Why we should care about moral foundations when preparing for the next pandemic: Insights from Canada, the UK and the US

Lizette Pizza; Samuel Ronfard; John D. Coley; Deborah Kelemen

doi:10.1371/journal.pone.0285549

Abstract

Health behaviors that do not effectively prevent disease can negatively impact psychological wellbeing and potentially drain motivations to engage in more effective behavior, potentially creating higher health risk. Despite this, studies linking “moral foundations” (i.e., concerns about harm, fairness, purity, authority, ingroup, and/or liberty) to health behaviors have generally been limited to a narrow range of behaviors, specifically effective ones. We therefore explored the degree to which moral foundations predicted a wider range of not only effective but ineffective (overreactive) preventative behaviors during the COVID-19 pandemic. In Study 1, participants from Canada, the United Kingdom, and the United States reported their engagement in these preventative behaviors and completed a COVID-specific adaptation of the Moral Foundations Questionnaire during the pandemic peak. While differences occurred across countries, authority considerations consistently predicted increased engagement in both effective preventative behaviors but also ineffective overreactions, even when controlling for political ideology. By contrast, purity and liberty considerations reduced intentions to engage in effective behaviors like vaccination but had no effect on ineffective behaviors. Study 2 revealed that the influence of moral foundations on U.S participants’ behavior remained stable 5-months later, after the pandemic peak. These findings demonstrate that the impact of moral foundations on preventative behaviors is similar across a range of western democracies, and that recommendations by authorities can have unexpected consequences in terms of promoting ineffective—and potentially damaging—overreactive behaviors. The findings underscore the importance of moral concerns for the design of health interventions that selectively promote effective preventative behavior.

Citation: Pizza L, Ronfard S, Coley JD, Kelemen D (2023) Why we should care about moral foundations when preparing for the next pandemic: Insights from Canada, the UK and the US. PLoS ONE 18(5): e0285549. https://doi.org/10.1371/journal.pone.0285549

Editor: Luu Duc Toan Huynh, Queen Mary University of London, UNITED KINGDOM

Received: December 6, 2022; Accepted: April 26, 2023; Published: May 12, 2023

Copyright: © 2023 Pizza et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The data that support the findings of this study are publicly available in the OSF repository (https://osf.io/r7sfj/)."

Funding: This research was conducted under National Science Foundation Award (NSF) #2028140 to DK and SR and under NSF #2028109 to JDC. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Funder website: https://www.nsf.gov/.

Competing interests: The authors have declared that no competing interests exist.

Introduction

As a result of continuous globalization, national borders have become more diffuse, and infectious diseases have spread more rapidly across countries [1, 2]. In the last two decades, the world has seen multiple disease outbreaks emerging in different locations (e.g., H1N1 pandemic, Ebola, COVID-19), provoking profound health and social-economic crises along with widespread fear. Thus, it is critical that governments develop and implement effective strategies to control and prevent the emergence of infectious diseases [3]. This is difficult given that many factors influence how resources are allocated to support public health. Psychological factors that promote effective public health behaviors [4–6] and reduce deleterious or ineffective behavioral change are especially important.

Given its magnitude, the COVID-19 pandemic created an opportunity for understanding the impact of various psychological factors related to preventative health behavior change. Although variables like trust in science, misinformation, and scientific knowledge about the virus have been shown to predict COVID-related health behavior [7, 8], we argue here that special attention should be directed to the role of moral concerns. Moral concerns implicitly impact people’s judgments, play an important role in their health decision-making [9–11], and robustly predict behavior [12]. In fact, political ideology—which is often treated as a proxy for moral values—is sometimes more predictive of intentions to modify health behavior than scientific knowledge [13].

Although several researchers have found evidence of the important role of moral values in predicting participants’ engagement in preventative behaviors, to our knowledge no research has assessed whether moral concerns also influence participants’ engagement in less desirable behaviors, that is, behaviors that are ineffective like wearing a mask when alone. Specifically, although a study associated with this one found that U.S. participants with greater biological knowledge were less likely to engage in ineffective behaviors [14], it remains unknown whether moral values also influence people’s tendency to engage in ineffective behaviors. This is important because effective public health interventions need to reduce behaviors that are irrelevant to illness transmission but may have other deleterious effects (e.g., heightened anxiety, increased social isolation) while promoting behaviors that actually reduce illness transmission. In this research, we draw on Moral Foundations Theory [15], to explore the influence of people’s abstract moral considerations (i.e., “moral foundations”)—concepts such as personal and interpersonal harm, ingroup loyalty, and authority—on their self-reported health effective and ineffective behaviors when reasoning about various pandemic-related contexts and at different pandemic timepoints.

In Study 1, we examined whether specific moral foundations predicted people’s engagement in preventative behaviors and their intentions to vaccinate in the US and two other western countries (Canada and the UK) during the initial peak of the COVID-19 pandemic of December 2020 and January 2021 when daily mortality rates were high and COVID vaccinations were not yet available. In Study 2, we explored the stability of our earlier results at a later time point in the US, specifically, once the pandemic was past its initial deadliest peak and after vaccine programs started being rolled out for adults.

Moral foundations and health behaviors

According to Haidt and Joseph [15], moral foundations are innate, universal, affectively-based psychological systems that constitute the building blocks of morality with cultures varying in the weight accorded to each of them [16–18]. Five moral foundations have been identified: The harm foundation focuses on concern about others’ suffering as well as personal welfare, and the fairness foundation focuses on concerns about equal resource distribution, although these two foundations are usually clustered together as individualizing moral foundations. The remaining three foundations involve concerns that bind social groups (i.e., binding foundations). The authority foundation concerns the importance of showing obedience to rules and to others within a social hierarchy; the ingroup foundation relates to the virtues of loyalty and willingness to sacrifice for the group; and the purity foundation includes concerns about preserving naturalness or spiritual and physical cleanliness. Because—via the emotion of disgust—the purity foundation promotes avoidance of contamination by potentially invisible harms, it has been proposed as an evolutionary adaptation for avoiding disease transmission [18]. It is therefore particularly relevant in a pandemic situation. A sixth moral foundation—liberty—has also been recently proposed. It captures concerns related to an individual’s right to self-determine (i.e., be left alone) and resist oppression. Liberty is unique in that it can act as a sole guiding value at the expense of other moral foundations [19]. Given its significance, plus its relatively understudied influence on decision-making, it was a focus in this research.

Prior research on COVID-19 has found that, in general, people who value foundations such as care for others or fairness tend to comply with effective, recommended, COVID-19 preventative behaviors [10, 20–22]. This is presumably because those individuals prioritize staying healthy to protect those whose health is more vulnerable and tend to trust scientific experts’ recommendations. In contrast, in the US, non-compliance is associated with preferences for liberty over equality presumably because some individuals weigh their right to personal freedom over their sense of duty to others’ health leading them to downplay the severity of the pandemic [9]. Additional research—mainly conducted in the US—on people’s reasoning about public health measures like vaccination has also revealed associations with moral foundations. The individualizing foundations, as well as ingroup loyalty, have been found to be related to higher pro-vaccination attitudes [23, 24], while vaccine hesitancy has been related to liberty concerns—insofar as government vaccine mandates are seen to violate personal rights and civil liberties—and especially to purity considerations presumably due to individuals’ assumption that vaccines are artificial and that their components might contaminate the body and disturb the pure process of naturally building up immunity [25].

Most of the previous research has measured people’s personal weighing of different moral foundations using the Moral Foundations Questionnaire (MFQ) [26]. The MFQ asks participants to rate the relevance of foundation-related concerns to their general decision-making. However, given that the pandemic was an unexpected and novel situation, the MFQ might be too general and, therefore, unable to identify specific contextual patterns of the situation. For instance, the MFQ does not differentiate between concerns about harming others versus harming the self—a distinction that is highly relevant during a pandemic because these variables might play different roles in individuals’ preventative behaviors. Consequently, we created a modified version of the MFQ that focused specifically on the pandemic context for this study.

Furthermore, prior research on preventative behaviors and moral foundations has tended to cluster together concerns like ingroup, authority and purity [23] because they are binding foundations that often act similarly in context of other topics (e.g., incest). However, in context of a pandemic-relevant preventative behavior like vaccination, these binding foundations (e.g., authority and purity) might operate in profoundly different and potentially opposite ways. To understand these subtleties of prediction in the health context in the present study, we therefore analyzed the predictive role of each foundation independently.

Finally, and importantly, prior research demonstrating that moral concerns play an important role in predicting infection prevention behaviors has not explored their role in promoting ineffective behaviors, that is, actions that may be consistent with specific moral foundations (e.g., wearing a mask when completely alone; compulsive clothes washing, extreme social isolation) but that public health research suggests are not helpful in transmission prevention and which potentially represent dysfunctional hyper-reactive overgeneralizations of public health guidance. Insofar as they reflect and promote anxiety, these ineffective overreactions have potentially deleterious effects on long-term mental health [27]. While our prior research has shown that lower levels of biological knowledge may predict such behaviors [14] it is an open question how moral considerations influence them.

Moral foundations and cultural considerations

The ways in which moral concerns influence behavior might not always generalize even across relatively similar cultures, since human moral systems result from specific social and cultural experiences as well as adaptive pressures [16]. Before the pandemic, most studies exploring relationships between health behaviors and moral foundations tended to be conducted in the US (but see Rossen et al. [28]). This is problematic because the US is singular even among western nations. Relative to other western societies, the US tends to have higher levels of individual religiosity, and places stronger emphasis on individual freedom [29, 30]. U.S. findings therefore might not generalize to other countries. Fortuitously, the global context of the COVID-19 pandemic created an opportunity to explore whether different countries faced with similar acute challenges showed parallel patterns of response to them.

Prior research has been equivocal over whether patterns found in the US during the pandemic have generalizability beyond its boundaries [31, 32] (but see Schmidtke et al. [33]). One reason for this lack of consistency is the distinctive and marked pattern of polarization that occurred in the US relative to other even closely related western countries (e.g., the UK): For example, in the US, conservatives reported resistance to a whole range of behaviors including mask-wearing, hand washing, and willingness to vaccinate [13, 34]. They tended to minimize the threat posed by the pandemic and in turn were more likely to contract COVID-19 [35]. By contrast, in a non-US international sample, conservatives were generally more likely to endorse protective behavior like mask-wearing than liberals [36] and so the same COVID-19 transmission pattern did not hold. Given the strong association between political ideology and moral foundations in the US—with conservatives usually weighing the binding foundations of ingroup loyalty, authority, and purity as more relevant than liberals [17]—it is possible that correlations between political ideology and health behaviors in US reflect a culturally distinctive pattern in underlying moral reasoning.

Although Canada and the UK are Western nations connected to the US via shared language, history, and traditions, they differ from the US in ways that might impact the kind of abstract foundations weighed in reasoning about illness, health, and the pandemic. In addition to having a reduced narrative emphasis on individual rights and liberties [30], both differ from the US by having long-established government-funded universal healthcare systems rather than privately-funded healthcare. Furthermore, although both the US and UK–unlike Canada–had politically conservative government leadership during the pandemic, the rhetoric around the pandemic in both the UK and Canada was far less politicized and overtly polarized than in the US [13, 37]. For example, in the UK and Canada, political leaders strongly promoted compliance with scientifically recommended preventative behaviors and acknowledged the serious health threat posed by COVID-19. Canada and UK therefore represented highly relevant contexts in which to examine the generalizability of moral foundation effects on health behaviors during an international emergency.

The present research

The current studies examine how moral foundations relate to self-reported effective and ineffective preventative behaviors. In our preregistration of this research, we hypothesized that moral foundations would relate to effective preventative behaviors in each of three western countries (Canada, the UK, and the US), but it was an open question which foundations would be most predictive in each cultural context. By contrast, when considering ineffective preventative behaviors—overreactive behaviors that have not been studied to this point—and vaccine intentions, we anticipated a special role for the purity foundation given its relationship to disgust and pathogen avoidance [38]. Specifically, we hypothesized that purity beliefs might predict increases in ineffective behavior due to heightened fears about infection but decreases in intentions to vaccinate due to concerns about the tainting effects of vaccination ingredients. To test these hypotheses, in Study 1, we conducted an online study in the US, UK, and Canada at the peak of the pandemic, using an adapted contextualized version of the MFQ and questions about participants’ engagement in effective and ineffective preventative behaviors.

Because vaccination against COVID-19 was not publicly available during Study 1, our analyses separated participants’ reports of actual preventative behavior (e.g., handwashing) from self-reports of their notional preventative behavior (i.e., their intentions to vaccinate). Furthermore, given that moral foundations are argued to represent culturally-adapted abstract moral principles, we hypothesized that the relationships found in Study 1 would be relatively stable across time. To test this hypothesis, we conducted Study 2 in the US after the pandemic peak using materials adapted from Study 1. The only change we expected concerned the influence of the authority foundation on preventative behaviors given the relaxing of some official mandates as COVID-19 deaths dropped.

The specific analyses conducted for this paper were preregistered at: https://osf.io/ucn2m/?view_only=42fd4324a0f3421885c13b4519c7ec90. The surveys for Study 1 and Study 2 were part of a larger online study of adults’, specifically, parents’ reasoning about COVID-19 that is preregistered at: https://osf.io/wqvyr?view_only=6390c34605b6498fa7db91a5605d6da2 (Study 1); https://osf.io/sedx5?view_only=1f07463e8fab45a5836062f0bb935c39 (Study 2).

Study 1

Method

Participants.

Participants were recruited through the online platforms Prolific and Centiment in Canada (N = 176), the UK (N = 139), and the US (N = 138). Sample size was the same as for the overall project of which this study was a part. For the project, it was estimated from power analyses that a sample size of 131 participants in each country would achieve 80% power in the analyses. To be included, participants needed to be residents of each country, and not have participated in previous related studies with our labs. Because this study was part of a broader project on reasoning about COVID-19 among parents, all participants had elementary-school-aged children, and we included questions to tap into moral foundations guiding adults’ preventative reasoning concerning children. For participants recruited through the Prolific recruitment platform, the larger Qualtrics survey included five attention checks, and any participant who failed more than one was excluded from the final sample. However, Canadian participants (44%) recruited through the Centiment platform had two attention checks and were excluded when failing one of them. This difference occurred because we needed to use more than one platform to recruit a big enough sample of parents in Canada but each platform had different attention check and retention criteria. Additionally, we excluded participants who failed at least one of four additional checks in the COVID-19 Moral Foundations Questionnaire (adapted from original MFQ attention checks). In total, 44% of participants in Canada, 8% in the UK, and 15% in the US were excluded for failed attention checks. Although the criteria set by the recruitment platform in Canada resulted in a higher exclusion rate, the Canadian sample did not significantly differ from our other samples in key demographics (see Table 1). Participants leaned liberal politically on a 1–5 composite score scale (5 = high socio-economic conservatism: US M = 2.6, SD = 1.1; Canada M = 2.7 SD = 1.0; UK M = 2.6 SD = 0.8), and generally had at least a college degree (69% in the US, 80% in Canada, and 63% in the UK). All participants provided informed written consent and the survey was approved by Boston University institutional IRB (Approval number: 2350E).

Download:

Table 1. Demographics of each sample in Studies 1 and 2.

https://doi.org/10.1371/journal.pone.0285549.t001

Data collection started in late December 2020 just as the number of COVID-19 cases (approximately 4 million) and deaths (over 69,000) were at a peak globally since the onset of the pandemic in Fall 2019 [39]. At the time, it was public knowledge that several COVID vaccines had been developed, with the WHO authorizing emergency vaccine use for select adults, although child vaccination was not authorized yet [40, 41].

Finally, in addition to focusing on three related Western countries, for this study, we also collected data via Centiment in India to include a non-western country. However, once our pre-registered exclusion criteria were applied, an unexpectedly large number of participants had to be excluded, leaving only a small sample (N = 48) from which it is hard to generalize. The results from this smaller sample are therefore not presented in the paper and instead are included as exploratory data in Tables K to M of the S1 File (see supporting information section).

Materials and procedures.

The present study presented three sections of a larger Qualtrics survey in fixed order. These focused on (1) participants’ self-reported engagement in effective and ineffective COVID-19 preventative behaviors, (2) their intentions to vaccinate, and (3) their answers to a COVID-19 moral foundations questionnaire. The full survey took about 25 minutes to complete.

Engagement in effective and ineffective preventative behaviors.

Effective (helpful) and ineffective (non-helpful, overreactive) preventative behaviors were designated based on scientific information about the COVID transmission mechanism established by December 2020. Specifically, it had been confirmed by then that the coronavirus was transmitted via droplet infection, with indoor mask use and social distancing known as the best strategies to decrease viral spread [42–44]. In contrast, wearing a face mask in outdoor spaces when alone or capable of maintaining more than 6 ft. distance from others was not considered helpful.

Participants’ engagement in effective and ineffective preventative behaviors was assessed with yes/no questions about mask-wearing in seven situations. In four of the situations, mask-wearing was effective (e.g., “when grocery shopping”) while in three others it was ineffective (e.g., “when I am outdoors alone”).

Next, participants answered nine rating scale questions about social-distancing behaviors which were reverse coded. Five of the behaviors were effective (e.g., “(avoid) shaking other people’s hands”), and four of them were ineffective (e.g., “(avoid) borrowing tools/objects from my neighbor”). Participants rated the frequency of engagement with those behaviors relative to pre-pandemic times from 1 (much less frequently) to 5 (much more frequently).

Finally, we asked participants eight questions about washing behaviors, half of them were effective (e.g., “Wash my hands for 20 seconds”) and the rest were ineffective (e.g., “Wash my children’s clothes”) and they rated the frequency of engagement in them relative to pre-pandemic levels, on the same scale as the social-distancing behaviors.

Items in each section were presented in random order. From these sets of questions, two composite scores were generated for effective and ineffective preventative behaviors respectively. The composites were obtained after standardizing each item and obtaining the item mean for each behavior. Before creating the composites, four items (one ineffective and three effective) were removed because they had been difficult for participants to interpret. After exclusions, we had 10 effective preventative behavior items and 10 ineffective preventative behavior items. See Table A in S1 File for a list of all items, including exclusions.

Vaccine intentions.

Participants were asked two questions about their intentions to vaccinate, concerning themselves and their child. In both cases, they answered a forced-choice question (e.g., “Now that vaccines are becoming publicly available, are you intending to take the vaccine?”).

COVID-19 moral foundations questionnaire.

Participants were presented with four hypothetical decision-making scenarios about COVID-19 behaviors. The scenarios concerned mask-wearing, social distancing, intentions to personally vaccinate, and intentions to vaccinate their child (see a sample scenario in Table 2). Participants rated the degree to which they would weigh each of the 6 moral foundations as relevant in each of the 4 scenarios from 1 (not at all relevant) to 6 (extremely relevant). The moral foundations probed were: harm to others, harm to self (personal risk), authority, ingroup loyalty, purity, and liberty. The scenarios were presented in a fixed order, but the rating items were presented in random order, and we calculated an average score for each moral foundation across all four scenarios. Importantly, although moral foundation phrasings were based on the original MFQ [26], our measure departed from the MFQ in that participants’ rating of the personal relevance of each moral value was not abstracted and general (e.g., “people should not do things that are disgusting, even if no one is harmed”) but contextualized to each COVID-19 scenario being considered (see Table 2; specific wording of all scenarios and items is available in Table B in S1 File). Contextualizing the MFQ allowed us to identify which moral foundations have the most relevance for preventative behaviors during a novel worldwide health emergency that triggered very different governmental reactions and levels of public polarization in different countries.

Download:

Table 2. Sample COVID-19 MFQ mask-wearing scenario and its rating response items.

https://doi.org/10.1371/journal.pone.0285549.t002

Other departures from the original MFQ were the inclusion of ratings for moral concerns about liberty, because we expected them to be relevant [19], but the exclusion of fairness concerns because we could not easily adapt the MFQ’s original fairness item to each of our four COVID-19 scenarios. Also, we included ratings for concerns about harm to self so we could distinguish them from concerns about harming others within the COVID-19 context. The distinction is important to investigate given that previous research has suggested that framing health campaigns in terms of harm to the self versus harm to others can lead to differential effectiveness [45, 46].

Results

Cronbach’s alpha confirmed the reliability of the scores for effective and ineffective preventative behaviors (ranging from .73 to .89) and moral foundation scores (ranging from .78 to .92), which showed acceptable to excellent levels of internal consistency in each country. An exception to this was for the purity foundation composite. However, excluding one item resulted in acceptable reliability levels in each country (ranging from .67 to .74), so we retained that composite given our preregistration.

Moral foundations across situations.

Average ratings per country for each moral foundation across the four scenarios are plotted in Fig 1. Repeated measures ANOVAs were conducted to assess within-subjects differences in moral foundation scores in each country. Given that Mauchly’s test indicated a violation of sphericity (ps < 0.001), degrees of freedom were corrected using Greenhouse-Geiser estimates (US: ε = .54; Canada: ε = .64; UK: ε = .62). We found significant differences in the relative endorsement of different foundations within each country (US: F(2.69,368.95) = 179.64, p < .001, η_p² = 0.57; Canada: F(3.20,560.64) = 257.03, p < .001, η_p² = 0.60; UK: F(3.12,430.83) = 231.65, p < .001, η_p² = 0.63). Posthoc tests with Bonferroni adjustment revealed that in all countries, harm to others and harm to self were rated as more important than the other foundations (ps < .001) with the US alone weighing harm to self as more important than harm to others (p < .001). In all three countries, the authority foundation then surpassed all other foundations (ps < .001), (except for liberty in the US, p >.05), with no differences found between ingroup and liberty (ps >.05). In Canada and the UK, the purity foundation was rated lower than all other foundations (ps < .05), while in the US, it was equivalent to ratings of the ingroup foundation (p >.05). Overall, these results suggest that there was a similar pattern of importance attributed to each foundation in the context of COVID-19 situations in each of the three countries. Only the foundation of authority received different importance ratings across the three countries, with the US attributing less importance to it on average.

Download:

Fig 1. Mean moral foundation scores across the four situations in Study 1.

Note. The Moral Foundation score ranged from 1 (not relevant) to 6 (extremely relevant).

https://doi.org/10.1371/journal.pone.0285549.g001

Spearman’s correlations were conducted to check whether our data showed the kinds of associations between moral foundations and political orientation found in prior research (e.g., Graham et al. [47]) No significant correlations were found in the UK. However, in the US and Canada greater conservatism was related to greater liberty (US: r(136) = .39, p < .001; Canada: r(174) = .18, p = .02), and purity considerations (US: r(136) = .27, p = .001; Canada: r(174) = .21, p = .006), but with reduced concern about harm to the self (US: r(136) = -.42, p < .001; Canada: r(174) = -.29, p < .001), and harm to others (US: r(136) = -.44, p < .001; Canada: r(174) = -.28, p < .001). These results suggest that, at least in the US and Canada, political ideology was associated with the weight given to different moral foundations, following patterns found in previous research [17].

Moral foundations and COVID-19 effective and ineffective behaviors.

Consistent with our pre-registration, we conducted separate regressions to explore whether moral foundation scores predicted participants’ engagement in effective and ineffective preventative behaviors in each country. In the presence of outliers, we conducted Linear robust regression [48], and if normality assumptions were violated, we conducted Quantile regression (on the median). We computed the Variance Inflation Factor (VIF) for all the models and found that multicollinearity was not a concern (VIFs were below 3.8). Additionally, as preregistered, countries were not statistically compared given that we could not assume measurement invariance across our intended samples and sample size.

Raw scores indicated that adherence to effective preventative behaviors was relatively high in each country (see Table I in S1 File) and none of the moral foundations significantly predicted engagement in them in either Canada or the UK. However, in the US effective behaviors were positively predicted by greater considerations of harm to others and attention to authority but negatively predicted by concerns about personal liberty (see Table 3).

Download:

Table 3. Regression of effective and ineffective preventative behaviors on moral foundations in Studies 1 and 2.

https://doi.org/10.1371/journal.pone.0285549.t003

Regarding ineffective behaviors that have not been an emphasis of prior work, analyses revealed that they were positively predicted by concerns about authority in all countries—a pattern which contrasts that found for effective behaviors. In addition, in Canada alone, ingroup loyalty predicted lower engagement in ineffective behaviors.

Finally, we explored whether political ideology outweighed any of the moral foundations as predictors when added to the model in each country. However, we found that it was not a significant predictor and made no difference to the pattern of results.

Moral foundations and intentions to vaccinate against COVID-19.

Participants in all three countries tended to express intentions to vaccinate themselves (US: 68%; Canada: 80%; UK: 74%) and their child (US: 62%; Canada: 69%; UK: 63%). To examine the effect of moral foundation scores on intentions to personally vaccinate–an imminent decision–and to vaccinate their child—a hypothetical decision during the study—we conducted two separate logistic regression analyses in each country (see Table 4).

Download:

Table 4. Regression of intentions to vaccinate the self and the child on moral foundations in Studies 1 and 2.

https://doi.org/10.1371/journal.pone.0285549.t004

Intentions to vaccinate the self and a child were positively predicted by different foundations in each country: attention to authority in the UK, ingroup loyalty in Canada and, for vaccination of a child only, concerns about harm to others in the US. However, purity concerns negatively predicted intentions to vaccinate the self and a child with remarkable consistency in all countries. In Canada, intentions to vaccinate the self were also negatively predicted by concerns about liberty. Again, exploratory analyses found political orientation was not a significant predictor when it was added into models and did not change the pattern of findings.

Discussion

Consistent with expectations, moral foundations played a significant role in tendencies to engage not only in effective behaviors and intentions to vaccinate but also in ineffective preventative behaviors at the peak of the COVID-19 pandemic. Indeed, the influence of moral foundations outweighed that of political orientation despite prior findings (e.g., Gollwitzer et al. [34]; Pennycook et al. [13]) that the latter have significant effects (see S1 File showing that relationships between political ideology and preventative behavior were often mediated by moral foundations). Patterns in our results revealed cross-country similarities and differences in relationships between moral foundations and preventative behaviors.

Cross-country similarities.

Participants’ ratings of the importance of each of the foundations were similar in all three countries, with harm (to others or to self) most explicitly weighed when considering pandemic-related situations. It is clear then that people viewed themselves as trying their best to care for themselves and others amidst the pandemic. However, despite being explicitly rated as important, regression analyses revealed that these considerations were not consistently predictive of people’s compliance with effective preventative behaviors. Specifically, while concern about harming others was a significant predictor in the US, it was not predictive in the UK or Canada.

Importantly, two moral foundations consistently predicted health behaviors across all countries. One of these was the authority foundation which—while predicting helpful responses in the UK (intention to vaccinate) and the US (effective behaviors such as social distancing)—was most consistently predictive of ineffective behavioral responses. That is, across all countries, there was evidence that attention to public health mandates issued by authorities had the unintended effect of promoting indiscriminate and potentially fearful overreactions, presumably because of participants’ lack of underlying causal knowledge about disease transmission. The fact that strong adherence to mandates increased overreactions, such as complete social isolation, and possibly fear, is worthy of special note given that greater levels of the former are linked to diminished self-regulation skills [27] and greater levels of the latter are associated with heightened depression and anxiety symptoms [49, 50]. Furthermore, the negative effects of overreactions have potential public health system-level effects too insofar as they have the potential to further tax overloaded health care systems by increasing Emergency Department visits [51]. When considered alongside findings that individuals have lower propensities to engage in ineffective preventative behaviors when they have greater biological knowledge [14], these results counsel attention to the unplanned consequences of prescriptive public health messages that are issued—without explanation—in contexts without concerted public education about biology or disease mechanisms.

The second consistent predictor was the purity foundation which, contrary to hypothesis, did not increase ineffective preventative behavior but did have reliably negative predictive effects on intentions to vaccinate the self and children across all countries. This might stem from intuitive perceptions of the vaccine as unnatural and fears that it might taint the body even more than the virus [52]. In the US, these intuitive beliefs were potentially enhanced by presidentially sanctioned scientific misinformation that promoted mistrust in vaccinations—cultural dynamics that were less marked in the UK and Canada [13].

Culture-specific patterns.

Despite these similarities across all three western countries, culture-specific patterns were also observed. In the UK, the authority foundation had broader predictive effects than in other countries, revealing a particularly marked sense of obligation towards institutions and rules in this country [53]. In Canada, loyalty to the ingroup played a distinctive role, not only predicting increased intentions to vaccinate but reducing engagement in ineffective preventative measures, perhaps because the value placed on the group-binding function of engaging with friends and family acted as a counterweight to overreaching behavioral tendencies that might socially isolate. Finally, there was the unique pattern in the US: Consistent with previous research [10, 23, 24, 32, 54], concerns about harm to others played a distinctive role, predicting increased vaccination intentions and effective preventative behaviors like mask wearing in public spaces. Meanwhile, liberty concerns countervailed—amidst relatively low concerns about authority—to predict reduced engagement in effective behaviors, also predicting reduced vaccination intentions in our other North American country, Canada. As we later discuss, these region-specific patterns may not only reflect subtle differences between the three Western countries in the foundations entrenched in traditions and cultural narratives, but also specific features of the health campaigns and governmental responses at the height of the pandemic. In turn, they therefore also raise a question: how stable are the abstract guiding foundations that predict behavior within a culture when the situational conditions change?

In Study 2, we therefore explored this question at a later time point––the post-peak pandemic period of early summer 2021—in one of the countries. Specifically, we focused on the US during this period because of its rapidly changing context: COVID-19 cases and deaths were dropping [55], mandates in certain states were relaxing, and an increasing number of people returned to work/school in person after the rollout of mass vaccination programs targeting adults and adolescents (which was faster than in other countries). Thus, the US offered a substantial test of whether the predictive effects of moral values are generalizable across contexts. Based on the theorized stability of the moral foundations [15], we hypothesized that Study 2 would find similar patterns as Study 1, except for reduced influences of the authority foundation given reductions in mandates.