Adolescents' social and moral reasoning about COVID-19 public health behaviors

Examining the forms of social and moral reasoning adolescents use is important for understanding youth engagement with public health guidelines. The present work examined adolescents' perceptions of social norms and associated reasoning in the COVID-19 context. Participants (n = 127, M age = 17.00, SD = 0.71) negatively evaluated other teenagers who broke COVID-19 guidelines and reasoned about harm reduction to justify breaking these rules, but also recognised the importance of protecting mental health as one reason to spend time with friends counter to these rules. Further, adolescents reported that they were more likely to engage in public health behaviors compared to their peers or the average teenager, suggesting a social norm of lower engagement with these guidelines. Together, this evidence documents the importance of considering social norms and moral reasoning in framing communication efforts that target adolescents' adherence to public health guidelines.

Examining the forms of social and moral reasoning adolescents use is important for understanding youth engagement with public health guidelines. The present work examined adolescents' perceptions of social norms and associated reasoning in the COVID-19 context. Participants (n = 127, M age = 17.00, SD = 0.71) negatively evaluated other teenagers who broke COVID-19 guidelines and reasoned about harm reduction to justify breaking these rules, but also recognised the importance of protecting mental health as one reason to spend time with friends counter to these rules. Further, adolescents reported that they were more likely to engage in public health behaviors compared to their peers or the average teenager, suggesting a social norm of lower engagement with these guidelines. Together, this evidence documents the importance of considering social norms and moral reasoning in framing communication efforts that target adolescents' adherence to public health guidelines.
Although there are hopeful signs that the deadliest waves of the COVID-19 pandemic are behind us (McMahan et al., 2022) there are still likely to be further variants of the virus emerging, and future pandemics remain a concern (Carlson et al., 2021). With this in mind, understanding how individuals of all ages make decisions about measures to protect against such viruses will play a vital role in reducing the spread of future COVID variants and other zoonotic viruses. Despite debates about the odds of youth transmitting the COVID virus (Danis et al., 2020;Gaythorpe et al., 2021;Zhang et al., 2020), it is important to understand the ways in which adolescents think and reason about being asked to participate in public health practices that are designed to protect older, more vulnerable adults, when they themselves are at minimal risk. Researchers and theorists have highlighted the increased importance of personal autonomy in adolescence (Smetana, 2010) which may contribute to narratives that adolescents are likely to selfishly breach guidelines around public health practices related to COVID-19 (Andrews, Foulkes, & Blakemore, 2020). While theorists have examined the contribution of social norms to adolescent health and risk behavior, here we offer an extension to this work by exploring the use of moral reasoning amongst adolescents. We aimed to understand whether late adolescents see breaches of COVID-19 guidelines as a matter of personal choice or an issue dependent on moral principles and conventional health guidelines.

Social reasoning & COVID-19
A rich body of work has examined adolescents' decision-making in risky behavioral contexts. For example, researchers have utilised the Theory of Planned Behavior (TPB; Ajzen, 1991) to argue that adolescents' public health and risk decisions are informed by their attitudes, intentions, perceived behavioral control, and importantly, social norms. Perceptions of how others intend to behave play an important role in determining adolescents' own behaviors in these risky contexts. This model has been used to predict a wide range of adolescent health behaviors, from smoking to sexting (Harakeh, Scholte, Vermulst, de Vries, & Engels, 2004;Walrave, Heirman, & Hallam, 2014). Crucially, theorists have argued that the TPB has not always been successful in persuading adolescents to adopt healthy behaviors due in part to adolescent egocentrism (Lin, 2016). The argument that adolescents' selffocus may undermine the success of TPB interventions is reminiscent of popular narratives during COVID-19 that youth were most likely to breach guidelines around social distancing (Andrews et al., 2020). In the present paper, we draw on an alternative perspective to argue that considering adolescents' moral reasoning can offer a valuable extension to our understanding of adolescents' thinking and reasoning around public health guidelines.
Work with adolescents has documented the importance of social norms in relation to not only public health behaviors but also social and moral decision making (McGuire, Elenbaas, Killen, & Rutland, 2019). Here we argue that it is not only important to understand the extent to which youth are guided by perceived social norms (as in TPB), but also the ways in which they see public health behaviors as a moral issue, offering an alternative view to perceptions of adolescents as egocentric. Social Domain Theory (SDT; Turiel, 2015) is an important theoretical framework that has been used to understand the ways in which individuals reason about social decisions. So far, the SDT framework has not been applied to public health decision-making, but there are important reasons to expect this framework to play a role in how adolescents think about COVID-19 public health guidelines.
The SDT framework argues that decision making falls broadly into three domains. First, the moral domain (e.g., "does this decision cause harm to others?"), the social-conventional domain (e.g., "what is considered societally appropriate in this situation?"), and the personal domain (e.g., "do I have autonomy in this context?"). The personal domain grows in importance in adolescence, when decisions about personal relationships and how to spend one's time begin to move away from parental dictate towards adolescents' own control (Smetana, 2010). This increased focus on personal autonomy shares some features with ideas derived from the TPB, where researchers have argued that adolescence is a period of increased egocentrism or self-focus (Lin, 2016). This in turn may have consequences for viral transmission. For example, if public health behaviors are framed as an issue of personal autonomy (e.g., "it's my choice whether or not to see my friends"), this may impact decisions to adhere to guidelines. However, SDT also recognises the importance of moral and social-conventional thinking. In the COVID-19 context, moral (e.g., "not seeing my friends may help reduce harm to others who are at risk") or social-conventional (e.g., "other members of our society are not seeing their friends so I ought to do the same") reasoning may be more likely to lead to adherence to public health practices. Researchers have demonstrated that adolescents are more competent than children in balancing demands from the three different domains and use contextual information to decide when it is appropriate to favour oneself or when it is appropriate to prioritise moral concerns (McGuire, Rizzo, Killen, & Rutland, 2019). Although social conventions and concerns of personal autonomy are important to adolescents, moral concerns can be prioritised in certain contexts (Hitti, Mulvey, Rutland, Abrams, & Killen, 2014).
The first key aim of the present work was to assess the kinds of reasoning that adolescents use to justify their evaluations of two hypothetical vignettes where they were asked to evaluate adolescents who breached COVID-19 public health guidelines. Specifically, we aimed to understand whether adolescents would frame these issues using personal domain concerns (in line with expectations for adolescent egocentrism), or whether they would recognise the moral and socialconventional aspects of the situation (in line with Social Domain Theory). In the first vignette, participants were asked to evaluate a group of teenagers not following social distancing guidelines and in the second, another teenager not wearing a mask when asked to in a public place. In both cases, we asked participants to evaluate and provide an open-ended reasoning response. These two situations were chosen as they differ in their degree of personal cost. Whereas wearing a mask involves little personal cost, social distancing does involve personal cost for adolescents (i.e., not spending time with friends) and is not always easy to achieve (e.g., in crowded public spaces). Thus, we expected to see different patterns of reasoning across contexts. In the mask context, we expected participants to reason in the moral and conventional domains with reference to harm and breaches of rules. In contrast, in the social distancing context, we expected to still see references to moral and conventional concerns, but for these to be balanced with reference to personal concerns, including the right of the individual to spend time with friends.
Researchers examining adolescents' individual and perceived group attitudes in moral decision making have also documented that adolescents perceive their peers will behave differently to themselves in contexts like resource allocation (Rutland & Killen, 2017) and peer evaluation (Abrams, Palmer, Rutland, Cameron, & Van de Vyver, 2014). In general, adolescents perceive that they would engage in more inclusive or equitable behavior than their peers. In the present work we expected to observe similar differences between individual and group evaluations of public health guidance breaches, as well as accompanying reasoning. Specifically, we expected that when asked to evaluate a situation where a hypothetical adolescent breached COVID-19 guidelines, adolescents would report that they individually would evaluate breaking such rules more negatively than their peers would.

Social norms & public health practices
Recent evidence in the COVID-19 context has shown that subjective social norms (as in TPB) are strong predictors of social and physical distancing (Gibson, Magnan, Kramer, & Bryan, 2021;Yu, Lau, & Lau, 2021), adherence to COVID-19 prevention guidelines (Shanka & Gebremariam Kotecho, 2021) and vaccine uptake (Shmueli, 2021) in adults. This evidence emphasises the importance of subjective social norms in decision making about health behaviors. What is important to note is that these norms are subject to biases in perception.
Crucially, individuals often misperceive how others will think or behave, which in turn is related to a shift towards what we perceive others to believe and the creation of new norms founded on inaccurate beliefs (Prentice & Miller, 1993). Adolescents have also been shown to report that their peers would behave differently than they themselves would in various decision-making contexts (McGuire, Rizzo, et al., 2019). So far, less is known about perceptions of social norms in the COVID-19 context, especially amongst adolescents. If adolescents' public health behavior decisions are impacted by norms, and these norms are being misperceived, then we are likely to see behavior reflecting the perceived social norm, which may in turn lead to instances like superspreader events, even if youth are considering the moral consequences of such behavior. Therefore, the second key aim of the present work was to assess adolescents' perceptions of norms regarding COVID-19 public health behaviors (e.g., mask-wearing, social distancing).
In the context of COVID-19, adults in the U.S. reported that they individually believed adherence to guidelines to be more important than the average U.S. adult believed it to be (Lees, Cetron, Vollberg, Reggev, & Cikara, 2020). If what we think the 'average' person does informs our own behavior and creates a behavioral norm, then it is important to document whether youth also report differences between their own behavioral intentions and their perceptions of what their peers will do. Given previous work documenting such biases in late adolescence (Hines, Saris, & Throckmorton-Belzer, 2002) and in relation to COVID-19 (Lees et al., 2020) in the present work we expected that adolescents would report that their peers and the average teenager in the U.K. would be less likely to engage in COVID-19 preventive behaviors than they themselves would be. Together, evidence from reasoning tasks and social norm tasks should provide a new perspective regarding whether adolescents behave in line with concerns for personal autonomy and adolescent egocentrism (Smetana, 2010) or whether they coordinate these concerns with moral principles (e.g., harm aversion) and the importance of social conventions.

The present study
In the current study we examine adolescents' thinking about COVID-19 public health behaviors by asking them to evaluate and reason about breaches of these behaviors. Specifically, we explore whether adolescents focus on egocentric personal domain reasoning, or instead highlight the social-conventional and moral elements of COVID-19 public health decision making. Further, we examine the role of social norms in relation to COVID-19 health behaviors. Together, these tasks seek to provide evidence regarding the domains of thought that are pertinent to adolescents' decision-making, as well as perceptions of norms related to COVID-19. To do so, we presented adolescents with hypothetical vignettes where fellow adolescents break guidelines of mask-wearing and social distancing. Participants were asked to evaluate these behaviors and provide open-ended reasoning to justify their evaluation. Further, we asked participants to report their own engagement with COVID-19 preventive behaviors (e.g., social distancing, hand washing), as well as their perceptions of the engagement in these behaviors of their peers and the average teenager in the UK.

H1.
Participants were expected to report that their peers would evaluate breaches of public health guidelines portrayed in hypothetical vignettes (a. mask-wearing and b. social distancing guidelines) more positively than they themselves would.
H2. When evaluating a hypothetical scenario about a teenager who refuses to wear a mask, participants were expected to reason about harm reduction (moral concern) and adherence to guidelines (social-conventional concern) to justify their negative evaluation of the rule-break.
H3. When evaluating a hypothetical scenario about teenagers who were socialising despite social distancing guidelines, participants who negatively evaluated the rule-break were expected to reason about harm reduction (moral concern) and adherence to guidelines (social-conventional concern) to justify their negative evaluation of the rule-break. However, participants who less negatively evaluated the rule-break were expected to reference personal autonomy (personal concern) to justify their evaluation of the rule-break.
H4. Participants were expected to report that they engage in COVID-19 preventive behaviors at a higher rate than their peers and the average teenager in the U.K.

Participants
Participants (n = 127, minimum age = 16, maximum age = 19, M age = 17.00, SD = 0.71) were recruited from schools in the southwest of the United Kingdom. Participants included 48 male, 56 female, 2 gender non-binary, 1 other gender identity and 20 participants who did not report their gender. The sample included 100 White British, 3 Mixed Race/Dual-Heritage, 4 Other ethnicity and 20 participants who did not report their ethnicity. Participants attended schools serving low to middle socio-economic status populations (SES was estimated based on geographic area of schools). Power analyses based on our repeated measures analysis (described below) conducted in G*Power indicated that to detect a small effect size (f = 0.20) with an alpha of 0.05, power of 0.95, 1 group and 5 measurements (with estimated correlation between repeated measures of 0.5, and sphericity assumed), a minimum sample of 48 participants was required. Using opportunity sampling with the schools we worked with, we recruited above this number to account for possible drop out, and to enhance our mixed-methods analysis of the qualitative responses to the open-ended reasoning questions discussed below.

Procedure
All measures were approved by the [BLINDED] Ethics Committee and were completed using Qualtrics (Provo, UT) online survey software. The survey took approximately 15 min to complete. Participants completed the survey during school hours. All participants gave their own informed consent.

Data collection context
Data was collected in England during March and April 2021. During this period, the U.K. was emerging from a full national lockdown with significant restrictions remaining in place on households in England, which required adults to stay at home other than to attend work (if essential), for food shopping, and essential appointments (e.g., hospital appointments). It was permitted to meet one person from another household outdoors, and to engage in outdoor physical recreation. English schools had returned to in-person teaching at the beginning of March 2021. Half of the U.K. adult (over 18 years-olds) population had received a first dose of a COVID-19 vaccine. Participants were recruited from the upper year groups of schools in the UK, which are comparable to high school settings in the US, with older adolescents studying for more advanced qualifications. Teachers and authority figures are present in such settings and may act to enforce rules and guidelines such as those in place during the covid pandemic.

Public health breach vignette onemask wearing
Participants were asked to read two vignettes that included a breach of a government public health mandate. In the first, participants read: "Imagine that you are out walking around your local town and you see someone your age and gender inside a shop. This teenager is not wearing a mask or face covering. The shop has signs up asking customers to wear masks or face coverings. You hear the shop assistant asking whether this teenager has a health reason/exemption to not wear a mask. The teenager says no, they do not have an exemption from wearing a mask." This was followed by a binary evaluation question ("Is it okay or not okay for this teenager not to wear a mask?"; okay, not okay), a scale evaluation question ("How okay or not okay is it for this teenager not to wear a mask?"; 1 = really not okay, 6 = really okay) and an open-ended individual reasoning question ("Why is it okay or not okay for this teenager not to wear a mask?"). Further, we asked a perceived peer scale evaluation question ("How okay or not okay would other teenagers think it is for this teenager not to wear a mask?"; 1 = really not okay, 6 = really okay) and an open-ended peer reasoning question ("Why would other teenagers think it is okay or not okay for this teenager to not wear a mask?")

Public health breach vignette twoindoor gathering
In the second vignette, participants read: "Imagine that you are out walking around your local town and stop to have a drink. Imagine that we are in a tier where cafes are allowed to be open. You see a group of teenagers (who are the same age and gender as you) meet up with one another in a cafe. This group of teenagers are not from the same household or social bubble. They are sat close to each other in the cafe (less than two metres apart)." This was followed by binary evaluation, scale evaluation, openended individual reasoning, perceived peer scale evaluation and openended peer reasoning questions as per the first vignette.

Public health behaviorself & other
Participants were asked to respond to questions assessing engagement with five COVID-19 public health behaviors (adapted from Lees et al., 2020). These included hand washing, mask-wearing, social distancing when outdoors, avoiding outdoor gatherings, and avoiding indoor gatherings. In each case, the participants read: "In order to reduce the spread of COVID-19, there are different things people can do." This was followed by "how much do you agree or disagree with the sentences below", "how much do you think your friends would agree or disagree…" and "how much would an average teenager of your age and gender would agree or disagree…" The statements followed, for example: "I wear a face covering in public places where I'm supposed to (e.g. supermarket)" (1 = strongly disagree, 7 = strongly agree).

Data analysis plan
For our reasoning responses, we developed a coding framework that drew from Social Domain Theory (Turiel, 2015) and a grounded reading of the data (see Table 1 for the full framework). The first five codes aligned with the three domains of SDT. The sixth code ("Inconsistent Rules") was in reference to British government guidelines and derived from this grounded reading of the data. Two coders conducted the coding. Coders could select as many categories as were present in the participants' response (0 = no use of category, 0.5 = one of two categories used, 1 = category used alone). Inter-rater reliability procedures based on the final framework indicated total agreement between the coders across 16% (n = 21) of responses for both the individual response reasoning (Cohen's K = 1.00) and perceived group reasoning (Cohen's K = 1.00). Once the coding was complete, we noted that a minority of participants (individual distancing vignette reasoning n = 11, group distancing vignette reasoning n = 13, individual mask vignette reasoning n = 6, group mask vignette reasoning n = 6) used more than one category. For the analyses reported below we focused instead on participants who used one category, omitting those participants who used multiple categories from our analyses. To analyse differences in reasoning we conducted chi-square tests to assess whether there were differences in participants' use of reasoning codes as a function of their binary evaluation of the public health breach (i.e., Okay or not Okay). Where a given category included fewer than 5 responses, Fisher's Exact test was used instead of the Chi Square value.
We examined within-participants differences in perceptions of intentions to engage with the five public health behaviors at the self, peer, and average teenager levels using a two-way repeated measures ANOVA with the five behaviors (mask wearing, hand washing, social distancing, outdoor gatherings, indoor gatherings) as the first within-subjects factor and level (self, peer, average teenager) as a second within-subjects factor. Where appropriate, we used pairwise comparisons with Bonferonni corrections applied to adjust for multiple comparisons.

Individual reasoning
It was not possible to conduct our planned chi-square analysis based on binary evaluation as all participants who provided a reasoning response reported that they thought refusing to wear a mask was not okay. See Table 2 for category use. The most prominent categories included harm (e.g., "masks protect other people from the potential spread of pathogens so it's not okay as they are putting others at risk, not just themselves") and rules (e.g., "they are not following the rules set by the government").

Peer reasoning
For perceived peer reasoning regarding the breach of mask wearing guidelines, the non-significant chi-square analysis suggest there was no difference in reasoning responses provided by those who perceived their group would say this act was okay or not okay, Fisher's Exact Test(2) = 1.85, p = .54. See Table 2 for category use. Again, the prominent categories included harm (e.g., other teenagers would evaluate as 'not okay' because "they have family/relatives who are shielding") and rules (e.g.,

"they would agree that there is no excuse for not wearing a mask unless they have a valid reason which this teenager doesn't, because we all have to do it").
In this case, there were several responses that did not meaningfully fit into our coding framework, including references to the importance of masks for returning to normality (e.g., "want things to go back to normal so everyone should wear masks where they are told to") and pragmatic concerns (e.g., "because masks can be uncomfortable and not always work").

Individual reasoning
Reasoning analyses revealed significant differences in the use of reasoning categories as a function of participants' binary evaluation (okay, not okay) of the indoor gathering; Fisher's Exact Test(3) = 14.40, p = .002 (see Table 2). The harm category was referenced more by participants who reported the gathering was not okay (33%, 'a bit not okay' because "they could bring [COVID] back to their household and infect other people, if this spreads it'll up the R as each person is infecting more people") compared to those who said it was okay (7%). In contrast, participants who said the gathering was okay were more likely to reference mental states (26%, e.g. 'a bit okay' because "being a teenager is especially lonely and isolating during the pandemic and lockdown") than participants who said the gathering was not okay (2%).

Peer reasoning
Perceived peer reasoning analyses revealed a significant difference in category use as a function of perceived group evaluation, Fisher's Exact Test(3) = 21.21, p < .001 (see Table 2). Participants who reported their group would believe the gathering to be not okay were more likely to reference rules (45%, e.g. 'a bit not okay' because "they may believe we should follow the rules more rigidly") than those who perceived their peers would believe the gathering to be okay (16%). In contrast, participants who believed their peers would report the gathering to be okay referenced mental states (38%, e.g., 'okay' because "they want to socialise and it helps with reducing stress") more than those who believed their peers would perceive the gathering to be not okay (0%).

Self behaviors
Pairwise comparisons suggested that participants were more likely to report that they would wear a face mask (M = 6.75, SD = 0.69) compared to the four other behaviors (Distancing M = 3.73, SD = 1.75; Hands M = 5.24, SD = 1.26; Outdoor Gatherings M = 4.50, SD = 1.62; Indoor Gatherings M = 5.66, SD = 1.46; all p's < 0.001). Crucially, participants also reported they were less likely to physically distance from others compared to the four other behaviors (all p's < 0.001).

Peer behaviors
As per individual intentions, participants perceived their peers were more likely to wear a face mask (M = 6.37, SD = 0.93) compared to the other four behaviors (Distancing M = 3.36, SD = 1.61; Hands M = 4.47, SD = 1.51; Outdoor Gatherings M = 4.14, SD = 1.54; Indoor Activities M = 4.93, SD = 1.56; all p's < 0.001). Again, participants reported they perceived their peers were less likely to physically distance from others as compared to the other four behaviors (all p's < 0.001).

Average teen behaviors
Again, participants perceived the average U.K. teenager was more likely to intend to wear a mask (M = 5.84, SD = 1.01) compared to the other four behaviors (Distancing M = 2.52, SD = 1.36; Hands M = 3.85, SD = 1.49; Outdoor Gatherings M = 3.32, SD = 1.56; Indoor Activities M = 4.10, SD = 1.53, p's < 0.001). Again, participants reported they expected the average teenager to be less likely to intend to socially distance compared to the other four behaviors (all p's < 0.001).

Within behavior differences
Finally, we turn to examining differences between self-intentions, perceived peer intentions, and perceived average teenager intentions within each behavior.

Hand washing
Participants reported that they (M = 5.29, SD = 1.27) were more likely to engage in hand washing than their peers (M = 4.51, SD = 1.52), and in turn their friends were more likely to engage in this behavior than the average teenager (M = 3.89, SD = 1.46; all p's < 0.001).

Social distancing
Participants reported that they (M = 3.75, SD = 1.76) were more likely to engage in social distancing than their peers (M = 3.33, SD = 1.60, p = .02), and in turn their friends were more likely to engage in this behavior than the average teenager (M = 2.56, SD = 1.32; p < .001).

Mask wearing
Participants reported that they (M = 6.76, SD = 0.70) were more likely to wear a mask than their peers (M = 6.35, SD = 0.94, p = .002), and in turn their friends were more likely to wear a mask than the average teenager (M = 5.80, SD = 1.05; p < .001).

Outdoor gatherings
Participants reported that they (M = 4.48, SD = 1.63) were just as likely to avoid gatherings as their peers (M = 4.13, SD = 1.53; p = .17), but that both they and their friends (p's < 0.001) were more likely to avoid gatherings than the average teenager (M = 3.38, SD = 1.55).

Indoor activities
Finally, participants reported that they (M = 5.66, SD = 1.49) were more likely to avoid indoor gatherings than their peers (M = 4.90, SD = 1.55), and in turn their friends were more likely to avoid indoor gatherings than the average teenager (M = 4.17, SD = 1.50; all p's < 0.001).

Discussion
In the context of COVID-19, concerns have been raised regarding adolescents' likelihood of breaching guidelines (Andrews et al., 2020) and research from the Theory of Planned Behavior has highlighted adolescent egocentrism as an important correlate of decision-making (Lin, 2016). The current findings offer a counter-narrative by asking participants to evaluate breaches of mask-wearing and social distancing guidelines. In both cases, and in support of our hypotheses, participants expected their peers to evaluate a breach of social distancing or maskwearing guidelines less negatively than they themselves would. However, we also saw in both cases that these perceived peer evaluations of the breach were either explicitly negative (in the case of not wearing a mask without an exemption), or at least not positive (in the case of social distancing). Crucially, social reasoning findings point to the morally motivated reasoning processes that adolescents use in these instances. Specifically, participants referenced both the moral importance of harm reduction and the conventional importance of government guidelines, as well as seeing the connection between these ideas (i.e., that conventions are important to achieve harm reduction).
These findings are aligned with the Social Domain Theory framework (Turiel, 2015) and work that has asked adolescents to make judgments in other contexts where moral, conventional, and personal domain concerns are in conflict. For example, researchers have documented that when asked to allocate resources, adolescents will prioritise moral concerns for equity over concerns to privilege one's in-group (Hitti et al., 2014). Similarly, in the COVID-19 context outlined in the current study, adolescents negatively evaluated actions that would wilfully cause harm, such as refusing to wear a mask. Such judgments and associated reasoning suggest that here adolescents use advanced coordination abilities to prioritise moral domain concerns for welfare over personal domain concerns for autonomy of the individual. This was particularly evident in the mask vignette, where the unambiguous breach of the mask-wearing convention (without any medical exemption on the part of the character in the vignette) was related to harm, without any justification that would have led to the prioritisation of social-conventions or personal concerns. Here then, adolescents coordinated domain concerns and focused on either the moral domain, or a cross-domain response that understood the moral benefits of social conventions. This study adds to the Social Domain Theory literature by evidencing, for the first time, that domain knowledge and coordination is applied in the context of public health. Existing literature has evidenced the value of domain theory in understanding reasoning when adolescents are asked to make decisions about morally-relevant tasks like resource allocation and social exclusion (e.g., McGuire, Rizzo, et al., 2019). Here we show that these domains are also highly relevant when understanding novel emerging contexts like COVID-19 and ought to be considered in attempts to communicate the benefits of adhering to public health guidelines to adolescents.
In the social distancing vignette, judgments and reasoning were more ambivalent. Evaluations of the social distancing breach were less negative than in the mask vignette. However, participants' reasoning in this vignette offered some important qualifications to further counteract the traditional stereotype of the egocentric teenager. Specifically, participants stated that they understood there were mental state reasons (e. g., protecting one's mental health through spending time with friends) that might conflict with the potential harm of breaching social distancing guidelines. Further, adolescents referenced inconsistent guidelines as a justification for why other teenagers in the vignette may not have been socially distanced. Participants referenced the fact that at school they were unable to distance from their peers in classroom settings, and therefore it would be more acceptable to be in close contact with these same friends outside of school.
Together this reasoning evidence brings to the fore the complex coordination of domains that adolescents are engaging in when considering COVID-19 public health guidelines. Concerns for harm reduction through adherence to guidelines are dominant in the minds of adolescents when situations are unambiguous. However, where breaches of social distancing were evaluated less negatively, it was mental state concerns and adult-led rule inconsistencies that lay behind these evaluations. The Theory of Planned Behavior has been a valuable framework for predicting adolescents' engagement with different public health behaviors (Harakeh et al., 2004;Walrave et al., 2014), although its success in altering behavior has been more mixed, which some theorists have argued is in part due to adolescents' egocentrism (Lin, 2016). We believe our findings offer an extension to this perspective and highlight the importance of not only social norms but also moral concern in public health settings. Adolescents in our study framed the COVID-19 context as one that involves moral concerns, in particular the importance of reducing harm towards others. Further, there was very little personal autonomy or egocentric reasoning evident. Other forms of public health contexts involve harm towards others outside of the self. For example, smoking is not only harmful to the self, but also to others through second-hand smoke. One possibility is that communication efforts targeting adolescent public health ought to consider using morally-focused messages where behaviors do have secondary health consequences for other parties.
With regards to our social norm measure, we documented that teenagers believed their peers were less likely than themselves to wash their hands more frequently, to wear a mask, to socially distance and to avoid indoor gatherings. Teenagers also reported the average U.K. teenager was less likely than themselves to do these things and to avoid outdoor gatherings. When comparing across behaviors, participants reported that wearing a face mask was most likely, while participants reported the lowest intentions to socially distance. These findings provide insights about the formation and perception of novel social norms in a public health context. The ability to directly observe the behavior of others seems central to accurate perceptions of these norms. With maskwearing guidelines mandating wearing a face covering in indoor public spaces (including schools and universities), participants accurately reported that in general most people are likely to engage with this behavior. When behaviors are less directly observable (e.g., hand washing) or happening beyond the school gates (e.g., avoiding indoor gatherings of friends), the distance widened between individual and perceived peer perceptions.
These findings are troubling considering the adult literature on perceived social norms (e.g., Prentice & Miller, 1993) which indicates that individuals will shift their behavior towards such norms, along with more recent literature documenting the importance of social norms in the COVID-19 context (Gibson et al., 2021;Shanka & Gebremariam Kotecho, 2021;Shmueli, 2021;Yu et al., 2021). In future pandemics, this difference between observable and unobservable behaviors will be important to consider. Rectifying misperceptions of social norms ought to be targeted towards those behaviors where norms are less visible, for example, by using norm feedback interventions to educate youth about accurate levels of engagement with behaviors like avoiding indoor gatherings with friends.

Limitations & future directions
While these findings make a novel contribution to both understanding of COVID-19 public health guideline adherence in adolescence, and the relation between social norms and reasoning more generally, there are limitations that ought to be considered. First, this data was collected at one time point in one country. The COVID-19 pandemic has been a rapidly evolving global phenomenon, and as such we ought to show caution in extrapolating the views of adolescents in even one area of the U.K. too far beyond this geographic and temporal context. With three distinct waves of the pandemic in the U.K., it is likely that perceptions of engagement with public health guidelines changed over time and future work ought to attempt to capture this. The present study asked about engagement with guidelines using single-item measures, which has potential consequences for reliability. While we used several approaches (i.e., Likert-type questions, vignettes, reasoning) to approach the issue from multiple angles, future work ought to develop a reliable scale measurement of adherence to public health practices.
In our vignettes, participants read about either one teenager refusing to wear a mask, or a group of teenagers breaking social distancing guidelines. One possibility is that the presence of a group of teenagers in the social distancing vignette, as compared to one teenager alone in the mask vignette, was related to participants' evaluations. While we believe that the context of the vignette is more likely to account for the differences in evaluation (as evidenced in the content of the reasoning), future work should be conducted that presents an equal number of teenage characters in both types of vignette. Similarly, in our social norm task we asked about intentions of individuals, peers, and average teenagers in the UK. It is worth considering the possibility that the size of the group in question (i.e., small peer group versus teenagers at the national level) may be related to beliefs about diversity of opinion. For example, participants may account for a wide array of beliefs at the national level by aggregating and reporting a lower overall intention to engage with public health guidelines.
Further, understanding the kinds of information adolescents are using to inform these evaluations would be a valuable next step. For example, one possibility is that adolescents perceive wearing a mask to be a more effective means to reduce COVID-19 spread than social distancing, which could in part explain their stronger negative evaluation of the mask vignette. Including a measure of perceived effectiveness of different COVID-19 public health guidelines would be a valuable next step. Finally, the mean age of participants in the present study was around 17 years old. It is important to consider differences in perceptions of adolescents and children in different developmental periods outside of this age range. Older adolescents and emerging adults who are studying in university, for example, may see avoiding indoor gatherings as more of a challenge as they are thrust into living together in dormitory settings. This may in turn have important consequences for behavior shifts towards more damaging norms.

Conclusion
In future pandemics (as well as other public health contexts), public health authorities ought to frame communication efforts using the kinds of reasoning that adolescents themselves utiliseespecially in contexts where adolescents are acting to protect the wellbeing of others when they themselves are not at risk. Studying reasoning is essential for more clearly understanding why adolescents may be at risk of breaching guidelines. Even if a shift towards a perceived norm is occurring, the evidence presented here demonstrates that we can use the morally motivated thinking of adolescents to guide individuals towards engagement with public health practices. As adolescents competently weigh moral concerns, public health conventions and concerns for mental states, we ought to use these abilities when framing information to promote more accurate understanding of perceived social norms and engagement with public health behaviors.

Data availability
Data will be made available on request.