive emotional response in the conservative-dissonant condition, though without as much variation between conditions as conservatives.

Turning to motivated resistance to persuasion (see Figure 3), conservatives had significantly greater motivated resistance to persuasion than liberals in the conservative-dissonant condition, whereas liberals reported significantly greater resistance than conservatives in the liberal-dissonant condition. Interestingly, in each condition, both conservatives and liberals resisted dissonant science messages more than their counterparts in the ideologically neutral condition regardless of their ideology. In the neutral conditions, there was no significant difference in motivated resistance between liberals and conservatives.

The second step of our analysis was testing whether affective experience and motivated resistance predicted trust in the scientific community and thus mediated the relationship between dissonant science message exposure and institutional trust (see Table 2). The first model in Table 2 presents the equation predicting institutional trust without moderation or entry of the mediating variables. Beyond controls, institutional trust was lower in both the conservative-dissonant (climate change/evolution) condition, and the liberal-dissonant (fracking/nuclear power) condition as compared to the ideologically neutral condition. In other words, there is a net drop in trust of the scientific communication around politically contentious science in both contexts compared with the more neutral science topics. The second model presents the results with the moderating influence of ideology and accuracy of prior beliefs without the mediators in the equation. Ideology does not moderate the unmediated relationship between exposure to the conservative-dissonant condition and institutional trust, but conservatism does amplify institutional trust in the liberal-dissonant condition. The treatment effect of the liberal-dissonant condition (fracking/nuclear power) as compared to the referent condition (geology/astronomy) significantly reduced trust in the scientific community among liberals and moderates, but not conservatives. In comparison, for both the conservative-dissonant and liberal-dissonant conditions, prior belief accuracy amplified trust in the scientific community as compared to the ideologically neutral condition.

The last model in Table 2 includes our measures of metacognitive affective experience and motivated resistance to persuasion, which both significantly reduced institutional trust in the scientific community. Adding these factors also lessened the direct effect of condition on institutional trust. The effect of the liberal-dissonant condition on trust dropped below significance and the magnitude of conservative-dissonant treatment effect substantially decreased. Thus, meta-cognitive affective experience and resistance to persuasion significantly dampened institutional trust in the scientific community, seemingly absorbing most of the influence of the different ideologically charged science messages.

To further test the mediating roles of metacognitive affective experience and motivated resistance to persuasion, we used PROCESS to estimate the conditional indirect effects of exposure to conservative- and liberal-dissonant science messages on trust as compared to the ideologically neutral condition at differing levels of political ideology and prior belief accuracy (not shown; see Tables A1 and A2 in the appendix). Indirect effects of exposure to the conservative-dissonant message on science trust through metacognitive experience were consistently negative, though how negative the effects were varied significantly by ideology and belief accuracy. That is, the conservative-dissonant condition significantly reduced trust in the scientific community as compared to the neutral condition via negative affect for both conservatives and liberals, but the effect was most pronounced among those with the lowest belief accuracy. In contrast, in the liberal-dissonant condition, metacognitive experience significantly dampened institutional trust among liberals with low (b = −.10, p ≤ .001) or moderate (b = −.06, p ≤ .001) belief accuracy, while significantly increasing institutional trust among conservatives with moderate (b = .04, p ≤ .01) or high (b = .08, p ≤ .001) belief accuracy.

Examining indirect effects of dissonant science on trust in the scientific community via motivated resistance to persuasion, we find that the influence of belief accuracy is comparable to its influence through affect: individuals with less accurate initial beliefs are more likely to engage in motivated resistance, which in turn reduces trust in science. The influence of ideology, however, is reversed across the two conditions. For instance, in the conservative-dissonant condition, there were no significant indirect effects for high-accuracy liberals (b = −.06, n.s.), but significant indirect effects for high accuracy conservatives (b = −.24, p ≤ .001). A mirror image of this pattern is evident in the liberal-dissonant condition: the indirect effect of condition on trust in the scientific community among high accuracy conservatives (b = −.06, n.s.) was not significant, but was significant for high accuracy liberals (b = −.33, p ≤ .001).

Discussion

Limitations of the study

Before we discuss the implications of our findings we wish to note some limitations to the study. One limitation is that the experiment was not fully crossed in terms of conditions—participants were not all asked the same belief questions prior to the science communication—and we employed a reference condition, not a traditional control. Employing a neutral science reference condition and using stimulus sampling of science topics within the conditions increases the external validity of the study, though internal validity may have been only modestly reduced as participants were not all asked the same scientific belief questions in each condition. It also would have been better to have conducted a separate pretest in which we measured accuracy of beliefs across all topics and institutional trust prior to exposure to the stimulus with a significant time interval (e.g., a week) between exposure to the dissonant messages and the following posttest to assess within-subject change in addition to between-subject change, but that was not possible in this study with the resources available. We see these as future opportunities to further explicate audience reactions to dissonant science messages as well as to better model the psychological dynamics of salient controversial science issues by conducting a series of repeated exposures to dissonant science messages across different topics to evaluate changes in institutional trust over time.

Implications for understanding audience reactions to dissonant science messages

Audiences that are exposed to politically controversial science communications often experience negative emotions and cognitive resistance as they process these messages, both of which have negative consequences for institutional trust in the scientific community. These processes are politically biased, influencing conservatives and liberals alike, and they are amplified by prior beliefs and value orientations. Our experiment demonstrates that both liberals and conservatives respond more negatively to ideologically dissonant science communication, which indirectly leads to lower trust in the scientific community—findings consistent with the contextual thesis and inconsistent with the intrinsic thesis. Unpacking the intervening processes between message exposure and institutional trust makes it clear that these reactions also vary by science context.

One important open question concerns why the magnitude of the negative affective response and motivated resistance to persuasion was greater in the conservative-dissonant condition than was the case in the liberal-dissonant condition. One possible explanation for these differences is the greater attitude polarization between liberals and conservatives on issues of climate change and human evolution as compared to fracking and nuclear power. For instance, post hoc analysis shows significantly higher levels of ideological differentiation in the accuracy of beliefs for the two conservative-dissonant issues (liberals more accurate and conservatives less accurate) as compared to liberal-dissonant science issues. We suspect that these observed differences help us to explain the conditional indirect effects reported here.

Climate change and the debates over teaching evolution generally have had a higher profile in media and political discourse than those over fracking or nuclear power, as have associated partisan identity markers. By virtue of these topics’ prominence, individuals know what (liberal and conservative) opinion leaders believe, and they know a great deal about the debate. Together, these factors leave individuals better equipped to counterargue and more likely to react negatively to the messages, which may help to explain the somewhat greater magnitude of the indirect effects on trust via motivated resistance for the conservative-dissonant condition as compared to the liberal-dissonant condition.

Just as polarized media and political discourse have contributed to greater affective polarization between liberals and conservatives (Garrett and Stroud, forthcoming; Iyengar, Sood, and Lelkes 2012), the greater polarization in beliefs around issues like climate change and human evolution may also invoke a stronger emotional response in conservatives and liberals who are confronted with these topics. Additional post hoc analysis confirms the participants were generally angrier about climate change and human evolution topics than the other four topics (fracking, nuclear power, geology, and space), and these high levels of anger translate into greater distrust of the scientific community for everyone.

These results suggest, distressingly, that political polarization around science has the potential to depress trust in science, regardless of where one lies on the ideological spectrum. In our experiment, institutional trust was significantly lower for all participants, including liberals, when considering conservative-dissonant issues, because of the psychological response of all participants to messages about science that are deeply ideologically polarized.

What leads some science issues to be more ideological polarized than others? Our data do not address this question, but we suspect that this is where the institutional processes and political dynamics outlined in the contextual thesis come into play. For example, liberal-dissonant issues such as hydraulic fracking, nuclear power, or genetically modified food currently have relatively low public salience compared to conservative-dissonant issues such as climate change and human evolution. The low salience of liberal-dissonant science policy issues contributes to the relative stability of institutional trust in the scientific community among liberals over the last 30 years. Liberal audiences have not been exposed to repeated liberal-dissonant messages, which we have shown would drive down institutional trust in the scientific community. Conservative audiences, in contrast, have faced a steady stream of dissonance-inducing science messages—a process augmented by the emergence of partisan news outlets that vary greatly in their treatment of issues such as climate change, and lead to greater polarization about the issue among viewers (Hmielowski et al. 2013; Feldman et al. 2012).

Our findings illustrate an important step in what we suspect is a much larger cycle of science communication. In the short term, audiences’ affective experience of and motivated resistance to a dissonant message may dampen trust in the scientific community. However, the possible long-term ramifications are more problematic because audiences are exposed to dissonant science messages repeatedly in political and media discourse. A cycle of science communication that includes a reinforcing spiral of source derogation toward the scientific community over time, as exampled by the issue of climate change, becomes difficult to break. The perceived trust and credibility of the messenger constrains how audiences receive and respond to future communications, to the point that trust in the scientific community becomes the basis on which messages are accepted or rejected, leading people to largely ignore the arguments or evidence (Brehm and Brehm 1981; Byrne and Hart 2009; Chaiken 1987; Petty and Cacioppo 1986). As a “trust gap” develops, it significantly influences the ability of the scientific community and other associated institutions to be effective communicators and advocates for science (Gauchat 2012; Hmielowski et al. 2013; Priest, Bonfadelli, and Rusanen 2003). It also provides incentives for mobilizing strategic actors to widen the trust gap and ideological cleavages around controversial science issues to increase mobilization.

Furthermore, our findings illustrate that advocates of the intrinsic thesis (e.g., Mooney 2012) have done science communicators a disservice. By promoting the idea that there are inherent psychological differences between conservatives and liberals when forming attitudes and making judgments about science, they are effectively—and ironically—contributing to the very political polarization of science they decry and thereby inhibiting more effective science communication. Some who adhere to this thesis rightly point out that “leading with values,” not with evidence, in a science message may be an effective communication strategy when communicating dissonant science to conservative audiences (Mooney 2011). However, by targeting conservatives specifically as somehow uniquely deficient when it comes to science, the overall framework of the intrinsic thesis lends itself to focusing on ideological countermobilization and/or a conversion of worldviews (“If only everyone were liberal!”), rather than to bridging ideological gaps (because the intrinsic thesis holds that they cannot be bridged). Demonizing a third of the population in a science policy debate by claiming they have an insurmountable psychological deficit does nothing to promote a solution to the challenges of effective science communication. And, as we have shown here, it is not empirically justified.

Footnotes

Appendix

Table A2

Total Conditional Indirect Effects on Trust in Science via Motivated Resistance to Persuasion

Ideology	Prior Belief Accuracy	Indirect Effect for Conservative-Dissonant Condition	Indirect Effect for Liberal-Dissonant Condition
Liberals	Low Accuracy	−.44***	−.53***
	Moderate Accuracy	−.25***	−.43***
	High Accuracy	−.06	−.33***
Moderates	Low Accuracy	−.53***	−.39***
	Moderate Accuracy	−.34***	−.29***
	High Accuracy	−.15**	−.20***
Conservatives	Low Accuracy	−.62***	−.26***
	Moderate Accuracy	−.43***	−.16***
	High Accuracy	−.24***	−.06

NOTE: Unstandardized coefficients reported. Bootstrapped confidence intervals employing 5,000 samples were calculated 99 percent and 99.9 percent.

p ≤ .01, ***p ≤ .001.

Notes

Erik C. Nisbet is an associate professor in the School of Communication, the School of Environment and Natural Resources (by courtesy), and the Department of Political Science (by courtesy) at the Ohio State University. His research includes the intersection of political and science communication, focusing on how people engage in motivated reasoning when processing scientific information and forming policy preferences.

Kathryn E. Cooper is a doctoral student within the School of Communication at the Ohio State University. Her research interests include science communication, media effects, decision making, and narrative persuasion.

R. Kelly Garrett is an associate professor in the School of Communication at the Ohio State University. His research interests include how people’s exposure to and perceptions of corrective or debiasing information are related to their political beliefs.

References

Braman

Donald

Kahan

Dan M.

Peters

Ellen

Wittlin

Maggie

Slovic

Paul

. 2012. The polarizing impact of science literacy and numeracy on perceived climate change risks. Nature Climate Change 2:732–35.

Brehm

Sharon S.

Brehm

Jack W.

1981. Psychological reactance: A theory of freedom and control. New York, NY: Academic Press.

Byrne

Sahara

Hart

P. Sol

. 2009. The boomerang effect: A synthesis of findings and a preliminary theoretical framework. Communication Yearbook 33:3–37.

Carney

Dana R.

Jost

John T.

Gosling

Samuel D.

Potter

Jeff

. 2008. The secret lives of liberals and conservatives: Personality profiles, interaction styles, and the things they leave behind. Political Psychology 29 (6): 807–40.

Chaiken

Shelly

. 1987. The heuristic model of persuasion. In Social influence: The Ontario symposium, eds. Zanna

Mark P.

Olson

James M.

Herman

C. Peter

, 3–39. New York, NY: Lawrence Erlbaum Associates.

Cobb

Roger W.

Elder

Charles D.

1983. Participation in American politics: The dynamics of agenda-building. Baltimore, MD: Johns Hopkins University Press.

Cohen

Geoffrey L.

2003. Party over policy: The dominating impact of group influence on political beliefs. Journal of Personality and Social Psychology 85 (5): 808–22.

Dillard

James P.

Shen

Lijiang

. 2005. On the nature of reactance and its role in persuasive health communication. Communication Monographs 72 (2): 144–68.

Dodd

Michael D.

Balzer

Amanda

Jacobs

Carly M.

Gruszczynski

Michael W.

Smith

Kevin B.

Hibbing

John R.

2012. The political Left rolls with the good and the political Right rolls with the bad: Connecting physiology and cognition to preferences. Philosophical Transactions of the Royal Society of London 367 (1589): 640–49.

10.

Dunlap

Riley E.

McCright

Aaron M.

2008. A widening gap: Republican and Democratic views on climate change. Environment 50:26–35.

11.

Dunlap

Riley E.

McCright

Aaron M.

2011. Organize climate change denial. In Oxford handbook of climate change and society, eds. Dryzek

John S.

Norgaard

Richard B.

Schlosberg

David

, 144–60. Oxford: Oxford University Press.

12.

Feldman

Lauren

Maibach

Edward W.

Roser-Renouf

Connie

Leiserowitz

Anthony

. 2012. Climate on cable: The nature and impact of global warming coverage on Fox News, CNN, and MSNBC. The International Journal of Press/Politics 17 (1): 3–31.

13.

Festinger

Leon

. 1957. A theory of cognitive dissonance. Stanford, CA: Stanford University Press.

14.

Gaines

Brian J.

Kuklinski

James H.

Quirk

Paul J.

Peyton

Buddy

Verkuilen

Jay

. 2007. Same facts, different interpretations: Partisan motivation and opinions on Iraq. Journal of Politics 69 (4): 957–74.

15.

Garrett

R. Kelly

Stroud

Natalie J.

Forthcoming. Partisan paths to exposure diversity: Differences in pro- and counter-attitudinal news consumption. Journal of Communication.

16.

Gauchat

Gordon

. 2012. Politicization of science in the public sphere: A study of public trust in the United States, 1974 to 2010. American Sociological Review 77 (2): 167–87.

17.

Hart

P. Sol

Nisbet

Erik C.

2012. Boomerang effects in science communication: How motivated reasoning and identity cues amplify opinion polarization about climate mitigation policies. Communication Research 39 (6): 701–23.

18.

Hayes

Andrew F.

2013. An introduction to mediation, moderation, and conditional process analysis. New York, NY: Guilford Press.

19.

Hilgartner

Stephen

Bosk

Charles L.

1988. The rise and fall of social problems: A public arenas model. The American Journal of Sociology 94 (1): 53–78.

20.

Hindman

Douglas Blanks

. 2009. Mass media flow and the differential distribution of politically disputed beliefs: The belief gap hypothesis. Journalism & Mass Communication Quarterly 86:790–808.

21.

Hmielowski

Jay D.

Feldman

Lauren

Myers

Teresa A.

Leiserowitz

Anthony

Maibach

Edward

. 2013. An attack on science? Media use, trust in scientists, and perceptions of global warming. Public Understanding of Science. doi:10.1177/0963662513480091.

22.

Iyengar

Shanto

Sood

Gaurav

Lelkes

Yphtach

. 2012. Affect, not ideology: A social identity perspective on polarization. Public Opinion Quarterly 76 (3): 405–31.

23.

Jacks

Julia Zuwerink

Cameron

Kimberly A.

2003. Strategies for resisting persuasion. Basic and Applied Social Psychology 25 (2): 145–61.

24.

Jasanoff

Sheila

. 1987. Biology and the Bill of Rights: Can science reframe the Constitution? American Journal of Law and Medicine 13:249–89.

25.

Johnson

Blair T.

Eagly

Alice H.

1989. Effects of involvement on persuasion: A meta-analysis. Psychological Bulletin 106 (2): 290–314.

26.

Jost

John T.

Amodio

David M.

2012. Political ideology as motivated social cognition: Behavioral and neuroscientific evidence. Motivation and Emotion 36:55–64.

27.

Jost

John T.

Glaser

Jack

Kruglanski

Arie W.

Sulloway

Frank J.

2003. Political conservatism as motivated social cognition. Psychological Bulletin 129 (3): 339–75.

28.

Kahan

Dan M.

2013. Ideology, motivated reasoning, and cognitive reflection. Judgment and Decision Making 8 (4): 407–24.

29.

Kahan

Dan M.

Braman

Donald

. 2006. Cultural cognition and public policy. Yale Law & Policy Review 24 (1): 149–72.

30.

Kahan

Dan M.

Peters

Ellen

Wittlin

Maggie

Slovic

Paul

Ouellete

Lisa Larrimore

Braman

Donald

Mandel

Gregory

. 2012. The polarizing impact of science literacy and numeracy on perceived climate change risks. Nature Climate Change 10 (2): 732–35.

31.

Knowles

Eric S.

Linn

Jay A.

2004. The importance of resistance to persuasion. In Resistance to persuasion, eds. Knowles

Eric S.

Linn

Jay A.

, 3–11. Mahwah, NJ: Lawrence Erlbaum.

32.

Krosnick

Jon A.

Holbrook

Allyson L.

Visser

Penny S.

2000. The impact of the fall 1997 debate about global warming on American public opinion. Public Understanding of Science 9:239–60.

33.

Kruglanski

Arie W.

2004. The psychology of closed mindedness. New York, NY: Psychology Press.

34.

Kunda

Ziva

. 1990. The case for motivated reasoning. Psychological Bulletin 108 (3): 480–98.

35.

Leeper

Thomas J.

Slothuus

Rune

. 2014. Political parties, motivated reasoning, and public opinion formation. Advances in Political Psychology 35:129–56.

36.

Lewandowsky

Stephan

Ecker

Ullrich K. H.

Seifert

Colleen M.

Schwarz

Norbert

Cook

John

. 2012. Misinformation and its correction: Continued influence and successful debiasing. Psychological Science in the Public Interest 13 (3): 106–31.

37.

Lodge

Milton

Taber

Charles

. 2000. Three steps toward a theory of motivated political reasoning. In Elements of reason: Cognition, choice, and the bounds of rationality, eds. Lupia

Arthur

McCubbins

Matthew D.

, 183–213. New York, NY: Cambridge University Press.

38.

Lord

Charles G.

Ross

Lee

Lepper

Mark R.

1979. Biased assimilation and attitude polarization: Effects of prior theories on subsequently considered evidence. Journal of Personality and Social Psychology 37 (11): 2098–2109.

39.

McCright

Aaron M.

Dentzman

Katherine

Charters

Meghan

Dietz

Thomas

. 2013. The influence of political ideology on trust in science. Environmental Research Letters 8:1–9.

40.

McGuire

William J.

1964. Inducing resistance to persuasion. In Advances in experimental and social psychology, ed. Berkowitz

Leonard

, 192–229. New York, NY: McGraw-Hill.

41.

Mooney

Chris

. 2005. The Republican war on science. New York, NY: Basic Books.

42.

Mooney

Chris

. May/June 2011. The science of why we don’t believe science. Mother Jones.

43.

Mooney

Chris

. 2012. The Republican brain: The science of why they deny science and reality. Hoboken, NJ: Wiley.

44.

Moyer-Gusé

Emily

Nabi

Robin L.

2010. Explaining the persuasive effects of narrative in an entertainment television program: Overcoming resistance to persuasion. Human Communication Research 36:25–51.

45.

Nabi

Robin L.

Moyer-Gusé

Emily

Byrne

Sahara

. 2007. All joking aside: A serious investigation into the persuasive effect of funny social issue messages. Communication Monographs 74:29–54.

46.

Nam

H. Hannah

Jost

John T.

Van Bavel

Jay J.

2013. “Not for all the tea in China!” Political ideology and the avoidance of dissonance. PLoS One 8 (4).

47.

National Science Board. 2014. Science and engineering indicators 2014. Arlington, VA: National Science Foundation. Available from www.nsf.gov.

48.

National Science Foundation. 2012. Science and technology: Public attitudes and understanding. Science and Engineering Indicators 2012. Washington, DC. Available from http://www.nsf.gov/statistics/seind12/pdf/c07.pdf.

49.

Nelkin

Dorothy

. 1975. The political impact of technical expertise. Social Studies of Science 5:34–54.

50.

Nelkin

Dorothy

. 1992. Science, technology, and political conflict: Analyzing the issues. In Controversy: Politics of technical decisions, ed. Nelkin

Dorothy

, vii–ix. Thousand Oaks, CA: Sage Publications.

51.

Nisbet

Erik C.

Cooper

Kathryn E.

Ellithorpe

Morgan

. Forthcoming. Ignorance or bias? Evaluating the ideological and informational drivers of communication gaps about climate change. Public Understanding of Science.

52.

Nisbet

Matthew C.

2005. The competition for worldviews: Values, information, and public support for stem cell research. International Journal of Public Opinion Research 17 (1): 90–112.

53.

Nisbet

Matthew C.

2011. Climate shift: Clear vision for the next decade of public debate. Washington, DC: American University, School of Communication. Available from http://climateshiftproject.org/wp-content/uploads/2011/08/ClimateShift_report_June2011.pdf.

54.

Nisbet

Matthew C.

Huge

Mike

. 2006. Attention cycles and frames in the plant biotechnology debate: Managing power and participation through the press/policy connection. International Journal of Press/Politics 11 (2): 3–40.

55.

Nisbet

Matthew C.

Markowitz

Ezra M.

2014. Understanding public opinion in debates over biomedical research: Looking beyond political partisanship to focus on beliefs about science and society. PLoS One 9 (2). doi:10.1371/journal.pone.0088473.

56.

Nyhan

Brendan

Reifler

Jason

. 2010. When corrections fail: The persistence of political misperceptions. Political Behavior 32 (2): 303–30.

57.

Oreskes

Naomi

Conway

Erik M.

2010. Merchants of doubt: How a handful of scientists obscured the truth on issues from tobacco smoke to global warming. New York, NY: Bloomsbury Press.

58.

Petty

Richard E.

Cacioppo

John T.

1986. Communication and persuasion: Central and peripheral routes to attitude change. New York, NY: Springer-Verlag.

59.

Pew Research Center. 2013a. Continued support for Keystone XL Pipeline. Washington, DC: Pew Research Center.

60.

Pew Research Center. 2013b. GOP deeply divided over climate change. Washington, DC: Pew Research Center.

61.

Pew Research Center. 2013c. Public’s views on human evolution. Washington, DC: Pew Research Center.

62.

Priest

Susanna Hornig

Bonfadelli

Heinz

Rusanen

Maria

. 2003. The “trust gap” hypothesis: Predicting support for biotechnology across national cultures as a function of trust in actors. Risk Analysis 23 (4): 751–66.

63.

Rains

Stephen A.

2013. The nature of psychological reactance revisited: A meta-analytic review. Human Communication Research 39 (1): 47–73.

64.

Sarewitz

Daniel

. 2013. The rightful place of science. Issues in Science and Technology 25 (4): 89–94.

65.

Scheufele

Dietram A.

2006. Messages and heuristics: How audiences form attitudes about emerging technologies. In Engaging science: Thoughts, deeds, analysis and action, ed. Turney

Jon

, 20–25. London: The Wellcome Trust.

66.

Schwarz

Norbert

. 2011. Feelings-as-information theory. In Handbook of theories of social psychology, eds. Van Lange

Paul A. M.

Kruglanski

Arie W.

Higgins

Tory

, 289–308. Thousand Oaks, CA: Sage Publications.

67.

Schwarz

Norbert

Sanna

Lawrence J.

Skurnik

Ian

Yoon

Carolyn

. 2007. Metacognitive experiences and the intricacies of setting people straight: Implications for debiasing public information campaigns. Advances in Experimental Social Psychology 39:127–61.

68.

Shook

Natalie J.

Fazio

Russell H.

2009. Political ideology, exploration of novel stimuli, and attitude formation. Journal of Experimental Social Psychology 45 (4): 995–98.

69.

Slater

Michael D.

Rouner

Donna

. 2002. Entertainment-education and elaboration likelihood: Understanding the processing of narrative persuasion. Communication Theory 12:173–91.

70.

Slothuus

Rune

de Vreese

Claes H.

2010. Political parties, motivated reasoning, and issue framing effects. The Journal of Politics 72 (3): 630–45.

71.

Smith

Mary John

. 1977. The effects of threats to attitudinal freedom as a function of message quality and initial receiver attitude. Communication Monographs 44 (3): 196–206.

72.

Taber

Charles S.

Cann

Damon

Kucsova

Simona

. 2009. The motivated processing of political arguments. Political Behavior 31:137–55.

73.

Taber

Charles S.

Lodge

Milton

. 2006. Motivated skepticism in the evaluation of political beliefs. American Journal of Political Science 50 (3): 755–69.

74.

Watson

David

Clark

Lee Anna

Tellegen

Auke

. 1988. Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology 54 (6): 1063–70.

75.

Weeks

Brian E.

Garrett

R. Kelly

. Forthcoming. Electoral consequences of political rumors: Motivated reasoning, candidate rumors, and vote choice during the 2008 U.S. presidential election. International Journal of Public Opinion.

76.

Wells

Gary L.

Windschitl

Paul D.

1999. Stimulus sampling and social psychological experimentation. Personality and Social Psychology Bulletin 25:1115–25.