The COVID-19 pandemic brought unprecedented scrutiny to governmental science advice, placing institutions and individual experts under intense public observation. The Scientific Advisory Group for Emergencies (SAGE), the UK government's primary source of scientific advice during emergencies since 2009, became a central focus of this scrutiny. For the first time during an ongoing emergency, SAGE's meeting minutes were released publicly, offering a unique opportunity to analyze their communication practices. This study addresses four key themes identified in the literature: transparency, plurality of expertise, the science-policy boundary, and the communication of consensus while addressing uncertainty. The research questions are: 1) Did SAGE's approach to transparency and expertise plurality change over time? 2) How is SAGE's role constructed in the meeting minutes? 3) How are consensus and uncertainty communicated in the minutes? The study employs a mixed-methods approach, analyzing both the metadata (e.g., meeting and publication dates, attendance) and the linguistic choices (e.g., use of hedges, boosters, attitude markers, self-references) within the minutes. The analysis focuses on the period from January 22, 2020 to May 13, 2021, covering the initial uncertainty of the pandemic's emergence to the beginnings of easing lockdown measures.

The paper draws on Science and Technology Studies (STS) literature, particularly Hilgartner's 'Science on Stage' framework, which emphasizes the performance aspect of science advice and the distinction between 'front stage' (public) and 'backstage' (internal) processes. The review highlights the importance of credibility, transparency, and diverse expertise in science advisory committees. It also discusses the complex relationship between science and politics, the challenge of defining the boundary between them, and the tension between the neutrality of scientific advice and its usefulness for policymakers. The literature emphasizes the importance of honesty about uncertainty and the social weight of consensus in shaping public trust and policy decisions. Previous studies on SAGE’s role in previous crises advocate for increased transparency and greater inclusion of diverse expertise in discussions. The review incorporates insights from reports by the Institute for Government (IfG) and the House of Commons Science and Technology Committee (STC) on UK science advice during the pandemic. Additionally, the paper cites existing research on linguistic markers of certainty and uncertainty, particularly Hyland's work on stance in academic writing, which provides a framework for analyzing the expression of authorial commitment and authority.

The study uses the first 89 publicly available SAGE meeting minutes as its primary data source. The researchers employed a mixed-methods approach. Method 1 focused on metadata analysis to assess changes in transparency and expertise plurality. This involved tracking publication delays, identifying attendees, and categorizing their roles (scientific expert, observer, secretariat). The study period was divided into five timeframes based on key moments in the pandemic's media coverage. Method 2 involved linguistic analysis, drawing on Hyland's stance framework, to examine how SAGE's role was constructed and how consensus and uncertainty were communicated in the minutes. This focused on identifying self-references, attitude markers (expressing judgments and opinions), boosters (enhancing certainty), hedges (downgrading certainty), and the use of confidence intervals to explicitly mark uncertainty. Frequencies of these linguistic markers were calculated and normalized per 1000 words. Statistical analysis (Kruskal–Wallis tests and post-hoc comparisons with Bonferroni correction) was used to compare marker frequencies across the timeframes. The study acknowledges the ethical considerations involved in analyzing publicly available data, particularly regarding the anonymization of individuals and institutions.

The analysis revealed several key findings: First, although SAGE’s transparency increased over time with the publication of minutes and names of experts, significant delays in publication remained. Second, a core group of experts dominated discussions, despite an overall increase in the number of attendees over the study period. Third, SAGE's meeting minutes presented a largely consensus view, with limited explicit acknowledgment of disagreements, despite the potential for diverse viewpoints. Fourth, the frequency of linguistic markers of certainty (boosters) and uncertainty (hedges) both increased over time. This reflects a commitment to communicating precise information, including limitations and uncertainties. However, the frequency of explicit self-references to SAGE decreased significantly over time, suggesting a potential shift in how SAGE positioned itself in relation to policy decisions. Fifth, the use of explicit confidence intervals to mark uncertainty increased from midway through the study period, showcasing a growing recognition of the importance of clearly communicating uncertainty.

The findings demonstrate a complex picture of SAGE's communication during the pandemic. While transparency improved, the dominance of a core group raises concerns about potential biases or groupthink. The presentation of a largely consensus view, despite the involvement of numerous experts with varying perspectives, requires further investigation into the internal deliberation processes. The increase in both certainty and uncertainty markers underscores the challenge of communicating complex and evolving scientific information accurately, while the decrease in self-references might reflect a cautious attempt to maintain scientific neutrality in the face of intense public and media scrutiny. The study reveals a potential tension between the need to provide clear and useful advice and the desire to maintain scientific objectivity. The use of explicit confidence intervals demonstrates a growing awareness of communicating uncertainty, but more explicit acknowledgment of dissenting opinions might further enhance transparency.

The study contributes to understanding science communication in crisis contexts and highlights the value of linguistic analysis in exploring the strategies used by scientific advisory bodies. The findings underscore the need for ongoing refinement of communication protocols in science advisory groups, particularly regarding the explicit acknowledgment of disagreements and uncertainties. Future research should investigate the internal decision-making processes within SAGE, explore the interaction between SAGE and policymakers in more detail, and examine how different communication strategies affect public trust and understanding of scientific ambiguity. It is recommended that future guidelines for science advisory groups consider the communication of both consensus and dissent, together with clear explanations of the degree of certainty associated with scientific findings.

The study is limited by its reliance on publicly available meeting minutes, which might not fully capture the complexity of SAGE's internal deliberations and informal interactions. The analysis focuses on a specific period and doesn’t encompass the entire duration of the pandemic, nor does it account for events such as the emergence of new variants. The linguistic analysis relies on a predefined set of markers and might miss other subtle indicators of stance or uncertainty. The study also does not assess the impact of media coverage and other external factors on SAGE's communication strategies, and it does not account for possible variations in the minute-takers' approaches from meeting to meeting. Further research should explore these factors to obtain a more comprehensive view.