key: cord-0862394-yeq5zlaz
authors: Chen, Kaiyuan; Zhang, Jinghui; Ao, Xiang; Ramdass, Jeff
title: The burden of being certain: National identity certainty predicts support for COVID‐Related restrictive measures and outgroup conspiracy beliefs
date: 2022-03-03
journal: J Appl Soc Psychol
DOI: 10.1111/jasp.12868
sha: 8fd3a8632ea8e85e2417ae34c144ce7d81fbea4e
doc_id: 862394
cord_uid: yeq5zlaz

In dealing with the COVID‐19 pandemic, government officials often encounter two concurrent concerns: they have to enforce necessary public health and safety measures to manage COVID‐19. Meanwhile, they also have to mitigate conspiracy beliefs about COVID‐19. To shed light on these issues, we conducted two studies to investigate national identity certainty (i.e., the extent to which people are certain about their national identity) as a predictor of (a) support for restrictive measures to curtail COVID‐19 and (b) conspiracy beliefs about an outgroup as the culprit of COVID‐19. Study 1 was a three‐week longitudinal study (N = 301) where we investigated the relationships both on a between‐person level (differences between individuals) and on a within‐person level (week‐by‐week fluctuations of the same individual). We found that individual differences in national identity certainty predicted increased support for restrictive measures and increased outgroup conspiracy beliefs. These relationships emerged, even when we controlled for national identity positivity, that is, the extent to which people see their national identity in positive light. In Study 2 (N = 316), we used a cross‐sectional correlational design and replicated the findings of Study 1. Moreover, we found that the relationships were explained by distinct threat perceptions: realistic threat explained the increased support for restrictive measures, whereas symbolic threat explained the increased outgroup conspiracy beliefs. Overall, our findings suggest that support for restrictive measures and outgroup conspiracy beliefs can be seen as attempts of people high in national identity certainty to address the distinct threats of COVID‐19.

, peoples' adherence to these measures varies considerably (Allcott et al., 2020; Graham et al., 2020) . Moreover, some people want more restrictive measures than what were enacted in their communities, such as stringent isolation protocols for anyone infected with or travel ban at the borders, while others want less. For instance, in a recent survey by Devlin et al. (2021) for the Pew Research Center, 56% of U.S. adults thought there should have been more restrictions on public activity whereas 26% thought there should have been fewer restrictions. The researchers reported similar splits in public opinions in other advanced economies (e.g., France, 40% vs 24%).

Meanwhile, many people believe in conspiracy theories about COVID-19 (Douglas, 2021; Enders et al., 2020; Romer & Jamieson, 2020; Sternisko et al., 2021; Uscniski et al., 2020) . Examples of these beliefs include the idea that COVID-19 is humanmade or that the pandemic is less serious than its portrayal by public health and government officials. Conspiracy beliefs may negatively affect people by leading them to defy public health and safety measures such as social distancing (Biddlestone et al., 2020; Bierwiaczonek et al., 2020; Freeman et al., 2022; Romer & Jamieson, 2020) .

Successful management of COVID-19 could thus involve rallying people around restrictive measures (when there is a need) while mitigating conspiracy beliefs about COVID-19. To address these concerns, it is essential to understand why people may support restrictive measures and COVID-related conspiracy theories. To date, relevant research has been emerging. For example, research has attributed the support for restrictive measures to perceptions of COVID-19 as a realistic threat (Kachanoff et al., 2020) . COVIDrelated conspiracy theories have, on the other hand, been attributed to dispositions and ideological motivations or beliefs about a nation's greatness (Sternisko et al., 2021) . The current research adjoins prior research by identifying a factor that could contribute to these issues but has received relatively less attention:

individual difference in national identity certainty. Integrating previous research (e.g., Chen et al., 2004; DeMarree et al., 2007; Godinic et al., 2020; Jaspal & Nerlich, 2020; Kachanoff et al., 2020) , we contend that people high in national identity certainty should be more likely to support restrictive measures and believe in conspiracy theories that blame COVID-19 on an outgroup.

People struggle to feel certain about the world (Festinger, 1954) .

Feelings of uncertainty, particularly those about themselves, can be unsettling (Jonas et al., 2014) . One thing people strive to be somewhat certain about is their social identity. According to social identity theory (Abrams & Hogg, 2010; Tajfel & Turner, 1979) and its recent development, uncertainty-identity theory (Hogg, 2007) , people categorize each other in group terms (e.g., "us" vs. "them"), forming social identities at various levels (gender identity, national identity etc.). Social identities prescribe people's behaviors and are said to be a viable source of certainty. Indeed, research suggests that when people feel uncertain about themselves, they tend to strengthen the bond with their social identities (Grieve & Hogg, 1999; Hogg & Mahajan, 2018) . The implication is that most people have a need to reduce feelings of uncertainty, which can be fulfilled by social identities.

The majority of work on uncertainty-identity theory focuses on feelings of uncertainty, a situationally induced, unsettling state (Choi & Hogg, 2020) . We, on the other hand, focused on individual differences in social identity certainty-and more specifically national identity certainty-as a relatively stable disposition or trait that varies between individuals. That is, individuals differ in the clarity and correctness in their metacognitive assessment of self-concepts (Baumgardner, 1990; Chen et al., 2004; DeMarree et al., 2007; Pelham, 1991) . For example, when asked about what it means by being an "American", some Americans may give answers with great confidence, whereas others may hold some reservations-they may consider to be "American" is to "support diversity", but they may admit they are unsure they are correct. Individual differences as such could stem from the fact that some people are dispositionally more tolerant of uncertainty (Hillen et al., 2017) or are less in need of certainty (Webster & Kruglanski, 1998) . Still, people, regardless of being high or low in national identity certainty, could experience feelings of uncertainty that fluctuate across different situations and engage in correspondent uncertainty-reduction. In this sense, our work operates on a different level of analysis (i.e., trait vs. state) than traditional uncertainty-identity theory research.

Research has suggested people high in identity certainty should in general invest more epistemically into that identity (Chen et al., 2004; DeMarree et al., 2007; Pelham, 1991) ; their views about the identity should be stable and resistant to change. Given their investment, people high in identity certainty should also be more motivated to sustain the self-views against threats that may result in self-view inconsistencies (Festinger, 1957; Swann & Read, 1981) . For instance, Chen et al. (2004) demonstrated that people who are highly certain about their group attributes prefer to interact with an ingroup partner who confirms (vs. disconfirms) these attributes-even when these attributes are negative. In contrast, those who are less certain about group attributes do not selectively attend to ingroup partners in a similar manner. Built on prior research, we propose that under the ongoing pandemic, people high in national identity certainty should experience more acute threat from COVID-19. This threat could take the specific form of threat to the health or mortality of group members (including the individual), identity-related norms, or economic opportunities.

To elaborate, people may perceive COVID-19 as an identity threat (Godinic et al., 2020; Jaspal & Nerlich, 2020; Kachanoff et al., 2020) . The virus prompts measures of social distancing that change people's daily routines and disrupt their sense of self (Jaspal & Nerlich, 2020) or their nation's norms (e.g., individualism and anti-statism, Bazzi et al., 2020) . In addition, COVID-19 has delivered a heavy blow on regional and global economies (Bartik et al., 2020; Carlsson-Szlezak et al., 2020) and consequently rendered the economic prospect of a nation uncertain (Godinic et al., 2020) . Any of these threats can prompt people to question the certainty of their national identity, rendering the latter in the brim of collapse.

Because people high in national identity certainty epistemically have more at stake, they should be more threatened by Those high in national identity certainty should be more motivated to take actions to resolve that threat. One action they could take is to support restrictive measures. Even if restrictive measures can be costly to social and personal life, they are powerful tools that could quell the immediate threat of the pandemic (e.g., by stopping the spread, Arshed et al., 2020; Breitenbach et al., 2020) . We thus predict:

Those high (vs. low) national identity certainty should be more inclined to support restrictive measures to manage COVID-19 (H1).

Meanwhile, an alternative for people high in national identity certainty is to believe in conspiracy theories, particularly those that depict an opposing outgroup as the culprit of the pandemic. Outgroup conspiracy beliefs cater to epistemic needs (Marchlewska, et al., 2018; Van Prooijen & Douglas, 2017) . That is, conspiracy theories provide a narrative of how and why major threatening events, such as the pandemic, happen (as secretly plotted by malevolent entities). In salient intergroup conflicts, appeal to an opposing outgroup can be particularly conducive to this epistemic process, because the outgroup is a ready target to blame (Van Prooijen, 2020; Van Prooijen & Song, 2021) . For our investigations, we note outgroup conspiracy beliefs could explain away some of the threats that COVID-19 may pose and thereby alleviate or even forestall the collapse of a national identity. For example, the pandemic may introduce behavioral and social changes in conflict with national norms. By ascribing to conspiracy beliefs, people (particularly those high in national identity certainty) may be able to see such changes as a necessary move against a hostile outgroup (as opposed to a threat to their national identity). We thus predict:

Those high (vs. low) national identity certainty should be more inclined to hold outgroup conspiracy beliefs (H2).

It is crucial to clarify, however, that we are not contending that people high in national identity certainty necessarily possess a general disposition toward conspiracy thinking (i.e., a tendency to believe in any conspiracy theories regardless of what the conspiracy theories are about). Not all conspiracy beliefs could serve the purpose of alleviating the identity threats-at least to an equal degree. For example, conspiracy beliefs about an ingroup as the culprit may not be as effective. Such beliefs may even backfire because they entail strong associations between one's ingroup identity (national identity) and COVID-19 (threat).

Accordingly, our prediction is that people high in national identity certainty should turn specifically to outgroup conspiracy beliefs because doing so directly addresses the identity threat of COVID-19.

One other line of research suggests that outgroup conspiracy beliefs help safeguard social (national) identity positivity, that is, a positive social (national) identity that people generally cherish Van Prooijen, 2020; Van Prooijen & Song, 2021; Van Prooijen & Van Vugt, 2018) . A robust finding across studies is that collective narcissism (i.e., unrealistic, exaggerated beliefs about ingroup positivity) predict increased conspiracy beliefs (e.g., Cichocka et al., 2016; De Zavala & Cichocka, 2012; Golec de Zavala & Federico, 2018) , including those about the COVID-19 (Hughes & Machan, 2021; Sternisko et al., 2021) . While research on conspiracy beliefs and national identity positivity is relevant to our investigation, our research diverges from this line of work. Based on the perspective of attitude and attitude certainty (e.g., Tormala & Rucker, 2018), we draw a difference between national identity certainty and national identity positivity (see DeMarree et al., 2007 for a parallel between self-concepts and attitudes), with our focus on the former: conceptually, national identity certainty hinges on a sense of clarity (as opposed to the positive valence) of one's national identity. It differs from how positive people view their national identity in that people can deem their national identity rather positively with some degree of uncertainty. People can also deem their national identity negatively with full degree of certainty. Admittedly, it is possible that national identity positivity somewhat contributes to our predicted relationships involving national identity certainty.

Still, we suspect that national identity certainty could predict restrictive measures or outgroup conspiracy beliefs over and above national identity positivity. This is because not every threat posed by COVID-19 involves the positivity of a nation. For instance, the virus could make the future of a nation uncertain, regardless of whether the image of that nation is considered as positive or negative. Our hypothesized effect of national identity certainty could therefore be irreducible to the effect of national identity positivity.

To summarize, we see both support for restrictive measures and outgroup conspiracy beliefs as ways that people-particularly those high in national identity certainty-use to cope with the identity threats posed by the COVID-19 Pandemic. Importantly, we do not think that to cope with the threat of COVID-19, people can only turn to restrictive measures or conspiracy theories. Nor do we consider restrictive measures and conspiracy beliefs as fully interchangeable for that purpose-they address different aspects of threats that COVID-19 may pose.

Drawing from Integrated Threat Theory (Stephan & Stephan, 2000) , recent research examined how peoples' actions and beliefs differed based on whether they saw COVID-19 as a realistic threat or a symbolic threat (Kachanoff et al., 2020) . Realistic threat refers to a concrete assault on physical or material well-being of individuals and their ingroups (Esses et al., 1998) . Symbolic threat refers to an assault on the norms and values of the group. Perceived realistic and symbolic threat (of COVID-19) can have distinct implications on individuals' behaviors and attitudes. Whereas perceived realistic threat predicted individuals' increased adherence to social distancing, perceived symbolic threat predicted decreased adherence (Kachanoff et al., 2020) . Kachanoff and colleagues attributed these patterns to social distancing simultaneously mitigating realistic threat (e.g., preventing sickness or mortality from and inflating symbolic threat (e.g., by weakening a sense of national identity).

Drawing from Kachanoff and colleagues' work, we argue that the relationship between national identity certainty and support for restrictive measures should be explained by perceived realistic (but not symbolic) threat. Restrictive measures aim to reduce the spread of the virus and should thus cater to realistic threat only.

Such measures hardly resolve symbolic threat. If anything, in nations with individualistic cultural orientation such as the United States, restrictive measures could by itself become a symbolic threat to be resolved. On the other hand, we expect the relationship between national identity certainty and outgroup conspiracy beliefs be explained by perceived symbolic threat but not realistic threat. Conspiracy beliefs hardly address the actual threat of the pandemic (e.g., the spread of COVID-19 and the declining economy). However, they do allow people to rationalize and explain away some of the threats COVID-19 may pose (on group norm etc.) by appealing to a hostile outgroup. We thus advance two additional hypotheses on the role of threat perception in people high in national identity certainty's support for restrictive measure and outgroup conspiracy beliefs:

Perceived realistic (but not symbolic) threat should explain the relationship between national identity certainty and support for restrictive measures (H3).

Perceived symbolic (but not realistic) threat should explain the relationship between national identity certainty and outgroup conspiracy beliefs (H4).

In sum, we predict that people high in national identity certainty should be more motivated to support restrictive measures against COVID-19 (H1) and believe in conspiracy theories that blame COVID-19 on an outgroup (H2). The increase in support for restrictive measures should be explained by perceived realistic (but not symbolic) threat (H3); the increase in outgroup conspiracy beliefs should be explained by perceived symbolic (but not realistic) threat (H4). To test these ideas, we conducted two studies. Both studies were set in the context of the United States, where China was often portrayed as the outgroup antagonist in COVID-related conspiracy theories.

Study 1 was a three-week longitudinal study around a projected peak of COVID-19 cases. We explored the co-variation between national identity certainty and support for restrictive measures or outgroup conspiracy beliefs, both between-person and withinperson. Our framework (H1 and H2) would be supported by significant between-person relationships. Namely, people high (vs. low) in national identity certainty should be more likely to support restrictive measures and/or believe in outgroup conspiracy theories across the three weeks. Meanwhile, uncertainty-identity theory could predict significant within-person relationships such that people should support restrictive measures and believe in outgroup conspiracy theories during weeks they report being less certain about national identity certainty. Presumably, for the same person, week-by-week fluctuations in national identity certainty reflect how much uncertainty the person feels at any particular week. On weeks people feel uncertain, they could be inclined to support restrictive measures and/or believe in outgroup conspiracy theories to mitigate such feelings. In addition, we also explored if any of the relationships would emerge when national identity positivity was controlled for. This allowed us to tease apart the potential effect of national identity certainty from that of national identity positivity.

Built on the findings of Study 1, Study 2 examined perceived realistic/symbolic threat as mediators of hypothesized relationships (H3 and H4). We also attempted to manipulate national identity certainty. We expected people high in national identity certainty to feel threatened both realistically and symbolically by COVID-19, which differentially explained their support for restrictive measures and outgroup conspiracy beliefs. We did not have predictions over whether and how certain aspect of threat (realistic or symbolic) was more dominant among people high in national identity certainty. The study materials and data sets are available at: https://osf.io/2ja6g/ ?view_only=6453f 46fca c44df a84fa 47040 9b4267c 2 | S TUDY 1

Data of Study 1 were collected from April 18, 2020 to May 6, 2020 when COVID-19 was estimated to reach its first peak in most regions of the United States (Murray, 2020) . We recruited 301 U.S. participants (172 males, 126 females, 3 preferred to self-describe or did not disclose; M age = 40.05 years, SD = 13.38; 71.1% White) from Amazon Mechanical Turk (MTurk). Each wave of survey took about 5 minutes. Participants were paid $0.50 per wave of survey completion and an additional bonus $0.50 if they completed all three waves.

The sample size of Study 1 was determined a priori. A power analysis via the software Optimal Design (Raudenbush, Spybrook, et al., 2011) revealed that a minimum of 200 participants would be needed to detect a small-to-medium effect size of δ = 0.25 (roughly equivalent to an r of .12) at p = .05 (two-tailed) with a power of .80 within individuals (assuming effect size variability σ δ 2 = 0.05) or between individuals (assuming intra-class correlation rho = .05).

The study consisted of three consecutive waves of weekly surveys.

Each wave was delivered online via Qualtrics software during the weekend. We matched data points by anonymous IDs generated by MTurk. Overall, the survey response rate was satisfactory with 89% of the participants completing at least two waves of surveys and 71.4% of the participants completing all three waves. Regardless of how many waves of surveys were ultimately completed, participants were included in subsequent data analyses. Missing data were listwise deleted.

Each Similarly, we aimed to ensure that participants' ratings of support for restrictive measures were not affected by their ratings of outgroup conspiracy beliefs (or vice versa).

Based on Wagoner and colleagues' work (2017) and past research that measured attitude certainty (see Tormala & Petty, 2002) , we developed two items to measure national identity certainty ("I am certain about what 'Americans' are supposed to be" and "I am certain about what it means by being an 'American'"). Participants were informed that the items concern their beliefs about Americans (inhabitants of the United States) and rated their agreement with the items on 9-point Likert scales (1 = Strongly disagree, 5 = Neither agree nor disagree, 9 = Strongly agree). Responses on the two items were highly correlated across waves of measurement and were collapsed into one composite (M = 6.49, SD = 1.88, r spearman = .81).

We measured national identity positivity with a 9-point semantic differential scale (Snider & Osgood, 1969) . Participants were asked to rate "Americans" as a whole on a total of six word pairs: uncaringcaring, inefficient-efficient, unintelligent-intelligent, incompetentcompetent, insincere-sincere, and untrustworthy-trustworthy (α = .96 across waves of measurement). We collapsed the scores into one composite (M = 6.38, SD = 1.69). Higher scores reflect more positive national identity.

We developed four items to measure support for restrictive measures. The items covered examples of potential public health policies that were commonly referred to by media and scholars as "draconian", "drastic" or "strict" (see Conway et al., 2020; Parmet & Sinha, 2020; Ren, 2020 , for examples)-some of these policies had already been to some degree implemented by the U.S. government when the study was launched. The four items were "Close the border and let no one in," "Enact Martial Law nationwide," "Lockdown major cities and restrict transportation to those areas," and "Make it mandatory that anyone exposed to COVID-19 cases be taken into special quarantine facilities." Participants rated to what extent they agree with the United States should enact these items/policies on 9-point Likert scales (1 = Strongly disagree, 9 = Strongly agree). Responses on the four items were internally consistent (α = .75, across survey waves) and were collapsed into one composite (M = 5.22, SD = 2.03).

To test the idea that people high in national identity certainty turn specifically to outgroup conspiracy beliefs (relative to ingroup conspiracy beliefs), we measured both conspiracy beliefs that blame virus is deliberately engineered in a Chinese/U.S. government laboratory" and "The Chinese/U.S. government has tried to withhold key information about COVID-19 treatment from the rest of the world").

Participants rated to what extent they believed the items were true et al., 2000) . Below, we focused on fixed effect estimates. More details of data analyses (e.g., covariance structure) can be found on OSF.

For the relationship between national identity certainty and support for restrictive measures, we first estimated an unconditional model (i.e., a random intercept model) to calculate the intraclass correlation coefficient (Snijders & Bosker, 1999) . The analysis revealed that most of the variance in support for restrictive measures (80.1%) was at Level 2 (between individuals), while there was a small amount of variance (19.9%) at Level 1 (within individuals across waves). We proceeded by including the person-mean centered national identity certainty at Level 1 to capture purely within-person relationships (Bryk & Raudenbush, 1992) ; at Level 2, we adjusted the intercept with the person mean of national identity certainty (grand-mean centered) to capture the relationship on a between-person level. The results are presented in Table 3 (Model 1). National identity certainty significantly predicted increased support for restrictive measures on a between-person level, but not on a within-person level.

Next, we controlled for national identity positivity. We personmean centered both national identity positivity and national identity certainty, and included both variables at Level 1 of the model. At

Level 2, we adjusted the intercept with the person means of both national identity positivity and national identity certainty (grandmean centered). As presented in Table 3 

We applied the same analytic strategy for outgroup conspiracy beliefs. Estimation of the unconditional model revealed that majority of the variance in outgroup conspiracy beliefs (79.2%) was at Level 2, with a small amount of variance (20.8%) at Level 1. Including national identity certainty into the model (Table 4 , Model 1), the analysis revealed that national identity certainty significantly predicted increased outgroup conspiracy beliefs on a between-person level, but not on a within-person level. Finally, we controlled for national identity positivity on both within-person and between-person levels ( Table 4 , Model 2). National identity positivity significantly predicted increased conspiracy beliefs on a between-person level but not on a within-person level. With national identity positivity controlled, national identity certainty still significantly predicted increased outgroup conspiracy beliefs on a between-person level-only the relationship got slightly weaker.

Study 1 provided initial evidence for our hypotheses H1 and H2.

National identity certainty predicted increased support for restrictive measures and outgroup conspiracy beliefs on a betweenperson level, but not on a within-person level. Stated differently, Although we found some supporting evidence for our H1 and H2

in Study 1, a major limitation of the study was that we failed to fully test out our rationales. We reasoned that people high in national identity certainty was drawn to restrictive measures or outgroup conspiracy beliefs to essentially cope with distinct aspects of threats of COVID-19. Elevated realistic (but not symbolic) threat should account for the relationship between national identity certainty and support for restrictive measures (H3); elevated symbolic (but not realistic) threat should account for the relationship between national identity certainty and outgroup conspiracy beliefs (H4). In Study 2, we directly measured the realistic/symbolic threat of COVID-19 and tested these rationales out. We also included a manipulation of national identity certainty to explore potential causal relationships. The sample size of Study 2 was determined by a power analysis via G*Power v3.1 (Erdfelder et al., 1996) . The analysis revealed that to detect a small-to-medium effect size of d = 0.35 (equivalent to TA B L E 3 Support for restrictive measures as predicted by measured variables on between-person and within-person levels (Study 1) .78

[−0.18, 0.14] r = .17) at p = .025 (assuming a power of .80), we need to recruit at least 314 participants (157 participants per condition). We ended up collecting slightly more than the analysis had suggested.

As in Study 1, the whole procedure was delivered online via should attribute the high/low certainty they experienced in the initial recall task to the aspect of identity (in this case, national identity) activated by the second task. All participants then completed the same measure of national identity certainty (r spearman = .82) as used in Study 1. This measure served as the manipulation check.

After national identity certainty manipulation, they completed the COVID-19 Threat Scale (Kachanoff et al., 2020) that assesses COVID-19 as realistic threat (α = .79) and symbolic threat (α = .91).

The scale presented participants ten aspects of their life that might be threatened by COVID-19 (e.g., "Your personal health" for realistic threat and "American values and traditions" for symbolic threat).

Outgroup conspiracy belief as predicted by measured variables on between-person and within-person levels (Study 1) 

Descriptive statistics and correlation coefficients are presented in Table 5 . We first conducted independent sample t-tests to examine the effects of national identity certainty manipulation on self-report national identity certainty (i.e., manipulation check), support for restrictive measures and outgroup conspiracy beliefs. The results are presented in Table 6 . Surprisingly, people in high and low certainty condition did not differ in self-report national identity certainty. This suggested that our manipulation of national identity certainty was not successful.

Given our manipulation was not successful, we examined instead, the correlations between self-report national identity certainty and other key variables. As seen in Table 5 , there were small yet positive correlations between national identity certainty and outcome variables (support for restrictive measures and outgroup conspiracy beliefs). National identity certainty was also positively correlated with COVID-19 as symbolic threat, and to a lesser extent, COVID-19 as realistic threat, p = .051. Moreover, symbolic threat was positively correlated with outgroup conspiracy beliefs whereas realistic threat was positively correlated with support for restrictive measures.

We proceeded by examining whether realistic and symbolic threat mediated the relationship between self-report national identity certainty and outcome variables. We used the PROCESS macro for SPSS v 3.4 (Hayes, 2017 , Model 4, with 10,000 bootstrap samples) to estimate the indirect effects. Both realistic threat and symbolic threat were entered into the models as mediators so that the indirect effects could be teased apart from one another. As presented in Figure 1 , the analyses revealed a significant indirect effect of national identity certainty on support for restrictive measures via realistic (but not symbolic) threat. This suggested that people high in national identity certainty experienced more acute realistic threat of COVID-19, which explained their increased likelihood in supporting restrictive measures. The analyses also revealed (see Figure 2 ) a significant indirect effect of national identity certainty on outgroup conspiracy beliefs via symbolic (but not realistic) threat. This suggested that people high in national identity certainty experienced more acute symbolic threat of COVID-19, which explained their increased outgroup conspiracy beliefs. These patterns were consistent with H3 and H4.

Across two studies, we found evidence that individual difference in national identity certainty predicted both increased support for restrictive measures to curtail COVID-19 (H1) and increased conspiracy beliefs about an outgroup (vs. ingroup) being the culprit of the pandemic (H2). These relationships held true on a between-person level across three weeks around a projected peak of COVID-19 cases (Study 1) and were replicated months later in a subsequent cross-sectional study (Study 2). Moreover, these relationships emerged even when national identity positivity was statistically controlled for (Study 1). We also found that the relationships between national identity certainty and support 

Taken together, our work joined emerging research (e.g., Kachanoff et al., 2020; Sternisko et al., 2021; Uscinski et al., 2020) to unfold the psychology behind people's beliefs and attitudes toward COVID-19

and related policies. The findings highlighted the role of national identity certainty in these issues. Whereas some research suggested that social (national) identity could help reinstate certainty and reduce feelings of uncertainty (Hogg, 2007) , our work suggests if someone is too certain about a social (national) identity, the person could be vulnerable to the identity threats of COVID-19 and are more motivated to resolve such threats.

Growing research suggests that peoples' feelings toward a shared identity play a key role in collective action during or following a natural disaster (Drury, 2018; Van Bavel et al., 2022; Vezzali et al., 2015) . In the context of COVID-19, a strong sense of shared national identity is proposed to promote solidarity and is found predictive of support for public health policies (Van Bavel et al., 2022) .

The current research, on the other hand, provided a more complex portrait of national identity in a pandemic. Our work suggested the same identity-related variable (i.e., national identity certainty) is actually related to a broader variety of attitudes and beliefs, including measures that are somewhat controversial and costly (i.e., restrictive policies such as lockdown) and conspiracy beliefs that blame an outgroup. Moreover, past research suggested that conspiracy beliefs could undermine efforts of social distancing (Biddlestone et al., 2020; Bierwiaczonek et al., 2020; Freeman et al., 2022; Romer & Jamieson, 2020 ; also see Imhoff & Lamberty, 2020 for a more nuanced view). Under this lens, it could be unintuitive or even surprising that we found individual difference in national identity certainty positively related to both support for restrictive measures and outgroup conspiracy beliefs. In other words, the same group of people seemed to possess beliefs and attitudes of conflicting implications. This supported early research that symbolic and realistic threats can have divergent implications (Kachanoff et al., 2020) .

One theoretical implication of our research is that national identity could be a source of division. Within the same group (i.e., those high in national identity certainty), COVID-19 could pose more of a symbolic threat to some but more of a realistic threat to others, leaving people to respond differently (i.e., support restrictive measures or believe in conspiracy theories (Abrams & Hogg, 2010; Hogg, 2007) Theoretically, our findings also contribute to a line of research showing that the metacognitive certainty about cognitions (the self, judgment, attitude, etc.) could sway related behaviors and decisionmaking (e.g., Chen et al., 2004; DeMarree et al., 2007; Tormala & Rucker, 2018) . Under this lens, self-certainty reflects peoples' epistemic investment into particular or global self-concepts-something people seek to sustain and defend in the face of threat or inconsistencies (Festinger, 1954; Swann & Read, 1981) . Our work suggests that peoples' struggle to maintain a highly certain self-view may put them in a position where an identity threat seem more threatening so that they would need to support restrictive measures or believe in outgroup conspiracy theories (as a way to resolve the threat). Our findings suggest that people do not just turn to conspiracy beliefs to safeguard a positive image of national identity; they may also believe in conspiracy theories to safeguard the certainty of that image. Interestingly, a similar divergence between certainty and valence (i.e., positivity) has been documented in the study of attitudes (see Tormala & Rucker, 2018 for a discussion). It is likely that the divergence between certainty and positivity is a more general phenomenon, not restricted in peoples' attitudes or beliefs about identity. Future research should look into this possibility.

More broadly, our work has implications beyond the context of COVID-19. The ongoing pandemic is not the only thing that can threaten ones' invested social (national) identity. Diverse lines of research suggest that natural disasters in general (even an imagined tornado, see Wohl et al., 2010) , the incoming immigrants in a host nation (Stephan et al., 2005) Under such condition, people high in national identity certainty should be more likely to believe in conspiracy theories, instead.

The current research has several limitations to be addressed. First, our entire findings were built upon a two-item scale of national identity certainty. Scales with more items are of more measurement reliability and validity (Emons et al., 2007) . Moreover, the operationalization left us difficult to tease national identity certainty apart from the related construct of national identification, because we did not measure key aspects of national identification (e.g., the positive feelings associated with group membership, Cameron, 2004) in the studies. 4 Intuitively, people high in national identity certainty should identify with their nation more strongly. Our rationale (of COVID-19 as identity threat) could apply to national identification-the latter could in fact be more fundamental than national identity certainty.

It is also possible that national identity certainty is unique and more fundamental. Teasing apart these possibilities would require research on the relationship between national identity certainty and national identification (e.g., whether they reflect the same underlying psychological processes or latent structure)-research to the best our knowledge is lacking at the moment.

Second, we operationalized outgroup conspiracy beliefs as a difference score between conspiracy beliefs that blame COVID-19 on an outgroup and conspiracy beliefs that blame COVID-19 on an ingroup.

We sought to capture the idea that not all conspiracy beliefs serve the purpose of resolving identity threat to an equal extent-outgroup conspiracy beliefs should be particularly appealing for this purpose (relative to ingroup conspiracy beliefs). The tradeoff of our approach is that it left aside individual differences in conspiracy thinking (i.e., someone who scored high on both ingroup and outgroup conspiracy beliefs was practically indistinguishable from someone who scored low on both conspiracy beliefs). To examine the relationship between individual differences in conspiracy thinking and national identity certainty, future research could look into a broader variety of conspiracy beliefs (e.g., conspiracy beliefs that consider COVID-19 as a "hoax", Imhoff & Lamberty, 2020) and compute a composite (as opposed to a difference score) out of participants' ratings.

Third, we note a disconnect between our measure of national identity positivity and standard measures of collective narcissism.

We had participants rate their national identity (i.e., "American") on word-pairs of opposing valence (desirable vs. undesirable), an approach consistent with how attitudes are typically measured (see, for example, Tormala & Petty, 2002) . In contrast, collective narcissism is traditionally measured by having participants rate how much they agree with statements such as "I will never be satisfied until the United States gets the recognition it deserves" (see, for example, Golec de Zavala & Federico, 2018). It would be ideal for follow-up research to use standard measures of collective narcissism. Still, given an exceedingly positive national identity theoretically lies at the heart of collective narcissism, we expect our findings to be replicable when standard measures of collective narcissism are used.

In addition to measurement issues, our work also left some intriguing future directions. For instance, our work used exclusively samples from the United States. Given COVID-19 is a global public health crisis, future research could test the generalizability of our findings in other nations or groups. Relatedly, future research could explore other approaches of coping with COVID-19 as an identity threat (e.g., getting vaccinated). Future research can also advance our understanding by validating the relationship between national identity certainty and symbolic (vs. realistic) threat and exploring potential moderators. For example, among conservatives (vs. liberal) in United States, the threat of COVID-19 to national identity could be predominantly symbolic (vs. realistic). Alternatively, some people may construe an identity threat in more abstract, non-pragmatic terms (Trope et al., 2007) . Their national identity certainty may well be related to symbolic threat.

Last, we recommend future research to examine whether our findings hold true on a more situational, state level. Will momentary increase in outgroup conspiracy beliefs actually makes people feel more certain about their national identity? To test this possibility, follow-up studies may have to capture meaningful within-person changes in support for restrictive measures or outgroup conspiracy theories. Given support for restrictive measures and outgroup conspiracy theories appears rather stable in our research, we recommend future research to conduct longitudinal studies over a longer time span or use stronger experimental manipulations.

The COVID-19 pandemic has prompted controversies regarding the usage of restrictive measures and the conspiracy theories about the virus's origin. Understanding why people may support or endorse these issues is vital to the management of COVID-19. Our research suggests that it is people who are more certain about national identity that are more likely to support restrictive measures-in an attempt to curtail the realistic threat of COVID-19. These people also tend to believe in conspiracy theories that blame COVID-19 on an outgroup, for a related but different purpose: to cope with symbolic threat. To effectively manage the ongoing pandemic, we recommend policymakers to pay attention to people high in national identity certainty and the type of threat involved in messaging.

All authors contributed to the study conception and design. The first author performed most of the material preparation, data collection and analysis. We would like to thank anonymous reviewers for their suggestions on how to improve the fluency of our hypotheses and broaden the implication of our work.

The authors declare that they have no conflict of interest.

The data that support the findings of this study are openly available 1 We conducted the same analysis to examine the relationship between national identity certainty and Chinese-government conspiracy belief only, with or without US-government conspiracy beliefs controlled as a covariate. The results of the two studies remained mostly unchanged. Details can be found in Supporting Information.

2 For exploratory purposes, we computed a difference score between realistic and symbolic threat and examined its correlation with national identity certainty. The analysis did reveal a significant negative correlation, r = −.12, p = .03, suggesting those high in national identity certainty could experience more symbolic (over realistic) threat of COVID-19.

The SAGE handbook of prejudice, stereotyping and discrimination

Polarization and public health: Partisan differences in social distancing during the coronavirus pandemic

Empirical assessment of government policies and flattening of the COVID 19 curve

Seriousness checks are useful to improve data validity in online research

The impact of COVID-19 on small business outcomes and expectations

To know oneself is to like oneself: Selfcertainty and self-affect

Rugged individualism and collective (in)action during the COVID-19 pandemic

When identity marketing backfires: Consumer agency in identity expression

Cultural orientation, power, belief in conspiracy theories, and intentions to reduce the spread of COVID-19

Belief in COVID-19 conspiracy theories reduces social distancing over time

Efficiency of healthcare systems in the first wave of COVID-19-A technical efficiency analysis. Munich Personal RePEc Archive

Hierarchical linear models in social and behavioral research: Applications and data analysis methods

A three-factor model of social identity

What coronavirus could mean for the global economy

Self-verification motives at the collective level of self-definition

Self-uncertainty and group identification: A meta-analysis

They will not control us

Social psychological measurements of COVID-19: Coronavirus perceived threat, government response, impacts, and experiences questionnaires

Collective narcissism and anti-Semitism in Poland

Self-certainty: Parallels to attitude certainty

People in advanced economies say their society is more divided than before pandemic

COVID-19 conspiracy theories

The psychology of conspiracy theories

The role of social identity processes in mass emergency behaviour: An integrative review

On the consistency of individual classification using short scales

The different forms of COVID-19 misinformation and their consequences

GPOWER: A general power analysis program

Intergroup competition and attitudes toward immigrants and immigration: An instrumental model of group conflict

A theory of social comparison processes

A theory of cognitive dissonance

Coronavirus conspiracy beliefs, mistrust, and compliance with government guidelines in England

The "shaken self": Product choices as a means of restoring self-view confidence

Effects of economic uncertainty on mental health in the COVID-19 pandemic context: Social identity disturbance, job uncertainty and psychological well-being model

Belief in Conspiracy Theories

Collective narcissism and the growth of conspiracy thinking over the course of the 2016 United States presidential election: A longitudinal analysis

Faith in Trump, moral foundations, and social distancing defiance during the coronavirus pandemic

Subjective uncertainty and intergroup discrimination in the minimal group situation

Introduction to mediation, moderation, and conditional process analysis: A regression-based approach

Tolerance of uncertainty: Conceptual analysis, integrative model, and implications for healthcare

Uncertainty-identity theory

Domains of self-uncertainty and their relationship to group identification

It's a conspiracy: Covid-19 conspiracies link to psychopathy, Machiavellianism and collective narcissism

A bioweapon or a hoax? The link between distinct conspiracy beliefs about the Coronavirus disease (COVID-19) outbreak and pandemic behavior

Social representations, identity threat, and coping amid COVID-19

Threat and defense: From anxiety to approach

Measuring realistic and symbolic threats of COVID-19 and their unique impacts on well-being and adherence to public health behaviors

Addicted to answers: Need for cognitive closure and the endorsement of conspiracy beliefs

Forecasting the impact of the first wave of the COVID-19 pandemic on hospital demand and deaths for the USA and European Economic Area countries

Covid-19-The law and limits of quarantine

On confidence and consequence: The certainty and importance of self-knowledge

HLM 7: Hierarchical linear and nonlinear modeling (Version 7.03) [Computer software

Pandemic and lockdown: a territorial approach to COVID-19 in China, Italy and the United States

Conspiracy theories as barriers to controlling the spread of COVID-19 in the US

Contrasts and effect sizes in behavioral research: A correlational approach

Semantic differential technique: A sourcebook

Multilevel analysis: An introduction to basic and advanced multilevel modeling

The effects of feeling threatened on attitudes toward immigrants

An integrated threat theory of prejudice

National narcissism predicts the belief in and the dissemination of conspiracy theories during the COVID-19 pandemic: Evidence from 56 countries

Self-verification processes: How we sustain our self-conceptions

An intergrative theory of intergroup conflict

What doesn't kill me makes me stronger: The effects of resisting persuasion on attitude certainty

Attitude certainty: Antecedents, consequences, and new directions

Construal levels and psychological distance: Effects on representation, prediction, evaluation, and behavior

Why do people believe COVID-19 conspiracy theories?

National identity predicts public health support during a global pandemic

An existential threat model of conspiracy theories

Conspiracy theories as part of history: The role of societal crisis situations

The cultural dimension of intergroup conspiracy theories

Conspiracy theories: Evolved functions and psychological mechanisms

Feeling like a group after a natural disaster: Common ingroup identity and relations with outgroup victims among majority and minority young children

Out-group leadership and subgroup schisms: An examination of the 2016 US presidential election

Social identity uncertainty: Conceptualization, measurement, and construct validity

Cognitive and social consequences of the need for cognitive closure

European review of social psychology

Perceiving your group's future to be in jeopardy: Extinction threat induces collective angst and the desire to strengthen the ingroup

Additional supporting information may be found in the online version of the article at the publisher's website.