Frames in Crisis: A Quantitative and Qualitative Analysis of Risk Frames in Governmental Communications of European Countries

(1)

I

Frames in Crisis:

A Quantitative and Qualitative Analysis of Risk Frames in Governmental Communications of European Countries.

Research Project

Date of publication: 06/2020

Name: L. J. Huijser

Student number: 10351787

Study: Political Science, Public Policy and Governance track … University: University of Amsterdam

Supervisor: Dr. A. M. C. Loeber Second Reader: Dr. L. W. Fransen

(2)

I

(3)

II

Lists of Tables

Table 3.1. Number of documents reviewed ... 26

Table 4.1. Start of each phase per country ... 32

Table 4.2. Risk frames analyzed in official documents ... 33

Table 4.3. Most applied frames ... 34

Table 4.4. Most frequent used frames per country ... 36

Table 4.5. Excess mortality rates for each phase per country ... 37

Table 4.6. Relationship risk management factors, frames and excess mortality rates ... 39

List of figures

Figure 2.1. Four categories of risk (Douglas & Wildavsky, 1982, p. 5) ... 17

Figure 2.2. Four types of policy problems. (Source: Hisschemöller & Hoppe, 1995, p. 44) . 20 Figure 3.1. Coding template ... 24

(6)

5

Chapter 1. Introduction

What started off in November 2019 as disturbing news of an irregularly high numbers of people with pneumonia complaints in Wuhan, China, has become a pandemic that affects all continents except Antarctica (World Health Organization, 2020). This pandemic, caused by the coronavirus SARS-CoV-2, has impacted lives globally in an unprecedented fashion. Indeed, according to the International Monetary Fund, it is the worst economic crisis since the Great Depression.

While the World Health Organization (WHO) has given advice on governments should response to the novel coronavirus, a plethora of different measures have been taken around the globe. However, beyond the different measures that have been adopted by various

countries, the way in which the virus has been framed by governmental authorities also varies tremendously. The coronavirus has been referred to as a “challenge comparable to that of World War II” (Castelfranco, 2020) by Italian Prime Minister Conte , while the Swedish administration wrote the following: “Keep your distance, we normally say in traffic. But that also applies to social life now” (Edwards, 2020). Meanwhile, the President of the United States has referred to the coronavirus as the Democratic Party’s “new hoax” (Cook & Choi, 2020).

The manner in which world leaders are framing the corona crisis illustrates that biological phenomena cannot and should not be seen as “neutral” occurrences. The way in which people respond to such events has a significant impact on the effect of such natural events”?.

Evidence has shown that the framing of an issue can be a powerful tool in terms of shaping policy outcomes (D’Angelo & Kuypers, 2010, p. 7). This is why it is crucial to investigate both how different countries have framed SARS-CoV-2 and the impact of such framing. Considering the responses to the COVID-19 crisis indicates that even within federal and confederal systems, there has not been a unified approach to addressing this pandemic In this research project, I seek to answer the following research question:

How do (ex) member states of the European Union frame their public health policies intended to address the COVID-19 crisis, and what do these findings indicate about the current state of the risk management literature?

(7)

6 To answer this question, I have formulated the following sub-questions:

SQ1: Which risk frames can be identified in the official communications of the analyzed countries?

SQ2: What are the excess mortality rates (EMR) in the analyzed countries?

SQ3: How do the risk frames, as well as other factors identified as relevant in the risk management literature, relate to the excess mortality rates?

The quantitative framing analysis based on these research questions yielded certain odd results that could not be explained by quantitatively analyzing the data. For this reason, a qualitative framing analysis was conducted following the quantitative framing analysis. The qualitative framing analysis was intended to provide a better understanding of the central research question while also attempting the following additional sub-questions

SQ4: How can we explain the differences in excess mortality rates between Germany and France given that they employ the same frames and have similar public

healthcare configurations?

SQ5: How can we explain the lenient lockdown measures of the Netherlands and Sweden in light of their framing strategy?

SQ6: what do these findings regarding SQ4 and SQ5 imply for the risk management literature?

Answering these questions is crucial to understand the differences in the policy approaches adopted by these countries and can serve as a foundation for future research analyzing why some countries have been more severely impacted by the corona pandemic than others, and show why the current literature does not provide helpful resources with which to respond to the COVID-19 crisis. To answer the research question, an explanation is given as to what the current state of the literature on risk management is, how it related to policy and metacultural frames , why there are reasons to believe that metacultural frames play a crucial role in policy development, and how the differences in terms of frames among the four countries are

identified.

I answer the central research question by conducting framing analysis, both quantitative and qualitative. The official communications of the researched countries are analyzed to

(8)

7 numbers of people have become affected by the virus. This research shows that quantitative research analysis based on current risk management literature give insight in how frames work, but it cannot show the full picture. For that, it is crucial to analyze the overarching story by conducting a qualitative framing analysis

This study aims to further the academic understand of how policy frames are employed in circumstances characterized by uncertainty and in which immediate action is required. A better understanding of how these frames are used work can help us to prepare for future crises, which could potentially mitigate economic hardship or even save lives.

Currently, most researchers are focusing on understanding the virus as a biological phenomenon, with researchers attempting to determine where the SARS-CoV-2 virus originated, identify which groups are at risk, and develop a vaccine for the virus. Such

research is crucial, but the impact of government actions intended to address social aspects of such crises also merits extensive investigation. There is a gap in knowledge concerning how frames are being utilized by governments when a wicked problem that requires immediate action emerges.

Chapter 2 of this research project will lay the theoretical framework to answer the research question and the subsequent sub questions. It will elaborate on how frames are used in academic discourse, what risk management literature is currently saying, the link between frames and risk management, and solutions proposed by risk management to ‘wicked problems’ such as the COVID-19 crisis. Examples of risk frames utilized by governmental agencies during past pandemics are presented to demonstrate the similarities and, more importantly, the differences between the responses to the SARS-CoV-2 pandemic and

pandemics of the past. In Chapter 3 of the research project, I elaborate on the methodological decisions made in this research. Here, I explain the research approach I adopted and the reasoning behind this choice, how the data was collected, and how I analyzed this data. In addition, I explain the rationale for making a particular choice regarding the phases. Chapter 4 presents the findings of the research, focusing on the policy and metacultural frames that have been used by the governments of the selected countries during different phases of the pandemic. Thereafter, Chapter 5 discusses the findings presented in Chapter 4. This thesis concludes by discussing the limitations and implications of this research and explores options for follow-up research.

(9)

8

Chapter 2. Theoretical Framework

This chapter explains the importance of including the concepts of framing and frame analysis in this research to better understand the social realities in the analyzed countries. An

overview of the literature on frames to public policy and communication is provided.

Thereafter, I focus on risk and risk management; I also consider scholars’ theories concerning how to address problems regarding which we lack knowledge, and the shortcomings of these theories. Thereafter, I explain why this pandemic can be viewed as a wicked problem and why the current strategies current strategies intended to address the COVID-19 crisis would fail.

2.1 The COVID-19 health crisis

What exactly is the COVID-19 crisis, and why can we classify it as an unstructured or wicked problem? In the following sections, I first explain the origins of the virus, the terminology, and the most common symptoms.

COVID-19 – Terminology

As the nature of this crisis continues to be characterized by uncertainty and ambiguity at the time of this writing, it is important to establish a clear definition of what is meant by the term the “corona crisis.” It is important to note that the majority of the information presented here was current as of May 2020; as such, some of the information presented here may change as new evidence emerges. What began as a number of people in Wuhan, China, with

pneumonia-like symptoms rapidly became a pandemic. These patients were actually infected with a novel corona virus, severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2). This new corona virus, which is closely related to the original SARS-CoV (SARS), causes the infectious coronavirus disease 2019 (COVID-19). The first case has been traced back to November 17th 2019 (European Centre for Disease Prevention and Control , 2020a). The most common symptoms are fever, cough, loss of appetite, fatigue, shortness of breath, and muscle and joint pain.

Major differences between SARS-CoV-2 and its predecessor include the case fatality rate and the symptoms exhibited. During the 2002–2004 outbreak of the SARS virus, incidence rate was 8.422 cases, with a case fatality rate (CFR) of 11%. The CFR of the SARS virus differed by age group, as is the case for the new coronavirus: Patients under 24 years of age had a less than 1% chance of dying, whereas patients of above 65 years of age had a CFR of over 55%

(10)

9 (Chan-Yeung & Xu, 2003, p. 8). In addition, the symptoms patients with SARS exhibited were almost never asymptomatic: when a person contracted SARS, he or she would almost certainly feel ill. The COVID-19 disease seems to work differently. For some, symptoms might be mild or not noticeable, while, for others, the disease may prove fatal. Whereas in the beginning of the COVID-19 crisis it was believed that the core symptom of the new

coronavirus was a fever, there have been numerous reports of people infected with the disease not showing any fever symptoms. Apart from that, the CFR, although it cannot be accurately calculated before the pandemic is over, seems to be much lower than that of SARS, at around 5.7% (Johns Hopkins University, n.d.).

The difference between a CFR of 5.7% and 11% might not seem significant, but, due to the uncertainty associated with and the characteristics of SARS -CoV-2, the importance of this difference cannot be underestimated. The CFR indicates the percentage of deaths from a disease compared to the number of diagnosed cases of that disease. As a number of people have tested positive for the new coronavirus without showing any symptoms, it is reasonable to assume that the number of actual cases of COVID-19 are substantially higher. In addition, most countries only count deaths that have been officially identified as being caused by the coronavirus (European Centre for Disease Prevention and Control, 2020a). This means that the actual mortality rate of COVID-19 will be lower than its CFR suggests. While this might be good news when one contracts COVID-19 as an individual, in combination with the asymptomatic traits of the virus and the somewhat contagious nature, the virus can become a pandemic.

2.2 Frames, framing, and reframing in policy discourse

Public policy and various countries’ political arenas have been drastically transformed in the last 60 years. In the 1960s and 1970s, there was a strong belief that the best policy outcomes could be achieved by a rationale debate between political actors (Frohock, 1972, p. 1039). Disputes should, according to dominant discourse of that time, be resolved by examining the facts of a situation and agreeing on said facts (Schon & Rein, 1994, p. 3). It is no surprise that the influential book The End of Ideology by Daniel Bell, which argued that politics should not be based on ideology but on logical reasoning, was published in this era (Bell, 1962). A positivist approach, which is based on perceived neutral and objective stances, was commonplace in public policy arenas (Gartrell & Gartrell, 2002, p. 654).

(11)

10 Since the constructivist turn that occurred within social and policy sciences during the 1990s, frame analysis has become a widely used methodology in the field of policy sciences (Fischer & Forester, 1993, p. 5). It differs from other constructivist approaches in that it focusses on the stability of frames, instead of a more fluid conceptualization of the policy arena. For instance, Hajer recognized the ambiguity in the policy arena: different discourses interact to form coherent policy. He coined the term discourse formation to refer to this phenomenon (Hajer, 1993, p. 66). Framing analysis, however, is based on the understanding that policies are structured within this frame that defines the policy problem (and solution) or where a multitude of frames contest each other (Schön & Rein, 1994, p. 24). In this sense, rather than being ambiguous themselves, policy frames exist to resolve ambiguity.

Public policy issues are rarely one-dimensional. For example, when public policy regarding a major health crisis are being discusses, this topic can be framed as public health issue, an economic issue, a security issue, or an issue of human suffering, among others. Those who can successfully frame a policy issue have a distinct advantage in public debates (Schön & Rein, 1994, p. 32).

Given the above discussion, what do we mean when we speak of frames and framing in the policy discourse? As frames and framing are commonly used outside of formal scholarly discourse, what frames actually are is often left to a tacit understanding between reader and researcher (Entman, 1993, p. 51). Thus, it is crucial to provide clear definitions of what frames and framing actually entail. While a variety of definitions of the terms frame and framing have been suggested in the academic literature, this study heavily relies on the work and concepts of Robert Entman (1993), Donald Schön and Martin Rein (1994), and Deborah Stone (2011) in developing its conceptual framework conceptual framework.

2.2.1 Frames and Framing

Policy frames are essentially a fractured representation of reality. As Entman puts it, to frame is to “[…] select some aspects of a perceived reality and make them more salient in a

communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described” (Entman, 1993, p. 52). In other words, frames are embedded in stories, in which some aspects may highlighted while others, whether intentionally or not, may be downplayed. Stories, according to Schon and Rein, are powerful tools that can shape public awareness of an issue (Schön & Rein, 1994, p. 25). In their book, these authors illustrate how conflicting frames

(12)

11 and stories are resistant to resolution by appeal to facts or reasoned arguments because the frames utilized determine what counts as a fact and which arguments are considered relevant and reasonable (Schön & Rein, 1994, p. 23).

Within policy frames, different strategies are utilized to internalize the process of framing described by Entman: the promotion of a particular problem definition, causal interpretation, moral evaluation, and justification for action. The internalization of a story implies that the story can only be descriptive, while calling for a prescription without ever doing so overtly. This jump from description frequently occurs in the real-world policy arena, and Rein and Schoen refer to it as “normative leap” (Rein & Schön, 1993, p. 146).

Policy frames differ from the frames that are deployed in other principles, such as media frames (Gamson & Modigliani, 1989) or framing in social movement studies (Druckman, Peterson, & Slothuus, 2013). The difference is that policy frames are conceptualized by most authors in the policy science field as coherent structures, while authors working in other disciplines acknowledge that frames can be dynamic given that they are jointly constructed and reconstructed by those using the frames and their target audience (Polletta & Ho, 2006, p. 192).

Schön and Rein distinguish two types of frames, rhetorical and action frames, and three different levels of frames, which they refer to as policy, institutional, and metacultural frames (Schön and Rein, 1994, p. 31). These distinctions are further elaborated on in following sections, as they form the foundation upon which the research conducted for this study was based.

The difference between rhetorical and action frames can be compared to the difference between policy debate and practice. Similarly, rhetorical and action frames are intertwined in much the same way as policy debate and practice.

Rhetorical frames

In policy debates, frames serve a rhetorical function. Persuasion, justification, and symbols are all rhetorical functions that are inherent to policy debate (Schön & Rein, 1994, p. 32). Stone notes that “groups […] portray issues deliberately in certain ways so as to win the allegiance of large number of people who agree (tacitly) to let the portrait speak for them” (2012, p. 310).

(13)

12 According to Stone, symbols are a powerful tool used in rhetorical frames; in this context, a symbol can be part of a story that not only describes a situation but provides a tacit

suggestion as to what should be done about it. A symbol stands for something else; the meaning of a symbol depends on how people interpret or respond to it (Stone, 2012, p. 158). Symbols can be anything. For example, an object such as the American flag can symbolize freedom; an individual such as George Floyd can symbolize police brutality and institutional racism; or a site such as the fields of Flanders symbolize remembrance and life. Given that symbols can represent almost anything, most constructivists claim that a particular situation is never “neutral” but is instead always embedded in the story that an actor wants to tell (da Vinha, 2017, p. 117).

Metaphors

Metaphors are important figures of speech in policy analysis, as they can be used to strategically represent a frame (Stone, 2012, p. 171). Superficially, they appear to draw a comparison between one thing and another, but, on another level, they imply a larger narrative and prescribe action. During the COVID-19 crisis, powerful metaphors were employed. Some leaders have said that there country is ‘at war’ with the virus. The view of waging war against a virus not only compares the current crisis with war, it also implied that, just as in war, extraordinary measures have to be taken to protect its citizens.

The symbolic devices are persuasive because, as Stone puts it, “[…] their story lines are subtle and their poetry so emotionally compelling that the normative leaps slip right past our rational brains” (Stone, 2012, p. 177).

Ambiguity

The most important feature of all of the symbolic devices used in rhetorical frames discussed above is arguably ambiguity (Stone, 2012, p. 178). A symbol can have more than one

meaning. For example, “free movement” can mean having no physical restrictions on movement, but it could also be interpreted as meaning that everyone will be provided with sufficient resources to move as they please People may interpret a symbol differently. Zygmunt Bauman (1991) notes that people become uneasy when they cannot “read” a situation: if we cannot read a situation as a coherent whole, we choose among alternatives to make a coherent story out of that situation” (p. 160). Bauman defines ambivalence (which can be used interchangeably with ambiguity) as the “possibility of assigning an object or an event to more than one category” (Bauman, 1991, p. 161). For example, to some people, the

(14)

13 flag of the Dutch East India Company may represent the “Golden Age” of the Netherlands, signifying the story of how the Netherlands grew to be the richest country around the world. For others, it could signify the Netherlands’ history of slavery and colonization.

Policymakers are believed to detest ambiguity. Stone (2012) argued that this is because of the uneasiness felt by policymakers when a situation does not fit in one particular category or cannot be understood in a particular way (p. 157). This uneasiness is because it is assumed that policymakers can employ classification: under which group of problems does a certain phenomenon fall? Mary Douglas (1986) observed that “institutions do the classifying” (p. 14). This means that the process of classifying a problem (problem definition) is an

institutional device for ordering. This ordering or classification is guided by routine (Douglas, 1986, p. 14). How, then, can we explain ambiguity in the policymaking process, in particular in risk management? In the current literature, there are two explanations as to why we can observe ambiguity in policy narratives. I discuss both explanations, as both are subsequently used in this research to attempt to identify the frames that have been used to depict the COVID-19 crisis. The first explanation of why we encounter ambiguity in policies is the role played by the concept of “bounded rationality

In the 1980s and 1990s, the rational actor model, which represented an attempt to explain human behavior, became subject to increasing criticism. Even in behavioral organization theory, behavioral decision theory, and economics, there was a notion that the rational choice model was a poor fit for describing human behavior and political decision-making (Jones, 1999, pp. 297-298). Bryan Jones’ (1999) answer to the question of whether this meant that people behave irrationally was “not at all” (p. 298), ). Instead, he suggested that irrational behavior can be explained with reference to the concept of bounded rationality. Bounded rationality assumes that both collective and individual decision-makers are inherently rational but have to deal with limited information and limited means of processing such information to make sense of complex policy issues (Jones, 1999, pp. 312-314). Due to the limited information that may be available concerning a policy issue, ambiguity can be a suitable rational solution. When more information becomes available or cognitive limitations to processing information have been lifted, policymakers can shift their emphasis to a more coherent story that is (more) unambiguous. Baumgartner and Jones viewed bounded

rationality as an inherent aspect of the policymaking process (Baumgartner & Jones, 1993, p. 10). According to the theory of bounded rationality, one is particularly like to find ambiguity

(15)

14 in policies related to issues on which limited information was available, such as the COVID-19 crisis.

The second explanation as to why ambiguity is found in the policymaking process is based on the interpretivist tradition. According to Stone, ambiguity can serve the strategic goals of policymakers (Stone, 2012, pp. 178-181). It can help to create alliances around a common policy by obscuring disagreement about specific details; it enables social movements and interest groups to unite people around broader objectives; it unites groups that would benefit or suffer in different ways as a result of a particular policy; it allows policymakers to address the concerns of both sides in a conflict; it allows political leaders to describe themselves as successful to constituents with different definitions of success; and it facilitates collective action (Stone, 2012, p. 181). Dvora Yanow (1997) also shows how politicians use ambiguity deliberately to resolve conflict and forge political coalitions (p. 226). All of the

abovementioned advantages mean that ambiguity is inherent to the political and policy arena and, in contrast to the notion of bounded rationality, which considers ambiguity as a strategy by which to overcome a lack of information, is perceived as a powerful tool from the

interpretivist perspective. In the constructivist paradigm, ambiguity is not viewed as a sign of weakness or lack of information but is rather considered a powerful and useful tool in the hands of political and policy practitioners.

Action frames – policy, institutional, and metacultural frames

In contrast to the rhetorical frames that are prevalent in policy debate, action frames inform policy practice (Schön & Rein, 1994, p. 32). According to Schön and Rein, a policy frame is “[…] the frame an institutional actor uses to construct the problem of a specific policy situation” (Schön & Rein, 1994, p. 33). Such frames are used to assign responsibility to certain actors and call for a specific policy response. Frames may identify specific objective, a type of solution, and the instruments that are deemed to be most effective to solve the issue (Weiss, 1989, p. 116). A number of authors have noted that these frames are coherent. As Maarten Hajer and David Laws (2006) put it, “[..] the frame is the internally coherent constellation of facts, values and action implications” (p 257).

This constellation of facts, values, and action can explain stability or changes in the policy domain based on the relative stability or contestability of a frame in the particular political discourse (Hajer & Laws, 2006, p. 257). Actors deliberately use frames when engaging in framing. Framing is seen as the deliberative act of “signifying” a frame to encourage others to

(16)

15 follow the same patterns of significations (Steinberg, 1998, p. 856). Policymakers, in this conceptualization, are signifying agents who try to push others to follow the same kind of “patterns of signification” (frames) (Steinberg, 1998, p. 856). This indicates that framing is an international and strategic activity.

Engagement in this strategic activity presumably involves the assumption that there is an ongoing struggle between the frame that is dominant in a particular political discourse and those that challenge the prevailing frame (Gamson & Modigliani, 1989, p. 7). As stated previously, frames are not proven or disproven by facts or evidence (Schön & Rein, 1994). This is because frames are closely linked to beliefs. By denouncing a previously adopted frame, we open ourselves up to doubt. Doubt is, according to Charles Peirce,

[..] an uneasy and dissatisfied state from which we struggle to free ourselves and pass into the state of belief; while the latter is a calm and satisfactory state which we do not wish to avoid, or to change to a belief in anything else. On the contrary, we cling tenaciously, not merely to believing, but to believing just what we do believe. (Peirce, 1877, as cited in Levi, 1991, p. 5).

The view that doubt is an uneasy state whereas belief is a satisfying state could explain the tenacity of dominant frames, even when faced with conflicting evidence. This could explain why it takes years or even decades for new evidence that challenges very dominant frames to be accepted. Consider, for example, the evidence concerning the detrimental effects of smoking on health. The findings were met with fierce opposition, and, to this day, some people still argue in the dominant frames that depict smoking as a question of freedom of choice.

These broad, culturally shared systems of belief are what Schön and Rein call metacultural frames. Metacultural frames are at the root of policy stories: they shape both rhetorical and action frames and are organized around generative metaphors (Schön and Rein, 1994, pp. 33-34). An example of a powerful metacultural frame that is also seen in the COVID-19 crisis is the nature versus nurture debate. Our perceptions of what can be done regarding a specific situation are affected by our perceptions of it as being “natural” or “environmental” in nature or not.

Certain authors consider pairs of metacultural frames as being ideologically opposed (e.g., nature versus nurture, disease versus cure, natural versus artificial, wholeness versus fragmentation). I argue that those opposites are ends of a spectrum.

(17)

16 In the next section, I will elaborate on risk and risk management, and show the linkages it has with the constructivist field and frames and framing in specific.

2.3 Risk and risk management

“Risk” is a concept the meaning of which is heavily dependent on the frame in which it is employed. Risk is essentially a neutral term referring to a probability or likelihood of an event occurring (Lupton, 1993, p. 425). It can be both positive or negative in its meaning: one can “risk” winning the lottery when buying a lottery ticket. However, as Mary Douglas and Aaron Wildavsky show in their book Risk and Culture (1983), in contemporary Western societies, risk has become interchangeable with the term “danger.” Thus, when we talk about “high risk,” we are actually talking about “significant danger”. Furthermore, the concept of “good risk” seems to be self-contradictory (Douglas & Wildavsky, 1982, p. 3).

The definition of an “acceptable” risk is open to interpretation. For some, the economic opportunities associated with the mining of coal outweigh the health and ecological risks. Such individuals would argue that the (economic) risk of putting a stop to these activities would not be worthwhile. In addition, as risk involves the probability of danger, it is

inherently political as a concept. Because involves probability, it can fit in most frames (if an that was deemed unlikely occurs, it could be it could be described as an instance of “bad luck,” meaning that one would not need to reconsider the assumptions of a particular frame. As Jerome Ravetz explains,

[…] risks are conceptually uncontrollable; one can never know whether one is doing enough to prevent a hazard from occurring. Even after a hazard has occurred, one is still left with the question of how much more action would have been necessary to have prevented it, and whether such action would have been within the bounds of reasonable behavior. (Ravetz, 1979, p. 299).

Thus, on the one hand, should a hazard arise despite policymakers previously having been warned of it, they could still claim that it would not have been reasonable to prepare for what they might choose to frame as a low-risk situation. On the other hand, policymakers can never prove to someone with a different frame that their actions actually prevented a hazard from occurring. This attribute of risk assessment is why, for example, antivaccination groups exist: there is no (ethical) way to prove the effectiveness of a vaccination program when a person does not share the same frame. In a sense, the overwhelming success of the

(18)

17 no longer felt in most countries. Presumably due to the antivaccine movement, since 2015 there has been an increase in outbreaks of vaccine-preventable infection diseases across Europe and the United States (Phadke, Bednarczyk, Salmon, & Omer, 2016, p. 1155). In the risk assessment theories stemming from the 1980s, risks in terms of public policy were generally grouped into the following four categories (Douglas & Wildavsky, 1982, p. 2):

1. Foreign affairs: the risk of foreign attack; war; loss of influence; prestige, and power; 2. Crime: internal collapse; failure of law and order; violence and white-collar crime; 3. Pollution: abuse of technology; fears for the environment; and

4. Economic failure: loss of prosperity

Theorists of this period recognized the fact that people in general and legislators in particular did not worry about each type of danger equally. In a survey administered in the 1980s, a clear distinction could be identified between what citizens and legislators considered to be the greatest dangers faced by the United States at the time (with citizens focusing on economy and energy and legislators emphasizing foreign affairs and crime rates) (Douglas &

Wildavsky, 1982, pp. 2-3). Douglas and Wildavsky also assumed that ranking risks, which is an inherent part of risk assessment, is a sort of zero-sum game: they observed that people who are, for instance,

more concerned about foreign attacks would likely be less worried about pollution (Douglas, Wildavsky, 1982, p. 3). As is shown in Figure 2-1, Douglas and Wildavsky recognize four different risk problems that all required a different way to manage. A risk

where the knowledge is certain

Figure 2.1. Four categories of risk (Douglas & Wildavsky, 1982, p. 5)

(19)

18 and the consent is complete, the way to manage the risk is those were consent is complete, but knowledge is certain or uncertain, and contested consent where knowledge is either certain or uncertain. As is shown in the figure, there are no clear solutions to managing risks that involve uncertainty and where knowledge is contested (the bottom-right quadrant). These kinds of risks have been called a wicked problem in the literature.

2.3.1 Wicked problem

The discourse around wicked problems emerged in the late 1960s and the early 1970s as a critique of the dominance of rational-technical approaches to decision-making in the planning and implementation of social policies. The academics in risk management argued for a more critical understanding of the significance of complex policy problems and the unforeseen consequences of policy interventions in areas characterized by risk and uncertainty (Head & Alford, 2013, p. 712). The understanding of a wicked problem is that it occurs within a system of problems; as problems tend to be interrelated. Authors in the field of risk management argued that one cannot solve problems independently of each other, as attempting to do so would only create a bigger “mess” and may produce unforeseen

consequences in other problem areas. Those who advocated the systems theory argued for the need to adopt a holistic approach to complex problems (Ackoff, 1974, p. 21). The second critique of the rational-technical approaches was based on scholars’ arguments that social issues should be considered from value perspectives (frames), not as technical issues. To these constructivist writers, “more information” To these constructivist writer, obtaining “more information” was not always the priority, and they did not consider obtaining more information as the only way to resolve complex issues (Rein, 1976, p. 10).

Wicked problems are generally associated with 1) social pluralism, 2) institutional complexity (e.g., they may involve interorganizational cooperation and multilevel governance), 3) and scientific uncertainty (Head & Alford, 2013, p. 716). In 1973, Horst Rittel and Melvin Webber concluded in their highly influential work Dilemmas in a General Theory of Planning that there are no criteria with which to assess the success of different policy approaches to wicked problems and that there is no way to “learn” in the policymaking process (Rittel & Webber, 1973, p. 167). Since then, different authors have advanced various ways of addressing wicked problems that seem more promising than. In the following

sections I elaborate on the adaptive strategies proposed by Donald Kettl and Matthijs Hisschemöller and Rob Hoppe’s notion of policy as learning.

(20)

19 Adaptive strategies

In his book The Next Government of the United States: Why Our Institutions Fail Us and How to Fix Them, Donald Kettl argues that “many of the most important problems we face simply do not match the institutions we have created to govern them” (Kettl, 2008, p. 25). He identifies many new challenges and risks faced by contemporary societies, such as terrorism, natural disasters, climate change, and pandemics There is one commonality in human

responses to these wicked issues: there are no single organizations dedicated to responding to individual issues (Kettl, 2008, p. 26). Problems are defined based on the institutions that can deal with them, instead of the other way around. In other words, the existence of frames for a particular issue depends on the current institutional settings in which the frames are situated. Generally speaking, one ministry is responsible for a particular problem. For example, the Dutch Ministry of Economic Affairs and Climate is the chief actor responsible for addressing climate change. The fact that the issue of climate change is “assigned” to this ministry’ (and the fact that economic affairs and climate share the same ministry) has ramifications in terms of which solutions intended to mitigate the effects of climate change and prevent further use of carbon-based fuels can be proposed. A wicked problem such as climate change cannot be solved by one or a number of ministries. Instead, Kettle proposes reforming our institutions by focusing on results, collaboration, and the creation of “rocket science public leaders” to adapt to the problems our society is facing instead of attempting to adapt the problems to the designs of our institutions. He compares good policymakers with rocket scientists, as rocket science relies on collaboration, flexibility during uncertainty, individual reflexivity, and leadership (Kettl, 2008, p. 221).

Policy as learning

Hisschemöller and Hoppe built upon the idea of frame reflection as a way of resolving intractable controversies introduced by Schön and Rein in their book Frame Reflection (1994). Hisschemöller and Hoppe attempted to determine whether and under what conditions it is possible to successfully cope with intractable controversies (Hisschemöller & Hoppe, 1995, p. 42). As illustrated in Figure 2-2, they distinguished four problem types. They define wicked problems as those for which no technical tools capable of overcoming the issue exist: ‘If a problem is ill-defined, “wicked”[…], “messy” […], “ill-structured”, or unstructured, the term we prefer, technical methods for problem solving appear inadequate’ (Hisschemöller & Hoppe, 1995, p. 43).

(21)

20 Thus, according to Hisschemöller and Hoppe,

a wicked problem refers to a situation in which there is no consensus regarding the relevant norms and values and there is a degree of uncertainty regarding the

information that is available. This definition closely resembles the typology developed by Douglas and Wildavsky, shown in Figure 2-1 (although the table is flipped). However, whereas Douglas and Wildavsky did not

identify an apparent solution to a wicked problem, Hisschemöller and Hoppe propose the policy-as-learning method as a solution for

intractable controversies. They argue that policy as facilitates the identification of problems, as the input of citizens with conflicting frames and interests will likely produce a new vision, which is necessary to structure a wicked problem (Hisschemöller and Hoppe, 1995, p. 57). A crucial aspect of this strategy is ensuring that no participants are excluded from the

policymaking process. Hisschemöller and Hoppe argue that there is already a bias present in the problem definition. This bias then becomes an implicit justification for excluding

potential actors and issues from policy debates (Hisschemöller and Hoppe, 1995, p. 57). They identify four conditions that need to be met to implement the policy-as-learning approach: 1) some segments of the official policy elite need to interact with those with alternative views on the problem; 2) all actors, especially authorities, should be willing to participate (this also means investing time and resources so that a broad range of options can be included in the discussion); 3) the problem should be discussed not in abstract but in concrete terms and with regard to the lived experiences involved; and 4) policy decisions should not be made before new insights concerning the problem and its potential solutions have been obtained.

Both the adaptive strategy proposed by Kettl and the policy-as-learning strategy described by Hisschemöller and Hoppe provide helpful insights into how we can deal with wicked

problems. However, both strategies require a resource that was not available during the outbreak of the COVID-19 crisis: time. Both strategies require considerable amounts of time, as the first requires a substantially restructuring of institutions (which one could refer to as institutional hardware), while the policy-as-learning strategy requires changing policy process

Figure 2.2. Four types of policy problems. (Source: Hisschemöller &

(22)

21 (institutional software), and implementing either of these strategies successfully requires a fairly significant amount of time.

2.3.2 COVID-19, a wicked problem

The asymptomatic nature of the coronavirus not only influences its CFR numbers but also means that the virus is shrouded in uncertainty. The symptoms of SARS-CoV-2 were unclear from the initial outbreak. There were reports that stated that a fever was necessary to test positive for the coronavirus, while others reported asymptomatic infected people (Gandhi, Yokoe, & Havlir, 2020, p. 2158). It remains unknown how many people have actually contracted the coronavirus at the time of writing, and the likelihood that researchers will be able to make a precise estimate in the future is slim. In addition, it remains debatable which measures are effective against the new virus. The initial response was the same as that adopted during the 2003 SARS outbreak: symptom-based case detection and subsequent testing were used as guidelines for isolations and quarantine, as the new virus is highly genetically related to SARS (Gandhi, Yokoe, & Havlir, 2020, p. 2158). However, it soon became clear that SARS-CoV-2 possesses features that significantly differ from those of SARS-CoV.

This uncertainty, particularly during the first few months in which the virus spread across the globe, led to a great deal of misinformation and a number of conspiracy theories being propagated A number of false claims were made; for example, one false claim was that, if people could hold their breath for 10 seconds, they did not have the coronavirus (Landsverk, 2020). Hydroxychloroquine, an anti-malarial drug that has been touted as a cure to the coronavirus, has even being promoted by the President of the United States, Donald Trump (D. J. [realDonaldTrump] Trump, 2020). This medicine can potentially cause severe or even deadly side effects (Taylor, 2020).

Thus, when we consider the theories of both Douglas and Hisschemöller and Hoppe, we can conclude that the COVID-19 virus falls into the category in which (relevant) knowledge is uncertain/not available at the moment. In addition, we can see that countries have exhibited a wide divergence in terms of policy measures taken. This is probably due to not only the uncertainty concerning the validity of the relevant knowledge but also contested values and norms. As the virus is spreading and more severe restrictions on social life are being imposed, arguments have been made as to whether the measures taken are worthwhile considering the substantial damage that has been done to many countries’ economies,

(23)

22 whether the lockdowns that have been imposed in many countries are legal, and whether the risks of the coronavirus justify infringing on the citizens’ right to privacy. All of these factors mean that the COVID-19 crisis can be categorized as a wicked problem. Currently, most (constructivist) answers to wicked problems involve deliberation or reconstructing institutions that are currently incapable of solving the policy issues associated with such problems. However, due to the nature of the COVID-19 crisis, these approaches do not seem to be appropriate in this specific case, as the crisis demands immediate action, and

governments are still faced with significant uncertainty.

Quantitative framing analysis in risk management – other variables

In the previous sections, the link between frames and the risk management literature has been shown. In the literature, variables have been acknowledged that can impact the way that the government communicates on risks to their citizens. Candlin and Candlin (2002) illustrate that the amount of money spend on healthcare can impact the discourses around risks and risks management. Governments that spend relatively more on healthcare, talk more in risk prevention, while governments that spend relatively less on healthcare talk about risk avoidance (Candlin & Candlin, 2002, p. 128). Böhm et al. show give a classification of the different types of healthcare systems. These healthcare systems have different fiscal incentives to deal with risk, which might be a contributor to a different approach to risk (Böhm, Schmid, Götze, Landwehr, & Rothgang, 2013, p. 262). The last factor that is of interest in a quantitative study, is if the responsible authority has the means or knowledge to reduce risks. In the case of the COVID-19 crisis, knowledge is uncertain, but governments can severely restrict public life. To understand how they manage risks, it is of importance to also take into account the restrictions they impose (Liu & Kim, 2011, p. 242).

(24)

23

Chapter 3. Methodology

In the research project, I have utilized both a quantitative and a qualitative framing analysis. In the first part of the methodology explains the methods and data used for the quantitative analysis, which variables have been included based on the theoretical framework, and why excess mortality rates have been chosen to express the severity of the COVID-19 pandemic in each country. The second part shows on which parts the qualitative analysis has focused to deepen our understanding of the usage of frames during the COVID-19 crisis.

3.1 Quantitative framing analysis 3.1.1 Methods

In the first part of this analysis, I analyze 9 countries all of which are (ex) European Union (EU) member states; Belgium, Denmark, France, Germany, Italy, the Netherlands, Spain, Sweden, and the United Kingdom. All official communication from these countries’

government agencies have been selected, and those that contained one of the following terms were included in the framing analysis: “Corona”, “Coronavirus”, “Corona crisis”, “Health crisis”, “COVID-19”, “Sars-CoV-2”. I constructed a coding template that I applied to categorize the data. This coding template was constructed based on previous literature that analyzed framing in risk management debates. The coding template is shown in Figure 4.1. All documents were reviewed manually. Using some kind of automatic approach to find keywords was considered, but this approach was not adopted given that most texts were translated, which could mean that different words might be used to describe the same frame. In other words, relying on keywords would have increased the likelihood of meaning being lost in translation, which would impact both internal and external validity of the research.

(25)

24

Figure 3.1. Coding template

Frames Description Count

Responsibility Who is to blame for the COVID-19 crisis or outbreak Economic consequences Focusses on the negative

economic effects of the crisis Human aspect Focusses on the stories of

individuals or communities that are affected by COVid-19 Collaboration Emphasizes the importance to

collaborate during the crisis Severity Focusses on the potential or

actual damage during the crisis, emphasizing the critical

situation.

Action Discusses the past and present response to COVID-19 New Evidence Discusses the discovery of new

evidence (whether it be how the virus works or how to avoid risks to contract the

virus)

Reassurance Telling the public to not worry, as the crisis is under control or the organizations in charge are

capable to deal with it. Uncertainty Discusses the uncertainty about

the crisis. This includes cause, the cure and how it is being

spread.

Disease detection Explains how the virus can be detected and what the

symptoms are Disease prevention Discusses how to prevent the

health problems and minimize risk in the future Healthcare services Discusses the health system in

general and the healthcare workers

Lifestyle risk factors Focusses on the personal behavior that affects the individual and community during the COVID-19 crisis. Other Any other frame that might be

communicated Figure 3.1. Coding template

3.1.2 Data

All written official announcements made by the chosen countries between the 31st of January and the 1st of May of 2020were analyzed. Then, documents containing short notifications of the numbers of patients who have officially tested positive for COVID-19 and events that

(26)

25 have been called due to the coronavirus have been filtered out and are not included in this analysis. Some countries (such as the Netherlands) reported extensively on these issues, while others did not. Communications that summarize previous press conferences were also not included in the analysis. It would not be interesting to analyze summaries of current affairs in a policy frame analysis, as they do not propose new policies or provide advice. As some countries produce vastly more communications surrounding COVID-19, the analysis of each phase will identify the most commonly used frames. Otherwise, there will be a discrepancy between countries not on the basis of content, but the basis on documents; some countries (e.g., Germany), publish most of their policy-related communications in a limited number of documents, while others produce shorter texts to inform the public (e.g., the United Kingdom held daily press conferences on weekdays from the 12th of March onwards). Thus I have limited my quantitative framing analysis to not include every single announcement. All links to the analyzed documents are laid out in Appendix I.

The statistical data presented in this thesis concerning the coronavirus was obtained from the European Centre for Disease Prevention and Control (ECDC) (European Centre for Disease Prevention and Control , 2020), and Johns Hopkins University’s website (Johns Hopkins University, n.d.). I have divided the course of the COVID-19 crisis in three phases to demonstrate the different stages of the novel coronavirus: phase I, phase II, and phase III. These stages are based on a categorization made by the WHO and other authors who categorize pandemics (see World Health Organization, 2020a; and Pan & Meng, 2016) Phase 1 started in every country on the 31st_{of January. On the 31}st_{of January, the first}

confirmed patients tested positive for the new corona virus on the European continent. In Italy, two Chinese tourists were tested positive (Poggioli, 2020). Phase I ended when

cumulatively, one per million people tested positive for COVID-19 in a country. The increase of people testing positive for COVID-19 signaled the beginning of phase 2. When a rate of one person per million inhabitants is achieved, it is likely that a virus is being transmitted within communities. This rate signals that an exponential growth in people contracting COVID-19 is likely to occur. Phase 3 begun when, cumulatively, the confirmed ratio of corona-related deaths is 0.1 per million inhabitants. In this phase, the pandemic is having severe effects on public health. Phase III ended when the amount confirmed COVID-19 cases per day is lower than 7 days prior, using a rolling 7-day average. This development

(27)

26 potentially1_{means that the exponential growth has been halted, and the outbreak can be}

considered to be “under control”.

As table 3.1 shows, there is a great difference between countries and the amount of press releases they give. A critical note here is that these numbers are for official communication in either English or Dutch (for the Netherlands and Belgium). Italy, for instance, has almost no communication about COVID-19 in English before the month April: most communication is done in Italian. As the third phase starts in Italy around March 28th, April has not been

analyzed on communication from Italian authorities. But all other countries besides Italy have analyzed documents in at least two out of three phases, as shown in table 3.1. Most countries have communication in phase II, although France and Italy are the exceptions here. For France, this is likely due to the fact phase II only lasted five days.

Table 3.1. Number of documents reviewed Table 3.12_{. Number of documents reviewed}

Country Phase I Phase II Phase III3 _Total

Belgium 0 1 8 9 Denmark 0 4 1 5 France 2 0 6 8 Germany 2 5 2 9 Italy 0 0 4 4 The Netherlands 3 2 27 29 Spain 8 4 35 47 Sweden 0 2 10 12 The United Kingdom 11 4 57 73 Total 26 21 149 196

There is one important reason why the different phases are demarcated using either the variable “tested positive for COVID-19” or by the variable “official death rates”. For phase 1, the death rate is not particular useful, as this rate has a lag: on average people die after 18.5 days (Knapton, 2020). This is why for phase I, it would not have been useful to use the death rate, it has a two-and-a-half-week lag: the exponential growth in the number of cases has already started during this time period. The reason why official death rates are measured in

1_{It doesn’t necessarily means that the exponential growth has halted. Other factors, such as testing facilities or}

the amount of available tests can determine this as well.

2_{The different colors are to enhance the readability of the table. It does have any statistical meaning.}

3_{For Sweden and the United Kingdom, all documents up to the 1}st_{of May have been reviewed, as phase 3 has}

(28)

27 phase II is because deaths result in a greater likelihood of political messages being sent and policy measures being adopted, as citizen deaths are almost always a major trigger for public outrage (Shubik, 1991, p. 14). Phase III utilized the variable ‘tested positive for COVID-19’ again for the same reasons as it was used to describe phase 1: there is a two-and-a-half-week lag, so death rates will decrease slower than the amount of people getting infected by the COVID-19 disease.

Frames were recognized by key words and phrases. Some frames were easy to identify, based on a few sentences, while other frames required an understanding of the entire document. The economic consequences frame, for instance, has been easy to identify. In document [118]4, for example, the Swedish government “[…] is now presenting additional measures to mitigate the financial impact of the virus outbreak.” [118]. This is a clear example of an ‘economic consequences frame’: the document describes that there is a financial impact due to the novel coronavirus. However, a ‘severity’ frame is a bit more difficult to frame. It does not entail that the coronavirus is an event that has to be taken serious: most countries

acknowledge the seriousness of the virus in phase II and onwards. However, this is not what the ‘severity’ frame tries to capture. To find a ‘severity frame’ in a document, I looked for language comparing the COVID-19 crisis with other disastrous events and concluding it is as bad/worse than the mentioned crisis, or language underlining the extreme urgent situation. The COVID-19 crisis has been compared with other major events such as the 2008 financial crisis [80, 124 ], World War II [25], and the Great Depression [195].

Information on additional factors that have been identified in risk management theory as having a potential impact on the effectiveness of policy measures during a public health crisis has also been gathered.. These additional categories are: public health regime (based on the categorization of Katharina Böhm (2013)), current health expenditure as percentage of GDP (data acquired from The World Bank Group (World Bank Group, n.d.)), and the excess morality rate (data acquired from the ECDC (2020b)).

3.1.3 Excess mortality rates

Official corona diagnostics and death rates were used in the quantitative framing analysis. Thus, the following question arises: why use excess mortality rates to measure the impact of the frames? The reason for this approach is that the official death rates and the numbers of

(29)

28 people who tested positive for the coronavirus depend on the methodology adopted, capacity of tests, and other political factors. In other words, these figures do not give an accurate impression of the real impact of the coronavirus. For instance, most countries only count a patient’s death as being corona-related if that individual had been officially diagnosed. However, Belgium and France, for example, takes into account all suspicious deaths that might be corona-related. Some countries have been testing rigorously, while others have only been testing those who are admitted to the hospital and have COVID-19 symptoms. This means that those countries that test more extensively will report in all likelihood a higher amount of infected individuals per one million inhabitants than those that do not.

However, if this is the case, why do we measure official death rates and the number of people who tested positive for the novel coronavirus? The answer is that, in all likelihood, politicians and policy makers do not respond to the real impact of a crisis, which might be obscured at the moment measures have to be taken, but to the political reality. Thus, when investigating the frames used by government agencies, considering the political reality is more helpful in terms of explaining the choices made in government communications than the real impact, which is most likely still obscured at the time of a press release.

To measure the actual impact, I am comparing EMR between countries. The WHO defines “excess mortality” as

Mortality above what would be expected based on the non-crisis mortality rate in the population of interest. Excess mortality is thus mortality that is attributable to the crisis conditions. (World Health Organization, 2014).

Using excess mortality during the COVID-19 crisis in this thesis means considering the numbers of registered deaths during phases I, II and III, and comparing them to the average numbers of deaths recorded in prior years. Focusing on the EMR is not a flawless approach either, as no statistical method is, but I argue that it is more accurate than considering official death and infection rates and more relevant than figures indicating economic decline. In this case, considering EMRs is an approach adopted to gauge the impact of the coronavirus over the short term.. I argue that already trying to determine the impact on the long-term is not fruitful during the pandemic, as the overall impact of the virus is still uncertain. Table 4.5 presents the EMRs per country per phase based on the data obtained from Eurostat.

(30)

29 Other factors

As described in the theoretical framework, the risk management literature suggests a

multitude of different factors that can have an impact on the effect of risk management. The literature suggests that the following factors will have an impact on the way risks can be managed: healthcare regimes, percentage of GDP spent on healthcare, risk frames, and actual measures taken. The categorizations of healthcare systems is based on the work of Böhm et al. (2013), the percentage of GDP spent on health care is based on data from the World Bank, and the categorizations of measurements being taken is based on an analysis from Politico (Hirsch, 2020). The severity of the lockdowns due to the COVID-19 crisis have been put on a scale, where 1 is having the least stringent measurements, and 5 having a severe lockdown which is being enforced by law . This has been a categorization based on a rapport of the European Commission (European Commission, 2020).

As described in the theoretical framework about risk management, certain factors need to be taken into account to understand how risk frames function. The literature suggests that the following factors have an impact on the way in which risks can be managed: healthcare regimes, percentage of GDP spent on healthcare, risk frames, and actual measures taken. The categorizations of healthcare systems is based on the work of Böhm et al. (2013), the

percentage of GDP spent on health care is determined on data from the World Bank, and the categorizations of measurements adopted by the analyzed countries is based on an analysis published by Politico (Hirsch, 2020). Table 4.6 displays the information gathered from these different sources, which includes my own analysis. To enhance the clarity of the table, certain categories have been collapsed. The fifth column highlights the three frames most commonly used by each country. The fourth row, severity of measures, attempts to rate the measures taken against COVID-19 on a scale, where 1 indicates the least stringent measures and 5 a severe lockdown. The measures were categorized based on a report published by the European Commission (European Commission, 2020).

3.2 Qualitative framing analysis 3.2.1 Methods

Some results of the quantitative framing analysis are hard to understand within the

framework of the quantitative framing analysis. To get a better understanding, I conducted a qualitative framing analysis. The qualitative framing analysis is, as is shown in the analysis, vital to induce the overarching story of the frames being utilized in the COVID-19 pandemic.

(31)

30 The goal of the qualitative framing analysis is to deepen our understanding of the results that have been generated by the quantitative analysis. The quality framing analysis involves repeated and extensive engagement with the documents and looks at the material holistically to identify frames (Connolly-Ahern & Broadway, 2008, p. 369).

3.2.2 Data

Based on the quantitative analysis, I have focused on four cases: Germany, France, the Netherlands, and Sweden. I compare Germany with France, and the Netherlands with

Sweden. For the qualitative framing analysis, I have focused on the speeches where the heads of states address their respective nation. This data can be found in Appendix II.

(32)

31

Chapter 4. Results

The results are divided into two parts, with the first focusing on the quantitative analysis and the second one on the qualitative analysis. In section 4.1, which is the quantitative research analysis I present the results of the frame analysis of all of the official communications that were published in the selected 10 countries between the start of phase I and May 1st. This chapter is divided into three sections, which are based on the sub-questions formulated in the introduction. The first section describes the particular frames in which the governments of the analyzed European countries communicated their COVID-19-related policies. The second section considers the excess mortality rates in these countries. Finally, the third section investigates the relationship between the excess mortality rate and the frames that these countries’ governments utilized.

In the second part, I explore why some countries use the same frames but have differences in terms of excess mortality rate. A qualitative framing analysis is conducted to investigate the frames that have been employed by the Dutch, French, German, and Swedish governments. The first section of this part shows that, contrary to what the quantitative analysis suggests, these countries did not adopt similar approaches to framing. When one considers the frames that have been utilized in detail, it becomes clear that the metacultural frames adopted by the chosen countries affect both the excess mortality rates and the policy measures being taken. The second section discusses the implications that these findings have on the use of framing analysis in the field of risk management.

4.1 Quantitative framing analysis

4.1.1 Official government communication regarding COVID-19.

As noted in the methodology section, the phases differ for each country, as COVID-19 did not affect all of the analyzed countries simultaneously. Table 4.1 categorizes the phases per country based on the information obtained from the ECDC.

(33)

32 Table 4.1. Start of each phase per country

Table 4.15_{. Start of each phase per country}

Country Phase I Phase II Phase III6

Belgium Jan 29th – March 4th March 4th – March 12th March 12th – April 16th

Denmark Jan 29th - March 4th March 4th – March 16th March 16th – April

15th

France Jan 29th_{- March 1}st _{March 1}st_{– March 6}th _{March 6}th_{– April 7}th

Germany Jan 29th - March 4th March 1st – March 15th March 15th – March

28th

Italy Jan 29th – Feb 23rd Feb 23th – Feb 25th Feb 25th – March 29th

The Netherlands Jan 29th – March 3rd March 3rd – March 9th March 9th – April 18th

Spain Jan 29th – Feb 29th Feb 29th – March 7th March 7th – April 3rd

Sweden Jan 29th - Feb 29th Feb 29th – March 12th March 12th - xxxxx

The United Kingdom

Jan 29th_{– March 5}th _{March 5}th_{– March 11}th _{March 11}th_{– xxxxx}

Table 4.1 shows that, in most of the analyzed countries (with the exception of Italy), phase one started at around the same time (between February 29th and March 5th). Phase 2 exhibits slightly more variation, but what is striking is that phase 3, the phase in which the number of patients who have officially tested positive for COVID-19 begins to decline, shows great variance among the chosen countries. Germany, for example, saw the number of infected people begin to decrease on the 28th_{of March, while the United Kingdom and Sweden had}

still not reached phase III by May 1st_{. While chance is likely a factor in the spread of}

COVID-19, these discrepancies in terms of the start of each phase suggest that policies do affect the trajectory of the pandemic.

Based on the coding scheme shown in the Methodology section, I have analyzed the documents and identified which frames were utilized in the official communications of the analyzed countries. Table 4.2 presents the frames that were identified during each phase. In each analyzed document, the coding scheme was used to see which frames were present in the documents. As the difference between the amount of frames observed and the number of documents analyzed suggests, more than one frame could be found in a text. Frames are, as discusses in the theoretical framework, a selection of a perceived reality. They promote a

5_{Note that the differences in color are to aid the readability of all the tables and do not have an analytical}

meaning. This is the case for all tables represented in this thesis.

6_{For Sweden and the United Kingdom, all documents up to the 1}st_{of May were reviewed, as phase 3 had not}

(34)

33 particular problem definition, causal interpretation, more evaluation and recommendation (Entman, 1993, p. 52).

The percentages shown in table 4.3 are the percentages of each frame present compared to all frames analyzed. So, for instance, in Phase II, the ‘human aspect frame’ was found 2 times, 3,6% of the time out of the 57 times a frame was identified. In total, the ‘human aspect frame’ was analyzed 32 times, which is 5,2% of 616, the total amount of frames that have been found in the analyzed documents.

Table 4.2. Risk frames analyzed in official documents Table 4.2. Risk frames analyzed in official documents

Frames Phase I Phase II Phase III Total

Responsibility 0 (0%) 0 (0%) 5 (0.9%) 5 (0.8%) Economic consequences 0 (0%) 6 (10.7%) 56 (10.6%) 62 (10.1%) Collaboration 2 (6.9%) 6 (10.7%) 35 (6.6%) 43 (7%) Human aspect 0 (0%) 2 (3.6%) 30 (5.7%) 32 (5.2%) Severity 0 (0%) 2 (3.6%) 43 (8.1%) 47 (7.6%) Action 2 (6.9%) 6 (10.7%) 115 (21.7%) 122 (19.8%) New Evidence 0 (0%) 2 (3.6%) 19 (3.6%) 21 (3.4%) Reassurance 10 (34.5%) 7 (12.5%) 14 (2.6%) 31 (5%) Uncertainty 1 (3.4%) 8 (14.3%) 56 (10.6%) 65 (10.6%) Disease detection 2 (6.8%) 5 (8.9%) 19 (3.6%) 26 (4.2%) Disease prevention 4 (13.8%) 3 (5.4%) 35 (6.6%) 42 (6.8%) Healthcare services 5 (17.2%) 6 (10.7%) 44 (8.3%) 55 (8.9%) Lifestyle risk factors 3 (10.3%) 4 (7.1%) 54 (10.2%) 59 (9.5%) Other 0 (0%) 0 (0%) 5 (0.9%) 5 (0.8%) Total 29 (100%) 57 (100%) 530 (100%) 616 (100%)

As Table 4.3 illustrates, government communication have deployed considerably more frames during phase III than they did previously. The documents analyzed in phase I

contained around one (1.1) frame, while the documents analyzed in phase II averaged around 3.5 frames per document. Furthermore, the most common frames in the analyzed documents differs from phase to phase. Table 4.4 shows the most applied frames for each phase.