A time for moral actions: Moral identity, morality-as-cooperation and moral circles predict support of collective action to fight the COVID-19 pandemic in an international sample

Abstract

Understanding what factors are linked to public health behavior in a global pandemic is critical to mobilizing an effective public health response. Although public policy and health messages are often framed through the lens of individual benefit, many of the behavioral strategies needed to combat a pandemic require individual sacrifices to benefit the collective welfare. Therefore, we examined the relationship between individuals’ morality and their support for public health measures. In a large-scale study with samples from 68 countries worldwide (Study 1; N = 46,576), we found robust evidence that moral identity, morality-as-cooperation, and moral circles are each positively related to people’s willingness to engage in public health behaviors and policy support. Together, these moral dispositions accounted for 9.8%, 10.2%, and 6.2% of support for limiting contact, improving hygiene, and supporting policy change, respectively. These morality variables (Study 2) and Schwartz’s values dimensions (Study 3) were also associated with behavioral responses across 42 countries in the form of reduced physical mobility during the pandemic. These results suggest that morality may help mobilize citizens to support public health policy.

Note: This article is part of a special issue entitled [Morality and Ethical Conduct (or their absence) in Groups, Politics, and Society], guest edited by [Stefano Pagliaro (lead guest editor), Theofilos Gkinopoulos and Maria Giuseppina Pacilli].

Significance Statement:

A central question during a pandemic is why people in groups fail to follow recommended physical hygiene habits and do not support public policies that promote public health. Across three international studies conducted at the beginning of the pandemic, we found that different moral dispositions were related to why some people and nations engaged in the collective actions necessary to fight the pandemic and supported policies to promote health, whereas others did not. Our results suggest that support for measures to combat the pandemic are linked to individuals’ morality through concerns about cooperation, benevolence, and universalism.

Introduction

Dealing with the pandemic has required people worldwide to modify their habits and engage in new behaviors, such as physical distancing and physical hygiene, or support new restrictive public policies designed to reduce the spread of the novel coronavirus (Van Bavel et al., 2020; Ruggeri et al., 2022). A key feature of these public health behaviors and measures is that they are not purely selfish in a pandemic—they protect oneself as well as other people. Therefore, it is not obvious whether they are primarily driven by self-interest or other-interest. This confusion also emerges in public communication, as the same health behavior is in some cases framed as self-protective while in others it is framed as other-protective (Yong, 2021). As such, the current article examines the role of morality in support of public health measures during the COVID-19 pandemic.

There is a growing body of evidence for a positive relationship between concern for others and behaviors to combat the pandemic. For instance, greater prosociality is associated with more physical distancing and protective behaviors for both the self and social well-being (Twardawski et al., 2021; Nelson-Coffey et al., 2021). Likewise, beliefs in pure goodness are related to prosociality and pandemic coping behavior (Schiffer et al., 2021). People who behaved prosocially in economic experiments years before the pandemic had greater preventive intentions (Jordan et al., 2021). Similarly, prosociality is associated with preventative measures (Campos-Mercade et al., 2021) and empathy for the most vulnerable predicts physical distancing and mask use (Petrocchi et al., 2021; Pfattheicher et al., 2020). Importantly, prosocial tendencies (in particular, helping someone without knowing who one is helping) are related to greater adherence to protective behaviors, whereas selfishness (antisocial tendencies) is negatively related to compliance (Dinić & Bodroža, 2021). In a similar vein, individualism negatively predicts adherence to physical distancing rules, while collectivism positively predicts adherence (Biddlestone et al., 2020). These findings offer a growing body of evidence that prosociality is linked to public health behavior during the pandemic.

Other studies have explored the association between morality and preventative measures. Chan (2021) found that the care and fairness moral foundations are associated with staying-at-home, mask-wearing, and social distancing, while the sanctity foundation is associated only with mask-wearing and social distancing. Pagliaro et al. (2021) found that endorsing the moral principles of care and fairness (vs. authority and loyalty) is associated with trust in science, which, in turn, is associated with intentions to engage in various prescribed and discretionary behaviors. Bor et al. (2022) found that vaccination and compliance with guidelines are highly moralized issues, although moral condemnation of non-vaccination and non-compliance is best predicted by self-interest. Interventions based on emphasizing that cooperating is the morally right thing to do have found an overall positive effect on intentions to engage with preventative measures in surveys (Ruggeri et al., 2022), thus suggesting that morality may play a role in determining adherence to public health measures. Finally, deontological arguments enhance the participant’s interest in sharing health messages (Everett et al., 2020).

However, other studies have provided mixed evidence for the role of morality in public health. For instance, Capraro and Barcelo (2020) and Favero and Pedersen (2020) found that showing people prosocial and empathy messages does not increase intentions to practice social distancing. Hacquin et al. (2020) found that messages emphasizing the duty to help are not superior to other messages (e.g., personal risk) in their effects on behavioral intentions. As such, there is inconsistent evidence that prosociality is associated with preventative behaviors during the pandemic. Indeed, a comprehensive assessment of the literature on COVID-19 recently revealed that messages highlighting benefits to others or the importance of protecting others had mixed effects on preventative behaviors (Ruggeri et al., 2022).

In the current investigation, we examine the link between several aspects of morality in support for physical distancing, physical hygiene, and support for public policies. We present three studies based on several large international samples, using both measures of intentions to engage in preventative measures and real-world behavior obtained through Google mobility data, and using individual-level morality measures as well as national-level moral values obtained through the World Value Survey, to investigate the role of morality on individual actions that foster public health. Specifically, in Study 1 we take a closer look at the importance of moral identity, cooperative morality, and the scope of the “moral circle” (i.e., the boundary around those regarded as worthy of moral consideration) on social distancing, physical hygiene, and policy support. In Study 2, we will verify if these morality measures correlate with people’s movement changes during the pandemic compared to pre-pandemic time. Finally, to verify the robustness of the relationship between morality and mobility, we will correlate data from the World Value Survey’s Wave 6 (WVS) and COVID-19 Google mobility data in Study 3. Mainly, we selected the WVS items most closely related to our study’s moral variables: Benevolence, Conformity, and Universalism.

Moral Identity

Maintaining physical isolation and other relevant public health behaviors during a pandemic should be more likely among those who see morally required actions as part of their core identity. Moral commitment reflects our moral identity (Aquino & Reed, 2002) or how important being a moral person is for our identity (Hardy & Carlo, 2011; Strohminger & Nichols, 2014). Moral identity includes two dimensions: internalization and symbolization (Aquino & Reed, 2002). Internalization refers to how critical moral virtues such as compassion, fairness, and care are to the individual’s self-concept (a private aspect of the self); symbolization refers to how important it is for the individual to express those characteristics to the outside world (a public aspect of the self).

Previous research suggests that moral internalization is a stronger predictor of prescriptive moral actions (committing good deeds) than symbolization. Internalization and symbolization are equally important for proscriptive moral actions (refraining from committing bad deeds) (Boegershausen et al., 2015). For example, internalization, but not symbolization of moral identity, is positively related to how much money people give to charitable organizations (Aquino & Reed, 2002), especially those that help outgroups (Reed & Aquino, 2003). Internalized moral identity is also positively associated with prosocial behavior and people’s views of humanity (Aquino et al., 2011). By contrast, internalization and symbolization both predict moral behaviors in response to perceived threats by others (Boegershausen et al., 2015).

In the current research, we hypothesized that moral identity would be positively associated with physical distance, physical hygiene, and support for public policies. Most of our measures were taken in a prescriptive format (e.g., “washing my hands longer than usual”) and none of the measures were framed in a strictly proscriptive format (e.g., “don’t do X”). Therefore, we also wanted to test whether internalized moral identity would predict these behaviors better than symbolized moral identity (Boegershausen et al., 2015).

Morality as Cooperation

The theory of morality-as-cooperation conceptualizes morality as a set of stable biological and cultural solutions to recurrent social challenges (Curry, Mullins, & Whitehouse, 2019; Curry, Jones Chesters, & Van Lissa, 2019). The theory posits that the most critical moral concerns are about actions that pose solutions to cooperative dilemmas. One of the theory’s pillars is viewing cooperative situations through the lens of non-zero-sum games. The pandemic response can be considered a non-zero-sum game as engaging in preventative behaviors results in positive collective results. For instance, prosocial people are more likely to engage in health behaviors. Messages highlighting the harm to the collectivity can effectively increase intentions to engage in preventive behaviors such as wearing a face mask (Capraro & Barcelo, 2020) and practicing physical distancing (Lunn et al., 2020; Pfattheicher et al., 2020), although some research found null or small effects, especially in the domain of physical distancing (Capraro & Barcelo, 2020; Jordan et al., 2021). An overall assessment of the available literature found that emphasizing that cooperating is the morally right thing to do has a positive effect on intentions to engage in preventative measures (Ruggeri et al., 2022). We, therefore, hypothesized that the extent to which people think of morality in cooperative terms might be associated with greater support for public health measures during the pandemic.

Moral Circles

Who deserves our consideration and care? The variation in such responses ranges from restricted family ties, usually involving kinship, to a broader view, including friends, strangers, animals, and even nature as a whole (Singer, 1981). These circles can be understood as concentric, ranging from a central circle whose moral consideration is narrow and parochialist to a broader and more universal circle that includes more distant members. If, on the one hand, parochial moral circles help to maintain the cohesion of family or other close-knit groups, universalist moral circles seem more likely to reflect concern for others and society as a whole during a pandemic. This is because the systems are not isolated. Quite the contrary, when it comes to the spread of disease, tight-knit circles are not isolated from one another but instead share a common fate. Based on this, we hypothesize that people who endorse a wider moral circle will also show a higher positive engagement on our outcome measures of pandemic response.

Overview

The current research examined whether individual and national differences in moral identity, morality-as-cooperation, and the size of moral circles are positively associated with the willingness to engage in public health behaviors (including physical hygiene, physical distancing, and policy support), across large international samples during the COVID-19 pandemic. Specifically, in Study 1 we took a closer look at the importance of moral identity, cooperative morality, and the scope of the “moral circle” (i.e., the boundary around those regarded as worthy of moral consideration) on social distancing, physical hygiene, and policy support. Our primary hypothesis was that people with higher moral identity, higher scores on morality-as-cooperation, and more universalist moral circles would be more likely to support all public health measures (Study 1). In Study 2, we verified if these morality measures correlate with people’s movement changes during the pandemic compared to pre-pandemic time, that is, we examine whether moral motivation has been associated with greater self-restrictions in physical movement during the pandemic. To do that, we examined the link between our three morality variables and real-world behavior using Google mobility reports (Study 2). Finally, to verify the robustness of the relationship between morality and mobility, we correlated data from the World Value Survey’s Wave 6 (WVS) and COVID-19 Google mobility data in Study 3. Mainly, we selected the WVS items most closely related to our study’s moral variables: Benevolence, Conformity, and Universalism. With this, we attempted to replicate our main results by linking mobility behavior to national-level moral values using the World Value Survey (Study 3).

Study 1: The Relationship between Morality and Support for Public Health Measures

Methods

This study is part of the International Collaboration on the Social & Moral Psychology of COVID-19 (ICSMP; https://icsmp-covid19.netlify.app/index.html). This collaboration was launched in April 2020 and brought together scholars worldwide to examine the psychological factors underlying the attitudes and behavioral intentions related to COVID-19. Massive multi-national samples were generated, and all data, codebooks, codes for analysis, and other materials are freely available here (Azevedo et al., 2023). For more methodological details such as sample size calculation, data exclusions, translation of instruments, see Van Bavel et al. (2022). The survey was approved by the ethics committee at the University of Kent.

Participants

Data from 46,576 participants from 68 countries were collected (M_age = 43.11, SD_age = 16.08; Gender = 24,186 females, 22,189 males, 163 others, 38 did not answer). Table S1 (Supplemental Material) presents a complete list of countries with respective sample sizes, mean age, and gender. Of the 68 countries where data were collected, representative or post-hoc weighted samples were achieved in 30, convenience samples were collected in 36, and mixed types of sampling (representative and convenience) were present in 2 countries. All the analyses reported in this article were repeated, controlling for differences in the sampling method. Although some coefficients differed as a function of the sampling procedure, none of these differences compromised or altered the main effects we reported.

Survey

Questionnaires were administered online. Each participant completed a series of psychological measures and self-reported public health behaviors (see the complete survey with all items in the Supplemental Material). Participants completed the scales in random order.

We focused on three potential predictors of public health support for the current paper. The first predictor is a 10-item Moral Identity inventory, which comprises two dimensions: internalization and symbolization (Aquino & Reed, 2002). Participants read and responded to the following instruction: “Listed below are some characteristics that might describe a person: caring, compassionate, fair, friendly, generous, helpful, hardworking, honest, kind. The person with these characteristics could be you or it could be someone else. For a moment, visualize in your mind the kind of person who has these characteristics. Imagine how that person would think, feel, and act. When you have a clear image of what this person would be like, answer the following questions.” Then, they responded to items measuring the two dimensions such as “It would make me feel good to be a person who has these characteristics” and “I often wear clothes that identify me as having these characteristics.” These measures used an 11-point slider scale with three labels items: 0 = “strongly disagree,” 5 = “neither agree nor disagree,” 10 = “strongly agree” (for details, see the Supplemental Material).

The second predictor is morality-as-cooperation (Curry, Jones Chesters, & Van Lissa, 2019), which comprises seven factors: family, group, reciprocity, heroism, deference, fairness, and property (one item each, though the factors are not orthogonal). Participants read and responded to the following instructions: “When you decide whether something is right or wrong, to what extent are the following considerations relevant to your thinking?” The items were: Family: “Whether or not someone helped a member of their family”; Group: “Whether or not someone worked to unite a community”; Reciprocity: “Whether or not someone kept their promise”; Heroism: “Whether or not someone showed courage in the face of adversity”; Deference: “Whether or not someone deferred to those in authority”; Fairness: “Whether or not someone kept the best part for themselves”; Property: “Whether or not someone kept something that didn’t belong to them”. These measures used an 11-point slider scale with three labels items: 0 = “strongly disagree,” 5 = “neither agree nor disagree,” 10 = “strongly agree.”

The third predictor is a one-item measure of the extension of moral circles (Waytz et al., 2019). Participants read and responded to the following instructions: “On this page, we would like you to indicate the extent of your moral circle. By moral circle, we mean the circle of people or other entities for which you are concerned about right and wrong done toward them. Please use the following scale to select the extent of your moral circle.” The scale ranged from 1 to 16 where 1 is “all of your immediate family” and 16 is “all things in existence.”

As outcome variables, we included three measures of public health support (for details, see Van Bavel et al., 2022). A physical distancing scale, consisting of four items (it was initially a five items scale, but one item was excluded due to reliability—see Van Bavel et al. (2022)—such as, “During the days of the coronavirus (COVID-19) pandemic, I have been staying at home as much as practically possible.” A physical hygiene scale, consisting of five items, such as, “During the days of the coronavirus (COVID-19) pandemic, I have been washing my hands longer than usual.” A policy support scale, consisting of five items, such as, “During the days of the coronavirus (COVID-19) pandemic, I have been in favor of closing all schools and universities.” As above, we used an 11-point slider scale anchored on 0 = “strongly disagree,” 50 = “neither agree nor disagree,” 100 = “strongly agree,” which was re-coded to a scale from 0 to 10. Data acquisition occurred over April and May 2020.

Results

We conceptualized our data as a multilevel data structure in which participants (level 1) were nested within countries (level 2). Accordingly, the data were analyzed with a series of multi-level random coefficient models (MLM) using the program HLM (Raudenbush & Liu, 2000). These analyses followed the guidelines offered by Nezlek (2010) for conducting multilevel modeling analyses of cross-cultural research.

The reliability for our measures that consisted of multiple items was also estimated using procedures described by Nezlek (2017). This procedure employed a three-level model: items nested with persons and persons nested within countries. These models estimated the multilevel equivalent of a Cronbach’s alpha. These reliabilities, as well as the descriptive statistics for our measures, are presented in Table 1. Person-level correlations between our measures were estimated using Mplus (Muthén & Muthén, n.d.), and they are shown in Table 1.

Table 1.

Summary statistics for person-level measures.

	Variance
	Mean	Between	Within	Reliability
Physical Contact	8.51	.21	2.87	.78
Physical Hygiene	7.93	.47	3.18	.79
Policy Support	7.86	.95	4.30	.87
Internalized moral identity	7.79	.31	2.66	.75
Symbolic moral identity	5.32	.54	4.18	.81
Moral circle	9.39	1.26	26.19	n/a
Moral cooperation	6.54	.17	2.50	.73

Note. The mean score for each scale is presented along with the estimated variance within and between participants and the scale reliability (alpha). There are no alphas for the moral circle scale as it is a one-item measure. Higher scores reflect greater support for each policy measure (0–10), stronger moral identification in the case of moral ID (0–10), larger moral circles in the case of moral circles (1–16), and stronger relevance attributed to each dimension of morality-as-cooperation (0–10).

The three COVID-19 protection measures (scores on the limiting contact, hygiene, and policy support scales) were treated as outcomes in these analyses. To test our hypotheses, we examined person-level relations between the three COVID-19 protection measures (modeled as outcomes) and moral identity internalization (MII), moral identity symbolization (MIS), morality-as-cooperation (MAC), and moral circle (MOC) (modeled as predictors). Predictors were defined as the mean scores for each scale. To account for relations among the predictors, all predictors were entered at the person level of the model. Predictors were entered group-mean centered and were modeled as randomly varying. Entering predictors group-mean centered meant that estimates of coefficients controlled for country-level differences in the means of predictors (Nezlek, 2010). In essence, a regression equation consisting of an intercept and a series of slopes was estimated for each country. These estimates were “passed up” to the country level, tested for significance. The model is presented below.

$\begin{array}{l}\mathrm{Level}1:{\mathrm{y}}_{\mathrm{ij}}={\beta }_{0\mathrm{j}}+{\beta }_{1\mathrm{j}}*\left(\mathrm{MII}\right)\\ +{\beta }_{2\mathrm{j}}*\left(\mathrm{MIS}\right)+{\beta }_{3\mathrm{j}}*\left(\mathrm{MAC}\right)\\ +{\beta }_{4\mathrm{j}}*\left(\mathrm{MOC}\right)+{\mathrm{r}}_{\mathrm{ij}}\end{array}$

$\mathrm{Level}2\left(\mathrm{intercept}\right):{\beta }_{0\mathrm{j}}={\gamma }_{00}+{\mu }_{0\mathrm{j}}$

$\mathrm{Level}2\left(\mathrm{slope}\mathrm{MII}\right):{\beta }_{1\mathrm{j}}={\gamma }_{10}+{\mu }_{1\mathrm{j}}$

$\mathrm{Level}2\left(\mathrm{slope}\mathrm{MIS}\right):{\beta }_{2\mathrm{j}}={\gamma }_{20}+{\mu }_{2\mathrm{j}}$

$\mathrm{Level}2\left(\mathrm{slope}\mathrm{MAC}\right):{\beta }_{3\mathrm{j}}={\gamma }_{30}+{\mu }_{3\mathrm{j}}$

$\mathrm{Level}2\left(\mathrm{slope}\mathrm{MOC}\right):{\beta }_{4\mathrm{j}}={\gamma }_{40}+{\mu }_{4\mathrm{j}}$

In this model, persons (denoted by the index i) are nested within countries (denoted by the index j). A set of coefficients (an intercept and two slopes) is estimated for each country, and the null hypotheses that mean coefficients (for countries) are not different from 0 are tested at level 2, the country level. Continuing this example, if the γ₁₀ coefficient is significantly different from 0, then the mean slope for MII (the relationship between the outcome and internalized moral identification) is significantly different from 0.

A summary of these analyses is presented in Table 2. MLM analyses estimate only unstandardized coefficients, representing the expected change in an outcome associated with a 1 point increase in a predictor. For example, in the endorsement of contact limitations analysis, the internalized subscale of moral identity (MII) coefficient was .20. Therefore, for every point that internalization increased, endorsement of contact limitations increased .20 points.

Table 2.

Relations between outcomes and predictors: moral identity internalization was the strongest predictor of all three COVID-19 public health support measures.

Outcomes
Predictors		Physical Contact	Hygiene	Policy Support
Moral ID	Internalization	.20* [0.17, 0.23]	.15* [0.13, 0.17]	.20* [0.18, 0.23]
	Symbolization	-.003 [-0.02, 0.13]	.10* [0.08, 0.12]	.01 [-0.004, 0.03]
Morality-as-cooperation		.10* [0.07, 0.10]	.12* [0.09, 0.15]	.12* [0.10, 0.15]
Moral circle		.013* [0.009, 0.016]	.01* [0.006, 0.014]	.008* [0.003, 0.013]
	Explained var.	9.8%	10.2%	6.2%

Note. Values are the coefficients and 95% CI from MLM analysis.

^*p < .001.

As predicted, moral identity internalization was significantly and positively related to limiting contact, improving physical hygiene, and endorsing COVID-19 public policies (see Table 2). The more individuals valued having a moral self-concept, the more they supported measures to combat the pandemic. Moral identity symbolization was positively related to improving hygiene only, suggesting that internalization or “private” moral identity had a broader impact on pandemic responses. Figure 1 shows the relation between moral identity and public health measures for all countries and territories.

Figure 1.

Relation between individuals’ moral concern and public health measures in 68 countries and territories (N = 46,576).

Note. The heat index depicts the slope coefficients in each country. Blueish colors mean negative associations between our predictors and our outcomes, while reddish colors mean positive associations (higher scores reflect stronger relationships between moral identity internalization and symbolization, limiting contact, physical hygiene, and policy support measures). Gray color represents countries for which we do not have data.

Additionally, morality-as-cooperation was significantly and positively related to endorsing COVID-19 public policies, limiting contact, and improving physical hygiene. Third and finally, the scope of the moral circle was also significantly and positively related to the three protective measures. On average, individuals with a broader moral circle supported the recommendations to fight the pandemic more strongly than individuals with a more restricted circle. Figure 1 and Figure S1 (Supplemental Material) show the relation between moral identity, moral-as-cooperation, and moral circle on public health measures.

Effect sizes were defined as the percent reduction in the person-level variance associated with the inclusion of a set of predictors. It is computed by comparing the residual level 1 variance from a model with predictors to one without predictors (the null model). For example, the level 1 variance of limiting contact is 2.868. When it is predicted by the morality measures, it is 2.586. The explained variance is computed by (2.868 - 2.586)/2.868 = 0.098; thus, our combined measures of moral motivation explain 9.8% of the variance in limiting contact (it could also be represented as a correlation by computing $\sqrt{0.098}$ = .31). As shown in Table 2, individuals’ morality accounted for 9.8%, 10.2%, and 6.2% of limiting contact, hygiene, and policy support, respectively.

Study 2: The Relationship between Morality and Reductions in Mobility

Methods

Study 1 relies on self-reported measures of public health support. To overcome this limitation, Study 2 examines whether our measures of morality are associated with actual behavioral change. Specifically, we test whether countries with higher average moral identity, morality-as-cooperation, and moral circles show more reduction in mobility during the pandemic.

The change in mobility measure was computed based on Google’s COVID-19 Community Mobility Reports (data available at www.google.com/covid19/mobility/), which indicate how people’s movement has changed during the pandemic. The reports show movement trends over time across different categories of places: retail and recreation, groceries and pharmacies, parks, transit stations, workplaces, and residential. Percentage change for each day is computed relative to a baseline, which is a median value, for the corresponding day of the week, during the 5-week period before the coronavirus outbreak (January 3–February 6, 2020). To compute an overall index, we computed average indices for each of the places over April and May 2020 (the period in which we collected most of the samples in Study 1). We then created a composite index of mobility (the lower the number, the lower the community mobility) by averaging mobility across all places, with residential mobility reverse-coded (alpha = .88, M = -35.21, SD = 16.24; i.e., a 35% reduction on movement relative to the baseline).

To control for the possibility of the effects on community mobility being influenced by the different containment measures adopted by each country, we included a stringency index in our statistical model. We use the Oxford Covid-19 Government Response Tracker (OxCGRT) to calculate the averaged index representing containment measures in each country over a given period. Since the index could change over time, we calculated the index in the specific data collection period for each country. In countries that did not record the collection date or were not present in the ICSMP data, we considered the average start and end date based on information from countries with this data. Information on how the index is obtained and calculated is described in the article by Hale et al. (2021) or at https://www.bsg.ox.ac.uk/research/research-projects/covid-19-government-response-tracker. This index ranged from 0 to 100. We ran Kendall partial correlation between moral identity, morality-as-cooperation, and moral circles and mobility (as indexed by Google), controlling for the effect of the lockdown stringency index. We conducted our analysis for a sample of all 64 countries that had available data for the morality variables, the mobility scores, and the lockdown stringency index.

Results

In line with our predictions, higher moral identity internalization and symbolization were negatively correlated with community mobility (r = -.57, p < .0001 and r = -.46, p < .001, respectively; see Figures 2A and 2B), higher moral circle was negatively correlated with community mobility (r = -.42, p < .001; see Figure 2C), and morality-as-cooperation was marginally negatively correlated with community mobility (r = -.29, p = .06; Figure 2D). After statistically adjusting for the effects of the lockdown stringency index, partial correlation analysis revealed that moral identity internalization and symbolization and moral circle were still negatively correlated with community mobility (r = -.24, p < .01, r = -.24, p < .01, r = -.22, p < .01, respectively; see Figures 2A, 2B, and 2C) while morality-as-cooperation was no longer correlated with community mobility (r = -.04, p = .66). Thus, after controlling the effects of different levels of lockdown across the countries, moral identity and the moral circle still predicted real-world behavior change during a pandemic.

Figure 2.

Relation between moral variables and community mobility.

Note. A = Correlation between moral identity internalization and google mobility index. B = Correlation between moral identity symbolization and google mobility index. C = Correlation between moral circle and google mobility index. D = Correlation between morality-as-cooperation and google mobility index. Google mobility is depicted as a mean change in mobility during April and May, i.e., blueish colors mean a higher reduction of mobility during this period, while reddish colors mean a smaller reduction of mobility.

Study 3: The Relationship between Values and Reductions in Mobility

Methods

To verify the robustness of the relationship between morality and mobility, we conducted an additional correlational analysis between data from the World Value Survey’s Wave 6 and Google’s COVID-19 mobility data (see the Methods of Study 2). In WVS’s Wave 6, 10 items related to the 10 dimensions of Shalom Schwartz’s Theory of Values (Schwartz, 2012) were collected: self-direction, power, security, hedonism, benevolence, achievement, stimulation, conformity, universalism, and tradition. We selected the WVS items most closely related to our study’s moral variables:

Benevolence: “It is important to this person to do something for the good of society”;

Conformity: “It is important to this person to always behave properly”;

Universalism: “Looking after the environment is important to this person; to care for nature and save life resources.”

For each sentence, participants from the WVS read and answered the following instruction: “Now I will briefly describe some people. Using this card, would you please indicate for each description whether that person is very much like you, like you, somewhat like you, not like you, or not at all like you? Very much like me (1) to Not at all like me (6).” We inverted the scale to keep the direction of the effects consistent with the other studies (higher scores mean strong support for these values: 1 = Not at all like me to 6 = Very much like me). We conducted our analysis for a sample of all 52 countries that had available data for the WVS and mobility scores.

To control for the possibility of the effects on community mobility being influenced by the lockdown adopted by each country, we followed the same procedures used in Study 2 based on the Oxford Covid-19 Government Response Tracker (OxCGRT). We ran Kendall’s partial correlation between Benevolence, Conformity, and Universalism and mobility (as indexed by Google), controlling for the effect of the lockdown stringency index. We conducted our analysis for a sample of all 52 countries that had available data for the WVS, the mobility scores, and the lockdown stringency index.

Results

In line with our predictions, higher scores on benevolence, conformity, and universalism were positively correlated with community mobility (r = -.51, p < .001, r = -.45, p < .001, and r = -.40, p < .01, respectively; Figure 3). After statistically adjusting for the effects of the lockdown stringency index, partial correlation analysis revealed that Benevolence and Universalism were still negatively correlated with community mobility (r = -.22, p = .03, r = -.20, p < .03, respectively; see Figure 3) while Conformity was only marginally correlated with community mobility (r = -.16, p = .09). Thus, after adjusting for the effects of different levels of lockdown across countries, Benevolence and Universalism still predict real-world behavior change during a pandemic offering additional support for our findings that different aspects of morality are related to a greater change in behavior in compliance with public health policies.

Figure 3.

Relation between human values and community.

Note. A = Correlation between benevolence and google mobility index. B = Correlation between conformity and google mobility index. C = Correlation between universalism and google mobility index. Google mobility is depicted as a mean change in mobility during April and May, i.e., blueish colors mean a higher reduction of mobility during this period while reddish colors mean a smaller reduction of mobility.

Discussion

A recurrent question researchers around the world have been asking is why do people get together at parties, do not wear face masks, and do not support public policies to combat the pandemic (Van Bavel et al., 2020)? These individual actions have collective consequences, putting others at risk of catching COVID-19. Across three large international studies, we found that various morality dispositions help explain why some people and nations engage in the collective actions necessary to fight the pandemic while others do not. Here we provide evidence for this theory by demonstrating that three aspects of morality predict support for policy measures to fight the pandemic: moral identity, morality-as-cooperation, and moral circles. Together, these moral dispositions accounted for 9.8%, 10.2%, and 6.2% of support for limiting contact, improving hygiene, and supporting policy change, respectively. Additionally, these three aspects of morality and the personal values of benevolence and universalism were correlated in the same direction with actual behavior, as measured by the COVID-19 Google mobility report (higher morality scores are correlated with higher reduction in community mobility). Taken together, these studies suggest that variation in morality might help explain some of the striking variations in public reactions to the pandemic across individuasl and nations.

As Study 1 used only self-report measures, we conducted two additional studies to investigate the relationship between morality and actual behavior in coping with the pandemic. These results confirmed those of the first study: morality was positively associated with active engagement against the pandemic, in this case, estimated by the reduction of community mobility. Notably, after controlling the effects of different levels of lockdown across the countries, moral identity and the moral circle still predict real-world behavior change during a pandemic. The robustness of these results is reinforced by observing a significant correlation in the same direction between personal values (benevolence and universalism) and community mobility. It is essential to mention that our data were collected in 2020, while the WVS data is from a few years earlier (between 2010 and 2014). Thus, it seems we are picking up on fairly stable tendencies characterizing the moral beliefs in the populations of the countries studied. Together, the findings of the three studies show the role of individuals’ morality not only in their interest in acting collectively but in their actual behavior to fight the pandemic.

Moral identity was the strongest predictor of support for public health measures during the pandemic. In particular, the internalization dimension of moral identity appeared to be more critical than the symbolization dimension of moral identity. Considering it essential for people to see themselves as caring, compassionate, and fair, predicts support for public health measures. These findings were broadly consistent across 68 countries with diverse cultures and social contexts and aligned with previous studies linking moral identity internalization to prosocial actions (Aquino et al., 2011; Aquino & Reed, 2002; Reed & Aquino, 2003). Actions to combat the pandemic require sacrifices and behavioral changes from each individual, resulting in better collective outcomes. In this sense, social changes and actions might be seen as non-zero-sum in which everyone wins with cooperation. This is in line with research showing that prosocial people are more likely to engage in health behaviors (Campos-Mercade et al., 2021; Jordan et al., 2021) and that messages highlighting that cooperating is the morally right thing to do increase intentions to engage in preventative measures (Ruggeri et al., 2022). Congruently, Nivette et al. (2021) found in a Swiss sample of young adults that “antisocial potential” (as described by “low acceptance of moral rules, legal cynicism, low shame or guilt, low self-control, high engagement in delinquent behaviors, and association with peers who exhibit social deviance”) was associated with non-compliance to physical hygiene and social distancing during the pandemic.

We also found that higher scores on morality-as-cooperation were associated with support measures to combat the pandemic. This result makes sense considering that the fight against the pandemic demands a collective perspective: each person’s action impacts the whole group (from a small family to a nation). Furthermore, they add to the results on internalized moral identity since some virtues related to morality-as-cooperation involve care, loyalty, solidarity, and trust (Curry, Jones Chesters, & Van Lissa, 2019).

Finally, the larger the moral circle of an individual, the greater the support for measures to combat the new coronavirus. These results suggest that promoting a more universalist perspective may positively tackle the pandemic. However, despite being accepted in ethical reasoning, the expansion of moral circles is still far from being taken at the practical level (Lunn et al., 2020). The defense of the interests of individuals and locals is part of the discourse of many political leaders. Moreover, parochialism versus universalism is related to political ideologies. A recent paper by Waytz et al. (2019) found that ideological differences are linked to parochialist and universalist moral circles, with the former being more conservative and the latter more liberal. During the COVID-19 pandemic, some studies have found that conservatives tend to support COVID-19 public health rules less than liberals in polarized contexts (Azevedo & Jost, 2021; Capraro & Barcelo, 2020; Gollwitzer et al., 2020; Van Bavel et al., 2022). Moreover, this partisanship effect in the United States was related to differences in limiting contact behavior and might have impacted the number of cases and deaths (Gollwitzer et al., 2020). The pandemic highlights the need to expand the circles of care beyond the members of our groups. The fact that we are immersed in this tragedy can serve as a clarion call to adopt more collectivist and universal norms.

The effect sizes of these moral dispositions were modest. As we noted above, these moral dispositions accounted for 9.8%, 10.2%, and 6.2% of support for limiting contact, improving hygiene, and supporting policy change, respectively, in our global study. Moreover, it is important to note that even small effect sizes can be important in a pandemic. As stated by Ellis (2010, p. 35), “in the right context even small effects may be meaningful.” Considering the exponential way a virus spreads, part of its control depends on different governmental actions, and behavior changes carried out quickly and in an organized manner. Many other behavioral changes taken together might be the most effective in maintaining R₀ below 1 (Haug et al., 2020). Such considerations were demonstrated by Haug et al. (2020). The authors noted that the implementation of social distance, travel restrictions, education and communication, and PPE provision are among the most effective measures to control the pandemic. As the authors themselves write, no non-pharmaceutical intervention works like a silver bullet. It will be the sum of different changes implemented that will have the most significant effect. In other words, the result of different behavior changes implemented in an organized way may be greater than the simple sum of their effects because when exponential growth (or decay) is in play, curves are geometric rather than summative. In a pandemic that is killing millions of people worldwide, even a small effect size has the potential to save the lives of thousands of people.

Our work does have some limitations. We collected data from 68 countries. However, some regions (e.g., Africa and the Middle East) were underrepresented; moreover, we were unable to collect representative samples with respect to sex and age in several countries. Therefore, our results apply to nations and samples for which we could collect data. Moreover, our data were collected during the early phase of the COVID-19 pandemic. So, whether these results apply to later stages of the COVID-19 pandemic or other pandemics remains an open question. Another limitation concerns the correlational nature of our study. The work on moral suasion suggests a causal link between morality and prosocial behavior (Brañas-Garza, 2007; Capraro et al., 2019; Capraro & Perc, 2021; Dal Bó & Dal Bó, 2014). Since our results indicate that pandemic response is seen as a prosocial behavior, we expect there to be indeed a causal link between morality and pandemic response. However, we cannot rule out the alternatives that pandemic response actually drives morality or that they are both driven by a third underlying variable. Yet another limitation regards the behavioral data from the Google mobility report, which, on the one hand, does provide a behavioral measure of reduced contact, but, on the other hand, it is unclear whether it relates to actual maintaining hygiene and policy support. Therefore, Study 2 and Study 3 do not allow us to conclude that morality is also associated with actual physical hygiene and policy support.

Future directions

Our results suggest that support for measures to combat the pandemic may be strengthened by fostering individual morality aimed at cooperation, benevolence, and universalism. Some studies have shown that situational cues can activate access to moral identity, inducing more moral behaviors (Aquino et al., 2009). In addition, expressing appreciation for prosocial actions increases prosociality in individuals with lower levels of moral identity (Winterich et al., 2013). Finally, promoting prosocial actions can be done with the mere presence of individuals recognized for their high moral identity. For example, individuals in social dilemmas interacting with people perceived as highly moral show more cooperative behavior regardless of their social value orientation (pro-self vs. pro-social (Lange & Liebrand, 1991). That is, the presence of moral individuals can induce a positive spiral of cooperative behavior in both prosocial and pro-self others. Thus, new studies should test messages’ effects to increase moral identity’s salience and whether such a manipulation enhances support of COVID-19 mitigating measures.

Conclusion

Fighting a global pandemic requires individual sacrifices for the good of the community. Individual actions contrary to health authorities’ guidelines can affect people who do not comply with the rules and others around them. Thus, engaging with behaviors to deal with the pandemic can be understood as a moral commitment in its cooperative dimensions. Our results highlight the connection between individuals’ morality in their support for public health measures and may provide a basis for leaders, institutions, and the media to consider or galvanize support for public policies limiting the spread of the virus. For instance, leaders and communicators may consider appealing to moral identity more explicitly in their messaging and public communication to more strongly target individuals disposed to support COVID-19 policies. Such support could have important follow-on effects on other members of the public by increasing the perception that “it’s easier to be and become virtuous among other people who care about your wellbeing and your character . . . who encourage you to do the right thing” (Alfano, 2016, p. 119).