Reducing Group Alignment in Factual Disputes? The Limited Effects of Social Identity Interventions

Abstract

Public divisions over contested science often arise due to individuals’ motivation to hold beliefs in line with others with whom they share a common social identity. To reduce the influence of threatened group identities, this research tests two potential interventions at the point of exposure to new information: self-affirmation and multiple identity salience manipulation. Experiments across dispute contexts—a toxic mining spill, a biofuels mandate, and gene editing technology—show limited support for either intervention reducing group alignment. Together, these results contribute to a growing body of work illuminating the limits of social identity interventions.

Keywords

polarization social identity motivated reasoning risk perception self-affirmation social identity complexity

In many cases, science communicators must compete with the leaders of political parties, religious groups, and others in “the competition of voices reaching the individual” (A. Campbell, Converse, Miller, & Stokes, 1960, p. 128). Members of a given group are often more likely to side with the “facts” offered by their side, resulting in meaningful gaps in beliefs about scientific findings (Pasek, 2017; Veenstra, Hossain, & Lyons, 2014) that are difficult to reduce once established (Nyhan & Reifler, 2010). This has led to both pessimism about public understanding of science (Rohn, 2018) as well as a search for strategies that can ultimately reduce group alignment (Hornsey & Fielding, 2017).

How might the influence of partisanship and other social identities on the uptake of this new information be reduced? Social identity theory (SIT) can help science communicators better understand this polarization among the public and potential ways it might be mitigated. Research derived from SIT suggests that unbiased processing is more likely when an individual’s self-concept is constituted in a way that is more resistant to dominance of a single identity.

The present research tests two interventions with this aim, drawing on SIT. First, self-affirmation, an intervention in which individuals reflect on personally relevant values, may reduce the need for defensive processing by bolstering an unrelated domain of self-worth (Cohen, Aronson, & Steele, 2000). Second, because the self is an exceedingly rich construct where numerous social identities overlap and contribute to one overarching structure (Roccas & Brewer, 2002), invoking this complexity may likewise reduce the capacity of any single social identity to dominate. To achieve this, multiple identity salience manipulation, an intervention in which many of an individual’s social groups are brought to mind simultaneously, is tested here for the first time.

The effects of self-affirmation and multiple identity salience manipulation at the point of exposure to new information are explored in three fictionalized dispute contexts. Results provide limited evidence of either manipulation reducing group alignment. Together, the results suggest that researchers should consider alternative means of navigating polarized disputes over science. This research note concludes with a discussion of ways science communication can instead use group norms to its advantage.

Social Identity and Information Processing

SIT provides the framework through which this research addresses the problem of polarized factual beliefs among the public. One of the social sciences’ most influential theories, SIT informs much of the research examining politicized science (e.g., Hart & Nisbet, 2012; Nyhan & Reifler, 2018). A social identity is “the individual’s knowledge that he belongs to certain social groups together with some emotional and value significance to him of this group membership” (Tajfel, 1972, p. 292). These shared mental representations enable collective behavior and homogenized perception.

The most compelling account of social identity’s source is social comparison (Hogg, 2000; Tajfel, 1972). This account argues that groups’ norms, structures, and relationships to other groups derive from comparisons individuals make between themselves and others, specifically those in groups to which they belong and those in outgroups. These comparisons build into subjective definitions of groups and from these prototypes can emerge. This enables self-categorization, which “depersonalizes self by assimilating self to the in-group prototype,” and generates group perception and behaviors (Hogg, 2000, p. 411).

Self-categorizations and therefore their downstream effects (e.g., the endorsement of beliefs favoring a given social group) are highly context-dependent, because the self-concept exists to mediate adaptively between environment and behavior, regulating the latter (Turner, 1982). In many scenarios, unfortunately, such behavior can be deleterious to the individual and society, as in the case of partisan misperceptions about science, health, and the environment. In other words, when seeing themselves as members of a given group, individuals’ factual beliefs about scientific issues can become aligned with those sharing the common group identity regardless of the actual weight of evidence. This group-alignment in misperceptions may arise due to social identity threat (Nauroth, Gollwitzer, Kozuchowski, Bender, & Rothmund, 2017). When scientific findings reflect poorly on a given social group, or the implications conflict with the group’s values, for instance, the findings represent a symbolic threat to the self-worth an individual group member derives from their membership. When confronted with negative information about a group with which they identify strongly, people respond defensively to degrade the threat, biasing the assimilating new information (de Hoog, 2013).

Self-Affirmation

The potential for psychological interventions of defensive processing has attracted scholars across the social sciences. Recently applied in political domains (e.g., Binning, Brick, Cohen, & Sherman, 2015; Nyhan & Reifler, 2018), self-affirmation originated as a general theory about approaches to self-threat (Steele, 1988). Drawing on many of the same premises as identity protective cognition, self-affirmation researchers suggest that closed-mindedness to challenging information is driven by self-regard. Because a sense of worth is derived from beliefs and their accuracy, conflicting information can threaten self-worth, while defensive rejections can preserve self-integrity (Cohen et al., 2000). Affirmation theorists see defensive responses as adaptations that reduce self-integrity threats (Sherman & Cohen, 2006). These can be maladaptive, though, in blocking learning from threatening information. However, these responses can be displaced by a different adaptive response that does not work through distortion to render threats less challenging. As an indirect psychological adaptation, self-affirmation alternatively allows for both the restoration of self-integrity and accurate perception (Sherman & Cohen, 2006).

The self-affirmation theoretical framework hinges on the premise that the self-system is flexible. Global self-integrity is the product of many domains, and the fungibility of the system allows for threats to one domain to be offset by bolstering another (Steele, 1988). Individuals can be affirmed through activities that make salient other core aspects of their self-worth. Threats then lose their potency as individuals reduce the prominence of the threatened domain, putting the threat into broader perspective and eliminating the need for ego protection. While much self-affirmation research addresses situations where information challenges personal identity, threats to social identities also can be buffered through indirect adaptions. This leads to less anchoring of group evaluations in one’s self-concept, allowing individuals to evaluate groups independent of their self-evaluation.

However, while the literature on self-affirmation effects is long, recent studies have begun to question the uniformity of the results, particularly in domains of politics and science (Kotcher, 2016; Levendusky, 2018; Nyhan & Reifler, 2018; Reavis, Ebbs, Onunkwo, & Sage, 2017; Stroud, 2015). Reavis et al. (2017), for example, report a failed intervention in the context of vaccination intention, and in a study of self-affirmation’s potential to mitigate selective exposure, Stroud (2015) writes that “it seems probable that self-affirmation effect is not as consistent as currently reported in the research literature” (p. 10). Indeed, recent large-scale replication studies in social psychology suggest that affirmation interventions are at best “fragile” (Hanselman, Rozek, Grigg, & Borman, 2017; Protzko & Aronson, 2016).

Social Identity Complexity and Relative Prominence

There may be other ways to alter self-perspective to reduce defensiveness. Social identity complexity—individuals’ subjective representation of their multiple identities’ interrelationships (Roccas & Brewer, 2002)—may provide leverage against group alignment. Individuals “probably possess a greater amount of information about the self than any other cognitive domain” (Linville, 1985, p. 96). A unitary structure could not handle the demands of this rich self-relevant data. A more complex representational structure of the self is necessary, particularly to manage the multiple social identities individuals possess, which may overlap to different degrees. The dynamic nature of these representational structures may provide opportunities to alter an identity’s effects on information processing.

Only one study has manipulated social identity complexity, what the authors refer to as “the perceived relationship among and thus prominence of constituent identities” (Grant & Hogg, 2012, p. 539). Grant and Hogg (2012) propose that both identity uniqueness and overlap are aspects of relative prominence and affect uncertainty-reducing group identification through the same logic. Both perceived uniqueness and low perceived overlap aspects contribute to increase group prominence within the self-concept.¹ These authors show that when an identity has to compete with more salient groups in the cognitive repertoire, that identity will be less prominent and bear less on self-conception and subsequent cognition (Grant & Hogg, 2012). This research applies this logic to identity protective cognition. By raising the salience of multiple identities at once, a multiple prominence intervention may reduce the sway of particular group commitments.

Hypotheses

The theoretical underpinnings of both self-affirmation and social identity complexity point to the potential for reducing biased assimilation of social identity-threatening information. Prior studies (see Sherman & Cohen, 2006) suggest this should be most likely among those most strongly affiliated with their group. Therefore, this research predicts both interventions will be effective in reducing group-aligned beliefs, and effects will be moderated by strength of affiliation. These effects should hold across issues and social identities, and so hypotheses are posed for Studies 1 to 3.

Hypothesis 1: Self-affirmation will reduce group alignment in factual beliefs.

Hypothesi 2: Multiple identity salience will reduce group alignment in factual beliefs.

Hypothesis 3: Self-affirmation will reduce group alignment more among strongly affiliated group members than among weakly affiliated members.

Hypothesis 4: Multiple identity salience will reduce group alignment more among strongly affiliated group members than among weakly affiliated members.

Overview of Studies

Each hypothesis was tested across three studies. The stimuli of these studies varied in order to address different contexts in which social identity threat may impede effective science communication. Across the studies, the research examines two kinds of factual beliefs: attribution of blame (Study 1) and risk/benefit perception (Studies 2 and 3). The studies also vary the orienting social identity, from partisanship (Studies 1 and 2) to religious orientation (Study 3).

Drawing on the technological-environmental risk paradigm (Baxter & Eyles, 1999), Study 1 looks at partisan attribution of blame for a mining spill. Study 2 examines perceived benefits and risks in a partisan dispute surrounding advanced biofuels, and Study 3 examines perceived benefits and risks in a religious/secular dispute surrounding gene editing technology. As this research focuses on the biased uptake of new information, fictionalized dispute events were used in each stimulus, which also allowed greater experimental control over the inclusion of group cues (e.g., Lyons, 2017; Hart & Nisbet, 2012). Each stimulus dispute was modeled on a real-world dispute, however (e.g., Cama, 2018; Joseph, 2016; Palmer, 2015).

In each study, participants were randomly assigned to one of four conditions: self-affirmation, multiple identity salience, single group identity salience, or a control.

Study 1

Data for each study were collected using Amazon Mechanical Turk’s (mTurk) panel of online workers, who complete surveys and other intelligence tasks. All participants were compensated monetarily.² Studies also show comparable effects between nationally representative samples and those gathered through mTurk (Coppock, 2018; Mullinix, Leeper, Druckman, & Freese, 2015).

257 participants³ were recruited for Study 1 in October 2015. Demographics are reported in the supplementary material (Table A1). Age, race, gender, income, education, party, affiliation strength, and political interest did not vary significantly across conditions.

Design and Procedure

Participants in the self-affirmation condition were provided with a list of personal values. After selecting the most personally important value, they were instructed to write about why the value is important to them. The list (Personal Values Questionnaire II, Cullen, 2014) is similar to most common affirmation instruments (e.g., Nyhan & Reifler, 2018) but gives greater focus to internal values.

Multiple identity salience was manipulated with a prompt that asked participants “Which identities do you consider important in providing you with a sense of who you are?” To illustrate the concept, participants were provided with a list of examples, and asked to describe how any identities selected “applies to you and what it means to you.” This method echoes the numerical identity prominence manipulation used by Grant and Hogg (2012), which asked participants to write about either 1 or 3 social identities “important in providing a sense of who they are” (p. 540), while also intending to stimulate a larger repertoire of accessible identities, and thus lower prominence of any one. Of the two methods employed by Grant and Hogg (2012), the numerical approach was chosen to decrease prominence instead of the overlap approach because the logic by which it reduces prominence can more easily be extended to a larger repertoire and hypothetically, lower prominence of a potentially threatened group.

For contrast with the multiple identity salience condition, a political identity salience condition (i.e., a single group) introduced by Klar (2013) was included. Participants in the control listed everything they had to eat or drink in the past 48 hours (Nyhan & Reifler, 2018).

Following the writing prompt, participants completed a battery of manipulation checks.⁴ Next, they were informed that they would be asked about

a recent event in the news. An investigation by the United States Environmental Protection Agency at an Alaskan mine went badly awry earlier this year, triggering a spill of zinc, iron, copper, and other heavy metals into the water supply.⁵

Next, participants read a news story on the spill that reported two statements from each group (Republicans vs. Democrats and the Environmental Protection Agency [EPA]) that blamed the other for the spill. Participants then reported their belief in each of four claims.

Measures

Group-aligned belief was computed based on the following factual beliefs: “The EPA probably allowed the toxic spill to occur on purpose” (M = 3.02, SD = 1.69); “The EPA probably is not being held to the same standard they would apply to a private business” (M = 4.57, SD = 1.58); “Republican lawmakers were probably willing to risk the spill in order to discredit the EPA” (M = 4.10, SD = 1.73); and “Republicans are probably using the spill to undercut the Obama administration’s rollout of emissions regulations” (M = 4.47, SD = 1.69). Beliefs that blamed Republicans were reverse coded so that all beliefs ranged from Democrat-aligned (1) to Republican-aligned (7). These were combined,⁶ recoded with a midpoint at 0, and crossed with respondent party to create a final group-aligned belief measure that excluded Independents (M = 0.56, SD = 0.87, n = 191). This measure was converted to a standardized z score.

To test for interventions’ effects across levels of group identification, strength of party affiliation (31.5% strong, n = 81) was also measured.

Results

Hypotheses were tested using an analysis of variance (ANOVA) model with condition, strength of party affiliation, and their interaction term as factors and party-aligned belief as dependent variable.⁷ Strength of affiliation had a significant main effect on party-aligned belief, F(1, 183) = 17.85, p < .001, $η_{p}^{2}$ = .08, while condition’s effect was not significant, F(3, 183) = 2.14, p = .097, $η_{p}^{2}$ = .03. The interaction was not significant, F(3, 183) = 2.39, p = .070, $η_{p}^{2}$ = .04.

To build on Study 1, two parallel studies were simultaneously conducted.

Study 2

A total of 598 participants were recruited via mTurk in March 2016. Demographics are reported in the supplementary material (Table A1). Age, race, gender, income, education, party, affiliation strength, and political interest did not vary significantly across conditions.

The Study 2 procedure replicated Study 1, except with the context altered to a partisan dispute over a fictional advanced biofuels mandate. The news story was presented as a 2-minute news video, with the transcript embedded below. The video included claims regarding the benefits of an advanced biofuels mandate, attributed to Democrats, and claims regarding risks, attributed to Republicans. Participants then answered questions about perceived risks and benefits.

Measures

Risk and benefit beliefs were measured on 7-point scales (1 = strongly disagree, 7 = strongly agree). These items were drawn from prior studies (see Fung, Choi, Scheufele, & Shaw, 2014), and covered economic, environmental, and social aspects of biofuels technology. Risk perception was measured with six items, Cronbach’s α = .85, M = 3.77, SD = 1.21. Benefit perception was also measured with six items, Cronbach’s α = .90, M = 5.21, SD = 1.13. Group-aligned beliefs were calculated by subtracting risk from benefits and crossing with respondent party, excluding true independents (M = 0.69, SD = 1.07, n = 507). This measure was converted to a standardized z score.

Participants also reported party affiliation strength (52% strong, n = 265) to be used as the moderator.

Results

Hypotheses were again tested using an ANOVA model with party-aligned risk-benefit perception as the dependent variable and condition, strength of affiliation, and their interaction as fixed factors. Strength of affiliation increased party alignment, F(1, 499) = 29.64, p < .001, $η_{p}^{2}$ = .06. Condition, F(3, 499) = .47, p = .706, $η_{p}^{2}$ = .00, and the interaction, F(3, 499) = .74, p = .527, $η_{p}^{2}$ = .00, were not significant in the model.

Study 3

A total of 399 participants were recruited via mTurk in March 2016. Demographics are reported in the supplementary material (Table A1). Study 3 pivots from partisan polarization to examine religious belief as the orienting identity. Because of the stimulus posed a debate between faith communities writ large versus secular, humanist, and nonreligious groups, the analysis used the World Values Survey (Inglehart et al., 2014) measure of religious (39.3%, n = 157) versus not religious and atheist (60.7%, n = 242) as the primary grouping for the construction of variables and analysis of group alignment.⁸

Age, race, gender, income, education, identification as religious, religiosity, and fundamentalism did not vary significantly across conditions.

The design and procedure were essentially identical to those in Study 2, but with religious cues removed from the writing prompt lists to minimize unintended effects. The wording of the identity salience prompt was slightly altered to religious rather than political terms. The stimulus materials featured a dispute between religious groups and secular humanist groups over the controversial science of gene editing technology, with religious groups stressing the risks and secular figures stressing benefits.

Measures

Beliefs about the disputed issue were assessed by measuring perception of a series of risks and benefits claimed by the opposing groups in the stimulus, on 7-point scales. These items were drawn from existing news coverage of the technology. Risk perception was measured with three items: “Gene editing’s unpredictable effects on future generations,” “Mis-targeted application of gene editing in individuals using current techniques,” and “Environmental risks of gene editing, such as the extinction of insect species,” Cronbach’s α = .85, M = 4.95, SD = 1.45. Benefit perception was measured with four items: Gene editing’s potential to “eliminate birth defects and genetic disorders,” “enhance our understanding of how embryos develop,” “reduce deaths from chronic diseases,” and “delay the aging process,” Cronbach’s α = .89, M = 5.37, SD = 1.37. A measure of group-aligned benefit-risk perception was constructed by centering the benefit and risk scales, subtracting risk from benefit to create a net measure, and then crossing with religious orientation to match the position taken by each group in the stimulus (religious = −1, not religious or atheist = 1). This measure was converted to a standardized z score.

Participants also reported religious affiliation strength (63% strong, n = 250) to be used as the moderator.

Results

Hypotheses were again tested using an ANOVA model with party-aligned risk-benefit perception as the dependent variable, and condition, strength of affiliation, and their interaction as fixed factors. Strength of affiliation increased group alignment, F(1, 389) = 18.95, p < .001, $η_{p}^{2}$ = .05. Conditions differed, F(3, 389 = 2.73, p = .044, $η_{p}^{2}$ = .02, but the interaction, F(3, 389) = .73, p = .537, $η_{p}^{2}$ = .01, was not significant. The means for each condition were as follows: M_multiple = −0.07, SD = 0.95, n = 90, M_affirmation = 0.09, SD = 0.91, n = 87, M_control = 0.12, SD = 1.00, n = 120, M_group = −0.16, SD = 1.10, n = 100. However, pairwise comparisons using Tukey’s HSD (honestly significant difference) adjustment showed no significant differences.

Studies 1, 2, and 3 Combined Analysis

Because of the largely parallel design and data structure of the three studies, a final analysis was conducted on a combined dataset (see Figure 1). An analysis of variance was conducted for this merged data (N = 1,095), with the data source (Study) as a covariate and condition and identity strength as fixed factors. While identity strength was a significant factor in group alignment, F(1, 1,086 = 59.80, p < .001, $η_{p}^{2}$ = .05, there were no significant differences among treatments, F(3, 1,086 = 2.14, p = .094, $η_{p}^{2}$ = .01, nor was the interaction term significant, F(3, 1,086) = 0.61, p = .611, $η_{p}^{2}$ = .00.⁹ A full summary of the results is presented in Table A4 in the supplementary material. For all subgroup means and standard deviations, see Table A5.

Figure 1.

Group alignment across condition and strength of affiliation (Studies 1-3). Error bars are 95% confidence intervals.

Discussion

This research tested two potential interventions aimed at reducing in-group bias in disputes surrounding environmental and scientific controversies. Both of these were intended to displace maladaptive defensive processing with an indirect psychological adaptation, allowing individuals to maintain self-integrity while more evenhandedly processing a given issue. Results showed neither successfully reduced the sway of group commitments. This outcome was consistent across each of the three contexts examined, in the combined model, and in the internal meta-analysis.

However, these results are a step toward better understanding the role of social identity in information processing. Together they suggest less contextual variation in identity salience effects on information uptake than we might otherwise assume (Turner, 1982)—at least under low-salience information processing conditions in survey experiments, and perhaps in news consumption outside the lab.¹⁰ Group identity clearly influences the formulation of factual beliefs in such scenarios—as evidenced by the significant role of strength of affiliation in all three studies presented here—but the influence of trait-based identity does not appear to be as sensitive to manipulation as theorized. While social identity interventions may be more viable in certain contexts, strategic depolarization will require a more consistent approach.

That the findings contradict prevailing research on self-affirmation suggest the uniformity of effects reported in the published literature so far may provide an incomplete picture. However, the findings reported here do not invalidate those of other studies. Rather, they may support Protzko and Aronson (2016), who suggest that “affirmation interventions, while clearly effective sometimes, may require certain conditions to be operative” (p. 506). It could be that the setup of the experiments reported here reduces the impact of affirmation. Nevertheless, these results contribute to a growing body of work finding little effect of self-affirmation on a range of outcomes governed by social identity including affective polarization (Levendusky, 2018), news selection (Stroud, 2015), and factual beliefs (Kotcher, 2016; Nyhan & Reifler, 2018; Reavis et al., 2017), some using preregistered analyses of nationally representative samples. For this reason, researchers have increasing reason to believe self-affirmation, and potentially related approaches to intervening on social identity, are less than ideal methods.

Kotcher (2016) provides an empirical explanation for the self-affirmation method’s inconsistent effects:

Self-affirmation not only can activate multiple psychological processes, but [. . .] it may activate both productive and counter-productive processes simultaneously. . . This suggests that self-affirmation is at the same time both more complex than previously understood [. . .] and less precise as a potential intervention than one might hope. (pp. 68-69)

Ultimately, then, self-affirmation interventions are not precise enough to allow science communicators to produce desired responses without the danger of concurrently eliciting negative responses. More broadly, it is important to remember that even empirically robust theoretical claims about psychological states “do not easily or automatically yield corresponding dependable generalizations about message design or communication effects” (O’Keefe, 2012, p. 15).

Limitations

As with any research, some limitations should be noted. One concern that the data cannot address is amount of overlap in respondents’ existing social circles. Whether or not an individual’s various social identities are distinct or mostly conterminous will likely moderate the effect of conjuring associations with those many different groups. More specifically, thinking of many different identities may be less reductive of a particular identity’s influence if those other identities are highly interconnected in the person’s mind (see Mason, 2016). This implies the effects detected in this research may be conservative. Future work should control for trait-based identity complexity (vs. state complexity induced in the manipulation) and may then find the effect of the intervention to be stronger.

More generally, understanding of social identity complexity and related relative prominence intervention is limited; only one other variation has been tested (Grant & Hogg, 2012). The results here might be extended by both simple replication as well as tests of variations of the instrument. In particular, Grant and Hogg identified two dimensions of relative identity prominence—uniqueness and overlap. It may be possible to amplify the uniqueness approach employed here, for instance asking participants to expend greater effort in considering each identity. It might also be the case that the overlap approach is more effective at reducing relative prominence, although Grant and Hogg found relatively symmetrical effects between the two.

Directions for Future Research

It is possible that further refinement to both interventions could result in more consistent effects. However, given that even robust laboratory findings can be less effective at scale (Stenhouse, 2017), these results also warrant greater caution on behalf of researchers studying polarization and practitioners implementing best practices. Instead, in scenarios when group-driven polarization cannot be prevented (Kahan, Jamieson, Landrum, & Winneg, 2017; van der Linden, Leiserowitz, Rosenthal, & Maibach, 2017), researchers and science communication practitioners may be better served by working with group identity.

Science communication in this vein attempts to show how scientific findings and the responses they may require are not a threat to one’s group identity. Rather than reducing the salience of identity altogether, science communicators can address solution aversion (T. H. Campbell & Kay, 2014), for instance, by framing the solution as identity-amenable (T. H. Campbell & Kay, 2014; Kahan, Jenkins-Smith, & Braman, 2011; Kahan, Jenkins-Smith, Tarantola, Silva, & Braman, 2015). In this case, free-market climate change solutions appeal to conservatives and have been found to reduce overall rejection of climate science.

Along the same lines, a long literature shows that employing messengers seen as credible by the target audience can be an effective route to depolarization (Kahan et al., 2011; Kahan et al. 2015; Lupia, 2013). Although there are some limits to this strategy as well (Landrum, Lull, Akin, Hasell, & Jamieson, 2017), in-group messengers are on balance more persuasive than their counterparts. In the context of climate change, for example, highlighting the Pentagon’s response could help leverage the high regard for the military among American Conservatives (O’Sullivan & Emmelhainz, 2014). Demonstrating agreement among a diverse set of experts can likewise reduce perceived identity threat (Kahan et al., 2011). In all these cases, science communication research shows that bitter division over scientific claims is avoidable if individuals are allowed to be themselves—and good group members—while also accepting the facts. Given the evidence base for interventions like self-affirmation, this may be a better route forward.

Conclusion

Ultimately, this research provides a step toward unraveling the pernicious effects of identity-protective cognition on individuals’ interactions with information by testing self-affirmation alongside a novel intervention based on the concept of social identity complexity, or relative identity prominence. Although neither of the interventions are silver bullets, the results detailed here point the way toward a more comprehensive understanding how social identity integrates with and influences the self in cognitive processes. Finally, they suggest science communication researchers and practitioners may be better served by seeking alternative paths to depolarization. We may be best served by harnessing group affiliation rather than trying to momentarily erase it.

Footnotes

Acknowledgements

The author would like acknowledge Aaron Veenstra, Scott McClurg, Wenjing Xie, Ken Mulligan, Yueh-Ting Lee, Joe Hilgard, and the anonymous reviewers and editor for feedback and assistance with this project.

Author’s Note

The data, code, and supplementary material can be found at osf.io/mtdc2.

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 project has received funding for Lyons’ time from the European Research Council (ERC) under the the European Union’s Horizon 2020 research and innovation programme [grant agreement number 682785].

ORCID iD

Benjamin Lyons

Notes

Author Biography

Benjamin Lyons is a research fellow in the Department of Politics at the University of Exeter studying political communication, science communication, and public opinion. His research focuses on misperceptions and fact-checking, particularly in the context of politicized science.

References

Baxter

J. W.

Eyles

J. D.

(1999). The utility of in-depth interviews for studying the meaning of environmental risk. Professional Geographer, 51, 307-320.

Binning

K. R.

Brick

Cohen

G. L.

Sherman

D. K.

(2015). Going along versus getting it right: The role of self-integrity in political conformity. Journal of Experimental Social Psychology, 56, 73-88.

Cama

(2018, March 8). Dem bill would overhaul ethanol mandate. The Hill. Retrieved from http://thehill.com/policy/energy-environment/377438-dems-bill-takes-aim-federal-ethanol-mandate

Campbell

Converse

P. E.

Miller

W. E.

Stokes

D. E.

(1960). The American voter. New York, NY: Wiley.

Campbell

T. H.

Kay

A. C.

(2014). Solution aversion: On the relation between ideology and motivated disbelief. Journal of Personality and Social Psychology, 107, 809-804.

Carey

J. M.

Nyhan

Valentino

Liu

(2016). An inflated view of the facts? How preferences and predispositions shape conspiracy beliefs about the Deflategate scandal. Research & Politics, 3(3). doi:10.1177/2053168016668671

Cohen

G. L.

Aronson

Steele

C. M.

(2000). When beliefs yield to evidence: Reducing biased evaluation by affirming the self. Personality and Social Psychology Bulletin, 26, 1151-1164.

Coppock

(2018). Generalizing from survey experiments conducted on Mechanical Turk: A replication approach. Political Science Research Methods. Advance online publication. doi:10.1017/psrm.2018.10

Cullen

(2014). A single case design study evaluating the impact of a values based positive self affirmations intervention on eating disorder symptoms in women with bulimia nervosa (Doctoral dissertation). Retrieved from http://uhra.herts.ac.uk/bitstream/handle/2299/14803/10280087%20Cullen%2c%20Ella-%20Final%20DClinPsy%20Submission.pdf?sequence=1&isAllowed=y

10.

de Hoog

. (2013). Processing of social identity threats: A defense motivation perspective. Social Psychology, 44, 361-372. doi:10.1027/1864-9335/a000133

11.

Faul

Erdfelder

Lang

A. G.

Buchner

(2007). G* Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175-191.

12.

Fung

T. K.

Choi

D. H.

Scheufele

D. A.

Shaw

B. R.

(2014). Public opinion about biofuels: The interplay between party identification and risk/benefit perception. Energy Policy, 73, 344-355.

13.

Grant

Hogg

M. A.

(2012). Self-uncertainty, social identity prominence and group identification. Journal of Experimental Social Psychology, 48, 538-542.

14.

Hanselman

Rozek

C. S.

Grigg

Borman

G. D.

(2017). New evidence on self-affirmation effects and theorized sources of heterogeneity from large-scale replications. Journal of Educational Psychology, 109, 405-424.

15.

Hart

P. S.

Nisbet

E. C.

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

16.

Hogg

M. A.

(2000). Social identity and social comparison. In Handbook of social comparison (pp. 401-421). Springer, Boston, MA.

17.

Hornsey

M. J.

Fielding

K. S.

(2017). Attitude roots and Jiu Jitsu persuasion: Understanding and overcoming the motivated rejection of science. American Psychologist, 72, 459-473.

18.

Inglehart

R., C.

Haerpfer

Moreno

Welzel

Kizilova

Diez-Medrano

. . . Puranen

. (2014). World values survey. Madrid, Spain: JD Systems Institute.

19.

Joseph

(2016, October 14). Gene-editing, religion and one scientist’s quest to reconcile the two. PBS. Retrieved from https://www.pbs.org/newshour/science/gene-editing-religion-scientist

20.

Kahan

D. M.

Jamieson

K. H.

Landrum

Winneg

(2017). Culturally antagonistic memes and the Zika virus: An experimental test. Journal of Risk Research, 20(1), 1-40.

21.

Kahan

D. M.

Jenkins-Smith

Braman

(2011). Cultural cognition of scientific consensus. Journal of Risk Research, 14, 147-174.

22.

Kahan

D. M.

Jenkins-Smith

Tarantola

Silva

C. L.

Braman

(2015). Geoengineering and climate change polarization: Testing a two-channel model of science communication. Annals of the American Academy of Political and Social Science, 658, 192-222.

23.

Klar

(2013). The influence of competing identity interventions on political preferences. Journal of Politics, 75, 1108-1124. doi:10.1017/S0022381613000698

24.

Kotcher

(2016). Reducing resistance to communication about climate change using an applied self-affirmation intervention (Doctoral dissertation, George Mason University). Retrieved from http://digilib.gmu.edu/jspui/bitstream/handle/1920/10545/Kotcher_gmu_0883E_11279.pdf?sequence=1&isAllowed=y

25.

Landrum

A. R.

Lull

R. B.

Akin

Hasell

Jamieson

K. H.

(2017). Processing the papal encyclical through perceptual filters: Pope Francis, identity-protective cognition, and climate change concern. Cognition, 166, 1-12.

26.

Levendusky

M. S.

(2018). When efforts to depolarize the electorate fail. Public Opinion Quarterly, nfy036.

27.

Linville

P. W.

(1985). Self-complexity and affective extremity: Don’t put all of your eggs in one cognitive basket. Social Cognition, 3, 94-120.

28.

Lupia

(2013). Communicating science in politicized environments. Proceedings of the National Academy of Sciences, U S A, 110(Suppl. 3), 14048-14054.

29.

Lyons

B. A.

(2017). When readers believe journalists: Effects of adjudication in varied dispute contexts. International Journal of Public Opinion Research, edx013.

30.

Mason

(2016). A cross-cutting calm: How social sorting drives affective polarization. Public Opinion Quarterly, 80(Suppl. 1), 351-377.

31.

Mullinix

K. J.

Leeper

T. J.

Druckman

J. N.

Freese

(2015). The generalizability of survey experiments. Journal of Experimental Political Science, 2, 109-138.

32.

Nauroth

Gollwitzer

Kozuchowski

Bender

Rothmund

(2017). The effects of social identity threat and social identity affirmation on laypersons’ perception of scientists. Public Understanding of Science, 26, 754-770.

33.

Nyhan

Reifler

(2010). When corrections fail: The persistence of political misperceptions. Political Behavior, 32, 303-330. doi:10.1007=s11109-010-9112-2

34.

Nyhan

Reifler

(2018). The roles of information deficits and identity threat in the prevalence of misperceptions. Journal of Elections, Public Opinion and Parties. Advance online publication. 10.1080/17457289.2018.1465061

35.

O’Keefe

(2012). From psychological theory to message design: Lessons from the story of gain-framed and loss-framed persuasive messages. In Cho

(Ed.), Health communication message design (pp. 3-20). Thousand Oaks, CA: Sage.

36.

O’Sullivan

T. M.

Emmelhainz

(2014). Reframing the climate change debate to better leverage policy change: An analysis of public opinion and political psychology. Journal of Homeland Security and Emergency Management, 11, 317-336.

37.

Palmer

(2015, August 25). Congress is as much to blame for the Gold King mine spill as the EPA. Pacific Standard. Retrieved from https://psmag.com/environment/congress-is-as-much-to-blame-for-the-gold-king-mine-spill-as-the-epa

38.

Pasek

(2017). It’s not my consensus: Motivated reasoning and the sources of scientific illiteracy. Public Understanding of Science, 27, 787-806. 10.1177/0963662517733681

39.

Protzko

Aronson

(2016). Context moderates affirmation effects on the ethnic achievement gap. Social Psychological and Personality Science, 7, 500-507.

40.

Reavis

R. D.

Ebbs

J. B.

Onunkwo

A. K.

Sage

L. M.

(2017). A self-affirmation exercise does not improve intentions to vaccinate among parents with negative vaccine attitudes (and may decrease intentions to vaccinate). PLoS One, 12, e0181368.

41.

Roccas

Brewer

M. B.

(2002). Social identity complexity. Personality and Social Psychology Review, 6, 88-106.

42.

Rohn

(2018, August 29). I was deluded. You can’t beat fake news with science communication. The Guardian. Retrieved from https://www.theguardian.com/science/occams-corner/2018/aug/29/science-communication-fake-news

43.

Sherman

D. K.

Cohen

G. L.

(2006). The psychology of self-defense: Self-affirmation theory. Advances in Experimental Social Psychology, 38, 183-242.

44.

Steele

C. M.

(1988). The psychology of self-affirmation: Sustaining the integrity of the self. Advances in Experimental Social Psychology, 21, 261-302.

45.

Stenhouse

(2017). Spreading success beyond the laboratory: Applying the RE-AIM framework for effective environmental communication interventions at scale. Environmental Communication, 11, 756-768.

46.

Stroud

N. J.

(2015, November). Self-affirmation, moral foundations, and disrupting partisan selectivity. Paper presented at the Conference of the National Communication Association, Political Communication Division, Las Vegas, NV.

47.

Tajfel

(1972). La categorisation sociale [Social categorization]. In Moscovici

(Ed.), Introduction à la psychologie sociale [Introduction to social psychology] (Vol. 1, pp. 272-302). Paris, France: Larousse.

48.

Turner

J. C.

(1982). Towards a cognitive redefinition of the social group. In Tajfel

(Ed.), Social identity and intergroup relations (pp. 15-40). Cambridge, England: Cambridge University Press.

49.

Van der Linden

Leiserowitz

Rosenthal

Maibach

. (2017). Inoculating the public against misinformation about climate change. Global Challenges, 1(2). doi:10.1002/gch2.201600008

50.

Veenstra

A. S.

Hossain

M. D.

Lyons

B. A.

(2014). Partisan media and discussion as enhancers of the belief gap. Mass Communication and Society, 17, 874-897.

51.

Viechtbauer

(2010). Conducting meta-analyses in R with the metafor package. Journal of Statistical Software, 36, 1-48.