"We're Not in That Circle of Misinformation": Understanding Community-Based Trusted Messengers Through Cultural Code-Switching

While social computing technologies are increasingly being used to counter misinformation, more work is needed to understand how they can support the crucial work of community-based trusted messengers, especially in marginalized communities where distrust in health authorities is rooted in historical inequities. We describe an early exploration of these opportunities in our collaboration with Black and Latinx young adult "Peer Champions" addressing COVID-19 vaccine hesitancy the U.S. state of Georgia. We conducted interviews engaging them with a social media monitoring and outreach dashboard we designed, to probe their understanding of their roles and current and potential use of digital platforms. With the concept of cultural code-switching as a framing, we found that the Peer Champions leveraged their particular combination of cultural, health, and digital literacy skills to understand their communities' concerns surrounding misinformation and to communicate health information in a culturally appropriate manner. While being positioned between their communities and public health research and practice motivated and enabled their work, it also introduced challenges in finding (mis)information online and navigating tensions around authenticity and respect when engaging those close to them. Our research contributes towards characterizing the valuable and difficult work trusted messengers do, and (re)imagining collaboratively designed interpretive digital tools to support them.


INTRODUCTION
Online mis-and disinformation, including though not limited to content on social media, have severely impacted public health.Understanding and addressing this issue-which encompasses information which is false or misleading, whether spread unintentionally or intentionally (with disinformation referring specifically to the latter) [135,142]-is of growing importance in computersupported cooperative work (CSCW) and social computing research.In the wake of the COVID-19 pandemic, various studies have shown the incredible volume of misinformation about COVID-19 online and how it has driven vaccine hesitancy and other risky health behaviors [124], with the World Health Organization (WHO) declaring the scenario an "infodemic" [1,123].These harms have especially affected marginalized communities who simultaneously are disproportionately impacted by the pandemic in terms of physical health and mortality, mental health, and economic instability [2].While much CSCW research focuses on misinformation on contemporary social media, we also recognize that mis-and disinformation in the broader information ecosystem has long capitalized on racialized and other oppressive narratives that reproduce and reinforce societal inequities [80].In the United States, researchers have shown how COVID-19 misinformation often draws upon narratives of institutional distrust and longstanding inequities that stem from the injustices Black and other racially minoritized communities have historically faced from health authorities [28,30,33,80,83].This notoriously includes the Tuskegee syphilis experimentation conducted from 1932 to 1972 by the Centers for Disease Control and Prevention (CDC), the U.S. national public health agency, that deceived and withheld treatment from the Black men in the study; but also commonplace discriminatory medical practices today [2,28].
Given these challenges, equity-oriented health communication and sociotechnical approaches to address vaccine hesitancy and misinformation are critical to the public health response especially in marginalized communities [72,78,129].As a key part of such approaches, public health researchers and practitioners have emphasized the importance of working with trusted messengers from within these communities: individuals who are knowledgeable of local contexts, have established community relationships, and are allied to public health causes [2,9,72,83].Young adults have been a particular focus of COVID-19 communication efforts because of their relatively lower COVID-19 vaccination coverage and intent [3] and the unique challenges they faced from the pandemic at a critical stage of life for social development [25].In tailoring health communication for young adults, researchers have recommended empowering young adults themselves to be peer advocates [110], and using social media given its popularity among young adults [55,115].
As part of a broader community-based participatory research (CBPR) project to address COVID-19 vaccine hesitancy among youth, we worked with Black and Latinx young adults in the U.S. state of Georgia who served as "Peer Champions, " and explored opportunities and challenges for digital platforms like social media and other technologies to support their work.Our collaboration with trusted messengers is complementary to other proposed interventions, such as innovations in online platform design to reduce the spread of misinformation [65,66,68] and efforts to promote media literacy [51,140], while recognizing the need for more community-engaged approaches to address the community-specific dynamics and inequities of misinformation [13,78,106].In bringing together young adults from these communities for this project, recognizing that Black and Latinx populations had the lowest COVID-19 vaccination rates in the state, we acknowledge the diverse histories and cultures of Black and Latinx communities in the U.S., and encouraged the Peer Champions to approach their work as appropriate for their specific communities.Furthermore, we seek to draw on opportunities for solidarity in taking a race-conscious and culturally-specific approach to address vaccine hesitancy, just as many scholars and activists of color have approached other issues of social justice [73,128].
In this study, we conducted semi-structured interviews with the Peer Champions to understand how they saw their roles as trusted messengers, how they used social media and other digital platforms, and potential use cases for social computing technologies.We engaged them in the interviews with a social media monitoring and outreach dashboard that we designed during the project, not with the goal of evaluating its design or effectiveness per se, but rather using it to probe these more formative questions.Three joint research questions informed this study: RQ1 How do young adult community-based trusted messengers understand their positions and roles in addressing vaccine hesitancy and health misinformation among their peers and community members?RQ2 How do young adult community-based trusted messengers conceptualize digital platforms including social media as related to health (mis)information, and what are their perspectives on how to use them for their work?RQ3 What opportunities are there for social computing technologies to further support the work of trusted messengers countering misinformation in their communities?
In answering these questions, we turn to sociolinguistic literature on cultural code-switching especially as practiced by people of color in the U.S. to provide a conceptual framing.Through this lens, we found that the Peer Champions leveraged the unique intersections of their cultural competency, health knowledge, and digital literacy skills to make sense of misinformation and vaccine hesitancy, learn COVID-19 expertise, and use these knowledges to engage with vaccine hesitant community members.While their joint commitment to their communities and public health motivated and enabled their work, it also introduced challenges in finding (mis)information online and tensions around maintaining authenticity in their communication and relationships.Our research contributes towards characterizing the important and often difficult work of trusted messengers and community advocates, and (re)conceptualizing how they currently do and potentially can leverage digital platforms and social computing tools for their work.We call for researchers and practitioners countering misinformation to center and uplift the work of community-based trusted messengers, and propose that digital tools to support them design for a more community-oriented interpretive and collaborative approach to addressing misinformation and vaccine hesitancy.

Community-Based Countering of Misinformation
There is a wide range of people who address online misinformation and its effects at the community level.Public health researchers and practitioners have emphasized the importance of working with trusted messengers-local leaders and advocates who have the trust of community members and the drive to promote public health-in engaging marginalized groups with culturally appropriate communication in community spaces [9,72], especially when it comes to issues as sensitive as racialized and politicized health misinformation [67,80].CBPR engagements to address COVID-19 vaccine hesitancy have worked with various community members as trusted messengers including local faith leaders [10,100], community health workers [17], pharmacists [54], community organizers [10], and bilingual members of immigrant communities [131].Youth participatory action research (YPAR) projects have also engaged young adults to be COVID-19 public health advocates and researchers [42,109,110,116].
Recent studies in CSCW, human-computer interaction (HCI), and computer-mediated communication have looked at how a range of different individuals discuss and counter online misinformation.While journalists and fact-checkers may report on misinformation as part of their professional work [45,70,88], many individuals informally take on these sorts of roles in their communities, such as young adults on family messaging app groups [84,98], local government officials and community health workers in rural India [126], group chat moderators in urban India [126], and medicine-shopkeepers and religious leaders in rural Bangladesh [120].While not formally tasked to be trusted messengers, these individuals' roles emerged due to the trust others place in them.How they engage others with (mis)information is often informed by community cultural values: for example, young adults trying to remain polite to elders [84], and villagers considering religious values in evaluating information [120].These approaches are reflected in their use of digital platforms, such as shifting sensitive conversations confronting misinformation off semi-public group chats to more private settings to maintain face [98], or on the other hand discussing possible misinformation in a group chat for collaborative situated fact-checking [120,126].Our research investigates the more-than-informal yet not quite professionalized work that Peer Champions did, and how their unique positioning and cultural sensibilities informed how they made sense of vaccine hesitancy in their communities and conducted outreach to address it offline and online.
Considering digital tools for misinformation countering at the community level, researchers and practitioners have emphasized the importance of monitoring social media for misinformation that might impact (health) behavior [34,37,123].Analysis of social media users who spread misinformation or vaccine hesitant content can also provide insight into their underlying attitudes and concerns [30,41,69,71], informing tailoring of messages to such individuals.Given this interest, many researchers and organizations have developed COVID-19 social media monitoring analytics and dashboards [43,99,113,127,134].However, such analytics-as well as similar tools for other health issues [79,94], city planning [27,125], and business intelligence [15,57,89,118]-are often meant to inform stakeholders and policymakers in higher-level administrative positions.Indeed, those who conduct community-based social media monitoring on issues such as elections misinformation [102,114], hate speech [95,102], natural disasters [138], and mental health on college campuses [136] point to the greater complexity of doing this work at the local level.Close partnership with various stakeholders is required to curate community-relevant and ethical data collection processes, and to coordinate appropriate in-community responses.Our work contributes to this important and emerging space of using social computing technologies to support communitybased misinformation countering.We take a step back to understand how trusted messengers like the Peer Champions understand and approach digital tools and strategies like social media monitoring, to (re)consider what community-oriented design of such technologies could encompass.

Community Advocates as Intermediaries Leveraging Social Computing
The growing use of social media and other social computing technologies for health communication has expanded even further during the COVID-19 pandemic, with many daily interactions shifted to digital platforms and widespread misinformation online [90].Given physical distancing measures, many trusted messengers in CBPR engagements have embraced digital channels for communication, such as social media posts [17], group chats on messaging platforms [132], webinars and virtual town halls [2,39], and email listservs [17].While there is a deep literature of CSCW research in healthcare [38], its engagement with public health and health communication at the community level has been emerging more recently.Researchers have looked at frontline community health workers in India, who are a key part of the Indian public health workforce, and characterized their role as one of an infomediary bridging disconnected situated knowledges between patients and clinic-based healthcare providers [61].To understand better how these frontline health workers then use digital communication channels for their work, Ismail and Kumar frame them as being at the intersections of these and other communities yet at the peripheries of each, pointing to how they do and can use social computing technologies to participate more fully in these communities and move towards empowerment [62].Outside of public health, research in CSCW and HCI has studied how other community advocates have leveraged social computing from a liminal position.Irannejad Bisafar et al. 's research on youth community organizers described them as a distinct "public" situated between other publics of staff at the youth empowerment organizations and the broader youth community, influencing their careful use of context-collapsed social media platforms [60].Considering staff of CBOs, Voida et al. 's study described how non-profit volunteer coordinators used social computing technologies to bridge between their organizations and in-community volunteers, identifying three main functions of social computing: transposing social computing tools to their organizational context, blurring boundaries between the personal and professional on social networking sites, and translating messages between the organization and volunteers [130].Sum et al. study workers at specifically Asian American and Pacific Islander CBOs, understanding their work as one of linguistic and cultural translation between their communities and structures including government institutions and health and social service systems, which serves to legitimate and resist the "othering" of their ethnic communities by these institutions and by technological systems [121].The Peer Champions we worked with occupied a similar liminal position as these various community advocates in their use of social computing technologies to work for social change.Our research contributes to this broader space, in particular taking the concept of cultural code-switching to frame the Peer Champions' positionality and actions as community-based trusted messengers in public health, and thus understand their current and potential use of social computing to advance health equity.

Cultural Code-Switching and Identity
While code-switching in linguistics refers to "the ability [. . . ] to alternate effortlessly" between languages or language varieties [18], more broadly construed, cultural code-switching involves "switching between the norms governing two communities," adapting behaviors including and beyond language use in response to different social contexts [93].In the American context, cultural code-switching has been studied in how people of color move between their home communities and predominantly white educational or professional contexts, changing not just how they speak but other ways in which they behave and present such as hairstyle, dress, and mannerisms [86,87,93].
Code-switching into socioeconomically dominant languages and cultural norms, at face value, can help members of minority communities to operate and succeed in the mainstream culture [93].However, the ability or opportunity to code-switch can vary by class, occupation, age or generation [49].While this may put code-switchers in a relatively more privileged position within their communities, researchers have also investigated the social, mental, and emotional strains they experience [87].For Black and other racial minorities in the U.S., this includes negative stereotyping for not acting "professionally" at school and at work [86,101]; or on the other hand, cultural invalidations from in-community peers like getting called "acting white" or pretentious [35].
Nonetheless, code-switching can be a valid strategy that people choose to take on for the benefit of themselves and their communities, even if it may feel like "a necessity born of unjust socioeconomic conditions" [93].Others, however, may choose instead to resist code-switching [49,137], such as Black students taking pride in African American Vernacular English (AAVE) and engaging with it socially and academically, whether at predominantly white schools [52] or at Historically Black Colleges and Universities (HBCUs) where they may feel less pressure to code-switch [64,97].Similarly, some may resist the notion of code-switching as inauthentic behavior but rather embrace and assert an authentically intersectional identity [44], for example framing biculturalism and "Spanglish" as central to growing up Hispanic American [19,139].Choosing to code-switch, or not, is thus a power-laden act that can serve to reaffirm community ties and cultural identity, but also poses risks of being misunderstood both outside of and within their community that could undermine interpersonal relationships [93].
In public health and medicine, researchers and practitioners have looked to code-switching as a part of engaging marginalized populations.Braithwaite et al. describe community health workers as "a trusted link between macroagencies and local residents" who "is savvy and serves as a 'code switcher' for advocating for the client with health and human service providers" [14].In particular, linguistically and culturally appropriate communication is highlighted as a critical skill of promotores (or promotoras), a community health worker role that originated in Latin America and is now also common for serving Spanish-speaking Latinx populations in the U.S. [21].Code-switching away from "doctor-speak" is also recognized as a common and often useful practice in medical settings, though requires attention to the identity of who is doing the code-switching and the risk of offending patients [133].
Scholars in CSCW and HCI have also studied code-switching, mainly as it pertains to language.Research includes examining code-switching (or lack thereof) and language variants on social media [8,76,105], and in developing natural language processing technologies such as conversational agents [5] and hate speech identification algorithms [48].Lingel et al. consider a broader form of code-switching in their study of transnational migrants' use of social media, describing "a kind of technology-enabled code-switching" that included language, emotional, and social shifts in performing online identities to face different geographies and communities [82].Harrington et al. specifically investigate cultural code-switching as part of the experiences of older Black American adults interacting with voice assistants for health information [47].They characterize cultural code-switching away from AAVE (which they recognize as a legitimate dialect of English) as something that Black Americans have to do in order to avoid communication breakdowns with voice assistants, and call for attention to this form of digital exclusion in the design of future voice technologies.While these latter two studies are more directly rooted in examining concrete acts of cultural code-switching in language use, whether as spoken or written online, our study considers cultural code-switching more as a conceptual analogy that provides a useful framework to understand the non-trivial work that Peer Champions do and the challenges they face in moving between the words and worlds of their home communities and public health institutions.

Project Background
This study was carried out in the context of a broader CBPR project engaging Black and Latinx young adults in the U.S. state of Georgia to promote COVID-19 vaccination in their college campuses and communities, which was led by the third and fourth authors who are public health researchers at a historically Black medical school.The young adult "Peer Champions" on the project were recruited from two public Historically Black Colleges and Universities (HBCUs, which are universities founded to serve primarily Black students prior to the Civil Rights Act of 1964 that racially desegregated schools), two minority-serving institutions (MSIs, which are universities that enroll a significant proportion of students from minority groups), and three Black-and Latinx-serving communitybased organizations (CBOs) in the state.We intentionally chose to partner with HBCUs, MSIs, and CBOs as they tend to be located in or near communities disproportionately impacted by health disparities, in part because many such institutions were founded in response to these communities' exclusion and marginalization by other institutions [122].Faculty and leadership at the partnering institutions, who also served as "Community Champion" advisors, helped recruit individuals who they believed would be capable and interested in the project as a community and research endeavor.These Peer Champions were part of the project for approximately a year starting in September 2021, and were hired and paid hourly as Student Workers by the lead institution, with the exception of those from one CBO which joined the project midway who were paid ad hoc.
The Peer Champions had several different responsibilities, including recruiting other young adults in their communities to participate in a survey about their COVID-19 vaccination experiences, knowledge, and attitudes; and ideating and implementing interventions as they deemed appropriate to reach out to their peers.Some interventions conducted include in-person events, online webinars, and creating public service announcement (PSA) videos to share on social media.We and other partners on the project also held several training and discussion sessions with the Peer Champions which focused on learning about racialized COVID-19 misinformation on social media and strategies to respond.

Context of Dashboard Design
For this study, we interviewed the Peer Champions while engaging them with a social media monitoring and outreach dashboard which we designed and developed as part of the project.The design of the dashboard began as part of a sister project involving the authors' institutions to address COVID-19 testing and vaccination hesitancy among Black and Latinx communities in the state [50], when various community partners raised concerns about misinformation spread on social media and expressed interest in engaging community members on social media.We then initiated this project with the Peer Champions, incorporating a focus on social media from the start (including continued work on the dashboard) as we believed it would be especially relevant for young adults.
As a starting point for the dashboard's design, we used prior work by the last author on an online platform for elections monitoring using social media [114], in particular to develop our topically-and locally-focused social media data collection and labeling pipeline.However, based on our project's goals and initial interactions with the Peer Champions, we recognized that our use case and users would be markedly different, with more browsing than in-depth analysis of social media content, and a greater emphasis on outreach as the end goal.The dashboard was then further designed and developed based on various interactions with the Peer Champions over the course of the project, including: group discussions early on about their intervention plans and personal social media use; a group online whiteboarding activity creating mock social media posts to address COVID-19 vaccine concerns; and presentations at general project meetings on the dashboard at different stages of its design and development where we elicited questions and feedback.We note that our ability to involve the Peer Champions in a more participatory manner in the design process was limited by project timelines, their availability given their various other responsibilities, and generally reduced interaction across the project team given COVID-19 safety measures and our geographic spread.
Our original goal was to co-design the dashboard with the Peer Champions and deploy it as a tool they could use for the project.However, as the project and its activities unfolded over time (as is often the case in CBPR [63]), in addition to the challenges outlined above, it became clear that doing so would be difficult.Furthermore though, we found that engaging the Peer Champions around the dashboard could serve as a means to probe deeper into their understanding of online health misinformation in their communities, their roles and activities as trusted messengers, how they used social media and other digital platforms at the time, and how they could use social computing tools including but not limited to our dashboard for their work.That is, while not originally designed with this intention, the dashboard served as a sort of technology probe [58], a method in HCI research (alongside other probe variations) that engages participants with artifacts to learn about how users interact with existing technologies or might interact with new ones, generating insights that may inspire future technology design [12].While we had discussed these questions with Peer Champions previously, we found that without an artifact to center the conversation around, such conversations remained at a high level without going into how they interacted with any particular platform or piece of social media content.An alternative method of in situ observation of the Peer Champions' use of digital platforms would have been challenging if not impossible given social distancing measures for COVID-19, as well as the complex and somewhat loosely defined nature of their work-for instance, they did not conduct (all) their work at regular times in any particular place, and "personal" usage of social media or search engines would have also influenced their work as Peer Champions.As such, we decided it would be valuable to conduct this formative research using the dashboard as such a probe.

Social Media Monitoring and Outreach Dashboard
The social media monitoring and outreach dashboard, when used for this study, consisted of two primary components: a live deployed social media monitoring page, accessible from a desktop web browser (Figure 1); and a prototype of a new page with curated health communication resources (Figure 2).We described its technical aspects in more detail in an earlier workshop paper [24].In the screenshots shared here, we have blurred account information and certain content and added an overlay to posts containing misinformation, following guidelines on best practices for reporting on misinformation [85,111].Below we describe its major features as pertinent to this study.The prototype resources page (Figure 2, screen on the right) demonstrated a new page for the dashboard which would contain curated health communication resources.These included COVID-related social media assets and fact-checked informational content which were created by or shared from sources trusted by our partnering organizations such as national and local health authorities, health promotion organizations, and our partners themselves.Similar to the social media monitoring filters, users would be able to filter content by topic, source, and resource format type.The prototype page, created in Figma, showed a mock website interface with several example resources, and allowed users to click through a limited number of filters.
We also prototyped a feature that would suggest relevant resources for responding to selected social media content from the monitoring page of the dashboard (Figure 2).While this was meant to demonstrate a potential semi-automated feature, for the prototype, we hand-selected several social media posts and accompanying relevant resources to gather the Peer Champions' perspectives on what would make for effective responses to vaccine hesitant social media content.For this study, we focus on the post shown in the screen on the left in Figure 2, a tweet from an account appearing to belong to a Black man which mockingly expresses doubt in the COVID-19 vaccine because of its politicization and rapid development.We chose three resources that varied in format and in how they addressed the tweet (Figure 2, screen on the right): 1) an Instagram post with a picture and description of Black scientist Dr. Kizzmekia Corbett and her role in the development of the vaccine, created by a partnering organization that ran a social media campaign addressing COVID-19 in Black communities; 2) a video entitled "COVID-19 Vaccines: Safe and Sound Science, " from the state Department of Public Health; and 3) an infographic on "The Journey of a Vaccine" describing the different phases of vaccine development and testing, created by national U.S. government health agencies including the CDC.

Interview Procedure
Our interviews with the Peer Champions combined think-aloud methods, walking them through the dashboard with specific tasks to perform, with semi-structured interview questions relating to and reflecting on their role and activities as Peer Champions.Interviews were conducted by the first and second authors.Interviews were conducted via video conferencing using Zoom, during which the interviewer provided the interviewee with links to the live dashboard and prototype websites to open on their laptop, and had them screenshare their browser window with the websites open during the interview.In addition to the interviewer, one research assistant joined each interview Zoom call to assist with notes and any technical issues.Each interview lasted approximately one hour, and was video recorded using Zoom's meeting recording feature.
For the dashboard's live social media monitoring page, we asked the Peer Champions to browse social media content that appeared and gave them specific tasks to use several different filters and the keyword search bar while doing so.For each task, we asked whether the content they saw was what they might have expected to see, and whether they had any questions regarding the dashboard features or content.After these initial tasks, we asked them to use any of the demonstrated features to find social media content about any COVID-related topic of interest to them.Follow-up questions included discussing why they chose particular topics, how they were using the dashboard for the task, their typical social media browsing and information seeking practices, and how useful or not they thought each feature would be for their work as a Peer Champion.
For the dashboard's resources page prototype, we similarly asked the interviewed Peer Champions to browse through communication resources displayed, providing specific tasks to try different filters.We asked whether the resources shown were what they might expect or want to see, and whether they had any questions about the prototype or resources.We also asked Peer Champions to read a social media post expressing vaccine hesitancy (left-most tweet in Figure 2, screen on the left), view the three selected resources for it (Figure 2, screen on the right), and discuss which one they thought was most relevant and why.
While we borrowed from HCI usability evaluation methods for the interview procedure, our primary goal was not evaluating usability or gathering feature-specific feedback for the dashboard per se, but rather leveraging the dashboard as a conversational starting point to discuss the Peer Champions' perspectives on their current and potential use of digital platforms and their broader work at large.Similar in spirit and goals to technology probes [12,58], our approach is close to that of Irannejad Bisafar et al. in their study conducting interviews with youth community organizers, where they used low-fidelity prototypes of hypothetical social network visualization tools "to spur conversations [. . .] that would help surface important areas for future inquiry and design" [59].Although the dashboard components used for our study were already fully functional or high-fidelity given their ongoing design and development, we made clear to the Peer Champions during the interviews and in general that work on the dashboard and potentially other future tools would be continuing and that we were open to change and alternative possibilities, readily acknowledging limitations and shortcomings of the existing live dashboard and prototype, to invite their frank feedback and participation.1).All were students, mostly undergraduate and masters students and one high school senior; and all were between the ages of 18-26, with the exception of two older students.Interviews were conducted between April and October 2022.
Based on the last author's conversation with their institutional IRB, we determined that this study did not require IRB approval as the Peer Champion interviewees were themselves project staff who we would be speaking to regarding ongoing project activities.(The broader CBPR project itself was approved by all authors' institutional IRBs.)However, we described the study in meetings and emails inviting them to participate, and obtained oral informed consent prior to beginning interviews.

Analysis
All interview sessions were video recorded and transcribed verbatim.Coding and qualitative analysis was conducted by the first and second authors, taking a grounded approach of using reflexive thematic analysis to inductively develop initial findings [16].For our first round of analysis, we used open coding without particular stipulations on length or frequency of codes, in addition to process coding that used verbs to identify particular processes and actions that Peer Champions described as part of their work (a grounded coding technique similar to [23]), and codes that identified interactions with the dashboard or described uses of other digital platforms.The first and second authors conducted this first round of coding in parallel, following which they compared and discussed their codes and then conducted a second round of analysis to refine the codes.We then discussed the codes again and iterated to collaboratively create higher-level themes based on groupings of these codes.One set of groupings characterized the higher-level functions the Peer Champions were performing, such as "listening," "researching," and "engaging;" a second overlapping set of groupings highlighted quotes that demonstrated the positioning of the Peer Champions, such as "connection with community" and "distance from community."Throughout the data collection and analysis process, the first and second authors debriefed with each other, comparing new data with preliminary themes.Preliminary themes were also shared for feedback with the other authors, and with several Peer Champions themselves at a project meeting following the completion of interviews.Initial findings were then revisited by the first author with the concept of cultural code-switching in mind, reinterpreting the Peer Champions' actions analogously to code-switching and their challenges as similar to those faced by code-switchers as described in sociolinguistic research.

FINDINGS
Our findings demonstrate how the Peer Champions, acting as trusted messengers, were not merely convenient information conduits between public health institutions and their communities.Rather, we found that they performed sophisticated work leveraging their cultural, health, and digital literacies that came from being members of both their home communities as well as the community of public health research and practice.We describe here the Peer Champions' various activities and strategies, operating from this particular in-between position, to prepare for and conduct outreach in their roles as trusted messengers: listening in their communities to understand vaccine hesitant perspectives; researching the latest public health information; and bringing together this community knowledge and health knowledge to engage those around them with health information in a culturally appropriate manner.We highlight for each of these sets of activities how they used or might use digital platforms including social media, search engines, and tools like our dashboard as spaces for (mis)information seeking and sharing, and their attitudes towards them.While the Peer Champions' joint commitment to public health and their communities informed and enabled their strategies offline and online, we show how it also introduced certain challenges and tensions.

Listening to Their Communities: Cultural Connection and Disconnections
To inform their outreach approach and issues to focus on, the Peer Champions actively attuned to what concerns and questions people in their communities had about COVID-19, accessing this information through their community presence, relationships, and cultural knowledge.In the process, they formed an understanding of why some of their peers and community members were vaccine hesitant while also recognizing a certain social and ideological distance from them.This distance also manifested as challenges they faced in being aware of misinformation, including on social media.As such, they noted the potential for social media monitoring to provide a better window onto diverse perspectives and misinformation that could be relevant to their communities, enabling them to develop a clearer culturally-informed understanding of vaccine hesitancy and thus be better prepared to respond.

Establishing a Critical Understanding of Vaccine Hesitancy Around
Them.An important part of the Peer Champions' roles was to understand the particular concerns and underlying attitudes driving vaccine hesitancy in their communities in order to inform outreach discussion points and strategies.The Peer Champions used different channels for sensing what concerns community members had about COVID-19.Many of them held project-sponsored events, such as on-campus tabling or online town halls, to engage others in conversation about COVID-19 vaccination.More often though, their knowledge came from casual interactions and being present in their communities, such as talking with close family members and friends, noting generally what they heard people talking about, or seeing people not wearing face masks-as P6 said, "we hear stuff in the air." Through these conversations and observations, the Peer Champions over time formed an understanding not just of the kinds of reasons people had for vaccine hesitancy, but also evaluated the kinds of people who might be vaccine hesitant relative to their own stances.
P9 for example shared how he talked about the vaccine a lot when it was first introduced with many friends and family members, and characterized what he thought of those who were vaccine hesitant or more strongly anti-vaccine: I think for the people who are more hesitant about it, they were just saying that you don't know what's in it, this could lead to more further problems with your body.I mean, some of them actually made some good points.But [. . .] there was some ignorance.You know, people were like, "COVID actually isn't a thing," you know, "this is just a facade, " or whatever.[. . .] Some people make good points, and then some people are just kind of ignorant about it, so.
While P9 was careful to respect and to some degree empathize with the concerns of vaccine hesitant people around him, he maintained a sense that he was more health-literate and discerning about information than the relatively more "ignorant" people around him.This respectful and caring understanding, but with a critical recognition of differences in health knowledge or even ideology, was a perspective that was consistently echoed by the other Peer Champions.
The Peer Champions also listened to different perspectives online, primarily on social media platforms.While the idea of social media monitoring was something we introduced to them during the project, several Peer Champions discussed their existing proclivity to pay attention to popular topics of discussion and different perspectives on social media.This included looking at content from people they did not personally know, but might be relevant to their community-P12, for example, often read through comments on local news stations' Instagram posts and evaluated the diverging opinions-and content from people they did know.
P8 also would look at comments on social media posts, particularly those containing misinformation.This was especially on Facebook, where she followed more people she knew "in real life" compared to on Twitter-though she noted that the people she saw sharing misinformation were "mostly old coworkers and acquaintances" rather than closer friends, recognizing some degree of social distance from them.While browsing posts under the "Known misinformation spreaders" filter on the dashboard's monitoring page, she commented on how she thought people she knew would react to them.She thought for instance that they would not engage with more text-heavy posts, as "a lot of people that I know that share misinformation, they're not gonna read that much." However, one tweet that misleadingly captioned a screenshot of a news headline caught her eye: But a post like this, I feel like I would see it on my Facebook feed of people that I know in my town [. . .] [sharing it] and being like, "so true." Just as P9 developed an understanding of vaccine hesitancy from his in-person interactions, P8 demonstrated her cultural attunement to what misinformation people in her community were attracted to and how they engaged with it online.Yet similarly, the way she talked about them indicated that she thought they were not very discerning and critical of misinformation online-unlike herself and the other Peer Champions.
4.1.2Negotiating Partial Access to Misinformation.While the Peer Champions' membership in their communities afforded them access to conversations offline and online about COVID-19 and thus cultural knowledge about the vaccine hesitant beliefs that resonated with people in their communities, some also recognized that they were not very aware of the content that vaccine hesitant community members might be consuming, especially online.P5 discussed her and her partner Peer Champion (P3)'s experience with this challenge, and how it made it harder to decide what issues to focus on: We just kind of brainstorm and think of things that we've heard.But it is a difficult process, because we're not in that circle of the misinformation.So it's kind of hard to think about what they could be thinking about or what they've been hearing.P5 recognized that they, as people who did not believe and engage with misinformation, were not as plugged into the online spaces with more of it, and thus did not have much direct visibility of misinformation influencing their communities, even for people with whom she had close personal relationships.For example, she knew that "a lot of the older generation" of Hispanic community members engaged with misinformation that was often in Spanish, including her father who had shown her posts from Facebook before that she recognized to be misinformation.When asked, however, whether she knew more specifically where he sees misinformation online, she explained: I don't use Facebook that often.So I'm not entirely sure where he finds information.I'm sure it is very likely from groups, and then it's shared on, like, people that he follows.So then he sees that information.But I'm not entirely sure where, exactly what kind of groups they are.While she had a general cultural awareness of the kinds of (mis)information sources people in her community went to, she had less direct familiarity with them.Other Peer Champions were similarly aware that their own social media feeds might not have much pertinent content for staying up to date on misinformation or other perspectives on COVID-19 developments relevant to their communities.Some also noted how this could vary by platform: for example, while P8 did see misinformation that people in her community shared on Facebook, she reflected that she probably did not see as much on Twitter, where she did not follow as many people she personally knew: I'm sure that there's misinformation on Twitter.But since I am someone who doesn't believe COVID misinformation, the people that I follow, I noticed they also [. . .] don't share misinformation often.But I think that's more just who I follow.Although the Peer Champions did not use the phrase in these interviews, their comments demonstrated an awareness of "filter bubbles" on social media, and how some combination of their own decisions (such as who they chose to follow) and the nature of the social media platform feeds themselves shaped the content they saw [26,143].In particular, they noticed that they were probably not in the filter bubbles where misinformation flourished [31,117], reflecting how they saw their social and ideological differences with vaccine hesitant people manifesting online.
To get around these challenges, the Peer Champions shared how they thought the dashboard's social media monitoring page could help them more easily find and stay up to date with misinformation, and thus be prepared to respond to people in their communities who could be exposed to that misinformation.P3 lamented how "there's new conspiracies every day, " and with social media monitoring, would want to "see what's out there, and then be able to debunk them."Her partner P5 added how she thought the dashboard putting potentially relevant content in one place would make it "a lot easier with finding [misinformation] things right away and just having to choose which ones we would want," compared to their experience of having to "keep searching, keep scrolling" elsewhere online.In particular, Peer Champions were interested in seeing vaccine hesitant or other perspectives that they otherwise would not see much on their own, and seeing content that was culturally relevant to their communities.P3, P5, and P8-the Latinx/Hispanic Peer Champions interviewed-all emphasized wanting to see content from Hispanic sources, including Spanish language content; and P6 similarly was interested in the "Black/African American" social media account filter on the dashboard.
Beyond simply seeing such content through social media monitoring, the Peer Champions demonstrated the critical eye they would have in evaluating it from both their personal perspective and as it might pertain to vaccine hesitant community members.For example, during their interviews, both P2 and P12 saw tweets on the dashboard by Candace Owens, a Black conservative commentator, declaring that she was unvaccinated, and expressed their deep disagreement and even contempt for her, with P2 asserting that "as a Black person, she doesn't really represent that culture, our culture." That being said, they both acknowledged that some people did share her views, with P12 saying that "hearing their stance on it is just as important, " such that she could be prepared to respond "if I am dealing with somebody who is like her [Candace Owens]." For P12, hearing vaccine hesitant perspectives that she disagreed with was an important part of engaging respectfully with others about COVID-19, where she wanted to understand and respect others' opinions just as she would like them to understand and respect hers.

Researching Public Health: Information Evaluation and Critique
In addition to attuning to what their communities believed about COVID-19, the Peer Champions also tried to learn what information they could about it.They leveraged their existing knowledge of public health and of credible health authorities, as well as information and digital literacy skills, to identify and interpret such information.Even so, they faced difficulties finding and having confidence in information online that they trusted and thought their communities would also trust.This was exacerbated by the flood of COVID-19 (mis)information and their overall skepticism of online information.To address these challenges, they noted how a curated digital platform with tools to assist in navigating and making sense of information could provide a helpful alternative to general search engines and other online platforms.

Becoming Critical Public Health Experts.
To prepare themselves to be reliable sources of information for community members, the Peer Champions sought to become local public health experts on COVID-19.In various ways, taking on this role aligned with their broader career and community-oriented goals.Many were active in the community-based organizations that the project recruited them through as well as other on-campus student organizations that focused on uplifting underserved groups in their communities.As students, many of the Peer Champions were studying public health, nursing, social work, public policy, and other health-and service-oriented subjects at their universities.Several of them said they were motivated to join the project in part because of an interest in research, with P2, P7, P9, and P10 noting in particular how their activities as Peer Champions learning about COVID-19 reminded them of how they looked things up for writing research papers.
The Peer Champions made efforts to stay abreast of the latest COVID-19 developments in a timely and community-responsive way, largely through seeking information online, such that they would be equipped with public health knowledge and information resources that they could share with their communities.P1, for example, wanted to look up information about vaccine boosters on the dashboard during his interview (which took place approximately three weeks after the CDC recommended a second vaccine booster dose for certain vulnerable adults) because, per his own judgement, "I honestly haven't done too much research on that one [. . .] and I just was curious on what does it mean in regards to health-like, does it make your chances of getting ill much more less likely, or is it still about the same?"Through the questions he asked, P1 demonstrated his own health literacy in terms of recognizing that the booster was aiming to reduce the risk of getting sick (versus absolutely preventing or treating COVID-19 infection), and his motivation to learn more.P8 also talked about how she would look things up that she heard others in her community-who were perhaps less health literate-ask about: There've been times where someone brings up, "well, have you heard that, like, someone said something about the vaccine doing [waves hands] something to something?"And no, [. . .] I really haven't heard of it.I could learn about it and see.
In describing how they looked up such information, the Peer Champions also demonstrated strong information and digital literacy skills [6], such as paying attention to the sources of information, checking URLs for signifiers like ending in ".gov," and using multiple sources.P2, for example, described the "cross-referencing" that he would do to verify information, including on social media: Even if I'm just doing regular social media stuff, I always try to see if it's published in more than one site, who are these sites.And then sometimes depending on how wide it might seem to me, I might look up onto something like Snoop [sic] or something.[laughs] [. . .] I've gone so far as to actually Google a site to see if it's actually trustworthy.And I know it's a little bit much compared to some of what my peers do.[laughs] But I just like to be on the front end of some good information.P2 took many deliberate steps in verifying information, even taking a bit of pride in his rigor compared to others.
In terms of their go-to sources for credible health information, nearly all the Peer Champions mentioned government health agencies as familiar and trusted sources that they would look to first, with all but two of them explicitly mentioning the CDC (among other agencies) in their interviews as a source they trusted.However, they were critical of how these institutions presented information to the public, and understood why others in their communities might not trust them anymore.P2 noticed for example: While P2 did not share these doubts in the CDC's credibility, understanding that guidelines were legitimately changing frequently, he nonetheless was empathetic to others' confusion and concerns about the CDC and would accommodate those doubts by providing them credible information from alternative sources.P10 also issued her own critique of health authorities' inconsistent public communications about booster shots, and what she thought would be better: If I was in their shoes-and I try to do this quite often-I would've said, "as of right now, we know this is gonna help the symptoms of COVID." [. . .] I wouldn't put such hard limits, like, "yes, you're only gonna need one dose, " and then come back and say, "well, we lied, basically, you're gonna need more than one dose."[. . .] I feel like that just makes more people more comfortable.In a time of panic, people were so panicked that they were taking anything.While she was harsh in her criticism of health authorities, P10 imagined herself "in their shoes" and wanted to do better as a public health communicator herself, demonstrating ultimately her connection to the public health sector.Indeed, the Peer Champions' criticism of health institutions' communications, especially their failure to maintain the trust of marginalized communities, was shared by many in public health [75,108,129], with the CDC itself recognizing its "confusing and overwhelming" COVID-19 guidance and announcing reforms following an internal review [20,81].

Digging for Trustworthy Information
Online.In discussing their experiences looking up information online, a predominant challenge many brought up was being overwhelmed with all the content they had to sift through.Several Peer Champions described searching the internet, as P6 put it, "like [being] in a big fishbowl." While this was in part due to the sheer amount of information and constant developments regarding COVID-19, they also accorded some of the blame to the online platforms they used, and expressed their suspicions about the motives of online platforms and sources to share relevant and credible information.P8 described having to "dig" through Google search results to find high-quality information resources past "the first things that pop up [that] are like, very clickbait-y news things." P4 also felt overwhelmed trying to sift through all the information she got from various digital platforms and channels: I think that the way I look up COVID now, it's a lot more sporadic and [. . .] unorganized just because I'm trying to figure out as much information as possible.And the information that comes to me, whether it's through Twitter, or my notifications, or just through watching a news outlet.It's just, they're trying to get really enticing headlines to make you watch or read that story.Or you're just looking at the most recent thing, but you're not necessarily getting the information that's the most important.
While their concern about the trustworthiness of online platforms and sources in providing better quality information is certainly founded and reflects their digital literacy, their worries often belied a distrust of ill-intended actors and their unchecked prevalence online, thus leading to a generalized skepticism of online content.P2, for instance, quipped that "everybody got a dot-org now" and preferred sticking with websites that he was personally familiar with.P6 shared this skepticism, even cynicism of sources she did not recognize.While browsing the dashboard, she commented on a tweet with an image promoting National Minority Health Month by the National COVID-19 Resiliency Network, a federally funded initiative promoting COVID-related outreach to vulnerable communities that she had not heard of before: I don't know, maybe it's just because I've worked in this field.[. . .] So many people just throw a name up there and put stuff out there.The fact that I've never heard of it, I wouldn't trust it, I would have to investigate it, I would have to Google it and see who they really are.[. . .] And the eye-catcher for this ad is "National Minority Health Month." So quite naturally, somebody like me would be like, "Oh, minority health month, let me see what they saying." It can always be a gimmick.The distrust expressed by P6-who was an older Peer Champion with previous experience working in health, and a Black woman-was not rooted in a distrust of public health institutions per se, but rather in a world-weariness of having seen people take advantage of minorities in the name of public health, especially online.
The Peer Champions' skepticism of online content and platforms reflects a broader trend of people, especially young people from marginalized communities, being worried about online misinformation [77].Our findings reflect prior research also conducted on COVID-19 information seeking that characterized a shift from (dis)trust in information sources to (dis)trust in online platforms and other information infrastructures [141], though our Peer Champions' evaluation of trust is also colored by their emerging self-identification with public health expertise.
In light of these challenges, when providing feedback on our dashboard, the Peer Champions emphasized how they felt the dashboard put things "in one place" (P8) and with the labels and filters appeared "organized" (P4) and "straightforward" (P1, P9), making it "easier" to find and identify pertinent information (P1, P5, P6).Several noted how they thought, unlike a general search engine, the narrower scope of our dashboard would be useful for their similarly focused needs as COVID-focused Peer Champions, with P5 remarking that compared to searching for information herself, "it's a lot easier to see and be like, oh, maybe I should also talk about that" when deciding what issues to focus on.
Besides helping them find relevant information, the Peer Champions also thought the dashboard, specifically the labels applied to social media posts and health communication resources, would help them make sense of the content and know what to trust.P1 explained that "I like how it [content labels] kind of give you the gist of what to expect under each article.When you Google things, you don't really know for sure."P11 especially found the "known misinformation spreaders" account label useful: Because everything on social media, it's either true or false.But most of the time, it's false.So if [. . .] something occurred from like, a social media page, and being able to come on the website [social media dashboard] and see that [. . .] it's mistrusted, then I would think, oh, it's not true.I should not take this as fact and continue to spread it to somebody else, that it's true.
Although they were skeptical of other online platforms, the Peer Champions were very interested in understanding from us how the dashboard worked such that they could trust its curation and labeling of content.In these as well as prior meetings, they asked us and other colleagues who helped develop the dashboard questions on things like how the social media data was collected, how labels were assigned, what different labels meant, where we got health communication resources from, and how we decided what was trustworthy.Beyond just wanting to know how to use the dashboard technically, their questions demonstrated a desire to establish trust in the dashboard through their connection with us.

Communicating Public Health Information to Their Communities: Engagement and Authenticity
In conducting their outreach, the Peer Champions did not want to just repeat and amplify the messages promoted by health authorities.Instead, the Peer Champions used the knowledge they gained about their communities' concerns and about COVID-19 public health expertise, together with their familiarity of cultural norms and preferences around communication, to provide information to their peers and community members in a reliable and culturally responsive manner both offline and online.In doing so, they wanted to authentically represent their belief in public health measures but also their care for community members-commitments that informed how they used online platforms to engage with others, but also introduced tensions around raising these sensitive topics especially with those closest to them.

Bringing Together Cultural
Knowledge with Health Knowledge.The Peer Champions' outreach strategies involved trying to attract community members to want to learn about COVID-19, shaped by their understanding of community members' concerns and what they would (and would not) be motivated by; and then providing credible information in an easy-to-digest and culturally appropriate format to their communities, using their understanding of COVID-19 public health expertise and of what people would be willing and able to read.In this way, the Peer Champions leveraged their knowledge of both community culture and public health in order to communicate effectively between their communities and public health expertise.
A primary challenge the Peer Champions faced was that their audience was, simply put, "tired of COVID" (as phrased by P1, P6, and P7) by the time of this project.Some Peer Champions attributed people's disinterest to a misled belief that COVID-19 was not that impactful anymore; or, that things kept changing confusingly such that people did not care to keep up anymore.This understanding motivated the Peer Champions' approach of sharing relevant, reliable information to get others to care again about COVID-19; or, as P12 put it, "you got to persuade people."P5 explained why, for example, she would want to share case numbers at their local hospital with others: 'Cause I know there's a lot of spikes here and there.And people don't see that because they're not looking those numbers up, and they just think that everything's going fine.So I think in seeing local news, and seeing that it's happening here, they're more likely to be receptive to that information.Besides feeling that many people were uninterested in and thus ill-informed about COVID-19, the Peer Champions also sensed that many of their peers and community members-especially those most susceptible to misinformation and vaccine hesitancy-were not the type of people who would, or would be able to, digest health information critically or in-depth (as also discussed in section 4.1.1).Thus, Peer Champions also saw it as their responsibility to collect and distill complex information about COVID-19 into an easily understood form that would be "captivating to a person's eyeballs" for the "people that don't like to read, that just want to believe the first thing that they hear all the time," as P6 described.The Latinx Peer Champions (P3, P5, and P8) also noted that it was critical for their communities to share information in Spanish.
To put together such content, P2 described how browsing through lay-audience-oriented health communication resources could be useful, not as information resources for himself to learn from per se, but "to help me frame it [COVID-19 information] in a more dynamic way, for people that don't want to spend the time." To dive more concretely into how the Peer Champions might leverage such resources to respond to someone expressing vaccine hesitancy online, we had them read a tweet from an account appearing to belong to a Black man caricaturing politicized narratives about the COVID-19 vaccine and select one of three resources that they thought would be most relevant in response (see Figure 2).Four Peer Champions selected the first resource, featuring an image of Black scientist Dr. Corbett; four selected the second, a video on "COVID-19 Vaccines: Safe and Sound Science;" and eight selected the third, an infographic on "The Journey of a Vaccine" and its development process (three Peer Champions chose two resources).
The Peer Champions who chose the video and infographic identified that the author of the tweet probably did not trust how the vaccine was developed so quickly and under highly politicized circumstances, and thought those resources addressed the author's concerns in an engaging format and from familiar trustworthy sources.As P8 saw it, the infographic explained "why they [the vaccines] were trustworthy to begin with" in an "easy to follow" visual manner, thus being able to address his doubts effectively.P12 also appreciated its neutral non-polarizing tone, that "it's not going at a certain party.It's not trying to make you feel bad for your choices.I think it's just trying to give you information." While the Peer Champions put great emphasis on sharing facts, they also thought people would appreciate hearing from or associating information with individuals they could identify with.P4 said of the first resource, the picture of Dr. Kizzmekia Corbett with a description of her role in the vaccine's development, that being able to "put an actual face to the information that they're getting [. . .] might seem more reliable, and [. . .] not just someone who's just talking [...] through a screen." P7 also gravitated to the image of Dr. Corbett, and how she thought it would resonate especially for other Black people, like the user in question: I feel like Black people can relate to other Black people.They may not relate to white people the same as they relate to another Black person.[. . .] So seeing someone that looks like you being involved in something that you're unsure about, or you're curious about, would make you be more open to seeing that, to being involved in it.Several Peer Champions also brought up examples of more relatable "testimonial"-like content to engage people; as P1 said, "I know a lot of people, including myself, we like to hear like what people exactly went through, versus always just reading straight facts." The Peer Champions noted the importance, however, of hearing from people they could actually relate to or cared about, recognizing the heterogeneity within communities and intersectionality of multiple identities beyond race, such as age and political leaning.P6 talked about how "just because somebody famous Black got tested and got vaccinated, that doesn't intrigue them [other Black people] to do the same thing, " such as for younger people hearing about older celebrities.P12 echoed this sentiment, saying how "posting a picture of a Senator getting the vaccine, if I don't like that Senator, I'm not gonna care, " and how she would rather see how "people my age are talking about this stuff." 4.3.2Negotiating Tensions in Relationships and Authenticity.While the Peer Champions' sense of what would most effectively engage their audience informed the content of messages they wanted to share, they faced tensions in deciding when and how to engage others about COVID-19 in a respectful and authentic manner without straining their relationships with others in their communities.Although they adopted certain outreach approaches to mitigate this, including strategies around self-presentation and communication on online platforms, these tensions often remained as an underlying anxiety.
The outreach approach of many of the Peer Champions was shaped by a strong belief in respecting every individual's autonomy to make informed decisions about their own health, and "treat people like I would like to be treated" (P10).Even while they might think others were ill-informed, they emphasized that their goal was to provide facts and enable others to do their own further research if desired and not force their pro-vaccine beliefs upon others, given the sensitivity and mistrust around the topic in their communities as well as their personal relationships with them.P12 shared for example how she would approach vaccine hesitant older relatives: My family in [other U.S. state], they don't really trust a lot of things.So they were like, "Okay, we're gonna pray it away, we're not gonna worry about a vaccine, the Lord has it, " which is cool.[. . .] It's harder talking to someone in an older generation [. . .] or who's always had just their mindset.They don't want to hear what [a] 20 year old has to say.And I just wanted different ways to approach it, not them feeling like I was talking at them, but trying to just give them knowledge of the information they needed and not overwhelm them or make them feel bad about their choice.Because at the end of the day, it's still their choice.P10 felt similarly, and as a nursing student, she also added further insight and motivation from her knowledge of health inequities, and her own professional aspirations to address them: I don't want to push my information or my doubts or anything on someone else.[. . .] Now, I know I'm becoming a medical professional, and that's different.But as I hope to practice, that I will give them the best information, and then they can make their own decision based off of that.[. . .] Even like in medical practices, when you're pushed into medicines that maybe you cannot afford, or maybe you just don't see the benefit of getting it [. . .]. Whatever your doubts [. . .], those are your doubts, and those are valuable doubts, and those are doubts to you, and nobody can understand your doubts but you.P10 had a strong sense of the kind of medical professional and communicator she wanted to be-breaking, essentially, from the historical precedent set by health authorities mistreating communities like hers.
Relatedly, many of the Peer Champions shied away from direct approaches of proactively bringing up COVID-19 and advocating for the vaccine.As part of their project responsibilities, most Peer Champions had conducted some project-sponsored outreach activities, like holding in-person events talking about COVID-19 and creating and sharing video PSAs (public service announcements) about the vaccine.However, in more informal settings, many Peer Champions tended to take a less direct approach that felt authentic to themselves and the nature of their interpersonal relationships.Several Peer Champions talked about how, especially with friends, they preferred talking about COVID-19 as it came up "in casual conversation" (P9) and in a way that felt more natural for them.
P4 described an example situation where, if COVID-19 and vaccine hesitancy happened to come up during a group chat's usual conversation, she might react by more discreetly sharing information in a one-on-one DM (direct message): Let's say I'm in a group chat, and we're talking about COVID and stuff like that.And I have a friend of mine who will say, "Well, I haven't gotten the vaccine yet, because I don't know a lot about it, " or something like that.Then that's when I go into my Instagram DMs and communicate further with that friend or send them a news article or something.This is similar to what Pearce and Malhotra find in how Indian young adults moved conversations correcting misinformation with older family members from group chats to more private settings out of concerns of politeness and face [98]-with P4 motivated by the particular cultural norms she had with her friends.
P10 also preferred to share information in a way that was authentic to her own personality and her typical tone with friends and peers-for example, resharing content that was informative but humorous on Instagram because "normally the people who follow me, they expect me to not be so serious about stuff."For instance, when browsing misinformational posts on the social media monitoring dashboard, she interpreted them through a humorous lens of how ridiculously false they seemed, and described how she might reframe and share one such post "in the sense of, 'I can't believe they said this' type of thing [. . .] something to say, like, basically watch out for stuff like this." Even with taking these careful approaches, Peer Champions worried about not being listened to, or potentially receiving backlash when engaging vaccine hesitant people-perhaps especially with people they were closer to.P5 spoke from experience of having been dismissed before for not having a satisfactory counter-explanation to misinformation: A lot of the time people, if they tell you misinformation [. . .] They'll be like, "oh, it's because of this," or it's a very medical thing.But [. . .] I can't explain it as well, because I'm not in that field yet.And so they're just like, "Oh, then you don't know what you're talking about," even though they also don't know what they're talking about, but they think they're right.[laughs dryly] P8's similar fear of being "shut down, " especially from people she knew, kept her from actively engaging others about misinformation.When asked whether she felt confident responding to people about vaccine misinformation, she explained that she did not: It's really, really hard.And nerve-wracking.Especially if it's someone you know.[. . .] If I respond, I'm scared of them shutting me down completely for everything else.It's very polarized, and it's very scary.I don't really do it.As much as I'd like to, or at all really.P13 also was worried about backlash to promoting the vaccine on his own social media accounts.P13 described himself as an active social media user who posted content across various popular platforms, and had participated before in outreach campaigns with local organizations about issues including voting and housing.However, with the COVID-19 vaccine-something that he himself used to be hesitant about, and saw as a politicized topic-he was more wary about "the backlash on information, " and not being the right person to speak confidently about the vaccine: We only got but so much information.I can't sit here and say I got all the information, just as well-informed experts can't say that.I'm not the scientist, I'm not the person that's actually developing these vaccination[s].As such, he explained that "I guess I just don't want to be the messenger" directly of scientific information per se, but rather made the distinction that he would "help contribute to spreading the message" that scientists have to share, with "us as a group, we do our research on that just to solidify what we're spreading, and then I could be one of the people that posts it on my social media platform." Ultimately though, he said that "I guess I don't want to just change my whole platform to be a spokesperson for COVID." These different approaches demonstrate how the Peer Champions varied in the extent to which they wanted to explicitly identify, and be identified by others, with public health as an institution-a tension arising from their in-between position of wanting to promote public health, but also wanting to respect and preserve relationships, especially given the degree of disinterest and distrust they knew some people had of COVID-19 public health guidance.

DISCUSSION
Many health communication efforts with marginalized communities, where people may distrust health authorities and be exposed to racialized and politicized misinformation, center on working with trusted messengers in the community.Part of the distinct role of these trusted messengers is to be present in and have care for both their communities and the realm of public health; and to move between these two domains in order to provide linguistically and culturally appropriate health communication, especially for community members who are less literate in or trusting of public health.Our findings showed how our Peer Champions experienced this in-between positioning and leveraged it in their roles as trusted messengers to address COVID-19 misinformation and vaccine hesitancy in their communities.In particular, we characterized their extensive use of digital platforms including social media and search engines to learn about vaccine hesitant perspectives, research COVID-19, and communicate with others; and how their positionality shaped how they interpreted the (mis)information they found online and how they engaged others about it.However, they faced difficulties and tensions that stemmed from this intermediary position and wanting to be authentic to both public health and those in their communities.These challenges also manifested online, such as their limited access to online misinformation and confidence in reliably identifying trustworthy online information; accordingly, many of the opportunities identified for digital tools including but not limited to our dashboard involved helping them mitigate these issues.
To delve deeper into our findings, we turn to cultural code-switching especially as practiced by people of color in the U.S. to provide a conceptual framing for the Peer Champions' experiences and work as trusted messengers.Through this lens, we can understand their in-between position as enabling them to skillfully "code-switch" to interpret and communicate health (mis)information between their home communities and public health institutions.Their challenges in being able to fully access these different domains and maintain their authenticity between them can also be seen as analogous to the tensions faced by code-switchers.In this section, we further discuss our findings on the role and position of trusted messengers using the concept of cultural code-switching, relating it to other literature on community advocates and discussing implications for (digital) health communication; offer several considerations for designing digital tools to support the work of trusted messengers; and lastly, reflect on the impact of the pandemic and infodemic on the motivation and morale of trusted messengers.

Trusted Messengers and Cultural Code-Switching
Just as Braithwaite et al. describe community health workers as code-switchers for their community members in interfacing with the formal healthcare system [14], we found that our Peer Champions also acted like code-switchers in advocating for public health in their communities.Their belonging in and cultural familiarity with their communities enabled them to listen to and understand the health concerns and misinformation in their communities, while their participation in and ability to parse the jargon of public health research and practice allowed them to research information about COVID-19-that is, they could speak and perform the cultural "codes" of both worlds to access and gain this knowledge about their communities and about public health.Their interpretation of health (mis)information through the lens of their communities and being able to craft linguistically and culturally appropriate health messages and outreach strategies in response further demonstrates their ability to "code-switch" and communicate between their communities and public health.It is this combined set of memberships and skills that uniquely enabled the Peer Champions to do the difficult work of bridging deeply rooted gaps of misunderstanding and distrust between their communities and public health institutions, setting them apart from less health literate community members or non-community-member public health professionals.
In seeing this in-between situation as a strength, we echo scholarship which recognizes codeswitching as a legitimate skill and strategy for bringing resources to one's community [49], rather than incorrect or inappropriate behavior [4]-even if the necessity for this strategy comes from the unjust present reality of structural inequities [93].Our work contributes to the wider body of literature which seeks to unpack and legitimize the work of "lay" people in public health, especially when engaging marginalized communities.In their work with frontline health workers in India, Ismail and Kumar similarly conceptualized the workers as being positioned at the intersection of different communities of practice-including the healthcare system, their families, the communities they served, and the online world at large-yet on the peripheries of each [62].However, they also choose to frame the frontline health workers' peripherality as a position of power, and point to the potential for digital technologies to support, leverage, and extend their efforts to take advantage of their intersectional situation.We showed how our proto-professional young adult Peer Champions' ability to "code-switch" from a similar in-between positioning for culturally appropriate health communication is especially important for dealing with the heavy and sensitive issue of misinformation and mistrust in marginalized communities, and begin to explore ways for social computing technologies to support people doing this kind of work.
Just as code-switchers from minoritized groups bear the risks of being labeled inauthentic and straining relationships [35,93], we also found that the Peer Champions experienced tensions in how to best present COVID-19 information to community members in a way that felt authentic to and respectful of their personal image, community values, and relationships.These tensions were heightened by the politicization of COVID-19 vaccination, historically-rooted distrust in health institutions, cultural norms of respecting older relatives, and people simply being tired of COVID-19.However, these are not necessarily "problems" to be "solved" [29], and stem in part from both long-standing societal inequities as well as the Peer Champions' intersectional identities and commitments as young adults, from these communities, in the pursuit of higher education, and with professional aspirations in health and other socially-oriented fields.In their research on how youth members of a youth empowerment organization use social media for community organizing, Irannejad Bisafar et al. similarly noted how the youth organizers experienced "shifting attachments" away from the broader community of youth, such as by being negatively stereotyped by their peers for working with the police, and began forming a separate "public" from them [60].In response, they carefully managed the visibility of their alignment in the form of likes and mentions on social media-strategies similar in sentiment to how some Peer Champions wanted to shape their promotion of COVID-19 vaccination on social media to be more or less overt.Rather than offer a technological solution, we strive for solidarity and creativity to support such advocates' navigation within these tensions and towards an alternate future.
While framing the Peer Champions' activities as code-switching captures many aspects of the skills they leverage as well as the challenges they face, we want to also consider how they may move beyond code-switching as a necessity to serve their communities [19,52,64], (re)creating public health to be more genuinely inclusive of marginalized groups.Our broader CBPR project-funded by a government public health institution, convened by an HBCU medical school with a longstanding dedication to community engagement [14], and comprising a team led by and predominantly made up of people of color (including the Peer Champions)-centered the perspectives and legitimate concerns of communities of color.While this paper primarily discusses how the Peer Champions communicated public health expertise to their communities, they also did significant work to bring their communities' perspectives to public health research.In addition to the Peer Champions' own processes of listening to and understanding the concerns and misinformational beliefs of community members, as described earlier, their work also included community surveys and key informant interviews with the goal of informing stakeholders such as local health departments and communications consultants to better target vaccination campaigns to Black and Latinx young adults in the state, and to serve as the basis of an academic publication in public health scholarship.Furthermore, the Peer Champions brought their own perspectives to public health research through their participation in project meetings and co-defining of project activities.This bi-directional flow of learning between local communities and academic research while building the capacity of community-based researchers are core tenets of CBPR [11].Our framing of the Peer Champions' work also resonates with Sum et al.'s research which characterizes Asian American and Pacific Islander CBO workers as doing translation work between their communities and hegemonic systems including government institutions and digital platforms, negotiating for legitimacy and resisting the narrative of their communities as "other" [121].We encourage future research and practice in health communication and social computing to employ and raise up community-based trusted messengers and other infomediaries as a key strategy to center marginalized voices and move towards digital health equity.

Considerations for Digital Tools for Trusted Messengers
Our social media monitoring and outreach dashboard is an early exploration of digital tools designed for community-based trusted messengers.In this study, we used it as a conversational probe in our interviews to elicit insights from our Peer Champions about their practices, challenges, and opportunities in using social media and other digital tools for community-based health promotion.Although we did not originally design it with the framing of cultural code-switching in mind, through our observations of the Peer Champions' interactions with the dashboard and conversations with them about it, we found that trusted messengers could leverage social media monitoring and communication resource curation to support their culturally informed efforts to learn about health (mis)information and to communicate effectively.We also offer several other possible functions and directions for social computing and digital tools for community-based trusted messengers.

Social
Computing to Facilitate Community-Oriented Understanding.A major application of social computing which we investigated was social media monitoring.The goal of social media monitoring is often to leverage social media as a distributed, real-time data source to provide awareness of issues to stakeholders and enable appropriate response [114].This includes activities like identifying emerging online misinformation to promptly inform debunking efforts [102,114]-something which the Peer Champions did see value in.Using the frame of cultural code-switching, we propose additionally that social media monitoring can help expand trusted messengers' visibility of diverse perspectives online relevant to their communities, and support their culturally-informed interpretation of online misinformation as we observed the Peer Champions doing.
One challenge the Peer Champions faced was their limited ability to use their personal social media accounts to find misinformation and vaccine hesitant content, even when they knew others in their communities were being influenced by such online content.Their limited participation in the same online communities as vaccine hesitant local community members points to the intersectional differences in identities and characteristics that they as code-switchers might have in health literacy, English literacy especially in Latinx communities, educational level, religiosity, age and generation, among other dimensions [49].To help first locate relevant online spaces for social media monitoring, prior research has explored the potential for social computing technologies and other strategies to make "filter bubbles" more visible to social media users [32,96,103], which may be useful for trusted messengers to better understand the online spaces they and their community members do and do not participate in.Also, as people in local communities are present in various online communities and platforms, it is important that social computing tools for trusted messengers also operate across platforms [104,114].Unfortunately, as others have also called out [83,95], various companies restrict researchers' ability to collect and analyze social media data from their platforms, including limiting information on semi-closed or smaller online groups, messaging apps, and hyperlocal content engagement metadata that would be particularly relevant for monitoring misinformation affecting minoritized communities.We thus call for platforms to enhance their transparency measures, and for researchers to explore more avenues for cross-platform work.
When viewing collected misinformation and other social media content, the Peer Champions did not only take note of the topics being discussed, but also interpreted the content through their communities' cultural lens, imagining how someone they knew or a hypothetical vaccine hesitant person might react to the content.The Peer Champions also took this interpretive approach when looking at health communication resources for responding to misinformation, evaluating them not just for accuracy and ease of reading, but based on how they might resonate emotionally and culturally with community members.In doing so, they demonstrated their ability to simultaneously leverage their familiarity with health (mis)information and community cultural "codes." We thus propose that social monitoring and other social computing tools for community-based trusted messengers should aim to support this culturally-informed interpretive work that they do.We contrast this more interpretivist approach with how social media monitoring is commonly canonized in public health as part of "infodemiology" and "infoveillance" [36], invoking the metaphor of predominantly positivist-oriented epidemiology and disease surveillance.Prior work for example has used quantitative methods including natural language processing techniques and analyses of online activity metadata to characterize social media users who spread and engage with misinformation [41,69].While the goal is often to more efficiently identify misinformation online at scale, such tools could also help trusted messengers pick up on linguistic and other online behavioral cues to develop a better contextual understanding of culturally-specific misinformation and vaccine hesitant perspectives.Technologies that leverage natural language processing should also take care to account for the multiple languages or language varieties like AAVE that are pertinent to communities of interest in order to better characterize culturally layered content like health misinformation [28,48,83].However, especially with the rise of generative large language models, it is important to recognize the centrality of trusted messengers in doing the nuanced and often sensitive work of cultural interpretation and health communication, and for digital tools to support rather than replace them [40,112].

Collaborative
Design and Curation of Digital Platforms.CBPR and other allied research approaches such as participatory design hinge upon equitable collaboration and trust between community partners and academic researchers [11,46,74].Based on our findings and experience with this project, we further highlight why these principles are especially important when working with trusted messengers dealing with misinformation in historically marginalized communities.
The Peer Champions pointed to our dashboard's straightforward labeling of content and easily accessible trustworthy resources as some of the most potentially useful features for their work, cutting down on the uncertainty of browsing the wider internet.While their cautious distrust of online content was driven by their skepticism of the intentions of online platforms and actors, especially from the perspective of marginalized communities, their trust in our dashboard's labels and content was predicated largely on their trust in us and sufficient knowledge of the dashboard's functioning.While our ability to involve Peer Champions in a more participatory manner throughout the dashboard's design was limited by several factors, we invited questions and feedback at various points (including these interviews), making clear that we valued open and exploratory collaboration.Common questions that specifically focused on trust included asking how social media accounts were determined to be frequent misinformation spreaders, and how resources from potentially unfamiliar organizations were deemed trustworthy.More broadly, members of the dashboard design and development team (including the authors) were regularly present at general project meetings, building rapport that enabled trust.
In addition to fostering trust through participation, leveraging the contextual knowledge of trusted messengers in curating the content of digital platforms-in our case, social media data collection, health communication resources, and the set of labels and filters available-is crucial to ensuring relevance to their communities and thus their work.Recognizing the significant variation in information needs and cultural preferences within communities, between individual trusted messengers, and as different situations require, having a diversity of content that can be systematically navigated allows for flexibility in the trusted messengers' tailoring of their listening, research, and outreach efforts.

Motivation and Morale
While the Peer Champions were strongly motivated by their desires to promote the health of their communities and participate in public health research, they too shared the feeling of being "tired of COVID."Some of this stress may stem from the same reasons they identified in their peers: experiencing loss and disruption in their own lives, and being overwhelmed with information and uncertainty which is itself shown to impact mental health [91].Other underlying reasons may be more tied to their roles as Peer Champions: being busy with this work on top of their other responsibilities as students and members of their communities, being discouraged seeing those close to them drawn to misinformation, and the anxieties they had regarding backlash and straining relationships.These burdens can be related to the social, mental, and emotional stresses faced by people of color who may code-switch for professional and community advancement [87].
Similar concerns over burnout and mental health especially over the COVID-19 pandemic have been raised of frontline health workers [107] and non-medical frontline workers like social workers [53] and teachers [7].This strain can be especially felt by those working in marginalized communities that they are from themselves, repeatedly encountering the inequities striking their own communities.In misinformation work, mental health has also been a noted challenge for journalists covering misinformation who are overwhelmed in their attempts to report trustworthy information [92] and social media content moderators who repeatedly encounter distressing content [119].Recommendations include provision and normalization of counseling and other mental health resources, adjusting workloads, and encouraging open discussion of these issues.We similarly call for attention to these considerations for trusted messengers as frontline workers in public health and misinformation countering in their own right.
Despite these challenges, the Peer Champions appreciated "being around like-minded people, " and hoped that the COVID-19 situation was improving in their communities and that they were playing some part in it.While this sense of connection to their own Peer Champion community and the broader mission cannot be solely relied upon to address these issues, it is an important consideration for the sustainability and resilience of CBPR alliances.This sentiment echoes bell hooks' writing of her experience being a Black woman from a working-class background in academia [56] Those of us who live, who "make it", passionately holding on to aspects of that "downhome" life we do not intend to lose while simultaneously seeking new knowledge and experience, invent spaces of radical openness.[. . .] For me this space of radical openness is a margin-a profound edge.Locating oneself there is difficult yet necessary.It is not a "safe" place.One is always at risk.One needs a community of resistance.

STUDY LIMITATIONS AND FUTURE WORK
Our study was conducted in a particular context of issues and partnerships which color the nature of our findings.First, our study-as with much recent work on misinformation-focused on the issue of COVID-19 vaccine hesitancy during the COVID-19 pandemic, a uniquely broad global crisis during which information and guidelines evolved rapidly.This shaped the challenges faced by the Peer Champions as trusted messengers, for instance potentially driving greater challenges of information overload and fatigue, and lesser problems of general awareness compared to other slower-moving or less prominent issues vulnerable to misinformation like parental hesitancy of childhood vaccinations.Future work with trusted messengers addressing different issues should attend to how varied contexts and informational environments affect their work.
The scope of our engagement with the Peer Champions was also limited in several ways.The broader CBPR project that this study was a part of was a one-year project.While we hope to continue collaborating to build on this work, our engagement with these particular Peer Champions was limited to this time period.Longer term engagement with a group of trusted messengers may allow for observing how trusted messengers' positionality and approaches evolve over time.We also did not yet have the chance during this project to engage the Peer Champions in sustained active use of a more fully developed dashboard.Future research involving deployment is needed to understand how trusted messengers would integrate social computing technologies into their practices.Additionally, future research aimed at evaluating the effectiveness of trusted messengers' outreach efforts and of social computing tools' ability to support their work should also involve gathering community members' perspectives.
Finally, we acknowledge that our findings and choice of cultural code-switching as a framing may be shaped by the particular Peer Champions that we worked with, and by the design of the dashboard that we engaged them with.Our findings however still provide insight into how trusted messengers may position themselves with respect to their communities and institutions, and how they might leverage social computing and information-seeking technologies in their work.We encourage wider exploration of the range of experiences of diverse trusted messengers and the potential of alternative digital platforms and tools.

CONCLUSION
Community-based trusted messengers are important partners in equitable health communication efforts to address vaccine hesitancy and misinformation in marginalized communities.Our study interviewing Black and Latinx young adult trusted messengers, who we worked with as part of a broader CBPR engagement, sought to better understand the nature of their work and how they did and potentially could use social computing and digital platforms, using a social media monitoring and outreach dashboard that we designed as a probe to do so.Using the concept of cultural codeswitching as a framing, we found that the Peer Champions leveraged their particular combination of cultural, health, and digital literacy skills to operate between their communities and public health research and practice, understanding and addressing vaccine hesitancy in a culturally responsive manner.However, they faced challenges that echoed their joint commitments to public health and community values, including their own limited exposure to online misinformation, skepticism of online information, and tensions around navigating authenticity and interpersonal relationships with vaccine hesitant community members.We thus point to opportunities for social computing technologies to support trusted messengers that take a more community-oriented interpretive approach to facilitate their culturally-informed communication efforts, collaboratively designed to invite their trust and agency.With the emerging promise of "lay" community advocates using social media to promote public health and social change more broadly, our research contributes towards characterizing the valuable and often difficult work they do, the challenges they face, and (re)imagining the ways digital platforms and social computing tools can figure in their work to advance health equity.

Fig. 1 .
Fig. 1.Screenshot of the live social media monitoring page of the dashboard, with filters (on the left hand bar), search (on the top bar), content and account labels (text below each post), and outreach-focused features (buttons below each post and right side bar).Account information and post content have been blurred and an overlay added to posts containing misinformation in this figure (not present in actual dashboard for the study).

Fig. 2 .
Fig.2.Two screens from the prototype of a health communication resources page of the dashboard and relevant resource recommendation feature.The screen on the left shows a modified version of the social media monitoring page, with one selected vaccine hesitant social media post shown.Clicking the "Look at Relevant Resources" button below the first social media post on this screen brings the user to the screen on the right.This screen shows a prototype of the resources page, with three selected resources for responding to the earlier social media post.Account information has been blurred and an overlay added to the post containing misinformation in this figure (not present in prototype for the study).

From
when I'm watching the news and hearing conversations, a lot of people [. . .] have lost trust in sources that I might think credible [. . .].So I think people would like to see [. . .] information that links to each other, they support each other, for more than one site, because a lot of people have lost faith in what maybe the CDC might say today and it might change tomorrow.[. . .] I've heard people say they [the CDC] don't even know what's going on.

Table 1 .
Study Participants 3.5 Interview ParticipantsAll Peer Champions were invited to participate in interviews with us, and were provided additional financial incentive.A total of 21 Peer Champions participated in the broader project; we recruited 13 of them for these interviews, with at least one Peer Champion from each partnering institution participating (Table