Introduction
This study examines policy narratives during the COVID-19 crisis to reflect on their effectiveness in assisting narrating policymakers to achieve their proposed policy solutions. Policy narratives provide coherence in policymaking by structuring complex issues, reinforcing views, and potentially converting opponents (Shanahan et al., Reference Shanahan, McBeth and Hathaway2011). By integrating narrative development into policy design and analysis, policymakers can enhance their understanding of policy issues and improve the effectiveness of their responses to complex challenges (Mintrom and O'Connor, Reference Mintrom and O'Connor2020). When attempting to persuade other policy actors of the merits of their policy proposals, policymakers invest effort and time in crafting their arguments, often employing rhetorical elements or emphasizing the rationale of a policy solution, its costs and benefits, and the unfounded nature of its opposers’ motives (Gray and Jones, Reference Gray and Jones2016; Blumenau and Lauderdale, Reference Blumenau and Lauderdale2024). These narratives incorporate purpose and intention, and are strategically constructed to support the adoption of preferred policy solutions (McBeth and Shanahan, Reference McBeth and Shanahan2004). Hence, this study takes a constructivist perspective on policy narratives, which sees framing as an essential tool for meaning-making in contentious political contexts. Like other social actors, policymakers use narratives to frame crises, justify solutions, and shape public and institutional responses. Della Porta and Tarrow (Reference Della Porta, Tarrow, Della Porta and Tarrow2005) highlight how social movements, for instance, rely on framing processes to construct collective identities and mobilize support. These processes illustrate how narratives are not just reflections of political struggles but tools to define crises, build legitimacy, and mobilize support. Similarly, Jasper (Reference Jasper, Jasper and Goodwin2011) focuses on strategic interaction and argues that political actors do not merely react to external opportunities but actively shape their environment through interpretation, framing, and tactical moves. In this sense, narratives are more than just passive reflections of political structures; they are strategic tools used to persuade allies, neutralize opponents, and redefine the stakes of political debates. Policymakers shape narratives that reinforce their authority and influence crisis-related discourse across governance levels.
Over the past two decades, several authors have focused their interest in policy narratives on the effectiveness of different communication strategies (e.g. Nelson, Reference Nelson2004; Coticchia, Reference Coticchia2015; Jung, Reference Jung2020). In this context, the Narrative Policy Framework (NPF) has emerged as a valuable approach for analyzing how policy narratives influence policy developments (Crow and lawlor, Reference Crow and Lawlor2016; Shanahan et al., Reference Shanahan, McBeth and Hathaway2011). Although widely applied to analyze narratives, the NPF has rarely assessed whether specific strategies help policymakers achieve preferred outputs (Blumenau and Lauderdale, Reference Blumenau and Lauderdale2024). This is particularly evident in times of crisis, when concurrent explanatory factors may come into play beyond rhetorical strategies. In crises, policymakers act as narrators: sourcing, shaping, and disseminating tailored narratives (Hand et al., Reference Hand, Morris and Rai2023). However, there is a paucity of studies analyzing policy narratives in relation to policymakers’ desired policy outputs in times of crisis, defined as policy solutions that policymakers aim to adopt (Howlett and Cashore, Reference Howlett, Cashore, Engeli and Rothmayr Allison2014). This article seeks to address this gap through a within-case comparative analysis of school policies and narratives during the COVID-19 pandemic in Italy, contrasting the strategies adopted and policy solutions proposed in these narratives with the policy solutions eventually implemented. Italy is a relevant case due to its significant COVID-19 exposure, political polarization, and chronic governmental instability. These conditions create a dynamic setting for observing the effectiveness of competing narrative strategies. In this context, the research question this analysis aims to answer asks: Are rhetorical narrative strategies more effective than non-rhetorical strategies for policymakers to achieve the adoption of the policy solutions they support? The analysis is conducted in light of both the literature on policy narratives and the literature on knowledge in times of crisis, thus connecting these two streams to better comprehend narratives during crises. It employs a within-case comparative research design and the READ method (Readying – Extracting – Analyzing – Distilling; see Online Appendix; Dalglish et al., Reference Dalglish, Khalid and McMahon2020) to analyze school closure/reopening decisions and public statements by five key actors involved in the in-school/online classes debate between the first COVID-19-related school closure measure (March 1, 2020) and the last day of the school year 2020/21 (August 31, 2021): prime ministers Giuseppe Conte (PM-C) and Mario Draghi (PM-D), ministers of education Lucia Azzolina (ME-A) and Patrizio Bianchi (ME-B), and minister of health Roberto Speranza (MH). The analysis will show that rhetorical narratives were no more effective than non-rhetorical narratives in enabling policy actors to achieve the implementation of their preferred policy solutions. It also highlights additional factors explaining whether preferred policy solutions were (not) adopted, namely the narrator’s reputation and the alignment of narratives with the relevant crisis stage and the available scientific advice.
The following sections elucidate the theoretical framework, present the research design and methods employed, illustrate and discuss the analysis findings, and conclude by highlighting the added value of this study and offering recommendations for further research.
Theoretical framework: combining narrative and crisis governance perspectives
Policy narrative theory sees humans as natural storytellers. Following the literature on frames and framing (Snow and Benford, Reference Snow and Benford1988; Caiani, Reference Caiani2023), this study conceptualizes policy narratives as structured interpretative devices that policymakers use to define problems, attribute responsibility, and justify solutions. Frames provide meaning to political issues by highlighting certain aspects while downplaying others, shaping both public perception and policy decisions. Within this framework, narratives can be understood as dynamic framing strategies that evolve throughout a crisis, responding to shifting contexts and actor positions. The presence of identity and oppositional frames (Lindekilde, Reference Lindekilde and Della Porta2014) suggests that policy actors may aim to construct narratives that delineate allies and adversaries in the policy arena. Narratives also help policymakers make sense of crises marked by uncertainty and polarization (Roe, Reference Roe1994; Hand et al., Reference Hand, Morris and Rai2023). In this context, the NPF provides a framework to analyze how policymakers construct and utilize narratives in the policy processes that develop during crises (Hand et al., Reference Hand, Morris and Rai2023). According to the NPF, narratives are defined by the coexistence of four elements: the context, which comprises the policy issues deemed worthy of attention at a given historical moment and the political and institutional contingencies (Shanahan et al., Reference Shanahan, Jones, McBeth, Radaelli, Weible and Sabatier2018); the characters, who can assume different roles (e.g., victims, heroes, etc.); the plot, which situates the characters and their relationships in time and space (Roe, Reference Roe1994); and the moral, consisting of the policy solutions proposed (Shanahan et al., Reference Shanahan, Jones, McBeth, Radaelli, Weible and Sabatier2018).
To ensure the prevalence of their narratives over those of others, policy actors employ a variety of strategies: those based on causal mechanisms, which strategically organize narrative elements to assign responsibility and blame for a political problem, thus causally linking different public issues; those based on the scope of the conflict, which actors expand or compress by exalting the costs and minimizing the benefits of the other side’s proposed policy solutions or by promoting the benefits and minimizing the costs of their own policy solution, depending on their self-representation within the policy conflict as losers or winners (McBeth et al., Reference McBeth, Shanahan, Arnell and Hathaway2007); and those based on devil and angel shifts, through which policy actors emphasize the negative value of their opponents’ motivations and actions or the positive value of their own problem-solving abilities (Shanahan et al., Reference Shanahan, Jones, McBeth and Lane2013).
The NPF offers strong analytical potential due to its comprehensive approach (Gray and Jones, Reference Gray and Jones2016). It particularly demonstrated efficacy in studying “wicked problems” characterized by intense value-based conflicts between policy coalitions (Veselková, Reference Veselková2014). In such situations, policy narratives help frame issues, propose solutions, and build support (Domorenok and Graziano, Reference Domorenok and Graziano2023). From this perspective, policy narratives can exhibit varying degrees of strength in aiding policymakers to persuade other policy actors of the merits of their preferred policy solutions (Brewer, Reference Brewer2021).
In this context, the research question underlying this study concerns the effectiveness of rhetorical strategies (Shanahan et al., Reference Shanahan, Jones, McBeth, Radaelli, Weible and Sabatier2018) and asks whether during crises rhetorical strategies are more effective than non-rhetorical strategies to help policymakers achieve the adoption of their preferred policy solutions. According to the NPF, a narrative is considered rhetorical when it employs strategies such as angel and devil shifts (Merry, Reference Merry2019), as well as metaphors, similes, and other rhetorical figures (Nicoll, Reference Nicoll, Simons, Olssen and Peters2009). By contrast, conflict scope and causal mechanism strategies are considered non-rhetorical, relying primarily on storytelling (Iversen, Reference Iversen, Hühn, Meister, Pier and Schmid2014). Previous literature identified rhetorical audience-as-hero narratives as more effective than non-rhetorical narratives in generating support for policies (Chalaya et al., Reference Chalaya, Schlaufer and Uldanov2024), and found that rhetorical policy narratives can reinforce congruent views and alter divergent views (Shanahan et al., Reference Shanahan, McBeth and Hathaway2011). However, in crises where a focusing event reveals a clear policy problem (e.g., the pandemic), technical solutions may take precedence over political conflict, and the expectations of policy actors and recipients might be that the communication of policymakers and the narratives they propose will be clear, objective, and reassuring, rather than characterized by an emphasis on emotion and passion typical of rhetorical strategies (Elster, Reference Elster and Elster1998). Particularly in the early phases of the crisis, policymakers might focus more on making sense of it (Boin et al., Reference Boin, Lodge and Luesink2020) than on attempting to convince others of something they are not yet certain of. Moreover, during a crisis such as a pandemic, scientific communication might predominate over emotional communication, considering the pivotal role of scientific/technical policy advice in these situations (Aagaard et al., Reference Aagaard, Easton and Head2024; Capano et al., Reference Capano, Casula, Malandrino, Terlizzi and Toth2024). Hence, factors influencing narrative effectiveness may include coherence with crisis trends and expert advice, as well as the narrator's reputation. Crises shape how policymakers communicate and frame policies, urging them to establish a sense of normality and foster collective learning to manage their reputation during crises (Boin et al., Reference Boin, ’t Hart, Stern and Sundelius2016), an outcome that is arguably not facilitated by rhetorical narratives, with their overreliance on passion and emotion (Elster, Reference Elster and Elster1998).
Building on these premises, the proposed model (Figure 1) illustrates the complementarity of three factors, i.e. alignment with the actual crisis trend, alignment with scientific advice, and the reputation of the narrating actors in understanding the (in)effectiveness of narratives. This analytical framework builds upon the NPF, which provides the core conceptual lens through which narrative strategies are identified and classified in this study. However, drawing on insights from the crisis management literature (e.g., Boin et al., Reference Boin, ’t Hart, Stern and Sundelius2016; Aagaard et al., Reference Aagaard, Easton and Head2024), the model expands the NPF by integrating additional contextual dimensions that are particularly salient in crisis settings for establishing legitimacy and fostering trust during emergencies. Hence, Figure 1 presents a theoretically grounded framework that bridges the NPF with crisis management scholarship, providing a comprehensive lens for assessing narrative effectiveness in times of crisis.
Figure 1. An analytical framework to understand narratives in times of crisis.
Beyond the NPF, other frameworks have attempted to investigate policy narratives and rhetorical strategies. For example, through Rhetorical Political Analysis (RPA), Finlayson (Reference Finlayson2007) found that successful political leaders craft narratives using rhetorical strategies that align with public sentiment. Still through RPA, Charteris-Black (Reference Charteris-Black2014) analyzed metaphor use in political rhetoric and found that emotional appeals (pathos) are highly effective in crises. Conversely, this article argues that during a crisis, public sentiment can be more likely to seek reassurance in science than in “noncrisis” times, thus making narratives supported by evidence stronger. On its part, the Advocacy Coalition Framework (ACF) (Weible et al., Reference Weible, Sabatier and McQueen2011) shows how coalitions use narratives to frame their opponents and gain policy traction, while Baumgartner and Jones (Reference Baumgartner and Jones1993) through Punctuated Equilibrium Theory (PET) show that dramatic narratives can trigger rapid policy change. In this context, this study adopts an NPF-based approach that complements existing frameworks by explicitly focusing on policymakers’ narratives and their effectiveness in linking strategy to output.
Research design and methods
In qualitative analysis, within-case comparison enhances validity by adding contextual richness and supporting theory-building (Ayres et al., Reference Ayres, Kavanaugh and Knafl2003). This article compares school policies and narratives in Italy during the COVID-19 pandemic, a crisis that witnessed the turnover of two governing coalitions (Malandrino, Reference Malandrino2022). Italy is an ideal case due to the clear polarization between in-school and online classes, as well as the intensity of COVID-19 exposure in the country. Moreover, this study applies the NPF to a rarely explored policy sector (education) and regional context (Southern Europe), thus filling a gap (Coticchia, Reference Coticchia2015; Terlizzi, Reference Terlizzi2021; Schlaufer et al., Reference Schlaufer, Kuenzler, Jones and Shanahan2022). The case also offers variation, as different actors held key positions across two coalitions, allowing comparison of reputations. The crisis lasted long enough to generate substantial narrative and policy data and offered a strong context for testing alignment with scientific advice (Mintrom et al., Reference Mintrom, Rost Rublee, Bonotti and Zech2021; Hadorn et al., Reference Hadorn, Sager, Mavrot, Malandrino and Ege2022). While the coding focused on narrative strategies as defined by the NPF, including rhetorical and non-rhetorical elements, additional contextual dimensions – namely the narrating actors’ reputation, alignment with the epidemiological trend, and coherence with scientific advice – were examined during the interpretive phase of the analysis, drawing on process tracing and qualitative triangulation (see Online Appendix).
Framing theory operationalizes frames as selective representations of reality that structure meaning and guide action (Snow and Benford, Reference Snow and Benford1988; Caiani, Reference Caiani2023). In this framework, narratives include diagnostic frames (defining problems), prognostic frames (proposing solutions), and motivational frames (mobilizing support). During the COVID-19 emergency, proposed and adopted policy decisions on school activities exhibited an oscillatory pattern between in-school instruction and online instruction, particularly in the 2019/20 and 2020/21 academic years. In support of the former, policy actors emphasized the socializing value of schooling; in support of the latter, policy actors highlighted the need to balance the right to education with the preservation of collective health. Consequently, different problem framings led to the proposal of different policy solutions conveyed through distinct and sometimes conflicting policy narratives.
The coding grid employed in this study combines NPF components with elements tailored to the research question. In particular, the categories named “Prevailing plot,” “Prevalent strategies,” “Prevalent narrative type,” “Proposed policy solution (moral),” and “Adopted policy solution (output)” are directly drawn from the NPF literature (e.g., McBeth et al., Reference McBeth, Shanahan, Arnell and Hathaway2007; Shanahan et al., Reference Shanahan, Jones, McBeth, Radaelli, Weible and Sabatier2018), which identifies key narrative elements and distinguishes between rhetorical and non-rhetorical strategies. Other categories, such as “Wave,” “School year,” “Executive in office,” and “Actor,” were introduced to contextualize the narratives within the institutional and temporal dynamics of the case and prepare the ground for the interpretive phase following the proposed theoretical model (Figure 1).
Data collection, preparation, and analysis were conducted in accordance with the READ method (Dalglish et al., Reference Dalglish, Khalid and McMahon2020), which is comprised of four phases: readying materials (which includes collecting them based on the PRISMA-S protocol), extracting data, analyzing data, and distilling findings. Sources were validated through press and specialized portal review. The analysis combined coding and process tracing – understood as a procedure for identifying steps leading to outputs in a particular historical context (Vennesson, Reference Vennesson, Della Porta and Keating2008) – through a detailed examination of legislative documents and press sources. The meso-level of analysis was combined with the micro level of citizens’ opinions in the distillation phase. More details on these steps can be found in the Online Appendix.
Findings: tracing strategies, plots, and outputs
The COVID-19 pandemic was a clear case of a problematic context becoming central in narratives (Shanahan et al., Reference Shanahan, Jones, McBeth, Radaelli, Weible and Sabatier2018). This was evident in the narrative reference to epidemiological trends and data as inescapable facts to be considered in the design and proposition of solutions, as well as the work of the Scientific Technical Committee established by the Civil Protection DepartmentFootnote 1 to elaborate and provide policy recommendations based on scientific data and professional experience (Hadorn et al., Reference Hadorn, Sager, Mavrot, Malandrino and Ege2022).
Synthetic reconstruction of the policies adopted (outputs)
In the first pandemic wave (latter half of 2019/20 school year), the decision was taken to close educational institutions and implement online learning concurrently, initially in areas with documented cases of infection and subsequently nationwide (lockdown). In the second wave, which coincided with the first semester and part of the second semester of the 2020/21 school year, the approved interventions involved a repeated alternation between in-school instruction and online instruction. Specific arrangements varied by school level and area-specific risk. In the third wave, which coincided with the final months of the 2020/21 school year and the inaugural period of the Draghi government, a more pronounced degree of restriction was initially implemented in the so-called red zones, characterized by the highest levels of risk. This was followed by a gradual reactivation of in-school instruction, initially in lower-risk areas and subsequently across the entire national education system (full reconstruction in Online Appendix).
Analysis of the narratives
Policies were shaped by competing arguments, with non-rhetorical strategies predominating (Figures 2 and 3). Rhetorical strategies were employed exclusively by the two Five Star Movement (5SM) area exponents (ME-A and PM-C), albeit to a lesser extent than non-rhetorical ones. This variety partly reflects the higher volume of ME-A’s statements.
Figure 2. Types of strategies by policy actor.
Figure 3. Narrative strategies by policy actor.
As illustrated in Figure 4, an overwhelming majority of the strategies employed can be classified as conflict compression. The following statement by ME-A exemplifies this phenomenon:
Any certainty has been shattered, the distance learning situations in Italy are very different, many schools were good at distance learning even before, but there are schools that are lagging behind. We certainly do not want to abandon students, we have put in place several tools, platforms, and webinar materials for teachers with more difficulties. (News8 - see Online Appendix for News references)
Figure 4. Narrative strategies employed.
Conflict compression (and reassurance) often co-occurred with causal mechanism strategies. In the following statement by ME-A, for example, the causal relationship is twofold. The first is an implicit relationship between collective behavior and the evolution of the policy problem (“I call on everyone to be as responsible as possible”). The second explicitly links the evolving problem to the solution (“everything depends on the evolution of these days, on the epidemiological scenario”) (News8).
Table 1 presents a summary of the narrative plots, strategies, and morals (supported policy solutions) adopted by the five governing actors, together with the policy outputs (adopted policy solutions). The table (row 5 including the header row) shows that in the only case where rhetorical strategies prevailed, the corresponding narrator was no more effective than her counterparts in achieving the desired policy solutions; on the contrary, this was the only case in which the narrator did not obtain the adoption of the desired policy decision.
Table 1. Actors, narratives, and outputs in context
Wave timeline based on data from Johns Hopkins University, in Our World in Data.
Source: Author’s elaboration.
First wave
Initially, in line with PM-C, ME-A advocated the choice to close schools in some regions as necessary, clarifying the efforts made by her ministry to support this choice in terms of technology and personnel made available to implement online instruction. The initial narrative balanced the right to health with the right to education and the need to ensure school continuity. As early as the beginning of March, ME-A was already supporting what she described as a difficult choice, i.e. the closure of schools throughout the country, which would remain the nationally adopted solution until the end of the school year, in line with the recommendations of the Technical Scientific Committee (News1). ME-A essentially supported the same solutions as PM-C. Meanwhile, MH had little presence in the school debate at this stage, subordinating school opening to the epidemiological trend.
Toward school year-end, ME-A had to defend a controversial plexiglass partition proposal aimed to limit contagions:
We have never spoken of children locked in security booths. We want to return to normality and will do so by finding the right balance between two sacrosanct rights, the right to education and the right to health (News9).
In June 2020, ME-A outlined what would be done for the safe start of teaching activities in the new school year, focusing on adapting desk distancing and classroom layout to schools’ diverse infrastructural conditions. The forecast measure for the new school year was therefore reopening schools and resuming in-school instruction, which also responded to the needs of parents, students and teachers expressed publicly in street demonstrations on 23 May and 25 June of that year:
We will return to school on 14 September, and I want to reassure all those expressing concern (News10; News11).
However, in August, daily contagion was already on the rise again. ME-A expressed her fear of a possible new lockdown. While MH emphasized conditions for reopening, which he defined as “the priority of the whole country” but only if the epidemiological situation permitted it (News13), ME-A hoped for a return to school and proposed to the regions the starting date of the upcoming school year 2020/21: September 14, although shortly afterward, on September 20–21, various local elections were scheduled to take place in Italian school premises, traditionally used as polling stations.
Second wave
The second wave (school year 2020/2021, first semester) emphasized the in-school-education-as-a-priority plot and, in contrast to the first wave, a more extensive use of rhetorical strategies and the presence of conflicting morals (i.e., proposed policy solutions). ME-A’s optimistic narratives were partly supported by PM-C, who reassured citizens about the school system’s readiness to handle possible contagion for a safe school year restart. ME-A’s angel narrative emphasized successes in spacing measures and adjacent domains (e.g., digital teacher recruitment to enable social distancing). The angel narrative was prevalent during this partial reopening phase. In the meantime, however, the first difficulties in managing in-school instruction began to be felt. School principals wrote to ME-A highlighting issues in handling symptomatic cases and staff allocation.Footnote 2 More radical criticism of the decision to reopen schools also came, on the occasion of an increase in school infections, from some regional presidents (News16). ME-A replied by insisting schools were safe, citing data whose biased interpretation was criticized by experts (News23). For their part, the trade unions commented on the total disconnection of ME-A’s reassuring narrative from the actual pandemic context and the reality of schoolwork in times of crisis. Meanwhile, ME-A attempted to compress the conflict by extolling her work based again on unclear references to data:
We have worked all summer, the data is proving us right, a reflection on whether schools are a priority should be made by everyone. (…) Schools must be the last to close, […] (News17).
However, infections continued to rise, and the government began to plan measures to contain them, including partial school closures, often mentioning the Technical Scientific Committee advisors to legitimize the proposed solutions (Hadorn et al., Reference Hadorn, Sager, Mavrot, Malandrino and Ege2022). ME-A reiterated the importance and safety of schools to support her narrative: “Education remains in-school because that is essential for everyone, from the youngest to those in upper-secondary school” (News6), also activating rhetorical strategies such as the devil shift:
Whoever thought that shifting the blame onto the school world was the solution this time has miscalculated. The contagion is elsewhere (News2).
At the intra-governmental level, the first signs of conflict were found in the mismatch between ME-A's reassuring declarations and the calls to “avoid travel where possible, and stay at home” (News14) by MH, in the face of the planned travels for thousands of candidates to take the tests of a teacher recruitment competition.Footnote 3 Evidence of the rising number of infections would soon lead, after some written tests were held in October 2020, to the suspension of that competition.Footnote 4 It is here that the rupture of the government front on the school issue became irreversible. From October 21, 2020, in the face of the worsening pandemic circumstances, online instruction was gradually reintroduced throughout the country. A few days later, PM-C adopted a new pandemic management system that divided Italy into three different risk zones based on 21 process and outcome indicators, involving the centralized definition of both the inclusion of a given region in a certain risk zone and the measures to be taken. Thus, in the less risky areas, primary and lower-secondary schools remained open, while upper-secondary schools activated a hybrid solution with a mix of in-school and online instruction. In the higher-risk areas, however, online education was also provided for the second and third years of lower-secondary school.
Before the Christmas holidays, ME-A’s narratives expressed a contradiction between closing schools and reopening businesses. Once again, ME-A activated devil shift strategies that saw students as victims of the regions:
We cannot imagine having crowded streets in the afternoon and closed schools in the morning in December. Students have the right to be in school as much as possible (News4).
It is interesting to note that, even at this stage, the clash between the positions of ME-A and MH was never direct, since the main statements against the opening of schools once again came from the Undersecretary of State at the Health Ministry, rather than from MH himself, a choice that can also be read as an expression of a desire to contain the conflict:
I do not think there will be any opening before Christmas, the intermittent opening of schools serves no one, it only serves to create flushes of contagion (News21).
And while ME-A declared that she was working “for a gradual return to school” (News5) the same health undersecretary commented:
[…] it would be better to open them in January, we must first make progress in the results because the objective (of lowering the number of infections) has not yet been achieved (News5).
The position of the Health Ministry was also supported by many virologists. After the Christmas break, indeed, some regions autonomously delayed the reopening of schools for fear of contagion. ME-A once again used a devil shift toward these regions:
The truth is that the young are always treated as the least important ones; (…) it was a little game made on purpose, without thinking of all those kids who are penalized by online instruction for the most diverse reasons (News3).
Meanwhile, the Technical Scientific Committee commented on the inadequacy of ME-A in proposing concrete solutions alongside her pro-opening rhetoric (News19). At the same time, the crisis of the Conte II government, ignited by a decrease in support from within the governing coalition, came to a head. As of February 13, 2021, the new prime minister was Mario Draghi, who kept Roberto Speranza at the helm of the Health Ministry and appointed Patrizio Bianchi as the new minister of education.
Third wave
The third wave (school year 2020/2021, second semester) was characterized by the school-as-a-priority plot, the limited use of rhetorical strategies, and the predominance of safe reopening morals. Just like ME-A, ME-B advanced a reopening narrative moral, although doing so at a different crisis stage and with different strategies that relied on his previous experience as a regional minister for education in the Emilia-Romagna region:
I want to bring teachers and students back into the classroom, as we did after the Emilia earthquake (News20).
Targeted interventions with mixed narrative strategies based on causal mechanisms and conflict compression were also proposed while displaying agreement with experts:
We will have to intervene on the groups that suffered most from distance learning, particularly adolescents in the Southern and inland areas (News18);
[…]We need to return to in-school education as soon as possible, which is why I met with COVID-19 Commissioner General Figliuolo and Professor Locatelli, spokesperson for the Technical Scientific Committee. […] (News24).
After Easter, many students returned to in-school classes and ME-B’s attention turned to the next school year:
We are taking all necessary action to ensure the return to the classroom with targeted and timely interventions (News12).
Overall, if we compare the narratives adopted by our five examined actors across the three waves, the desired policy solutions expressed therein, and the adopted policy solutions (see Table 1), we can observe that policymakers almost always achieved their preferred policy solutions; the only case where the achieved policy output deviated from the desired one was also the only case in which rhetorical strategies prevailed, thus calling for more elements to explain narrative success and failure in times of crisis (see next section).
Making sense of narrative (in)effectiveness
During the first wave lockdown, ME-A and PM-C were the primary participants in the discourse on school policies, while MH played a comparatively minor role. The debate subsequently expanded to include the Health Ministry and other stakeholders in the education sector, such as school heads, trade unions, and regional authorities, as well as the newly appointed ME-B and PM-D following the governmental transition in February 2021, which resulted from a crisis within the existing coalition. The slight thickening (Van der Heijden, Reference Van der Heijden2011) that occurred in the second wave, characterized by the Health Ministry’s increased involvement in the policy debate, coincided with an intensification of the divide between proponents of school closure and those advocating for reopening. During the first wave, the primary actors were largely unified and proactive in their efforts to balance public health concerns (addressed through school closures) with the right to education (ensured through remote instruction). This finding aligns with the perception of citizens, who overall recognized the government's ability to quickly adapt to changing circumstances in the first wave.Footnote 5 This alignment can be seen not only as a mere reflection of institutional choices but also as a factor that shaped these latter. Public support and widespread acknowledgement of the government’s responsiveness provided policymakers with greater legitimacy, allowing for a smoother implementation of restrictive measures. At the same time, the microlevel dimension of individual concerns filtered into the meso-level through public discourse, further influencing decision-making. Subsequently, particularly in the second wave, the debate regarding in-person versus online classes became more contentious and involved the utilization of rhetorical strategies by ME-A.
In this context, Table 1 shows that rhetorical narratives were no more effective than non-rhetorical narratives in enabling policy actors to achieve the implementation of their preferred policy solutions. Conversely, the sole instance in which the realized policy output deviated from the desired output was also the only case in which rhetorical strategies predominated, occurring in a context (the second wave) of heightened conflict over potential policy solutions. In contrast, during the first and third waves, the narratives of governing actors generally converged toward similar conclusions, attempting to make sense of the unfolding situation (Boin et al., Reference Boin, Lodge and Luesink2020) and adapting policy solutions to the evolving context shaped by improved understanding of the virus and vaccine availability (Casula and Malandrino, Reference Casula and Malandrino2023). The analysis results further indicate that the varying applications of rhetorical strategies are only partially attributable to changes in government. ME-A, who held the position of Minister of Education during two pandemic waves, modified her communication strategy between the first wave, characterized by limited use of rhetorical strategies, and the second wave, which saw a proliferation of such strategies. ME-B’s communication remained overall non-rhetorical, while ME-A’s limited recourse to rhetoric corresponded to the early stages of policy problem framing and her increased utilization of rhetorical strategies occurred during periods of greater conflict between differing proposed policy solutions. These rhetorical strategies, however, did not facilitate the achievement of her preferred policy solution. How can this be explained? The remainder of this section endeavors to offer some explanations.
Alignment with the epidemiological situation and scientific advice
The ineffectiveness of rhetorical narratives in the context examined can be attributed primarily to their incongruence with the trajectory of the crisis, specifically the epidemiological situation. While rhetorical narratives are not necessarily disconnected from a crisis trend, here the rhetorical emphasis appears to have been employed to compensate for the lack of consistency between those narratives and the data related to the increasing pandemic curve, as well as the scientific advice based on those data and predictions as a relevant element for public decision-making during the pandemic (Hadorn et al., Reference Hadorn, Sager, Mavrot, Malandrino and Ege2022; Capano et al., Reference Capano, Casula, Malandrino, Terlizzi and Toth2024). This study’s findings suggest that in times of crisis, the objectivity of narratives and decisions, along with their alignment with expert advice – especially in such countries as Italy, where citizens generally display a high level of trust in scientists and particularly in medical sciencesFootnote 6 – supersede the rhetoric of narratives. This highlights a reciprocal dynamic between the micro- and meso-levels: while institutional narratives aimed to reinforce legitimacy and compliance, they were effective only to the extent that they resonated with preexisting trust patterns at the individual level. In contexts where trust in science was lower, such as France,Footnote 7 public skepticism acted as a counterforce, prompting decision-makers to recalibrate their communication strategies and, in some cases, their policy choices, which were mirrored in less restrictive measures.Footnote 8 Thus, the crisis response can be seen as the result of a continuous feedback loop between citizens’ perceptions and meso-level governance structures, rather than a unidirectional process dictated by institutions. This finding corroborates previous studies that identified clarity and consistency with problematic reality as components of a successful narrative (Coticchia, Reference Coticchia2015).
The second wave was the most critical due to the high level of uncertainty combined with the need to restart socioeconomic activities, which may have contributed to the misalignment between narratives and policies. During the first wave, it rapidly became apparent that a substantial, novel, and unprecedented threat was being confronted; consequently, at that time, there was widespread consensus on the lockdown as a precautionary and temporary measure, as the emergency was not anticipated to persist for years, and similar policy solutions were being implemented in other countries. Furthermore, when a severe crisis emerges, policymakers often consult policy advisors to guide their actions, avoid blame, and share responsibility for their decisions (Aagaard et al., Reference Aagaard, Easton and Head2024). The analysis of narratives and policy documents revealed that during the first wave, when uncertainty was at its peak, most policy actors aligned their narratives with scientific advice and thus converged toward the same policy solutions, thereby strengthening support for those solutions. Thus, conformity to scientific advice reduced ideological polarization over a contentious policy issue. Likewise, in the third wave, the threat and corresponding containment solutions were already well-known, and the vaccination campaign had commenced, which undoubtedly contributed to making the school reopening a less disputed solution.
Narrating actors’ reputations
An additional explanatory factor was the reputation of the examined policy actors. First, it is indicative that the Health Ministry eclipsed itself completely from the debate on schools during the Draghi government,Footnote 9 showing full confidence in the decisions made by Draghi and Bianchi. This fact can also be explained considering that the reputation of Draghi and Bianchi, as measured before their respective government appointments, was different and, notably, better than that of Conte and Azzolina,Footnote 10 a function of the greater experience and notoriety of the former compared to the latter: Mario Draghi was the president of the European Central Bank between 2011 and 2019, while Patrizio Bianchi was the former minister (assessore) of education of the Emilia-Romagna region. Giuseppe Conte and Lucia Azzolina, coming from a movement (5SM) that does not prioritize in principle long-standing political careers, were not as politically experienced as the former two figures. But it should also be noted that Minister of Education Azzolina came into conflict even with Prime Minister Conte, who belonged to her own political movement, as shown by the sometimes-divergent narratives of the second wave. Therefore, even internally, claims about the school’s contribution to contagions and the distorted reading of existing data by ME-A, criticized by the experts themselves, may not have helped sustain her reputation. The 5SM governing actors’ appreciation was not stable over time, with notable peaks in the first wave and a constant decrease in the second wave.Footnote 11 In this context, ME-A in particular was subject to harsh criticism and experienced growing isolation, including from within her own governing majority.
To provide a concise overview of the configurations that emerged from the analysis, Table 2 summarizes the relationship between the type of narrative strategy, the contextual conditions in which it was deployed, and the extent to which policymakers achieved their preferred policy solutions.
Table 2. Narrative strategy type, contextual conditions, and achievement of preferred solutions
Conclusions
The analysis shows that in crisis contexts the use of rhetorical strategies may not be particularly useful and that, on the contrary, the use of rhetoric to compensate for misalignment from the evidence base and scientific policy advice may have negative effects on the eventual adoption of one’s preferred policy solutions, failing to establish a sense of normality and encourage collective learning (Boin et al., Reference Boin, ’t Hart, Stern and Sundelius2016). Conversely, the study suggests that relying on scientific policy advice can enhance the credibility of narratives and reduce their susceptibility to criticism, although in such a crisis as the COVID-19 pandemic, policy advisory systems themselves may encounter a high degree of uncertainty and the need to balance competing priorities, including health, the economy, and social interaction (Aagaard et al., Reference Aagaard, Easton and Head2024). Furthermore, in linking policy narratives and scientific advice, it is important to remember that the utilization of scientific evidence in these narratives can evolve strategically, as observed in New York City’s school shutdown debates during the pandemic (Klatt and Blum, Reference Klatt and Blum2024). Finally, following previous research (Coticchia, Reference Coticchia2015), the absence of effective counternarratives as well as the ability of policymakers to maintain awareness of the crisis can contribute to the efficacy of policy narratives, rendering them more resilient to criticism and, ultimately, more resistant to being disregarded when policy solutions are adopted.
The analysis conducted in this article provides support for the ineffectiveness of rhetorical narratives during a major crisis such as a pandemic and demonstrates how other factors become significant, namely the reputation of the narrating actors, consistency with the course of the crisis, and alignment with the recommendations of scientific advisors. Thus, this study makes a threefold contribution to the mentioned literature on narratives: (i) First, in a context dominated by NPF literature that focuses on how policymakers use narratives (e.g., McBeth et al., Reference McBeth, Shanahan, Arnell and Hathaway2007; Nicoll, Reference Nicoll, Simons, Olssen and Peters2009; Shanahan et al., Reference Shanahan, Jones, McBeth and Lane2013, Reference Shanahan, Jones, McBeth, Radaelli, Weible and Sabatier2018; Iversen, Reference Iversen, Hühn, Meister, Pier and Schmid2014; Gray and Jones, Reference Gray and Jones2016; Merry, Reference Merry2019), this study investigates the effectiveness of policy narratives in achieving the adoption of the intended policy solutions, thus adding evidence to the NPF research that focuses on the connection between narratives and policymaking (Shanahan et al., Reference Shanahan, McBeth and Hathaway2011; Brewer, Reference Brewer2021; Chalaya et al., Reference Chalaya, Schlaufer and Uldanov2024); (ii) Second, by using NPF in such an explanatory manner, the study adds an NPF perspective to the RPA and PET literature that aims to understand the power of narratives (Baumgartner and Jones, Reference Baumgartner and Jones1993; Charteris-Black, Reference Charteris-Black2014) and at the same time complements the RPA and ACF literature that focuses more on the way policymakers craft their narratives (Finlayson, Reference Finlayson2007; Weible et al., Reference Weible, Sabatier and McQueen2011); (iii) Third, the analysis focuses on narrative strategies in connection with other elements that concur to explain the narrative’s effectiveness, thus placing them in their context rather than examining them isolatedly; in particular, by contextualizing the analysis in the COVID-19 pandemic context, the study connects two streams of literature, namely that on policy and political narratives on the one hand, and that on knowledge in times of crises on the other hand.
From a practical point of view, this study suggests that especially in times of crisis, policymakers should craft their narratives so as to improve their own reliability and trustworthiness, including by correctly engaging with the existing evidence and scientific advice rather than bending it to their own ideological beliefs. These findings align with constructivist perspectives on political storytelling, highlighting that the effectiveness of crisis narratives depends not only on rhetorical strategies but also on their alignment with broader political and epistemic structures. They suggest that policymakers in crises must construct narratives that resonate with both scientific expertise and public expectations to achieve policy traction. Future studies on policy narrative effectiveness should consider these variables, and focus on the capacity of narratives to assist policymakers in pursuing their policy solutions during noncrisis periods. Additionally, further research could explore how different institutional settings and governance structures mediate the relationship between policy narratives and decision-making effectiveness, particularly in decentralized or multilevel governance contexts. Comparative studies across crises of different natures – such as economic, environmental, or security crises – could also provide insights into whether the findings of this study apply beyond public health emergencies. Finally, an examination of the interplay between policy narratives and media framing could offer a deeper understanding of how narratives gain traction or face resistance in the public sphere.
Funding
This research received no specific grant from any public or private funding agency.
Data
The replication dataset is available at https://doi.org/10.7910/DVN/N9ILJR.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/ipo.2025.12.
Competing interests
The author declares none.
Open access
