How Moral Threat Shapes Laypersons’ Engagement With Science

Abstract

Laypersons’ engagement with science has grown over the last decade, especially in Internet environments. While this development has many benefits, scientists also face the challenge of devaluation and public criticism by laypersons. Embedding this phenomenon in social-psychological theories and research on value-behavior correspondence, we investigated moral threat as a factor influencing laypersons’ engagement with science. Across three studies, we hypothesized and found that moral values shape the way laypersons evaluate and communicate about science when these values are threatened in a given situation and central to people’s self-concept. However, prior research on the underlying mechanism of moral threat effects cannot fully rule out value salience as an alternative explanation. To close this gap, we situationally induced value salience while varying the degree of moral threat (Study 3). Our findings indicate that moral threat amplifies the influence of moral values on laypersons’ evaluation of science above and beyond value salience.

Keywords

public engagement with science value-behavior correspondence value centrality moral threat value salience

Science benefits humankind by generating knowledge and technological inventions. Scientific innovations such as penicillin, the steam engine, or the printing press have improved human welfare by promoting public health, stimulating economic growth, spreading education, and making people’s everyday lives easier. Perhaps not surprisingly, public attitudes toward science are positive, and public interest in science is high (e.g., National Science Board, 2014). Nonetheless, there have been recent challenges to the science-society relationship. Specifically, scientific issues such as nuclear energy, stem cell research, and climate change have sparked controversial public debates. Even in the social sciences, public controversies have emerged on scientific issues such as the effects of violent video games on aggression.^1,2,3 In the course of these controversies, substantial parts of the public ignore, are skeptical about, or even devalue science. For example, large parts of the American public devalue research on anthropogenic climate change (Lewandowsky, Ecker, Seifert, Schwarz, & Cook, 2012) even though there is broad scientific consensus that climate change exists and is influenced by humans (Intergovernmental Panel on Climate Change, 2014). This devaluation of science by laypersons can have severe consequences for scientists, society, and even laypersons’ everyday lives. For example, the devaluation of science by large or influential parts of the public can undermine the reputation of scientists,⁴ pressure policy makers to cut funding for single research projects (e.g., Leshner, 2005) or entire research fields (e.g., Robertson, 2010), hinder societal growth, and endanger public health (e.g., Chigwedere, Seage, Gruskin, Lee, & Essex, 2008).

Arguably, the devaluation of science is most influential when laypersons express their criticism in public environments with large audiences, such as the Internet. Nowadays, the Internet has dramatically changed the science-society relationship by providing laypersons with the opportunity to evaluate and comment on scientific research on science blogs, video channels, twitter, online discussion forums, or via Facebook (Brossard & Scheufele, 2013). These forms of public engagement with science have grown tremendously in the past decade (Rödder, 2009). While this development can be seen as positive for encouraging dialogue between scientists and laypersons (Leshner, 2003), it also brings about new challenges, since public praise or critique on singular findings or entire research programs enter the public discourse unfiltered and can, ultimately, damage a scientist’s reputation. The question of when and why laypersons devalue or praise scientific evidence has received surprisingly little research attention, given the potential of online comments to influence and alter laypersons’ attitudes and knowledge about concrete scientific issues, as well as their trust in science and support for science funding (A. A. Anderson, Brossard, Scheufele, Xenos, & Ladwig, 2014; Lewandowsky et al., 2012). The present research aims to fill that gap by investigating factors that influence laypersons’ evaluation of and communication about scientific evidence in Internet environments. Specifically, we focus on the influence of moral values.

Moral Values and Public Engagement With Science

Moral values are particularly interesting as influencing factors, because many scientific issues are value-laden (Leshner, 2005). For example, scientific research on climate change or genetically modified foods touches upon laypersons’ environmental values (e.g., Honkanen & Verplanken, 2004). Likewise, scientific research on the deterrent effects of capital punishment or the harmfulness of violent video games touches upon nonviolence as a moral value (e.g., Rothmund, Bender, Nauroth, & Gollwitzer, 2015). Values can be defined as “trans-situational goals, varying in importance, that serve as guiding principles in the life of a person or group” (Schwartz et al., 2012, p. 664). Moral values differ from nonmoral values in that people tend to believe that their moral values are unconditionally and universally valid (Skitka, 2010). For example, people tend to perceive nonviolence (Mayton et al., 2002) and environmental protection (Steg & De Groot, 2012) as moral values.

Prior research on public engagement with science identified attitudes and beliefs as factors influencing the evaluation of science (C. A. Anderson, 1983; C. A. Anderson & Kellam, 1992; Greitemeyer, 2014; Lord, Ross, & Lepper, 1979). For example, proponents of capital punishment evaluate scientific evidence more positively when it confirms versus disconfirms the deterrent effects of capital punishment (Lord et al., 1979). Importantly, (moral) values differ from attitudes and beliefs in several conceptual features (for an overview, see Maio, Olson, Bernard, & Luke, 2003) and shape information processing and behavior simultaneously and independent of each other (Rothmund et al., 2015; Verplanken & Holland, 2002). For example, attitudes toward abusing the environment and environmental values uniquely predict voting for green parties (Verplanken & Holland, 2002). Thus, the present research aims to advance prior research on public engagement with science by investigating moral values rather than attitudes and beliefs as conceptually different person factors. There is initial evidence in the science communication literature for the link between values and laypersons’ reactions to science: For example, religious values and conservative values are negatively related to laypersons’ support for nanotechnology research (Brossard, Scheufele, Kim, & Lewenstein, 2009) and embryonic stem cell research (Ho, Brossard, & Scheufele, 2008). However, prior research on public engagement with science mainly focused on the effects on evaluation, neglecting behavioral outcomes such as publicly praising or discrediting science on the Internet. Moreover, little is known about the situational boundary conditions and underlying psychological processes of value influences on laypersons’ engagement with science. To investigate when and why—rather than whether—moral values influence science-related information processing and behavior, we draw on social-psychological theories and research on value-behavior correspondence.

Value-Behavior Correspondence

In line with a growing body of research, we argue that values shape information processing and behavior such as online behavior relating to science (for an overview, see Maio et al., 2003). Although the correspondence between people’s values and their actual behaviors is, on average, relatively low (Maio et al., 2003), there are a number of factors that amplify this correspondence, such as trait idealism (e.g., Wojciszke, 1987), time pressure (e.g., Darley & Batson, 1973), abstractness versus concreteness (Torelli & Kaikati, 2009), or temporal distance to the respective behavior (Eyal, Sagristano, Trope, Liberman, & Chaiken, 2009). These factors can also interact with each other. For example, Verplanken and Holland (2002) showed that situationally induced value salience promotes attention to value-relevant information and value-congruent behavior only for individuals high, but not low on value centrality. In line with this Person × Situation interaction approach, we argue that situationally induced moral threat increases value-behavior correspondence especially for individuals who define themselves in terms of the threatened moral value (Aquino & Reed, 2002).

Threats are generally defined as motivationally relevant discrepancies (Jonas et al., 2014) in the sense of “is-ought discrepancies with an aversive character” (Greve & Strobl, 2004, p. 194). Accordingly, we define moral threats as discrepancies between the motivations related to moral values and the observation of immoral behavior. For example, threats against the moral value of nonviolence (Mayton et al., 2002) are discrepancies between the motivation to reduce all forms of violence and observing violence in the world. Likewise, threats against environmental values (Steg & De Groot, 2012) are discrepancies between the motivation to protect the environment and observing environmental pollution or destruction.

Based on the general process model of threat and defense (Jonas et al., 2014) and the value protection model (Tetlock, 2003), we assume that moral values shape information processing and behavior especially when people experience moral threat. There are two reasons for this. First, laypersons are more attentive to scientific evidence that provides an explanation for a threatening event (Jonas et al., 2014). For example, scientific evidence confirming that violent video games cause real-life aggression provides an explanation for a particularly horrible violent event, such as a school shooting. Second, laypersons show heightened commitment to value-related motivations by engaging in value-congruent information processing and behavior (Jonas et al., 2014). For example, heightened commitment to environmental values may promote public praise for science that confirms the harmfulness of genetically modified foods because many people perceive this as congruent to their environmental values (Honkanen & Verplanken, 2004). Therefore, moral threat should increase value-behavior correspondence. In line with this theoretical assumption, research shows that threats to the sanctity of life promote the willingness to volunteer for organ donation campaigns (Tetlock, Kristel, Elson, Green, & Lerner, 2000, Study 2), that threats to equality promote the willingness to volunteer in pro-equality political rallies (Tetlock et al., 2000, Study 4), that threats to religious values promote intentions to participate in more religious activities (Tetlock et al., 2000, Study 5), that threats to environmental values promote global warming mitigation behavior (M. A. Ferguson & Branscombe, 2010), and that threats to nonviolence promote participating in demonstrations against violence (Lodewijkx, Kersten, & van Zomeren, 2008). Evidence for information processing effects comes from the broader finding that threats amplify the influence of worldview-relevant decisions on selective exposure to and evaluation of worldview-congruent information (Jonas & Greenberg, 2004; Jonas, Greenberg, & Frey, 2003). In summary, moral values shape information processing and behavior especially when they are threatened in a situation. Therefore, it can be assumed that moral threats amplify the influence of moral values on the evaluation of and communication about science in Internet environments.

In line with Person × Situation interaction approaches, moral threat should not promote value-behavior correspondence for every person to the same degree. We argue that threatening a moral value will shape information processing and behavior especially for individuals who define themselves in terms of the respective moral value (Aquino & Reed, 2002; Verplanken & Holland, 2002). Even though most moral values are widely shared (Rokeach, 1970), people still differ in the degree to which moral values are central to their self-concept (Aquino & Reed, 2002). The more central a specific moral value is to a person’s self-concept (i.e., the higher the value centrality), the more this value should influence how the person interprets a situation, processes information, and shows behavior (Verplanken & Holland, 2002). Therefore, moral threat should influence laypersons’ engagement with science more strongly for individuals high compared with individuals low on value centrality. There is some initial support for this assumption by Rothmund et al. (2015). The authors measured the degree to which participants endorsed nonviolence as a moral value, experimentally induced moral threat (using a video clip about military interventions), and gave participants the opportunity to search for scientific evidence on the harmfulness of violent video games. Confirming the Person × Situation interaction effect, results revealed that the more people endorse the moral value of nonviolence, the more they selectively search for scientific evidence that confirms rather than disconfirms the harmfulness of violent video games after a threat against nonviolence was experimentally induced (compared with not induced).

The present research goes a step further and focuses on behavioral outcomes such as publicly praising or discrediting science on the Internet. Moreover, the present research aims to advance our understanding of the underlying mechanism of the effects. That is, despite the growing evidence that moral threats amplify value-behavior correspondence, there is an important limitation to the previous findings. Specifically, there is an alternative explanation for why moral threats promote high value-behavior correspondence.

Value Salience as an Alternative Explanation

Based on previous research in the value-behavior correspondence literature, salience of the moral value might fully explain why moral threat amplifies value-behavior correspondence (Maio & Olson, 1995; Verplanken & Holland, 2002). From a theoretical point of view, threatening a moral value will inevitably increase the salience of the respective moral value. Therefore, value salience and moral threat are not entirely independent of each other. Rather, moral threat can be seen as a combination of value salience and an additional threat component (Jonas et al., 2014). At the same time, there is empirical evidence that value salience can promote value-behavior correspondence (Maio & Olson, 1995; Verplanken & Holland, 2002). For example, it was shown that the salience of altruistic values promotes donation behavior (Maio & Olson, 1995). Therefore, moral threat effects might be explained by value salience alone rather than by the combination of value salience and the additional threat component. By contrast, the moral threat hypothesis proposes that threat is a crucial component and a driving force behind high value-behavior correspondence. Thus, the moral threat hypothesis suggests that moral threat amplifies the influence of values on behavior above and beyond the effect of value salience.

The Present Research

The aims of the present article were twofold. From a public engagement with science perspective, we wanted to investigate how moral values shape laypersons’ engagement with science. In line with Person × Situation interaction approaches, we hypothesized that moral values are expressed in laypersons’ evaluation of and communication about scientific research especially if the respective research touches upon a moral value that is threatened in a given situation and central to the person’s self-concept (Hypothesis 1). By doing so, we extend prior research by investigating (a) moral values rather than attitudes or beliefs as conceptually different person factors, (b) situational boundary conditions for the effect, and (c) behavioral outcomes (i.e., communication behavior in Internet environments) in addition to evaluation. From the perspective of research on value-behavior correspondence, we aimed to further our understanding of the psychological mechanism that can explain when and why moral values shape information processing and behavior. We hypothesized that moral threat amplifies the influence of value centrality on the evaluation of scientific evidence above and beyond value salience (Hypothesis 2).

In three studies, experimental manipulations were used to induce threat against a moral value and value centrality was measured using self-report. Then, laypersons were presented with scientific evidence and given the opportunity to evaluate the research and to write an online comment about the research for a (fictitious) science blog on the Internet (Studies 1 and 2). In Study 3, we tested whether moral threat amplifies the influence of value centrality on the evaluation of scientific evidence above and beyond value salience.

Study 1

Study 1 provides a first test of the hypothesis that moral values affect laypersons’ engagement with science, especially when the value is threatened and central to their self-concept. We investigated this hypothesis in the context of environmental values and scientific research on the potential harmful effects of genetically modified foods. Specifically, we tested whether threatening environmental values affects the evaluation of and online communication about scientific evidence on the harmfulness of genetically modified foods for individuals with high value centrality of environmental values.

Method

Participants and design

We conducted an experimental online study with a 2 (threat against environmental values: yes vs. no) × 2 (direction of scientific results on genetically modified foods: harm-confirming vs. harm-disconfirming) between-subjects design with value centrality as a continuous between-subjects moderator. The study involved two occasions of measurement (average time between T1 and T2: 7 days), and participants were recruited via a professional German sampling agency (Respondi). The final sample of this study consisted of 175 participants (42.9% female; age: M = 45.8, SD = 14.1).⁵ The sample size was chosen to provide good statistical power (1 − β = .90) based on an effect size for information processing observed in a prior study (ΔR² = .060; Rothmund et al., 2015, Study 2).

Procedure, materials, and measures⁶

Participants were informed that the study was about people’s perceptions of scientific research on societal issues. One week prior to the online experiment, participants completed an online questionnaire with measures of value centrality, demographic variables, and other variables not immediately relevant for the present research question.⁷ Value centrality of environmental values was measured with an adapted 10-item version of the moral identity scale (Aquino & Reed, 2002; Cronbach’s α = .89). Specifically, participants were asked to imagine a person who sees environmental protection as a guiding principle in his or her life, who cares about environmental protection, and who actively promotes environmental protection. Then, participants indicated their level of agreement to each of 10 statements (e.g., “Being someone who has these characteristics is an important part of who I am”). Ratings were obtained on a 6-point scale from 1 = strongly disagree to 6 = strongly agree.

Approximately 7 days later, participants were invited to the second part of the online study. They were told that we were specifically interested in their opinions about science blogs on societal issues. First, an experimental manipulation was used to threaten environmental values. Participants were randomly assigned to one of two conditions. In the threat against environmental values condition (n = 85), participants were provided with a report that informed about a worldwide negative trend regarding environmental protection and addressed several problematic areas of environmental protection. In the control condition (n = 90), participants were not provided with a report. As a manipulation check, perceived threat against moral values was measured with six items on a 6-point scale from 1 = strongly disagree to 6 = strongly agree (Cronbach’s α = .93; for example, “I worry about the moral values in our society”). Next, participants read a scientific study about the effects of genetically modified maize on the development of beehives. Participants were told that the study would be published on a public science blog on the Internet. The results of the scientific study varied depending on the experimental condition. Participants were randomly assigned to one of two conditions. In the harm-confirming results condition (n = 84), the study provided evidence for the negative effect of genetically modified maize on the development and mortality rate of bees living in the maize field. In the harm-disconfirming results condition (n = 91), the same text was used with minor changes, so that in this condition, the study provided no evidence for a negative effect.

Next, participants evaluated the study based on 13 items (adapted from Nauroth, Gollwitzer, Bender, & Rothmund, 2015; Cronbach’s α = .95; for example, “I think that the study was a waste of public money” [reverse coded]). Response scales ranged from 1 = strongly disagree to 6 = strongly agree. Higher values indicate more positive evaluations of the scientific study.

Finally, participants were provided with the opportunity to write a comment about the study that would be posted along with the study on the science blog. Participants could choose between posting a positive (i.e., “pro” study) comment or a negative (i.e., “contra” study) comment. Then, participants completed a postexperimental questionnaire including a manipulation check for the direction of scientific results manipulation (“Please indicate the degree to which the scientific evidence confirms the harmfulness of genetically modified foods” on a scale from 1 = the scientific evidence provides no evidence for the harmfulness of genetically modified foods to 7 = the scientific evidence provides strong evidence for the harmfulness of genetically modified foods).

After completion of the online experiment, participants were thanked and fully debriefed, including information that the results of the scientific study do not represent the state of research and that the science blog was only fictitious.

Results

Manipulation check

To test the validity of the threat against environmental values manipulation, a t test for independent samples was conducted with perceived threat against moral values as the dependent variable and the threat manipulation as the independent variable. In support of the validity of the manipulation, perceived threat against moral values was higher in the threat against environmental values condition (M = 4.85, SD = 0.97) compared with the control condition (M = 4.52, SD = 1.04), t(173) = 2.14, p = .034, d = 0.33, 95% CI for d = [0.03, 0.62]. In support of the direction of scientific results manipulation, participants perceived the scientific study as stronger evidence for the harmfulness of genetically modified foods in the harm-confirming results condition (M = 5.37, SD = 1.71) compared with the harm-disconfirming results condition (M = 2.58, SD = 1.65), t(173) = 10.97, p < .001, d = 1.66, 95% CI for d = [1.31, 2.00]. Descriptive statistics and correlations for the variables of interest are shown in Table 1.⁸

Table 1.

Descriptive Statistics and Correlations Between Variables in Study 1.

Variable	Scale	M (SD)	Correlations
Variable	Scale	M (SD)	(1)	(2)	(3)
1. Online posting behavior	0-1	0.57 (0.50)	1
2. Evaluation of scientific evidence	1-6	3.94 (1.26)	.69***	1
3. Value centrality of environmental values	1-6	3.48 (1.01)	.05	.11	1

Note. N = 175.

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

Evaluation of scientific evidence

We expected that threats against environmental values would affect the evaluation of scientific evidence on the harmfulness of genetically modified foods for individuals with high (compared with low) levels of value centrality regarding environmental values. To test this, we conducted a moderated regression analysis with evaluation as the dependent variable and threat against environmental values, results of the described study, and value centrality as predictors. The threat variable was dummy-coded prior to the analysis such that, in the multiple regression model, the effect of the dummy variable will represent the contrast between the threat against environmental values condition and the control condition (0 = control, 1 = threat). The results variable was dummy-coded such that the dummy variable will represent the contrast between the harm-disconfirming results condition and the harm-confirming results condition (0 = harm-disconfirming results, 1 = harm-confirming results). Value centrality of environmental values was mean-centered prior to analysis (Cohen, Cohen, West, & Aiken, 2003). The overall moderated regression model accounted for 23.5% of the variance in evaluation (p < .001).⁹

Online posting behavior

We expected threats against environmental values to affect the communication behavior regarding scientific evidence on the harmfulness of genetically modified foods for individuals with high (compared with low) levels of value centrality regarding environmental values. Thus, we calculated a logistic regression analysis with online posting behavior as the binary dependent variable (0 = negative comment, 1 = positive comment).¹⁰ As predictors, we used threat against environmental values, results of the described study, and value centrality (mean-centered) as well as their respective interactions (i.e., saturated model; deviance of the overall model: 199.33, Nagelkerke-Index = 0.28). The analysis revealed a significant effect of the results variable, indicating that participants in the no threat control condition had a higher likelihood of posting a positive (compared with a negative) comment on the scientific study in the harm-confirming results condition compared with the harm-disconfirming results condition, B = 1.89, SE(B) = 0.51, p < .001, 95% CI for B = [0.90, 2.89]. In line with Hypothesis 1, this effect was qualified by a significant three-way interaction, B = 1.58, SE(B) = 0.78, p = .042, 95% CI for B = [0.06, 3.10], see Figure 1. More precisely, there was a significant two-way interaction between the results variable and value centrality in the threat condition, B = 1.45, SE(B) = 0.62, p = .019, 95% CI for B = [0.24, 2.67], but not in the control condition, B = −0.13, SE(B) = 0.47, p = .789, 95% CI for B = [−1.04, 0.79]. In the threat condition, individuals high (+1SD) on value centrality were more likely to post a positive online comment on the scientific study in the harm-confirming compared with the harm-disconfirming results condition, B = 3.77, SE(B) = 1.01, p < .001, 95% CI for B = [1.79, 5.76]. However, for individuals low (−1SD) on value centrality in the threat condition, the likelihood of posting a positive online comment did not differ between the harm-confirming and the harm-disconfirming results condition, B = 0.83, SE(B) = 0.63, p = .186, 95% CI for B = [−0.40, 2.07].

Figure 1.

Probability of posting a positive (compared with a negative) online comment on the scientific study on the harmfulness of genetically modified foods as a function of threat against environmental values, direction of results, and value centrality predicted by binary logistic regression (Study 1).

Discussion

Study 1 provides preliminary evidence that threatening a moral value affects the communication about scientific evidence for individuals high on value centrality. That is, in response to a threat against environmental protection, people who consider environmental values central to their self-concept were more likely to praise scientific research on a public science blog when the results of the scientific study confirmed rather than disconfirmed the harmfulness of genetically modified foods. However, we did not find evidence for effects on evaluation of scientific evidence. This could possibly be a result of the particular scientific topic we chose here: Genetically modified food is something that people in Germany are generally very skeptical about (Eurobarometer, 2010). This is mirrored by our result that even in the no threat control condition, participants strongly favored the harm-confirming over the harm-disconfirming finding. Study 2 aimed to replicate the findings in a different context.

Study 2

Study 2 further tests the hypothesis that moral values are expressed in laypersons’ engagement with science, especially when the value is threatened and central to their self-concept. To test the generalizability of our findings, we tested the hypothesis with a different moral threat (i.e., threat against nonviolence) and a different research domain (i.e., research on violent video game effects). The harmfulness of violent video games has been subject to very heated and controversial debates in the general public, especially in the aftermath of school shootings and other incidents of extreme violence. Regarding scientific evidence, comprehensive meta-analyses show that violent video games do indeed cause aggression (e.g., C. A. Anderson et al., 2010). However, there are also authors who have repeatedly argued against that claim (e.g., C. J. Ferguson & Kilburn, 2010). The fact that laypersons are confronted with scientific claims from both sides combined with the societal debate surrounding the issue makes it a particularly interesting context to test our predictions. Specifically, we tested whether threatening nonviolence as a moral value affects the evaluation of and online communication about scientific evidence on the harmfulness of violent video games for individuals with high value centrality regarding nonviolence.

Method

Participants and design

We conducted an experimental online study with a 2 (threat against nonviolence: yes vs. no) × 2 (direction of scientific results on violent video game effects: harm-confirming vs. harm-disconfirming) between-subjects design with value centrality as a continuous between-subjects moderator. The study involved two occasions of measurement (average time between T1 and T2: 9 days), and participants were recruited via a professional German sampling agency (Toluna). The final sample of this study consisted of 110 participants (64.5% female; age: M = 45.3, SD = 13.1).

Procedure, materials, and measures

Participants were told that the study was about societal issues in the media. One week prior to the online experiment, participants completed an online questionnaire with measures of value centrality and demographic variables. Value centrality of nonviolence was measured with an adapted version of the scale that we used in Study 1 (Cronbach’s α = .88).

Approximately 9 days later, participants were invited to the second part of the online study. They were told that we were specifically interested in their opinions about video clips and science blogs on societal issues. First, an experimental manipulation was used to threaten nonviolence as a moral value (see Rothmund et al., 2015). Participants were randomly assigned to one of two conditions. In each condition, they read a text and watched a 1-min video clip on a moral issue. In the threat against nonviolence condition (n = 53), participants were confronted with a text and a video clip on military interventions. In the control condition (n = 57), participants were confronted with a text and a video clip on same-sex marriage. Same-sex marriage was chosen because—like military interventions—it represents a moral issue, but one that is unrelated to violence. As a manipulation check, perceived threat against nonviolence was measured with three items on a 6-point scale from 1 = strongly disagree to 6 = strongly agree (Cronbach’s α = .93; for example, “I fear an increase in violence in our society”).

Then, participants read a scientific study on the effects of violent video game exposure on players’ desensitization toward real-life violence. Participants were told that the study would be published on a public science blog on the Internet. The results of the scientific study varied depending on the experimental condition. Participants were randomly assigned to one of two conditions. In the harm-confirming results condition (n = 53), the study provided evidence for a link between violent video game exposure and desensitization toward real-life violence. In the harm-disconfirming results condition (n = 57), the same text was used with minor changes, so that in this condition, the study provided no evidence for a link between violent video games and desensitization. Then, participants evaluated the study based on an adapted version of the scale that we used in Study 1 (Cronbach’s α = .94). Finally, participants completed the same measure of online posting behavior as in Study 1.

Results

Manipulation check

To test the validity of the threat against nonviolence manipulation, a t test for independent samples was conducted with perceived threat against nonviolence as the dependent variable and the threat manipulation as the independent variable. In support of the validity of the manipulation, perceived threat against nonviolence was higher in the threat against nonviolence condition (M = 4.82, SD = 1.09) compared with the control condition (M = 4.23, SD = 1.24), t(108) = 2.64, p = .010, d = 0.50, 95% CI for d = [0.12, 0.88]. Descriptive statistics and correlations for the variables of interest are shown in Table 2.

Table 2.

Descriptive Statistics and Correlations Between Variables in Study 2.

Variable	Scale	M (SD)	Correlations
Variable	Scale	M (SD)	(1)	(2)	(3)
1. Online posting behavior	0-1	0.48 (0.50)	1
2. Evaluation of scientific evidence	1-6	3.74 (1.11)	.74***	1
3. Value centrality of nonviolence	1-6	4.16 (0.86)	.06	.09	1

Note. N = 110.

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

Evaluation of scientific evidence

We expected that threats against nonviolence would affect the evaluation of scientific evidence on the harmfulness of violent video games for individuals with high (compared with low) levels of value centrality regarding nonviolence. To test this, we conducted a moderated regression analysis with evaluation as the dependent variable and threat against nonviolence, direction of scientific results, and value centrality as predictors. Threat against nonviolence was dummy-coded prior to the analysis (0 = control, 1 = threat). The results variable was also dummy-coded (0 = harm-disconfirming results, 1 = harm-confirming results). Value centrality of nonviolence was mean-centered prior to analysis (Cohen et al., 2003). The overall moderated regression model accounted for 24.3% of the variance in evaluation (p < .001).

There was a significant effect of the results variable, indicating that participants in the no threat control condition evaluated the scientific evidence more positively in the harm-confirming results condition than in the harm-disconfirming results condition, B = 0.52, SE(B) = 0.24, p = .033, 95% CI for B = [0.04, 1.00]. However, the effect of the results variable was qualified by a significant three-way interaction effect, B = 0.90, SE(B) = 0.42, p = .036, ΔR² = .034, 95% CI for B = [0.06, 1.73], see Figure 2. More precisely, there was a significant two-way interaction between the results variable and value centrality in the threat condition, B = 0.67, SE(B) = 0.27, p = .015, 95% CI for B = [0.13, 1.21], but not in the control condition, B = −0.22, SE(B) = 0.32, p = .489, 95% CI for B = [−0.86, 0.42]. In line with Hypothesis 1, individuals high (+1SD) on value centrality in the threat condition evaluated the scientific study more positively in the harm-confirming results condition compared with the harm-disconfirming results condition, B = 1.57, SE(B) = 0.34, p < .001, 95% CI for B = [0.90, 2.24]. For individuals low (−1SD) on value centrality in the threat condition, there was no difference in evaluation between the harm-confirming and the harm-disconfirming evidence, B = 0.41, SE(B) = 0.36, p = .257, 95% CI for B = [−0.30, 1.11].

Figure 2.

Evaluation of the scientific study on the harmfulness of violent video games as a function of threat against nonviolence, direction of results, and value centrality (Study 2).

Online posting behavior

We expected moral threat not only to account for effects on the evaluation but also on the commenting of scientific evidence. Thus, we calculated a logistic regression analysis with online posting behavior as the binary dependent variable (0 = negative comment, 1 = positive comment). As predictors, we used threat against nonviolence, direction of scientific results, and value centrality (mean-centered) as well as their respective interactions (i.e., saturated model; deviance of the overall model: 130.50, Nagelkerke-Index = 0.24). Again, the analysis revealed a significant effect of the results variable, indicating that participants in the no threat control condition had a higher likelihood of posting a positive (compared with a negative) comment on the scientific study in the harm-confirming results condition compared with the harm-disconfirming results condition, B = 1.17, SE(B) = 0.57, p = .041, 95% CI for B = [0.05, 2.28]. In line with Hypothesis 1, this effect was qualified by a significant three-way interaction effect, B = 2.25, SE(B) = 1.10, p = .042, 95% CI for B = [0.09, 4.41], see Figure 3. More precisely, there was a marginally significant two-way interaction between the results variable and value centrality in the threat condition, B = 1.39, SE(B) = 0.80, p = .080, 95% CI for B = [−0.17, 2.96], but not in the control condition, B = −0.85, SE(B) = 0.76, p = .262, 95% CI for B = [−2.35, 0.64]. Paralleling the findings for evaluation, individuals high (+1SD) on value centrality in the threat condition were more likely to post a positive online comment about the scientific study in the harm-confirming compared with the harm-disconfirming results condition, B = 3.29, SE(B) = 1.09, p = .003, 95% CI for B = [1.16, 5.42]. However, for individuals low (−1SD) on value centrality in the threat condition, the likelihood of posting a positive online comment did not differ between the harm-confirming and the harm-disconfirming results condition, B = 0.88, SE(B) = 0.84, p = .299, 95% CI for B = [−0.78, 2.53].

Figure 3.

Probability of posting a positive (compared with a negative) online comment on the scientific study on the harmfulness of violent video games as a function of threat against nonviolence, direction of results, and value centrality predicted by binary logistic regression (Study 2).

Discussion

Study 2 replicates and extends the findings from Study 1: Following a threat against nonviolence, people who define themselves in terms of nonviolence evaluated the exact same scientific evidence as more useful, its results more clear and reliable, its authors more objective and competent, and its methodology more sound when the results confirmed rather than disconfirmed the harmfulness of violent video games. Besides affecting the evaluation of scientific evidence, moral threat also influenced laypersons’ online commenting behavior. That is, in response to a threat against nonviolence, laypersons with high value centrality were more likely to publicly praise scientific evidence on a public science blog when the results confirmed rather than disconfirmed the harmfulness of violent video games. This finding lends further support to the hypothesis that moral values affect information processing and online behavior relating to science if they are (a) threatened in the situation and (b) central to the self-concept of the person. However, the findings of Studies 1 and 2 are also in line with the value salience explanation, which suggests that moral threat effects might be driven by value salience alone rather than a combination of value salience and threat (e.g., Verplanken & Holland, 2002). Study 3 investigates the underlying process of moral threat effects by testing whether moral threat effects can be explained by value salience.

Study 3

Study 3 provides a test for the hypothesis that moral threat amplifies the influence of value centrality on the evaluation of scientific evidence above and beyond value salience.¹¹ The study is based on the logic that situations with high threat should enhance the effects compared with situations with low threat, even if the value is salient in both situations. Thus, we situationally induced value salience by confronting people with a violence-related issue, but varied the degree to which the situation was threatening. Therefore, if value salience drives the effects, highly and lowly threatening situations should enhance the effects to the same degree. However, if moral threat drives the effects, situations with high (compared with low) threat should lead to stronger effects.

Method

Participants and design

We conducted an experimental online study with a single-factor randomized group design (value salience with high threat vs. value salience with low threat vs. control) and value centrality regarding nonviolence as a continuous between-subjects moderator. Participants were recruited via email distribution lists of German universities. The final sample consisted of 112 participants (65.8% female; age: M = 23.4, SD = 4.7).

Procedure, materials, and measures

First, participants were randomly assigned to one of three conditions. In the value salience with high threat condition (n = 36), participants read a two page report on violent crime in Germany, stating that violent crime had been increasing over recent years. In the value salience with low threat condition (n = 41), participants read the same report with minor changes, so that in this condition, the report stated that violent crime had been decreasing over recent years. Note that the two reports contained the exact same amount of violence-related words in order to keep value salience of nonviolence constant between the high and low threat condition. In the control condition (n = 35), participants were not provided with any report. Perceived threat against nonviolence was measured as a manipulation check with the same measure as in Study 2 (Cronbach’s α = .91). Next, participants read a scientific study that confirmed the harmfulness of violent video games (see Study 2) and evaluated the study (see Study 2, Cronbach’s α = .93). Note that in this study, the results of the scientific evidence were harm-confirming for all participants, that is, there was no harm-disconfirming results condition. Finally, value centrality regarding nonviolence was measured (see Study 2, Cronbach’s α = .85). After completion of the online experiment, participants were thanked and fully debriefed, including information that the scientific study was fictitious.

Results

Manipulation check

Perceived threat against nonviolence differed between the three experimental conditions, F(2,109) = 4.15, p = .018, η² = .07, 95% CI for η² = [0.01, 0.167]. In support of the validity of the experimental manipulation, post hoc tests (Tukey honestly significant difference [HSD]) revealed that perceived threat against nonviolence was higher in the value salience with high threat condition (M = 3.49, SD = 1.47) compared with the value salience with low threat condition (M = 2.60, SD = 1.45), p = .021, d = 0.61, 95% CI for d = [0.15, 1.06]. Perceived threat against nonviolence did not differ between the value salience with low threat condition and the control condition (M = 3.30, SD = 1.38), p = .089, d = −0.50, 95% CI for d = [−0.95, −0.03]. Surprisingly, perceived threat against nonviolence did also not differ between the value salience with high threat condition and the control condition, p = .849, d = 0.13, 95% CI for d = [−0.33, 0.60].

Evaluation of scientific evidence

In line with our hypothesis, we expected the influence of value centrality regarding nonviolence on the evaluation of the scientific study to be stronger in the value salience with high threat condition compared with the value salience with low threat condition and the control condition. A moderated regression analysis was conducted with evaluation as the dependent variable and threat against nonviolence and value centrality of nonviolence (mean-centered), as well as the respective interaction terms as independent variables. Threat against nonviolence was dummy-coded such that, in the multiple regression model, the effect of dummy1 will represent the contrast between the value salience with high threat condition and the control condition, and the effect of dummy2 will represent the contrast between the value salience with high threat condition and the value salience with low threat condition (dummy1: 0 = value salience with high threat, 0 = value salience with low threat, 1 = control; dummy2: 0 = value salience with high threat, 1 = value salience with low threat, 0 = control). The overall model explained 13.1% of the variance in evaluation (p = .010). In line with our assumptions, we found a two-way interaction between dummy1 and value centrality, B = −0.61, SE(B) = 0.29, p = .035, 95% CI for B = [−1.18, −0.04] (see Figure 4), indicating that the effect of value centrality on evaluation of the scientific evidence was stronger in the value salience with high threat condition compared with the control condition. Most importantly and in line with Hypothesis 2, we found a two-way interaction between dummy2 and value centrality, B = −0.59, SE(B) = 0.28, p = .036, 95% CI for B = [−1.14, −0.04], indicating that the effect was also amplified in the value salience with high threat condition compared with the value salience with low threat condition (ΔR² = .047 for adding both dummy-coded interactions to the model). Specifically, simple slope analyses revealed that value centrality was positively related to evaluation in the value salience with high threat condition, B = 0.77, SE(B) = 0.22, p < .001, 95% CI for B = [0.34, 1.20], whereas it was unrelated to evaluation in the value salience with low threat condition, B = 0.18, SE(B) = 0.17, p = .297, 95% CI for B = [−0.16, 0.52], and the control condition, B = 0.16, SE(B) = 0.18, p = .383, 95% CI for B = [−0.20, 0.53].

Figure 4.

Evaluation of the scientific study that confirms the harmfulness of violent video games as a function of condition (value salience with high threat or value salience with low threat or control) and value centrality (Study 3).

Discussion

In Study 3, we gained further insight into the psychological processes underlying the effects we found in Study 2. Our results indicate that moral threats amplify the strength of the influence of value centrality on the evaluation of scientific evidence above and beyond value salience. Specifically, the more central nonviolence was to people’s self-concept, the more positive they evaluated a scientific study that confirmed the harmfulness of violent video games when nonviolence was threatened in the situation. Most importantly, the influence of value centrality on evaluation was stronger in a situation with high threat compared with a situation with low threat, even though nonviolence was salient in both situations. Accordingly, Study 3 extends the findings of Studies 1 and 2, lending support to our Hypothesis 2 that moral threat amplifies the strength of the influence of value centrality on reactions to scientific evidence above and beyond value salience.

Despite these findings, the results of the manipulation check deserve critical attention. Specifically, perceived threat against nonviolence was higher in the value salience with high threat condition compared with the value salience with low threat condition, but not compared with the control condition. Nevertheless, high threat amplified value-behavior correspondence compared with both the low threat condition and the control condition. Notably, self-report measures of threat have been repeatedly challenged in the threat compensation literature, because (a) it is difficult to detect threat directly with self-report measures and (b) they typically fail to mediate the effect of threat manipulations on threat reactions (see Jonas et al., 2014). Indeed, our self-report measures of perceived threat were unrelated or only weakly related to evaluation (Study 1: r = −.05, p = .503; Study 2: r = .17, p = .070; Study 3: r = .19, p = .050) and online posting behavior (Study 1: r_pb = −.14, p = .063; Study 2: r_pb = .04, p = .691). Therefore, self-report measures of threat such as the one we used may fail to capture important aspects of the threat experience.

General Discussion

Public engagement with science has become increasingly popular during recent years (Brossard & Scheufele, 2013). Scientific societies like the American Association for the Advancement of Science (AAAS) have enthusiastically welcomed public engagement efforts and stressed that scientists “need to engage the public in a more open and honest bidirectional dialogue about science and technology” (Leshner, 2003, p. 977). Internet environments such as science blogs play a central role in these public engagement efforts by providing new opportunities for dialogue between scientists and laypersons. At the same time, it is important to be aware of the potential problems of these new science communication efforts such as the devaluation of science by laypersons and incivility in online comments about science (e.g., A. A. Anderson et al., 2014). The goal of the present research was to investigate how moral values can shape laypersons’ engagement with science, especially in Internet environments. To investigate this research question, we conducted three experimental studies, confronted laypersons with moral threats, and subsequently exposed them to scientific evidence on value-laden issues. In line with our assumptions, we found that moral values can shape the way laypersons evaluate and communicate about science when they are threatened and central to the self-concept. We provided evidence for the generalizability of these moral threat effects for two different moral values (i.e., nonviolence, environmental values) and two different research contexts (i.e., violent video games research, genetically modified foods research).

The present research extends prior research on public engagement with science (e.g., C. A. Anderson, 1983; C. A. Anderson & Kellam, 1992; Greitemeyer, 2014; Lord et al., 1979) in three ways. First, we focused on moral values as a conceptually different factor that can shape evaluation and behavior above and beyond attitudes or beliefs (Rothmund et al., 2015; Verplanken & Holland, 2002). In contrast to attitudes or beliefs, moral values possess the unique characteristic that they can affect laypersons’ reactions to seemingly unrelated issues. That is, the present studies show that people’s reactions toward violent video games research were affected even if they felt threatened by military interventions or violent crime statistics. Similarly, people’s reactions toward genetically modified foods research were affected even if they felt threatened by issues such as rising air or water pollution. Therefore, moral threat effects do not necessarily require a direct link between the threat issue (e.g., military interventions) and the response issue (e.g., violent video games). However, the two issues are indirectly linked because they both relate to the same moral value. Therefore, threats that were caused by one issue can affect the reactions toward another issue as long as the two issues are linked to the same basic moral values. This finding is particularly important given that laypersons can engage with science in situations in which moral threat was caused by unrelated issues. For example, people may be more likely to praise scientific evidence for the harmfulness of violent video games in the aftermath of a school shooting or evidence for the harmfulness of genetically modified foods in the aftermath of an environmental disaster.

Second, we extend prior science communication research on the influence of values on public engagement with science (Brossard et al., 2009; Ho et al., 2008) by investigating situational boundary conditions for the effect. Our findings imply that moral values shape public engagement with science especially when the respective value is threatened in the situation and central to the self-concept of the person. By doing so, the present research provides insights into when and why, rather than whether moral values shape laypersons’ engagement with science. These findings are in line with a growing body of research showing that moral values shape information processing and behavior especially when they are threatened (e.g., M. A. Ferguson & Branscombe, 2010; Lodewijkx et al., 2008; Tetlock et al., 2000). Notably, moral threat is a particularly relevant factor in the value-behavior correspondence debate, because numerous studies in that line of research investigate values in the moral domain (e.g., helpfulness, honesty, equality, nonviolence).

Third, we investigated behavioral outcomes in addition to evaluation of scientific evidence. So far, research has neglected actual science-related communication behaviors, especially in Internet environments (for an exception, see Nauroth et al., 2015). The present research fills that gap by showing that moral threats influence behavioral outcomes such as publicly praising or discrediting science on the Internet. These behavioral outcomes are particularly important, because they have the potential to profoundly change laypersons’ perceptions of science (A. A. Anderson et al., 2014; Brossard & Scheufele, 2013). Finally, the present research extends previous findings by showing that moral threat promotes value-behavior correspondence above and beyond value salience.

Limitations and Future Research

There are two limitations of the present research that deserve attention. First, one may argue that moral threat effects are driven by heightened cognitive support for a value (e.g., Maio & Olson, 1998). That is, moral threats can make people realize why a certain value is important, thus strengthening cognitive support for the value (Maio & Olson, 1998). For example, when people are confronted with a report on increasing violent crime, they may realize the importance of nonviolence. This heightened cognitive support for the value may then promote value-congruent information processing and behavior. Indeed, heightened cognitive support for a value may increase value-behavior correspondence above and beyond mere value salience (Maio et al., 2001). Thus, future research should investigate whether heightened cognitive support for a value is the mechanism underlying moral threat effects.

Second, there are two (mutually exclusive) psychological processes that can explain the effects of moral values on science-related information processing, and it remains unclear which of the two is driving the effects in the current studies. The first explanation is heightened attention to scientific information that provides possible solutions to the threat. For example, laypersons may positively evaluate scientific information that confirms the harmfulness of violent video games because this information provides an explanation and a possible solution to threats against nonviolence. According to the general process model (Jonas et al., 2014), this process is associated with activation of the Behavioral Inhibition System (BIS) and triggered immediately after the threat. The second process is heightened commitment to value-related motivations. For example, laypersons may positively evaluate scientific information that confirms the harmfulness of violent video games because they consider this information to be congruent with valuing nonviolence. This second process is associated with activation of the Behavioral Approach System (BAS) and triggered after a certain time delay (typically a few minutes). Since BAS activation automatically suppresses BIS activation, the two processes cannot happen simultaneously. Therefore, it remains an open question which of the two processes drives the effect of moral values on information processing. Future research could include a measure of approach and avoidance motivation to test which of the two processes drive the effects.

Conclusion

The present research shows that moral values shape the way laypersons evaluate and communicate about science in Internet environments if they are threatened in the situation and central to the self-concept of the person, and that moral threat is a powerful driving force for high value-behavior correspondence and shapes information processing and behavior above and beyond value salience.

Footnotes

Acknowledgements

We thank Luisa Horsten, Felicia Parusel, and Kevin Tiede for their help in conducting the studies.

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a German Research Foundation (Deutsche Forschungsgemeinschaft; no. RO 4248/1-2) grant to the second author.

Notes

References

Anderson

A. A.

Brossard

Scheufele

D. A.

Xenos

M. A.

Ladwig

(2014). The “nasty effect”: Online incivility and risk perceptions of emerging technologies. Journal of Computer-Mediated Communication, 19, 373-387. doi:10.1111/jcc4.12009

Anderson

C. A.

(1983). Abstract and concrete data in the perseverance of social theories: When weak data lead to unshakeable beliefs. Journal of Experimental Social Psychology, 19, 93-108. doi:10.1016/0022-1031(83)90031-8

Anderson

C. A.

Kellam

K. L.

(1992). Belief perseverance, biased assimilation, and covariation detection: The effects of hypothetical social theories and new data. Personality and Social Psychology Bulletin, 18, 555-565. doi:10.1177/0146167292185005

Anderson

C. A.

Shibuya

Ihori

Swing

E. L.

Bushman

B. J.

Sakamoto

. . . Saleem

(2010). Violent video game effects on aggression, empathy, and prosocial behavior in Eastern and Western countries: A meta-analytic review. Psychological Bulletin, 136, 151-173. doi:10.1037/a0018251

Aquino

Reed

II. (2002). The self-importance of moral identity. Journal of Personality and Social Psychology, 83, 1423-1440. doi:10.1037/0022-3514.83.6.1423

Brossard

Scheufele

D. A.

(2013). Science, new media, and the public. Science, 339, 40-41. doi:10.1126/science.1232329

Brossard

Scheufele

D. A.

Kim

Lewenstein

B. V.

(2009). Religiosity as a perceptual filter: Examining processes of opinion formation about nanotechnology. Public Understanding of Science, 18, 546-558. doi:10.1177/0963662507087304

Chigwedere

Seage

G. R.

Gruskin

Lee

T.-H.

Essex

(2008). Estimating the lost benefits of antiretroviral drug use in South Africa. Journal of Acquired Immune Deficiency Syndromes, 49, 410-415. doi:10.1097/QAI.0b013e31818a6cd5

Cohen

West

S. G.

Aiken

L. S.

(2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Hillsdale, NJ: Lawrence Erlbaum.

10.

Darley

J. M.

Batson

C. D.

(1973). “From Jerusalem to Jericho”: A study of situational and dispositional variables in helping behavior. Journal of Personality and Social Psychology, 27, 100-108. doi:10.1037/h0034449

11.

Eurobarometer. (2010). Special eurobarometer 341: Biotechnology. Retrieved from http://ec.europa.eu/public_opinion/archives/ebs/ebs_341_en.pdf

12.

Eyal

Sagristano

M. D.

Trope

Liberman

Chaiken

(2009). When values matter: Expressing values in behavioral intentions for the near vs. distant future. Journal of Experimental Social Psychology, 45, 35-43. doi:10.1016/j.jesp.2008.07.023

13.

Ferguson

C. J.

Kilburn

(2010). Much ado about nothing: The misestimation and overinterpretation of violent video game effects in Eastern and Western nations: Comment on Anderson et al. (2010). Psychological Bulletin, 136, 174-178. doi:10.1037/a0018566

14.

Ferguson

M. A.

Branscombe

N. R.

(2010). Collective guilt mediates the effect of beliefs about global warming on willingness to engage in mitigation behavior. Journal of Environmental Psychology, 30, 135-142. doi:10.1016/j.jenvp.2009.11.010

15.

Greitemeyer

(2014). I am right, you are wrong: How biased assimilation increases the perceived gap between believers and skeptics of violent video game effects. PLoS ONE, 9(4), e93440. doi:10.1371/journal.pone.0093440

16.

Greve

Strobl

(2004). Social and individual coping with threats: Outlines of an interdisciplinary approach. Review of General Psychology, 8, 194-207. doi:10.1037/1089-2680.8.3.194

17.

S. S.

Brossard

Scheufele

D. A.

(2008). Effects of value predispositions, mass media use, and knowledge on public attitudes toward embryonic stem cell research. International Journal of Public Opinion Research, 20, 171-192. doi:10.1093/ijpor/edn017

18.

Honkanen

Verplanken

(2004). Understanding attitudes towards genetically modified food: The role of values and attitude strength. Journal of Consumer Policy, 27, 401-420. doi:10.1007/s10603-004-2524-9

19.

Intergovernmental Panel on Climate Change. (2014). Climate change 2014: Synthesis report. Contribution of working groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva, Switzerland: Author.

20.

Jonas

Greenberg

(2004). Terror management and political attitudes: The influence of mortality salience on Germans’ defence of the German reunification. European Journal of Social Psychology, 34, 1-9. doi:10.1002/Ejsp.178

21.

Jonas

Greenberg

Frey

(2003). Connecting terror management and dissonance theory: Evidence that mortality salience increases the preference for supporting information after decisions. Personality and Social Psychology Bulletin, 29, 1181-1189. doi:10.1177/0146167203254599

22.

Jonas

McGregor

Klackl

Agroskin

Fritsche

Holbrook

. . . Quirin

(2014). Threat and defense: From anxiety to approach. In Zanna

M. P.

Olson

J. M.

(Eds.), Advances in experimental social psychology (Vol. 49, pp. 219-286). New York, NY: Elsevier.

23.

Leshner

A. I.

(2003). Public engagement with science. Science, 299, 977. doi:10.1126/science.299.5609.977

24.

Leshner

A. I.

(2005). Where science meets society. Science, 307, 815. doi:10.1126/science.1110260

25.

Lewandowsky

Ecker

U. K. H.

Seifert

C. M.

Schwarz

Cook

(2012). Misinformation and its correction: Continued influence and successful debiasing. Psychological Science in the Public Interest, 13, 106-131. doi:10.1177/1529100612451018

26.

Lodewijkx

H. F. M.

Kersten

G. L. E.

van Zomeren

(2008). Dual pathways to engage in “Silent Marches” against violence: Moral outrage, moral cleansing and modes of identification. Journal of Community & Applied Social Psychology, 18, 153-167. doi:10.1002/casp.916

27.

Lord

C. G.

Ross

Lepper

M. R.

(1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of Personality and Social Psychology, 37, 2098-2109. doi:10.1037/0022-3514.37.11.2098

28.

Maio

G. R.

Olson

J. M.

(1995). Relations between values, attitudes, and behavioral intentions: The moderating role of attitude function. Journal of Experimental Social Psychology, 31, 266-285. doi:10.1006/jesp.1995.1013

29.

Maio

G. R.

Olson

J. M.

(1998). Values as truisms: Evidence and implications. Journal of Personality and Social Psychology, 74, 294-311. doi:10.1037//0022-3514.74.2.294

30.

Maio

G. R.

Olson

J. M.

Allen

Bernard

M. M.

(2001). Addressing discrepancies between values and behavior: The motivating effect of reasons. Journal of Experimental Social Psychology, 37(2), 104-117. doi:10.1006/jesp.2000.1436

31.

Maio

G. R.

Olson

J. M.

Bernard

M. M.

Luke

M. A.

(2003). Ideologies, values, attitudes, and behavior. In Delamater

(Ed.), Handbook of social psychology (pp. 283-308). New York, NY: Kluwer-Plenum.

32.

Mayton

D. M.

Susnjic

Palmer

B. J.

Peters

D. J.

Gierth

Caswell

R. N.

(2002). The measurement of nonviolence: A review. Peace and Conflict: Journal of Peace Psychology, 8, 343-354. doi:10.1207/s15327949pac0804_03

33.

National Science Board. (2014). Science and technology: Public attitudes and understanding. Retrieved from http://www.nsf.gov/statistics/seind14/content/chapter-7/chapter-7.pdf

34.

Nauroth

Gollwitzer

Bender

Rothmund

(2015). Social identity threat motivates science-discrediting online comments. PLoS ONE, 10(2), e0117476. doi:10.1371/journal.pone.0117476

35.

Robertson

J. A.

(2010). Embryo stem cell research: Ten years of controversy. The Journal of Law, Medicine & Ethics, 38, 191-203. doi:10.1111/j.1748-720X.2010.00479.x

36.

Rödder

(2009). Reassessing the concept of a medialization of science: A story from the “book of life.” Public Understanding of Science, 18, 452-463. doi:10.1177/0963662507081168

37.

Rokeach

(1970). Beliefs, attitudes, and values: A theory of organization and change. San Francisco, CA: Jossey-Bass.

38.

Rothmund

Bender

Nauroth

Gollwitzer

(2015). Public concerns about violent video games are moral concerns–How moral threat can make pacifists susceptible to scientific and political claims against violent video games. European Journal of Social Psychology, 45, 769-783. doi:10.1002/ejsp.2125

39.

Schwartz

S. H.

Cieciuch

Vecchione

Davidov

Fischer

Beierlein

. . . Konty

(2012). Refining the theory of basic individual values. Journal of Personality and Social Psychology, 103, 663-688. doi:10.1037/a0029393

40.

Skitka

L. J.

(2010). The psychology of moral conviction. Social & Personality Psychology Compass, 4, 267-281. doi:10.1111/j.1751-9004.2010.00254.x

41.

Steg

De Groot

J. I. M.

(2012). Environmental values. In Clayton

(Ed.), The Oxford handbook of environmental and conservation psychology (pp. 81-92). New York, NY: Oxford University Press. doi:10.1093/oxfordhb/9780199733026.013.0005

42.

Tetlock

P. E.

(2003). Thinking the unthinkable: Sacred values and taboo cognitions. Trends in Cognitive Sciences, 7, 320-324. doi:10.1016/S1364-6613(03)00135-9

43.

Tetlock

P. E.

Kristel

O. V.

Elson

S. B.

Green

M. C.

Lerner

J. S.

(2000). The psychology of the unthinkable: Taboo trade-offs, forbidden base rates, and heretical counterfactuals. Journal of Personality and Social Psychology, 78, 853-870. doi:10.1037//0022-3514.78.5.853

44.

Torelli

C. J.

Kaikati

A. M.

(2009). Values as predictors of judgments and behaviors: The role of abstract and concrete mindsets. Journal of Personality and Social Psychology, 96, 231-247. doi:10.1037/a0013836

45.

Verplanken

Holland

R. W.

(2002). Motivated decision making: Effects of activation and self-centrality of values on choices and behavior. Journal of Personality and Social Psychology, 82, 434-447. doi:10.1037/0022-3514.82.3.434

46.

Wojciszke

(1987). Ideal-self, self-focus and value-behaviour consistency. European Journal of Social Psychology, 17, 187-198. doi:10.1002/ejsp.2420170206

How Moral Threat Shapes Laypersons’ Engagement With Science

Abstract

Keywords

Moral Values and Public Engagement With Science

Value-Behavior Correspondence

Value Salience as an Alternative Explanation

The Present Research

Study 1

Method

Participants and design

Procedure, materials, and measures 6

Results

Manipulation check

Evaluation of scientific evidence

Online posting behavior

Discussion

Study 2

Method

Participants and design

Procedure, materials, and measures

Results

Manipulation check

Evaluation of scientific evidence

Online posting behavior

Discussion

Study 3

Method

Participants and design

Procedure, materials, and measures

Results

Manipulation check

Evaluation of scientific evidence

Discussion

General Discussion

Limitations and Future Research

Conclusion

Footnotes

Acknowledgements

Declaration of Conflicting Interests

Funding

Notes

References

Procedure, materials, and measures⁶