Religion Protected Mental Health but Constrained Crisis Response During Crucial Early Days of the COVID-19 Pandemic.

This study demonstrates that religion protected mental health but constrained support for crisis response during the crucial early days of the COVID-19 pandemic. Data from a national probability-based sample of the U.S. population show that highly religious individuals and evangelicals suffered less distress in March 2020. They were also less likely to see the coronavirus outbreak as a crisis and less likely to support public health restrictions to limit the spread of the virus. The conservative politicization of religion in the United States can help explain why religious Americans (and evangelicals in particular) experienced less distress and were less likely to back public health efforts to contain the virus. We conclude that religion can be a source of comfort and strength in times of crisis, but-at least in the case of the COVID-19 pandemic-it can also undercut efforts to end the root causes of suffering.

Introduction

In his infamous “opiate of the masses” argument, Marx (1970 [1843]) suggested religion helps people through hardship but distracts them from addressing its material root causes. He pointed out that individuals often turn to religion in the face of adversity, but criticized it as masking the true nature of their suffering. A more generous vision of religion would acknowledge the full range of ways religion influences how individuals see the world. Rather than just distracting them, religion appears to actively shape people's outlook according to religious schemas—the teachings, norms, and ways of making sense of the world that inform social life (Ammerman 2020; Ogland and Bartkowski 2014). This study explores the ways religion operated as both a resource and a schema during the early days of the COVID‐19 pandemic in the United States, providing both a source of comfort and strength that people drew upon in the face of hardship as well as a frame of reference that structured how they viewed the world—a frame that could be seen as failing to address and even exacerbating the root causes of the suffering they experienced (Davis 1971; Schnabel 2021). In short, it appears religion protected mental health but endangered physical health during the COVID‐19 pandemic.

The early days of the pandemic were fraught with uncertainty yet crucial in terms of the spread of the virus. Fear of the virus itself joined with financial uncertainty, disrupted routines, and general societal upheaval to create a unique array of stressors (Van Bavel et al. 2020). Compounding matters was the fact that the public health measures necessary to prevent the spread of the virus robbed individuals of the social connection central to their well‐being (Bierman and Schieman 2020; Counted et al. 2021; Hagerty and Williams 2020; Kim and Jung 2021). Although social distancing and other measures posed a danger to mental health, failing to follow these measures jeopardized physical health and successfully containing the virus.

As in past crises, many Americans turned to religion for guidance and inspiration during the pandemic. Despite reports of growing secularism, the United States remains exceptionally religious (Schnabel and Bock 2017). For many Americans, their faith provides the lens through which they approach most matters of importance, and a health crisis on the scale of a pandemic certainly seems tailor‐made for turning to religion. Yet the religiosity of Americans poses a unique challenge during a pandemic, as the comfort religion offers is often premised on congregating in violation of social distancing protocols. This study therefore considers the role religion played in addressing the distress Americans experienced at the beginning of the pandemic, their emerging views about COVID‐19, and how religion shaped their thinking about public health restrictions.

Although high levels of distress among the U.S. population were noted early in the pandemic (Holingue et al. 2020), research suggests that the correspondingly high levels of religiosity could prove helpful for protecting against its negative mental health consequences (e.g., Schieman, Bierman, and Ellison 2013; Schieman, Bierman, and Upenieks 2018; Schnittker 2001). Religion can operate as a source of psychological compensation in the face of challenges such as the loss of a job (e.g., Hastings and Roeser 2020), providing an important resource that people can draw upon for comfort, strength, community, and a sense of control in the face of uncertainty and hardship (Kay et al. 2010; Laurin, Kay, and Moscovitch 2008; Schnabel 2021).

But in addition to being a resource people can draw upon, religion is also a powerful social institution and socializing force, with particular teachings and norms that provide a way of thinking about the world (Ogland and Bartkowski 2014; Schieman 2011; Schnabel 2021). Besides the connection and community it provides, religion's ideological structures—such as a sense of order to the world, an all‐loving and all‐powerful being overseeing one's fate, rules that provide clear guidance, and the hope of a future utopia—make it a potent source of comfort, strength, and meaning during hard times. Religion can be especially helpful in hard times (Hastings and Roeser 2020), and emerging research on religion and mental health outside the United States (Counted et al. 2021) and among subgroups in the United States (Pirutinsky, Cherniak, and Rosmarin 2020) suggest it may be particularly useful for coping with these unprecedented times. Yet as people draw upon religion as a resource, they further commit themselves to the ideological aspects of religion as a schema—and those ideological aspects (i.e., the content, teachings, and norms of a religion) can provide further comfort, strength, and certainty in a self‐reinforcing loop.

In theory, the ideological content of a religion could be largely irrelevant to social and political life. But religions develop and operate in the world and tend to promote particular ways of approaching, interpreting, and making decisions about these realms. At the theoretical level, the ideological commitments of religion could similarly span the political spectrum. But in the contemporary United States, religion has become politicized and is now viewed as entangled with conservative politics (Hout and Fischer 2014; Putnam and Campbell 2010). As a case in point, about 8 in 10 white evangelicals voted for Trump in both the 2016 and 2020 presidential elections, in line with expected voting patterns following the increased politicization of American religion (Margolis 2020). Intensely religious white people tend to embrace conservative partisan values and Republican politics (Schnabel 2021; Whitehead, Perry, and Baker 2018).1 As a result, in addition to socializing people into particular ways of seeing the world, religion is now a politicized identity that socializes people into partisan perspectives that may have relatively little to do with religion per se.

In the United States, the early response to the pandemic quickly became politicized, with some Republican politicians downplaying the impact of the virus, discouraging restrictions, and characterizing public health officials as overreacting (Newport 2020b; Perry, Whitehead, and Grubbs 2020). It is possible then that highly religious Americans were more likely to adopt a “conservative” response and downplay concerns about the pandemic (Fowler 2020; Hill, Gonzalez, and Burdette 2020). Because highly religious white Americans tend to align with the GOP, they may take cues regarding COVID‐19 from conservative politicians, media outlets, and religious leaders. Many of the latter claimed the coronavirus pandemic was not a real concern (Merritt 2020), and televangelists such as Jim Bakker and Kenneth Copeland told viewers they could be healed by touching their television screens or use prayer to halt the outbreak (Lemon 2020).

If the teachings, norms, and partisan values entangled with religion are not well‐suited to address a public health crisis such as COVID‐19, religion could then simultaneously alleviate distress about the pandemic and lead adherents to reject public health recommendations designed to curb it (Counted et al. 2021; Perry, Whitehead, and Grubbs 2021; Pirutinsky, Cherniak, and Rosmarin 2020). In sum, religion therefore could paradoxically buffer the hardship caused by the pandemic yet structure attitudes and orientations about public health and science in ways that ultimately increase it (Newport 2020a; Perry, Whitehead, and Grubbs 2020; Smith 2020).

Methods

Data

This study uses data from Pew Research Center's American Trends Panel (ATP), an online panel recruited through random sampling of residential addresses throughout the United States. Those without internet access were provided a tablet and Internet connection to complete self‐administered web surveys. This study uses data from the wave fielded March 19−24, 2020, shortly after the WHO declared COIVD‐19 a global health pandemic. Of the 15,433 sampled, 11,537 respondents completed the survey. More information about the panel can be found here: https://www.pewresearch.org/methods/u‐s‐survey‐research/american‐trends‐panel/.

Measures

Dependent Variables

Mental Distress

This study first considers mental distress, measured by a score derived from five items from the Pew ATP survey. These items ask respondents how often in the past 7 days they: (1) felt nervous, anxious, or on edge; (2) felt depressed; (3) felt lonely; (4) had trouble sleeping; and (5) felt hopeful about the future (reverse coded). Response options were “rarely or none of the time (less than 1 day)”; “some or a little of the time (1–2 days)”; “occasionally or a moderate amount of time (3–4 days)”; or “most or all of the time (5–7 days).” Following research using these items to estimate mental distress during the COVID‐19 pandemic (e.g., Cobb, Erving, and Byrd 2021; Holingue et al. 2020), these four response options were given weights of 1, 2, 3, and 4 to reflect increasing frequency of symptoms. A summary score for each person was derived by taking the sum of the five items (possible range 5−20, Cronbach's α = .73). Descriptive statistics for the mental distress scale and other measures are presented in Table 1.

Table 1

Descriptive statistics for key measures

Measures	Descriptions	N	Mean	SD	Range
Mental distress scale	Summative scale of five items, α = .73	11,369	10.07	3.40	5–20
Views on pandemic and public health response
COVID‐19 threat scale	Summative scale of four items, α = .58	11,407	10.26	1.45	4–12
People not taking seriously enough scale	Summative scale of six items, α = .71	11,189	12.37	2.13	6–18
Support for public health restrictions scale	Number of public health restrictions supported, α = .79	11,244	6.08	1.51	0–7
Social distancing scale	Number of social distancing behaviors, α = .72	11,371	3.30	1.47	0–5
Key independent variables
Attendance frequency	Never = 1 to more than once a week = 6	11,494	2.95	1.67	1–6
Affiliation: Evangelical	Religious affiliation = Evangelical	11,494	.22
Affiliation: Non‐evangelical Protestant	Religious affiliation = Non‐evangelical Protestant	11,494	.17
Affiliation: Catholic	Religious affiliation = Catholic	11,494	.22
Affiliation: Jewish	Religious affiliation = Jewish	11,494	.03
Affiliation: Nothing in particular	Religious affiliation = Nothing in Particular	11,494	.18
Affiliation: Agnostic	Religious affiliation = Agnostic	11,494	.07
Affiliation: Atheist	Religious affiliation = Atheist	11,494	.07
Affiliation: Other/missing	Religious affiliation = Other/missing	11,494	.06
Key covariates
Party: Republican	Self‐identifies as a Republican	11,494	.25
Party: Leans Republican	Does not identify as GOP or Dem, but leans Republican	11,494	.16
Party: Independent/other party	Identifies as independent, other, or refused and does not lean	11,494	.03
Party: Leans Democrat	Does not identify as GOP or Dem, but leans Democrat	11,494	.21
Party: Democrat	Self‐identifies as a Democrat	11,494	.36
Liberal political ideology	Very conservative = 1 to very liberal = 5	11,494	3.02	1.06	1–5
Have prayed to end COVID‐19	Have prayed for an end to the spread of the coronavirus = 1	11,494	.57
Have watched online or TV services	Have watched online or TV services instead of in person = 1	11,494	.28
Have attended less in person	Have attended religious services in person less often = 1	11,494	.32

Source: American Trends Panel, March 19—24, 2020 Survey.

Views on Pandemic and Public Health Response

This study also considers views on the pandemic and public health response. Descriptive statistics for each of these items are also presented in Table 1. First is a summative scale of four items about the extent to which respondents see the pandemic as a threat. For these items, respondents were asked how much of a threat, if any, the coronavirus outbreak is for: (1) the health of the U.S. population as a whole; (2) your personal health; (3) the U.S. economy; and (4) your personal financial situation. Response options were “not a threat”; “a minor threat”; or “a major threat.” The second measure is a summative scale of six items about whether the following groups are “overreacting to the outbreak”; “reacting about right”; or “not taking the outbreak seriously enough”: (1) your state government; (2) your local government; (3) your local school system; (4) ordinary people in your community; (5) ordinary people across the country; and (6) the people in your household.2

This study next considers support for a number of public health restrictions. For these items, the survey asked, “Thinking about some steps that have been announced in some areas to address the coronavirus outbreak, in general do you think each of the following have been necessary [1] or unnecessary [0]?” Respondents were then presented with the following topics:

Restricting international travel to the United States.

Requiring most businesses other than grocery stores and pharmacies to close.

Asking people to avoid gathering in groups of more than 10.

Canceling major sports and entertainment events.

Closing K‐12 schools.

Limiting restaurants to carry‐out only.

Postponing upcoming state primary elections.

Responses were combined for a total number of restrictions supported ranging from 0 to 7 (α = .79).

Finally, this study examines responses about social distancing behaviors. For these items, respondents were asked, “Given the current situation with the coronavirus outbreak, would you feel comfortable [0] or uncomfortable [1] doing each of the following?” Respondents were then presented with the following scenarios:

Visiting with a close friend or family member at their home.

Eating out in a restaurant.

Attending a crowded party.

Going out to the grocery store.

Going to a polling place to vote.

Responses were combined for a total number they would be uncomfortable with ranging from 0 to 5 (α = .72).

Key Independent Variables

The key independent variables are religious service attendance and religious affiliation. Attendance measures intensity of religious behavior and embeddedness in a religious community as a proxy for general religiosity. For this item, which was a part of the core profile measures and asked prior to the pandemic,3 respondents were asked, “Aside from weddings and funerals, how often do you attend religious services?” Response options included “never”; “seldom”; “a few times a year”; “once or twice a month”; “once a week”; or “more than once a week.”4

Religious affiliation addresses religious belonging and differences across religious subcultures. For this item, respondents indicated their general religious category. Then, for those who it might be relevant, a follow up question asked whether they identify as a born again or evangelical Christian. Affiliation is measured in eight categories: evangelical Protestant, non‐evangelical Protestant, Catholic, Jewish, nothing in particular, agnostic, atheist, and other/missing.

Covariates

Demographic variables include gender, race/ethnicity (non‐Latinx white, non‐Latinx black, Latinx, and other), age categories (18−29, 30−49,50−64, 65+), education (less than high graduate, high school graduate, some college or associate degree, college graduate, postgraduate), income ranges (<30k, 30–74.9k, 75k+), metropolitan status, and region (Northeast, Midwest, South, West). We also control for income or job loss due to the pandemic in three categories: (1) no income or job loss due to the pandemic, (2) they or someone in the household had to take a cut in pay or reduced hours due to the pandemic, or (3) they or someone in the household had been laid off or lost a job due to the pandemic.

Political party affiliation and ideology may mediate or otherwise explain the relationships between (1) religion and distress; (2) religion and views about the pandemic being a crisis; and (3) religion and support for actions to address it. Politics were measured with party identification (Republican, leans Republican, true independent/other party/refused, leans Democrat, Democrat) and political ideology (very conservative = 1 to very liberal = 5).

Analyses of distress consider prayer to end the coronavirus—a coping mechanism promoted by religious leaders as the virus emerged—as a potential mediator (Lemon 2020). For this item, respondents were asked if they had “prayed for an end to the spread of the coronavirus” with yes or no as response options. To address the fact that religious service attendance could have been disrupted by the pandemic and new virtual ways of drawing upon religion as a resource arose, we also consider measures of whether respondents have “attended religious services in person less often” or “watched religious services online or on TV instead of attending in person” (both measured as yes or no) due to the coronavirus outbreak.

Analytic Strategy

The sample was limited to the 11,494 respondents with data on frequency of religious service attendance. Missing data on covariates are imputed using chained equations. All analytic variables were used for imputation where the number of imputed data sets equals 20. Individual analyses vary slightly in sample size due to the number of cases with complete information on the outcome variables. OLS regression is used to first examine the relationship between religion and distress during the coronavirus pandemic. Then it is used to examine the relationship between religiosity and views of the pandemic and support for public health response to it. We present unweighted results because the measures used to construct the weights are a function of controls included in the models (Winship and Radbill 1994). Additional analyses with survey weights yielded substantively equivalent results.

Results

Religion as Resource During the COVID‐19 Pandemic

We first consider the relationship between religion and mental distress in March 2020.5 Model 1 of Table 2 shows that people who attend religious services more frequently report substantially less mental distress.6 We also see that the least distressed religious group are evangelicals (the reference category), with the highest levels of distress reported by Jewish and secular Americans. Additional analyses using General Social Survey panel data (see online supplement) indicate that the subjective wellbeing benefits of religion during the COVID‐19 pandemic operate over‐and‐above any pre‐existing benefits (at least on the general happiness measure available in the GSS).7 This suggests that religion, typically implicated in rates of distress, mitigated the increased distress most Americans were feeling in the early days of the pandemic.

Table 2

Religion predicting mental distress during the COVID‐19 pandemic

	Model 1	Model 2	Model 3
Attendance frequency	–.24***	–.17***	–.21***
	(.02)	(.02)	(.03)
Non‐evangelical Protestant (evangelical reference)	.62***	.29**	.30**
	(.10)	(.10)	(.10)
Catholic	.60***	.31**	.31**
	(.10)	(.10)	(.10)
Jewish	1.38***	.61**	.68**
	(.21)	(.21)	(.21)
Nothing in particular	.67***	.18	.24*
	(.12)	(.12)	(.12)
Agnostic	1.40***	.66***	.76***
	(.15)	(.15)	(.16)
Atheist	.82***	–.10	.00
	(.16)	(.16)	(.16)
Controls	Yes	Yes	Yes
Leans Republican (Republican reference)		.14	.14
		(.10)	(.10)
Independent/other party		.01	.02
		(.19)	(.19)
Leans Democrat		.93***	.94***
		(.10)	(.10)
Democrat		1.03***	1.04***
		(.11)	(.11)
Liberal political ideology		.36***	.37***
		(.04)	(.04)
Have prayed to end coronavirus			.23**
			(.08)
Have watched online or TV services			–.03
			(.08)
Have attended less in person			.17*
			(.08)
Constant	11.45	10.18	10.09
N	11,369	11,369	11,369

Notes: Controls include gender, race, age, education, income, metropolitan status, region, and job loss due to the pandemic. Models also include an “other/missing” affiliation category (not shown). Full model with coefficients and standard errors for control variables is presented in the Supporting Information. Standard errors in parentheses.

p < .05

p < .01

p < .001.

Source: American Trends Panel, March 19–24, 2020 Survey.

Model 2 introduces politics in our analysis. We see political party affiliation and general political ideology are both important factors in the distress reported during the COVID‐19 pandemic: Republicans and conservatives reported less distress than Democrats and liberals. The most obvious explanation for this pattern is the politicization of the pandemic and the fact that (as we will see below) Republicans and conservatives simply were not as concerned about the pandemic and less likely to think they needed to worry about social distancing, and so on. Evangelicals are much more likely to be Republican and conservative, and politics explains much of why evangelicals experienced less distress during the early stages of the pandemic: before accounting for politics, evangelicals experienced substantially less distress than all other groups, but after accounting for politics, all religious group differences are mitigated and the difference between evangelicals and atheists is fully explained by politics. Attendance differences are also reduced by about a third when accounting for politics.

Finally, Model 3 of Table 2 introduces additional factors that may help explain why more religious Americans experienced less distress during the COVID‐19 pandemic. But rather than explain away attendance or affiliation differences, if anything these additional factors act as a focusing lens to highlight the differences, especially for attendance. Perhaps this is not surprising, however. For example, we see that attending less in person is linked to more distress, and having attendance disrupted may have reduced (i.e., suppressed) some of the general benefits of attendance. Having prayed to end the coronavirus is also positively associated with distress, which we might assume is a function of selection rather than prayer being the cause of distress: religious people experiencing distress might use prayer to cope with that distress. Watching virtual services is not an important predictor one way or the other, suggesting they do not provide independent mental health benefits.

Religion as Schema During the COVID‐19 Pandemic

Religion has what appears to be a positive impact by limiting distress during the COVID‐19 pandemic. However, religion as resource and religion as schema operate in tandem, and people cannot draw comfort from religion without having it shape how they perceive and interact with the world around them—ultimately impacting their social and political views and actions. And as we have already seen, politics is part of why religious individuals (and evangelicals in particular) experienced less distress during the first days of the pandemic.

We now turn to the relationship between religion and both views of the pandemic and support for the public health response to it. In Table 3, we first see that evangelicals and people who attend religious services more frequently are less likely to see COVID‐19 as a threat to themselves and the nation. Politics entangled with religion explain why: people who are more religious (and especially evangelicals) are more likely to be Republican and conservative and less likely to see COVID‐19 as a threat. Politics even reverses the gap between evangelicals and atheists, with evangelicals less likely to see it as a threat before accounting for ideology and party affiliation and atheists less likely to see it as a threat after. Religious individuals and evangelicals are also more likely to think that people are overreacting (i.e., less likely to say people are not taking it seriously enough), and politics explain why.

Table 3

Religion predicting views on COVID‐19 pandemic and public health response

	COVID Threat Scale		People not Taking Outbreak Seriously Enough Scale		Support for Public Health Restrictions Scale		Social Distancing Scale
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
Attendance frequency	–.05***	–.03**	–.05**	.00	–.00	.02	–.03**	–.01
	(.01)	(.01)	(.02)	(.02)	(.01)	(.01)	(.01)	(.01)
Non‐evangelical Protestant (evangelical reference)	.16***	.02	.31***	.07	.12*	–.01	.10*	–.01
	(.04)	(.02)	(.07)	(.07)	(.05)	(.05)	(.05)	(.05)
Catholic	.30***	.18***	.36***	.15*	.17***	.06	.24***	.15***
	(.04)	(.04)	(.07)	(.06)	(.05)	(.05)	(.04)	(.04)
Jewish	.42***	.12	.60***	.07	.32**	.03	.47***	.24**
	(.09)	(.09)	(.14)	(.14)	(.10)	(.10)	(.09)	(.09)
Nothing in particular	.22***	.02	.41***	.06	.19***	.00	.23***	.08
	(.05)	(.05)	(.08)	(.08)	(.05)	(.06)	(.05)	(.05)
Agnostic	.25***	–.05	.72***	.21*	.40***	.12	.33***	.11
	(.07)	(.07)	(.10)	(.10)	(.07)	(.07)	(.07)	(.07)
Atheist	.15*	–.21**	.72***	.08	.34***	–.01	.37***	.09
	(.07)	(.07)	(.10)	(.10)	(.07)	(.07)	(.07)	(.07)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Leans Republican (Republican reference)		.02		–.09		–.19***		–.08
		(.04)		(.06)		(.04)		(.04)
Independent/other party		.38***		.32*		.28**		.17
		(.08)		(.13)		(.09)		(.09)
Leans Democrat		.48***		.71***		.31***		.26***
		(.04)		(.07)		(.05)		(.05)
Democrat		.51***		.68***		.32***		.27***
		(.04)		(.07)		(.05)		(.05)
Liberal political ideology		.09***		.22***		.13***		.11***
		(.02)		(.03)		(.02)		(.02)
Constant	9.22	8.82	11.21	10.44	5.23	4.87	2.65	2.32
N	11,407	11,407	11,189	11,189	11,244	11,244	11,371	11,371

Notes: Controls include gender, race, age, education, income, metropolitan status, region, and job loss due to the pandemic. Models also include an “other/missing” affiliation category (not shown). Full model with coefficients and standard errors for control variables is presented in the Supporting Information. Standard errors in parentheses.

p < .05

p < .01

p < .001.

Source: American Trends Panel, March 19–24, 2020 Survey.

Finally, Table 3 also shows that evangelicals are less likely to support public health restrictions to curb the spread of COVID‐19. However, once we account for politics there are no longer any significant differences between them and other religious groups. Both evangelical identification and religious service attendance are associated with being more comfortable engaging in several public activities (e.g., eating in restaurants, attending parties, visiting friends and family in their homes) that would violate social distancing guidelines, and again party affiliation and political ideology largely explain why. In sum, religion and religious schemas matter, but their impact is deeply entangled with particular political schemas that appear to be the driving force in perceptions of the outbreak, support for public health restrictions, and comfort with actions that could expose oneself and others to the virus.

Discussion

Religion limited the negative mental health impacts of the COVID‐19 pandemic in March 2020, with highly religious Americans and especially evangelicals experiencing less distress than more secular Americans. However, that mental health benefit came at the cost of less concern about and support for addressing an important real‐world problem: saving lives during a pandemic. Highly religious people and especially evangelicals tended to hold attitudes that ran counter to the policy recommendations of public health officials aimed at curbing the spread of COVID‐19.8 Secular Americans were more likely to support the public health response to the pandemic, whereas intensely religious Americans were less concerned about the pandemic, less likely to support public health guidelines, and more comfortable breaking social distancing protocols.

This study complements the literature on religion and coping, and especially work highlighting religion's role in protecting mental health in the face of hardship (Bradshaw and Ellison 2010; Hastings and Roeser 2020; Stratta et al. 2013). As we showed, religion proved helpful during the early days of exposure to new stressors created by the virus and the accompanying societal challenges, offering mental health protections to highly religious Americans and especially evangelicals. But it would seem that the mental health benefits were also accompanied by harmful views, reducing concern about the pandemic and lowering support for public health measures that were needed to curb its spread (Van Bavel et al. 2020; Perry, Whitehead, and Grubbs 2020). In fact, it appears the protective mechanism was not simply that religion provided a source of comfort in hardship as past research typically suggests; instead, religion—in large part due to its entanglement with politics—kept people from seeing the virus as a threat.

Protecting mental health is not automatically a wholly good thing if accompanied by attitudes and actions ineffective for containing a pandemic. Hill, Gonzalez, and Burdette (2020) showed that stay at home orders were less effective in more religious states. Stay at home orders, when actually followed, endanger mental health by disturbing routines, exacerbating financial uncertainty, and disrupting social connectedness. But they are also a key strategy for containing spread and protecting physical health. Paradoxically, therefore, it seems religion protected mental health but endangered physical health. While intensely religious Americans experienced comparatively less distress, more secular Americans faced elevated distress while embracing public health measures, thereby endangering their own mental health to protect the physical health of those around them.

The data for this study came from early in the pandemic and the exact extent to which religion will continue to protect mental health as the outbreak reaches its 1‐year anniversary remains to be seen. We hope future research will extend and improve on what we have presented here as new data become available, preferably with fine‐grained geographic measures making it possible to better control for factors like spread and death rates in particular locations as the pandemic progressed over time. It is likely that some religious partisans experienced increased distress when faced with tangible evidence that the virus is a threat such as getting sick themselves or losing loved ones. Yet as long as religion continues to provide a sense of comfort, support, and hope—and remains entangled with politics that reduce concern about and isolating action to address the pandemic—religious people will likely retain at least some of the mental health benefits documented here.

Religion can be a source of comfort and strength in times of crisis, but—at least in the case of the COVID‐19 pandemic—it can also undercut efforts to end the root causes of suffering. It appears religion then is a double‐edged sword—helping people cope with hardship yet perpetuating the hardship it is helping them through. In the case of this particular hardship, some of this “coping” involved avoiding the fact that the hardship even exists. Although effective for mitigating distress and protecting mental health, failing to acknowledge real danger is ineffective for protecting physical health.

Table S1: Descriptive statistics, full table

Table S2: Religion predicting mental distress during the COVID‐19 pandemic, full table

Table S3: Religion predicting views on COVID‐19 pandemic and public health response, full table

Table S4: Religion predicting unhappiness during the COVID‐19 pandemic controlling for prior unhappiness, 2016‐2020 and 2018‐2020 General Social Survey panels

Click here for additional data file.