The spokesperson matters: evaluating the crisis communication styles of primary spokespersons when presenting COVID-19 modeling data across three jurisdictions in Canada

Abstract Risk prevention measures are more likely to be accepted if people trust risk managers and their ability to handle the crisis, which often depends on who communicates with the public. During the COVID-19 pandemic, some Canadian officials became the main spokespersons of pandemic response in their jurisdiction, speaking almost daily to the public. We evaluated how the primary official for each jurisdiction chose to communicate about epidemiological modeling with the public and how they used modeling data to support their pandemic decisions. We conducted textual and visual analyses of press conferences held in British Columbia, Manitoba, and Ontario. Then, we asked focus group participants who they trusted the most and the least for information on COVID-19. We identified two main communication styles: compassionate-informative and condescending-evasive. Spokespersons following the former demonstrate a trust-building effort by providing straightforward answers, demonstrating expertise, while showing empathy and risk management competence. Those who followed the latter style predominantly offered superficial and defensive responses, engaging in blame-shifting and politicizing risk. Focus group participants trusted most the spokespersons who follow a compassionate-informative style are considered trustworthy, which could increase compliance with public health measures. However, those who use the condescending-evasive style were seen as less trustworthy. Our results underscore, first, the importance of disassociating political agendas from risk communication and emergency response during public health crises. Second, spokespersons should be trained in risk and crisis communication to engage with reporters and the public positively. Finally, crisis communication should emphasize the scientific evidence behind guidelines, while acknowledging scientific uncertainty.


Introduction
Since the beginning of the COVID-19 pandemic, provincial governments in Canada have engaged in intense and ongoing communication efforts, with almost daily briefings as officials informed the public about the rates of infection, epidemiological models, and new measures to contain the spread of infection. Adoption of recommended behaviours by the public -such as practicing hand hygiene, physical distancing, and getting vaccinated -is essential to control the spread of disease. Recommended risk reduction measures are more likely to be accepted and practiced if people trust risk managers and have confidence in their abilities. That trust often depends on who is communicating with the public.
As provinces began producing models to help them adapt their response to the spread of COVID-19, the issue of whether provincial authorities should make model data public became highly debated and contentious. Epidemiological models are complex mathematical and statistical calculations developed by scientists to track and predict the progress of infectious diseases within populations (Tariq, Haworth-Brockman, and Moghadas 2021). These predictions vary depending on the pathogen and the behaviour of the population. Models can help determine how effective an intervention will be, for instance, wearing facemasks or maintaining physical distance. These models are particularly important when we are dealing with a novel disease, such as COVID-19. The reliability of these models depends on several variables including how much is known about the disease's epidemiology and the amount of data available. COVID-19 epidemiological models in Canada gained much media attention and quickly became politicized as reporters, opposition leaders, and experts demanded provincial authorities publish their models and framed it as an issue of transparency ). However, many provincial governments were reluctant to share modeling data with the public at the beginning of the COVID-19 pandemic, arguing that without proper contextualization the models would create unnecessary concern. Eventually, one by one, the provinces started publishing modeling data on a regular basis.
Drawing on best practices from risk and crisis communication literature, we evaluate the verbal and visual communication strategies used by the primary spokespersons for three Canadian provinces, who were responsible for most of the messaging about the pandemic response in their jurisdictions. We analyse how these spokespersons communicated epidemiological modeling with the public and how they used modeling data to support their pandemic decisions. In Manitoba and British Columbia, the primary spokespersons were public health experts, whereas in Ontario, it was the political government leader that opted to serve as the primary spokesperson.
Risk research has focused on social trust in risk/crisis management and risk/crisis communication as a key condition for public acceptance and adoption of recommended behaviours (cf. kasperson, Golding, and Tuler 1992;Renn and Levine 1991;Slovic 2000). For example, Siegrist and Bearth (2021) found that risk perception and trust, along with values, shape public acceptance of COVID-19 preventive measures. In their Risk Information Seeking and Processing (RISP) Model, Griffin, Dunwoody, and neuwirth (1999) contend that individuals determine trustworthiness and usefulness of the message based on their assessment of the source. These assessments for trustworthiness and usefulness influence information seeking behaviour and the extent to which individuals may critically analyze the content of that risk information. Trust also has a central role in dual processing models for processing risk information, which feature intuitive/emotional and rational/calculative processes (Finucane et al. 2000;McAllister 1995;Slovic et al. 2004;Zajonc 1980). In their Trust, Confidence and Cooperation Model (TCC), Siegrist and Zingg (2014) conceptualize trust and confidence as determinants of how people engage with risk messages. In this model, trust is defined as the perception of shared values (i.e. a person may consider an elected official or health expert trustworthy if they believe the expert or official is sincere and has similar values, such as preserving public health). Confidence is understood as trustworthiness based on the source's past performance (i.e. how a previous health crisis was managed).
Risk researchers have long-argued that the choice of spokesperson should be carefully considered, as perceptions of trustworthiness can affect compliance with risk prevention guidelines (Abraham 2009;Freimuth et al. 2014;Siegrist and Zingg 2014). Public health guidelines are more likely to be accepted and adopted if people trust risk managers and have confidence in their abilities (kasperson, Golding, and Tuler 1992;Slovic, Flynn, and Layman 1991) and the sources of information are reputable. Lyu et al. (2013) found that spokespersons that demonstrate professional competence and have positive interactions with reporters are more trusted than those who are confrontational in press conferences, blame the media for reporting on their mistakes, and who lack authority to make decisions. Clear and consistent messages are also crucial for successful crisis communication. In their study of the anthrax crisis, Barrett et al. (2016) found that the Centers for Disease control and Prevention (CDC) had dozens of spokespersons talking to the media, which led to confusing and contradictory messages. Furthermore, trust and credibility are demonstrated and fostered by spokespersons by showing empathy and caring, competence and expertise, and honesty (Reynolds and Quinn 2008). The spokespersons' pre-existing reputations will also influence perceptions of credibility (Cairns, de Andrade, and MacDonald 2013;. Messages are more persuasive if the public likes the spokesperson (Tucker Smith and De Houwer 2014), they are perceived as 'better' or 'stronger' messages (Roskos-Ewoldsen and Fazio 1992), and they are more likely to be shared with others (Abu-Akel, Spitz, and West 2021). Researchers have equally shown that people are more likely to trust messages from reputable organizations, credible individuals, and physicians (Hackett 2008), but these sources must have a pre-established reputation (Chong 2006).
Conversely, trust and compliance can be negatively influenced if the risk message refers to individual responsibility and blame, inducing feelings of guilt, shame, and frustration (Coleman and Hatley Major 2014;Guttman and Ressl 2001), particularly when individuals may not have had the resources to fully comply with risk avoidance measures (Guttman 2017). Communication styles can also affect public trust, with paternalistic frames causing public rejection, while educational approaches can provide information but is not always effective (Bester 2015). Visual communication has proved effective for improving public understanding of risk communication (Sleigh et al. 2021), particularly when combining visual and linguistic representations (Houts et al. 2006).
Designing strategic risk management plans, including projections of future trends, and having consistent messages can increase trust and credibility (Cairns, de Andrade, and MacDonald 2013;Chong 2006). Additionally, clarity, openness, and honesty are recognized as key characteristics of effective crisis and risk messages, whereas if the spokesperson is perceived as dishonest, having poor emotional control, or having political motivations, these characteristics can undermine the message (Lee 2009;Lyu et al. 2013). In this study we examine how three Canadian officials communicated COVID-19 epidemiological modeling, a topic fraught with scientific uncertainty, and how they tried to promote trust. Previous studies have demonstrated that in risk or crisis situations spokespersons with expertise and credibility are more effective, and in this study, we evaluate the verbal and visual communications strategies used by the primary spokespersons who were responsible for providing the bulk of the communication messaging to the public in three jurisdictions. In Manitoba and British Columbia, the primary spokespersons were public health experts, whereas in Ontario, it was the political government leader that opted to serve as the primary spokesperson. We argue that beyond previous credibility of the spokesperson, there are specific communication strategies that can increase or decrease trust and credibility of the message. When two spokespersons have equal levels of expertise and credibility, using specific communication strategies to show empathy and care increase trust and likeability.

Methods
In this study, we examine how three Canadian officials, the 'face and voice' of the pandemic response in their jurisdictions, communicated about epidemiological modeling and how they used the models to increase support of their decisions. First, we did a content and visual analysis of press conferences held by three Canadian officials who were mentioned by focus group participants, in which they discussed epidemiological models. Second, we conducted focus groups as part of a wider study in which we asked participants who they trust as sources of information on COVID-19. We are guided by four main questions: RQ1: How were epidemiological models communicated? RQ2: Were pandemic-response decisions supported by the epidemiological models presented? RQ3: How did the speakers seek to promote trust? RQ4: What trust and confidence did spokespeople engender with the public?

Analysis of press conferences
We examined press conferences discussing epidemiological modeling in three Canadian provinces to assess the different communication styles adopted by the spokespersons. We focused our analysis on press conferences held by the authorities who regularly informed citizens about COVID-19 case numbers and measures adopted to contain its spread in Ontario, Manitoba, and British Columbia during the first and second waves of COVID-19 in Canada. These officials therefore became the 'voice and image' of the pandemic response in each of these three provinces. We chose to analyse the press conferences of spokespersons in Ontario, Manitoba, and British Columbia as these are the provinces where we also conducted focus groups as part of a larger research project on risk communication of COVID-19. One of the spokespersons is an elected official, while the other two are public health authorities.

Data collection
We collected videos from the press conferences held by Doug Ford, Premier of Ontario, Dr. Bonnie Henry, Provincial Health Officer for British Columbia, and Manitoba's Chief Public Health Officer, Dr. Brent Roussin. Data collection was performed through the online video platform YouTube. We chose to collect our sample from YouTube because the platform allows users to maintain robust video archives. network websites, conversely, tend to have limited archives, which would not have allowed us to access a complete sample to collect our dataset. We searched the YouTube channels of three Canadian news networks -CTV news, Global news, and the CBC -using combinations of keywords and the names of the three officials. The search terms were '[name] Epidemiological model' , '[name] Modeling' , and '[Province] model' .
For the first wave, data were collected within the time frame from when the respective jurisdiction declared a state of emergency to 30 April 2020. Both Ontario and British Columbia declared a state of emergency on 17 March 2020, and Manitoba followed suit on 20 March 2020. For the second wave, the data collected include press conferences between 1 november 2020 and 30 January 2021, which corresponds to the second wave of infection in Canada.
Relevant videos of spokespersons press conferences were included in the final dataset if one or more of the following criteria were met: (1) the video included the public health officer or elected official presenting modeling data at a press conference or answering questions on modeling data from reporters; (2) a clip from a press conference where the health officer or elected official spoke about epidemiological modeling was included in a story from the news outlet and the network provided commentary on the clip; or (3) the public health officer or elected official used modeling as a justification for public health measures. The sample also included videos of news stories about the models in which the officials were directly quoted or featured. These were produced by the national broadcasters CTV, Global, and CBC and included clips of the officials discussing the models in press conferences. We examined these news stories in terms of how they framed the officials' statements. Transcripts for these videos were automatically generated by YouTube and were, then, audio-reviewed for accuracy and corrected when necessary.

Content and visual analysis
Video transcripts were uploaded to nVivo12 and coded by two team members. We developed four principal codes: topic, tone, promoting trust, discouraging trust (see Table 1). Subcodes under 'topic' included discussion of the scientific process or evidence, shifting guidelines, discussion of hospital capacity (including capacity in general terms, capacity in critical care units, capacity in intensive care units, number of ventilators available, and personal protective equipment availability), comparing with national context, and comparing with international context. We coded 'speaker's tone' as neutral, defensive, doubtful, friendly, evasive, frustrated, and confident. 'Trust promotion' was assessed in how a spokesperson indicated a level of personal responsibility for the result of past decisions, if they referred to being transparent, if they showed expertise, if their message was consistent, and if they referred to visual aids when presenting the models. Finally, 'discouraging trust' was assessed if a spokesperson showed a lack of expertise, if there were various changes in speaker to answer one question, disorganized presentation, inconsistency, and assigning blame to others. We followed an open coding process allowing new codes to emerge from the data during the analysis.
Visual analysis of the videos focused on different categories: description, subject matter, form, medium, style, genre, and comparison (Schroeder 2006). We examined the corpus of videos for each spokesperson in order to assess their use of visual aids (e.g. slide presentations, graphs), their body language, and the visual organization of the press conference (e.g. people accompanying the spokesperson, whether speakers were switched, if they were wearing masks or keeping physical distance, etc.).
Coding disagreements were resolved through discussion between coders and other members of the research team. For the sample subset where double coding was completed, the final overall Cohen's kappa coefficient for intercoder reliability was 0.87.

Focus groups
We conducted 12 mixed-gender online focus groups with Canadians in the provincial capital cities and in the national capital city (Vancouver in British Columbia; Winnipeg in Manitoba; and Toronto and Ottawa in Ontario) from 8 to 22 December 2020. Participants were recruited using a market research firm using a variety of methods (e.g. emails to individuals signed up as part of existing panels, random digit dialing, ads posted on Facebook/Instagram). We conducted three focus groups in each city where participants were age-segregated into one of three groups (18-34 years, 35-54 years, 55+ years) for each city, where at least two people were recruited in smaller age groupings within each category. For example, in the 18-34 age group, the market research firm needed to recruit individuals between the ages of 18-24, 25-30, and 31-34 to ensure a better cross-section of participants. Participants received an honorarium of $70 for their time.
Focus groups were moderated by trained facilitators experienced in qualitative methods, with the lead researcher in attendance but off camera to avoid influencing the discussion. Occasionally the lead researcher suggested follow up prompts to the moderator using the chat function of Zoom. All focus groups were audio-recorded, transcribed verbatim, and audio-verified for accuracy. Transcripts were uploaded for analysis using nVivo12. To identify participants in the transcripts, we used the name they requested. We developed an initial thematic guide for the focus group sessions in which the moderator asked participants' opinions regarding implementation of public health guidelines, compliance with infection prevention measures, information seeking behaviour and trust, and attitudes towards immunization in general and specifically the COVID-19 vaccines. In this article, we discuss participants' opinions on which Canadian officials are deemed trustworthy sources (or not) of COVID-19 information. Two team members coded the transcripts, and two coding tests were performed with a third member of the research team to ensure intercoder reliability. Our kappa coefficient score was 0.90. Ethics approval was obtained from the University of Manitoba and Toronto Metropolitan University (previously Ryerson University). Whenever the speaker engages in the blame game saying we're in a terrible situation because of X or Y group. changing speakers Whenever a new speaker, i.e. not the four individuals chosen for analysis, is holding a press conference of a new speaker is introduced Disorganized presentation Whenever the presentation is disorganized, i.e. many speakers, no consistent message, conference starts late, etc. inconsistency When the message is inconsistent with what was previously said or done. lack of expertise Whenever the speaker is not an expert on the topic they are discussing 2. speaker's tone speaker's tone and/or attitude cheerful speaker is joking, smiling, laughing confident speaker is confident, assertive Defensive speaker defends their choices or replies to questions in defensively Doubtful speaker seems when answering a question or does not know who should be answering the questions evasive speaker evades questions, redirects conversation to another topic, spends time speaking of tangential or even irrelevant topics. friendly speaker appears friendly, uses encouraging words, is assertive but not defensive or offensive frustrated speaker expresses frustration over lack of compliance, criticism, or anything else neutral speaker maintains an informative tone 3. Trust promotion Whenever the speaker sought to build trust consistency speaker's message is consistent with previous messages expressed in previous press conferences or by other speakers expertise speaker is an expert in the topic they are presenting on Personal responsibility speaker assumes personal responsibility for public health response reference to transparency speaker refers to need for transparency or expresses commitment to transparency referring to visuals When the speaker makes explicit reference to the visuals provided

Results
We present our results by addressing, first, visual elements and general organization of the press conferences in each province; second, by addressing the speakers' tone and use of language; and third, we examine the clarity and consistency of the message and how the speakers demonstrated or not expertise in managing a public health crisis; and finally, we examine whether and how speakers used model data in their press conference to justify their decisions. Our final sample included 92 videos. We, then, present audience-reception perspectives from our 82 focus group participants in Ontario, Manitoba, and British Columbia (see Table 2 for a description of our sample). Drawing on examples of participants' comments we address how individuals interpreted the press conferences and whether they trust the spokespersons. For a summary of the results see Table 3.

Visual elements and organization
In Ontario, the pandemic response -including the presentation of epidemiological models -is communicated to the public through televised press conferences held by the Premier, Doug Ford, and other members of his government. Most of these press conferences are held at the Provincial Government Office in Toronto. Premier Ford walks into the room accompanied by members of his government. He stands behind the podium, facing the journalists and cameras, and his companions stand in a V formation behind him, against a backdrop of black curtains and provincial flags. The camera angle is wide so that we can see all participants. Premier Ford is always the first to speak, and then, other speakers take turns as the Premier calls them to make statements or answer questions. Occasionally the camera angle closes in on the speaker. The change of speakers during these press conferences is managed theatrically. Premier Ford takes a few steps back to occupy the space where, for example, Health Minister Elliot was. Minister Elliot also takes a few steps back to occupy the place of another person, who now walks to the podium. These exchanges seem rehearsed as cabinet members look down to find markings on the ground indicating where they should stand, while trying to keep 2 m between each other. This format remains during the second wave of the pandemic, with the addition of facemasks and the removal of the American Sign Language (ASL) interpreter.
In the second wave of the pandemic, some press conferences were held in different settings. For example, one conference was held in a school gymnasium with the Minister of Education, in which Premier Ford discussed plans for a safe return to school. In other press conferences, the Premier is sitting at a long black table with members of his government and medical experts. The aesthetic of these conferences remains nonetheless consistent, with a background of blue curtains and provincial flags. Additionally, there are signs on the table with the logo of the Ontario COVID-19 Science Advisory Table (SAT), the group of experts who develop the epidemiological models for the province, on the table. This layout denotes parity among the officials and experts, as opposed to the hierarchical formation, which remains the most frequently used during press conferences held by the Premier and members of his government during the second wave. Press conferences held by Premier Ford seldom include visual representations of the model data, that is, graphs projecting infection rates.
In British Columbia, the pandemic response is communicated by the provincial Minister of Health, Adrian Dix, and the Provincial Health Officer, Dr. Bonnie Henry. Their press conferences always happen in the same space, a large room with a podium and a background of blue curtains and six provincial flags. Mr. Dix and Dr. Henry walk into the room together. First, Mr. Dix briefly provides a schedule of briefings for the rest of the week. Then, he introduces Dr. Henry. As she walks to the podium In Manitoba, press conferences informing about the pandemic response are held by the Chief Provincial Public Health Officer, Dr. Brent Roussin, who provides the public health response, usually accompanied by Lanette Siragusa, Chief nursing Officer of Manitoba, who provides the health system response. Other times Dr. Roussin holds the press conference alone or accompanies Premier Brian Pallister. With a background of dark curtains and provincial flags, Dr. Roussin and other speakers sit at a long table, and the camera shifts from one speaker to the next. The camera angle starts wide and later closes to show just the speaker and, sometimes, also a sign on the table with a website address where to find more information about COVID-19. Dr. Roussin also uses a slide presentation to show different graphs and data to facilitate understanding of the models.

Tone and use of language
Premier Ford, in Ontario, announced in April 2020 that he would make the epidemiological projections public, arguing that Ontarians 'deserve to see the same data that I see when I'm making decisions' (2 April 2020). In doing so, the Premier presented himself as a transparent, trustworthy leader. When referring to epidemiological models, Premier Ford uses very colloquial language despite maintaining a grave tone. He warns about the seriousness of the situation by saying the models are 'very concerning' and 'stark' . The Premier speaks slowly with frequent pauses. He outlines the measures adopted by his government. He encourages people to comply, for example, by asking, 'Is a life worth having a few [beers] with your friends?' to encourage people to avoid gathering.
Additionally, Premier Ford frequently compares Ontario's performance to other provinces and countries, not necessarily providing context but highlighting that his government's pandemic response is successful. For example, he frequently expresses pride by saying, 'We are the first province to….' , 'We're doing everything we can' , and 'My government is ready to do anything necessary' . The Premier usually talks with optimism about Ontarians as resilient people, where Ontario is 'the greatest place in the world with the greatest people in the world' .
During the first wave of infection, Premier Ford gained Ontarians' trust by empathising with them and referring to the hardship of the pandemic with expressions such as 'my friends, the hard truth is…' and 'in these hard days…' . He also highlighted his willingness to adopt any necessary measures, for example, by saying 'we're doing everything we can' , 'we were the first province to take measures' , and 'my government is ready to do anything necessary' . In the second wave, however, the Premier's tone changed. He was no longer somber, nor was he optimistic. Premier Ford no longer spoke about him or his government as responsible for the pandemic response. The camera angle was narrower, framing the Premier from the waist up. He looked closer and bigger on the screen. Premier Ford appeared visibly frustrated by the model projections and blamed Ontarians for the second wave of infection, framing responsibility as shared between Ontarians and the government. For example, on one occasion, he scolded Ontarians for not staying home, saying, 'As Premier, I can't accept that, and I won't let that happen' (Global news, 2020, november 13). Premier Ford responded with frustration to reporters' questions. For example, on December 22, some reporters questioned his decision to put the province on lock-down after Boxing Day (December 26) instead of starting the lock-down immediately. Premier Ford responded defensively, saying, 'last [lock-down], you complained that I didn't give enough time' .
In British Columbia, Dr. Henry's presentations are highly structured, always following the same order. She begins with a summary of the previous press conference, which helps relate previous information with new information. Then, she gives an overview of new cases of infection, outbreak management and announces recent deaths. She offers her condolences to the families, communities, and caregivers of the deceased. After that, Dr. Henry explains the initiatives and measures being implemented, the province's goals, and the progress made. Her language is accessible; her tone is serious but calm. She expresses empathy but avoids using colloquial language and does not show frustration. Dr. Henry reiterates the public health guidelines, like keeping physical distance and isolating but does not blame people for not always complying. Instead, she credits British Columbians for the low and manageable increase in cases and encourages them to keep up the effort.
In Manitoba, Dr. Roussin's presentations also follow a pattern, but they are not rigidly structured. He usually begins by updating reporters on the number of infection cases and deaths; then, he offers condolences to the family and friends of the deceased. Dr. Roussin uses plain language when presenting the models and explaining what the data mean. When referring to successes, such as low hospitalization rates and low infection numbers, Dr. Roussin thanks Manitobans for 'stepping up and following public health guidance' and thanks the laboratory staff for processing the COVID-19 tests.
Dr. Roussin speaks calmly, and his facial expression is serious; there is no panic or frustration in his voice or face even when reporters challenge his decisions. During the second wave, Dr. Roussin's calm tone continues. Still, his message became more serious as he conveyed the severity of the situation, for example, by saying 'we've pushed our healthcare system to capacity' or that the death toll 'are not just numbers; these are Manitobans' .

Clarity and expertise
Despite holding press conferences about epidemiological modeling, Premier Ford never presents or explains any data or uses any visual graphs during his press conferences. Instead, he indicates when the Science Advisory Table will release the information. For example, on 8 January 2021, Premier Ford announced that new models would be published the following week, and he claimed that 'the situation is dire' . When reporters asked him to explain what he meant, he replied evasively, resorting to general phrases like 'We have to knuckle down and get through this together' , and then, called the Minister of Health to respond. By responding evasively, he fails to articulate and demonstrate an understanding of how the epidemiological models are produced or their implications in how they are used to support public health decisions. In these press conferences, the Premier also seems more comfortable when responding to non-healthrelated questions, like matters related to production, supply chain, and other industry concerns. He gives detailed answers regarding how his government is supporting businesses.
In British Columbia, Dr. Henry explains what the models are for and how they should be interpreted as probabilities instead of predictions. Then, Dr. Henry discusses the possible scenarios based on the data and explains the measures being taken to prevent the worse situations. She also provides clear timelines for the publication of new epidemiological data and when reporters will have access to it. After her presentation, she answers reporters' questions with lengthy, explanatory answers. Dr. Henry demonstrates her expertise in every press conference by explaining the models and answering questions clearly and in plain language. She also puts the numbers into context and explains how adopting different measures would change the projections. On one occasion, for example, a reporter asked whether it was possible to achieve community immunity to COVID-19. Dr. Henry replied by explaining what community immunity is, why it is good, and why it is not the case for COVID-19.
Dr. Henry, however, openly admits when she does not know something, projecting not only expertise but also honesty. On one occasion, she was asked how many people were infected with COVID-19 in a particular long-term care home. She responded that she still did not have a number but that her team was filling that information gap. Her tone is never defensive, even when answering challenging questions from journalists. This sense of calm and emotional control is reinforced when she ends each press conference by telling British Columbians, 'Be kind. Be calm. Be safe' .
In Manitoba, Dr. Roussin also demonstrates expertise by explaining clearly and in everyday language the projections and interpreting the graphs. He is also straightforward in his explanations, showing the data and not trying to soften the situation. For example, in a press conference during the second wave, Dr. Roussin explained: In this slide, we're presenting Manitoba's daily number of diagnosed cases, this is that black line with circles on top of our four projected scenarios. The scenarios reflect the combined effect of public health measures and public behavior. Proper measures with good compliance should lead to fewer cases, anything less than that will lead to more cases. By observing Manitoba's actual data, we can identify which trajectory Manitoba's epidemic is following according to the model, and that shows that right now we're at a critical place. We're following scenario one, extreme (Global news, 2020 December 4).
Dr. Roussin responds to reporters' questions calmly, even when his decisions are challenged or when responding to the same question several times by different reporters. Several times during the press conferences, and particularly when answering reporters' questions, Dr. Roussin refers to the epidemiological models as tools for planning the pandemic response. For example, on 29 April 2020, during his first presentation of model data, some reporters asked why the projections did not go further in time. Dr. Roussin explained that 'modeling, again, is an abstraction. We can't project all possible outcomes. It's not a crystal ball' . He also explained that projections shift depending on various factors, including individual behaviour. Dr. Roussin always explains technical terms in non-expert language, such as R-naught, 1 and why it is not helpful to generate epidemiological models for long periods.
Dr. Roussin constantly recognizes Manitobans' effort in following public health guidelines and staying at home. He also said he would keep updating Manitobans on new epidemiological model data. Still, he only did once for the first wave of infection, on April 29, and twice for the second wave, on november 27 and December 4. This lack of communication contrasts with the frequent press conferences held in other provinces to present similar model projections. Dr. Henry and Premier Ford held press conferences where they referred to the epidemiological models 3 or 4 times per week during the first two waves of infection.

Use of model data
Evidence-based decision-making can improve trust in risk messages and managers. Premier Ford, however, does not refer to the model data during his press conferences to explain or support public health decisions. It is clear that he does not have that expertise and when asked about his decisions, he usually responds that the Science Advisory Table will address those issues later or by asking the Minister of Health to reply.
In contrast, Dr. Henry consistently uses the model data to explain shifts in public health guidelines and refers to the graphs used in her presentations to address reporters' questions. For example, in one of the press conferences, she reminded people to avoid non-essential travel. She explained that people are more vulnerable to COVID-19 in small communities and do not have enough resources should the visitor get sick or make others sick. It is clear from these press conferences that the production and interpretation of the epidemiological models are directly informing the province's pandemic response. Additionally, in every press conference, Dr. Henry announced new measures and reviewed the effectiveness of those already implemented by referring to how the models had changed. She also uses the model data to explain measures that could be adopted soon. For example, several weeks before the holiday season in December, she asked British Columbians to 'start thinking how we can help people celebrate in a safe manner' . Similar to Dr. Henry, Dr. Roussin refers to the epidemiological models to support the introduction or discontinuation of public health restrictions. When reporters challenge his decisions, Dr. Roussin replies by showing the data supporting public health decisions and explains that new guidelines will be implemented only when the data justify it. His responses are clear, and he uses the case numbers and the projections to support the province's pandemic response measures.

Who do Canadians trust?
In our focus groups, we asked participants about who (an individual or organization) they trusted, and conversely distrusted, for information related to COVID-19. The provincial public health leads (Dr. Bonnie Henry and Dr. Brent Roussin), who were the primary spokespeople for the pandemic in their respective provinces, were among the most trusted sources identified by participants. By contrast, the primary spokesperson in Ontario (Premier Doug Ford) was frequently identified as a distrusted information source for COVID-19.
In Ontario, while some participants approved of Premier Ford's management of the pandemic, most of them referred to the Premier as untrustworthy because, as a politician and businessman, he is perceived as having a hidden agenda and lacking the necessary expertise. This distrust was more commonly expressed by participants aged 55 and older. One participant explained that 'politicians are for politics, this is a medical concern, so I listen to medical professionals' and that when it comes to COVID-19, 'I don't listen to the politicians regarding this matter' (Roland, Toronto 55+). Another participant in the same group added The one person that I don't trust is Doug Ford. He has forbidden the health people who are not politicians; they are not allowed to talk to us. Everything is filtered through him. He's a politician, he's a businessman, he is a mini [President] Trump (…) he's got his own agenda that sets the tone for what he tells us. He thinks we're stupid. The last four years have certainly taught us about politicians and their shenanigans (Barbara, Toronto 55+).
Others said they trusted Premier Ford at the beginning of the pandemic, but as the months went by, so did their trust in him. These participants highlighted that Premier Ford repeatedly ignored expert advice and had not implemented science-based policies: I really liked Doug Ford, how he handled this pandemic early on, but lately, he's done nothing with long-term care facilities. It's just talk, and there should have been more action because now people are dying in those places, and they haven't changed anything. They've had six months to do something, and they haven't. He's not doing what the medical professionals are saying; he's late (Bruce, Ottawa 55+).
Others said they distrusted the information provided by the Premier because it was inaccurate or distorted. One participant noted that '[health officials] may relay [information] to Doug Ford (…), and it's like a game of broken telephone. By the time it gets to you, it doesn't make sense' (Shehzad,. In all of the Vancouver focus groups, participants frequently expressed trust in Dr. Bonnie Henry, by name, with several describing themselves as 'fans' (Murray, 55+) of the public health officer. Dr. Henry was considered trustworthy because she is 'the most informed, accurate, and reputable' (Tammy, 55+). Furthermore, most participants said they trusted all information coming from Dr. Henry. One participant said, 'whatever Dr. Bonnie Henry says. I'm following that' (Alpine, Vancouver 35-54), and another one explained, Yeah, I trust her as well. Her job is basically to represent science. I don't know specific scientists who are doing the science around COVID and are coming up with the information that we know, but she seems to represent them, and I trust that. (…) Usually, if it says Bonnie Henry in it, it was good enough to click on or pay attention to (Laura, Additionally, some participants explained that their trust in Dr. Bonnie Henry is based not only on her expertise but also on her communication style, which was described as personal and empathetic. A participant in the 18-34 focus explained, I trust [Dr. Henry]. I feel when I hear her putting out information, I trust it, and I tend to agree with what she is saying (…) I kind of do like the way she puts things and seems to kind of make sense and speaks to the human side of people, and it seems like she kind of understands the difficulties and stuff. I feel like she is speaking to me as a human being (Mason).
In Winnipeg, focus group participants referred to Dr. Brent Roussin as a trusted source of information on COVID-19 due to his medical expertise and the perception that, as a medical expert, he is not swayed by political interests. For example, one participant explained that 'I trust Dr. Roussin and [Canada's Chief Public Health Officer] Dr. Tam (…) because they have the degrees, schooling, rigour, and experience behind it' (Monique, 35-54). Other participants described Dr. Roussin as honest and transparent, 'a straight shooter' (Donna, 55+), and 'coming from a place of real integrity' (Jan, 55+). Furthermore, some participants also considered that Dr. Roussin follows the scientific evidence and does not have a political agenda I trust Brent [Roussin] and Lanette [Siragusa] because they're health officers at the top of the system and are there regardless of government's political leanings, but as soon as you hear from someone like [Premier] Brian Pallister or [Health Minister] Cameron Friesen, they have obvious political leanings. They have to make it to the next political election cycle if they want to continue in government, so I feel they always have a bias (…) Whereas Dr. Roussin is much more likely to say that the situation is extremely concerning -not trying to sugar coat it (Christian,(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34).
I trust Dr. Roussin and the Chief nursing Officer, Lanette. I think they are both doing a fantastic job, and I trust them completely because they are looking at data and the science behind the data. I think they wait to receive the data before making conclusions and other statements. If there is not enough data, they hold back on what they say (Raffaele, 35-54).

Discussion
During a health crisis, such as the COVID-19 pandemic, timely, understandable, and empathetic communication is critical (Roeser 2012). Choosing spokespersons with scientific credibility and training them to be effective communicators could be critical in slowing down the spread of infection. People will be more likely to adhere to public health guidelines if they deem the speaker trustworthy and charismatic, and qualified (kasperson, Golding, and Tuler 1992;Slovic, Flynn, and Layman 1991). This was recognized as a critical factor in handling the 2003 SARS outbreak in Toronto (DiGiovanni et al. 2004) and the 2009 H1n1 influenza pandemic Fineberg 2014).
In our analysis, we identified several characteristics of spokespersons communication of epidemiological models, which we grouped into two communication styles: Compassionate-informative and condescending-evasive. The former refers to spokespersons who provide information in an organized manner while also expressing empathy, while the latter refers to spokespersons who focus on assigning blame to others, evading responsibility, and expressing a sense of superiority.
Spokespersons following a compassionate-informative style sought to build trust by providing straightforward answers to reporters' questions, demonstrating expertise while showing empathy and risk management competence. Those who followed a condescending-evasive style offered superficial and defensive responses, usually engaging in blame-shifting and politicizing risk management. Dr. Bonnie Henr y consistently followed the compassionate-informative style of communication by expressing empathy and providing information clearly (i.e. straight answers, visual aids, organized presentation). Dr. Roussin also followed the compassionate-informative style but to a lesser degree, communicating less often and not always expressing the same level of compassion and care. Premier Ford followed the condescending-evasive model, avoiding responsibility, answering reporters' questions evasively, and blaming others for negative outcomes.
Focus group data suggests that the compassionate-informative model of communication earned spokespersons the public's trust. Trust, in terms of how people perceived the expression of shared values (Siegrist and Zingg 2014), was fostered by Dr. Henry (Henry and Henry 2021) and Dr. Roussin by communicating clearly, demonstrating expertise, keeping a calm and professional tone, and -in some cases -showing empathy. Focus group participants also expressed trusting their messages because they were supported by modeling data grounded in scientific evidence. In British Columbia and Manitoba, epidemiological models were communicated as indicators of how COVID-19 was spreading in those jurisdictions and the projections on which the pandemic response was being based. Dr. Henry's and Dr. Roussin's expertise allowed them to explain the models in detail and to use that data to support public health measures, thus, placing scientific evidence at the centre of the pandemic response. In terms of tone, focus group participants highlighted the speakers' calm demeanor and straight answers as being in control of the situation. These positive perceptions could increase public confidence in Dr. Henry and Dr. Roussin in a future health crisis.
In contrast, the condescending-evasive style of communication reduced the sense of trustworthiness, potentially eroding public trust. Premier Ford, being a politician and not a scientist, could not explain the model data, and instead he deferred to the experts of the Science Advisory Table, who are not affiliated with the government. However, Premier Ford did not use the models developed by the experts to support his pandemic management either, which could lead to a detachment between expert recommendations and public health measures. Having a primary spokesperson who cannot speak to the data, while the experts who can explain it not addressing the public regularly, could potentially increase uncertainty. This effect could have been reduced by having Premier Ford join the experts of the Science Advisory Table in their press conferences. In his communications, the Premier kept an informal tone and, on many occasions, became defensive and shifted blame onto the Federal Government and individuals while praising his performance. Blaming others to various degrees is a common strategy used by elected officials when facing criticism (Hood 2002). Regional leaders tend to blame the central government when making controversial decisions, particularly when federal and provincial leaders do not belong to the same political party (Mortensen 2012). Shifting responsibility to individuals can induce guilt, shame, and frustration (Coleman and Hatley Major 2014;Driedger, Mazur, and Mistry 2014;Guttman 2017;Guttman and Ressl 2001), and reduces confidence in the message. This use of language and tone could be interpreted as political messages rather than public health risk messages and erode public trust and could have contributed to the plummeting of Premier Ford's approval rates in the second wave, which had substantially increased during the first wave (Eric 2021).
Crisis communication requires rapid response, constant dialogue, expertise, and the expression of empathy and care. Dr. Henry was an effective crisis communicator by being professional but still showing emotion, by engaging positively with reporters and by keeping constant communication. Premier Ford, despite having a team of experts, still was not trusted. This corroborates findings that having too many spokespersons may cause confusion (Barrett et al. 2016). Additionally, his reputation as a politician could have eroded the trust that his experts brought to the table. Focus group participants referred to the Premier's messages as political and did not trust the Premier to manage the pandemic response. However, the same participants expressed a higher level of trust in the experts who developed the epidemiological model, confirming the importance of developing science-based policies during a health crisis. Furthermore, Ford was very emotionally expressive during the press conferences, but to show frustration and blame others instead of showing empathy. This also corroborates previous findings (c.f. Lyu et al. 2013).
Dr. Roussin, as a spokesperson, is perhaps the most interesting case. He is an expert, and he has demonstrated ability to understand and explain the model data. However, his communication style did not always show the same level of empathy and care as Dr. Henry. He did not maintain a constant dialogue with reporters but held press conferences to discuss modeling data only occasionally, which could have reflected restrictions imposed on his availability by the provincial government. In Manitoba, as in Canada, while the Chief Public Health Officer has a number of designated powers and delegated authority under the Public Health Act to recommend courses of action, responsibilities for all decisions ultimately rest with the designated Minister and government. Further, while some studies have shown that, besides expertise, empathy, caring, openness, and dedication are also necessary to earn the public trust and credibility, (Reynolds and Quinn 2008). In Dr. Roussin's case it was precisely his expertise that focus group participants valued, despite not following other recommended crisis communication practices. The frequency of the press conferences on epidemiological models are decided by the provincial government, and Dr. Roussin generated a high level of trust -as much as Dr. Henry -despite having fewer opportunities to engage with the public. While Dr. Henry communicated about model data consistently since the start of the pandemic, Dr. Roussin needed to wait until the government of Manitoba approved the sharing of that information. Focus group participants' perceptions could also be influenced by gender, making them perceive Dr. Henry as more compassionate and caring because she is a woman, whereas Dr. Roussin's expertise was enough to inspire trust because he is a man. Future research could explore gendered perceptions of expertise and trust. These results highlight the importance of expertise when communicating with the public during a health crisis, but also some limits to how far that expertise can foster trust if political decisions about data sharing constrain the openness of that communication. While communicating frequently, which showed openness and dedication, and expressing empathy were valued by our focus group participants, the spokesperson must be recognized as an expert to foster trust.

Conclusion
Our study underscores the importance of jurisdictional decisions in choosing a spokesperson that emphasizes the connection between expert assessments and public health response during public health crises. Several crisis communication recommendations arise from this work. From an audience-reception perspective, structured presentations, use of visual aids, and consistent messaging are preferred, with a dedicated spokesperson who can become the 'face' of the pandemic response. To foster trustworthiness in spokespersons, public health officers should be trained in risk and crisis communication, that is, developing skills to communicate clearly, showing empathy, and in a timely manner. Empathy, in particular, emerged as a key characteristic of spokespersons who engage with reporters and the public in a positive manner. Crisis communication training should encourage personal reflection of what aspects of training best fit individual values, because then, those communication strategies come off more naturally offering a more consistent and authentic style. Ideally, only leaders and experts with effective communication skills should become the 'face' of the emergency response. While training can help to minimize communication deficiencies, experts who are more naturally genuine in their communication style will be able to draw on those traits when faced with more challenging circumstances. Finally, crisis communication should emphasize the scientific evidence behind public health guidelines and acknowledge scientific uncertainty to support public buy-in, particularly if that evidence changes.
One limitation of our study is that we focused on three spokespersons in Canada. While our results provide evidence of two communication styles and how these were interpreted by focus group participants, this analysis could be expanded to include more spokespersons, both Anglophone and Francophone. Additionally, our study focused on the first two waves of the pandemic in Canada, future research could address how trust and confidence in spokespersons evolved across waves of the COVID-19 pandemic. Furthermore, we did not examine how race, gender, age, and other intersectional characteristics impact trust in spokespersons. These are all aspects worth of analysis in future research.

1.
R-naught or R 0 is the basic reproduction number of a virus, i.e. how many new cases of infection can one infected person generate, when no one has any immunity and no interventions have been imposed to curb its spread (Chung 2020).

Code availability
Coding guides are available upon request.

Consent to participate
All participants gave informed consent to participate in this study, where publication was identified as one form of dissemination. Information and Study Consent Forms available upon request.