“Damned If You Do, Damned If You Don’t”: Media Frames of Responsibility and Accountability in Handling a Wildfire

Previous research on media framing of wildfire has chiefly been concerned with the nature of wildfire in the context of climate change and with framing effects on policy and public opinion. Empirical studies on media content, hence what is mediated to crisis managers and the public concerning authorities&#8217; and the public's response, seem to be largely missing. This is remarkable, given that the media represent main sources of information that may influence crisis management and shape public opinion. Thus, the aim of this study was to identify key media frames relating to portrayals of public and authority responses during and after a wildfire crisis. The study is based on media articles from two time periods: immediately after the fire and 1&#160;year later. We used a thematic method of analysis (TA), thus an inductive, &#8220;bottom&#8208;up&#8221; approach. A core frame, Responsibility/accountability is identified, underpinned by two sub&#8208;themes. One sub&#8208;theme relates to the causes of the fire and its escalation, revealing a number of different interrelated blame frames. The second sub&#8208;theme refers to management of the crisis, reflecting both authorities&#8217; and citizens&#8217; responses. The deficiencies of the former are implicitly suggested to have forced citizens to act to compensate for their inadequacy. The main theoretical contribution is the identification of an interrelationship between frames in relation to different groups of individuals responding to a crisis, pointing to a more complex view of framing effects. In addition, results show how media tend to assess crisis management based on idealistic criteria, inevitably making the evaluation negative. This contributes to an understanding of how media blame frames, thus &#8220;blame games,&#8221; may unfold. Practical implications of these results are discussed.

In this case, due to the magnitude of the event, for the first time the State (County Administrative Board) took over from local and regional actors and assumed operative responsibility. A total of 13,800 hectares of forest were affected, about 25 buildings were burned down or fire damaged. There was one fatality. The rescue operation was only officially finalized after 6 weeks, on the 11 September. At the time, the event received great attention in Swedish media reporting.
Media framing refers to media communication in the sense of media presenting and defining an issue. It is a matter of "culling a few elements of perceived reality and assembling a narrative that highlights connections among them to promote a particular interpretation" (Entman, 2007:164), a process that involves the production, content and use of different perspectives (de Vreese, 2005). By doing so, media raise the importance of certain ideas and encourage individuals to think, feel and decide in particular ways (Entman, 2007, p. 164). Entman (1993) suggests that fully developed frames typically perform four functions: problem definition, causal analysis, moral judgment and remedy promotion.
Previous research shows how media framing generally plays a significant role in crisis evolution (Schultz, Kleinnijenhuis, Oegema, Utz, & vad Atteveldt, 2012), for example by attributing meaning to the event (van der Meer & Verhoeven, 2013). Exposure to certain frames can activate knowledge, affect learning, interpretation and evaluation of issues and events, alter attitudes, stimulate stocks of cultural morals and values as well as create contexts (see, e.g., de Vreese, 2005). Media have been shown to shape and alter individuals' interpretations and preferences and are a powerful instrument in reflecting and shaping public opinion (Fan, 1988;Knight, 1999).
Thus, they may, for example, influence perceptions of who is responsible for causing or solving key social problems (Iyengar & Kinder, 1987). While the media may shape public risk perceptions, they also articulate public opinion and thereby play an important role in policymaking (Nelkin, 1987). There is evidence that media may influence crisis management, particularly when policy is unclear, media coverage is critically framed and focuses on suffering people, or when there is a lack of government consensus, as is common in the early stages of a crisis. Examples are the Cable News Network effect during humanitarian crises or the media invoking a rationale for humanitarian military intervention (Bloch-Elkon, 2007;Robinson, 2000). Consequences may thus be conceived on the societal level by the shaping of social level processes such as political socialization, decision-making and collective actions (de Vreese, 2005). In that regard, media may influence the distribution of power by determining who gets what, when and how (Entman, 2007).
While the way that media frame crisis (management) has been shown to affect both public opinion and the way managers respond to crisis, another line of research focuses on how media frame the public during crisis. The main results show how media tend to contribute to erroneous beliefs about human behaviour in stressful situations (Goltz, 1984;Wenger & Friedman, 1986). For example, a number of critical analyses of the reporting on Hurricane Katrina in 2005 demonstrate how media framing in terms of looting, social disorganization and deviant behaviour resulted in very negative consequences for the management of this disaster (Constable, 2008;Tierney, Bevc, & Kuligowski, 2006). The experiences from Katrina indicate links between media reporting, authorities' expectations of citizen behaviours, crisis management actions and real-life consequences for those affected by the disaster. In addition, a study by Nilsson, Alvinius, and Enander (2016) illustrates an often quite subtle bias inherent in the framing of the public during crisis that is constructed through media processes of identification, characterization and evaluation. These findings have several implications for the management of critical events in society, as the normative framing of certain public behaviours may influence citizen responses via implicit messages concerning vulnerability, as well as expected and appropriate/non-appropriate reactions. This in turn may influence negatively on societal planning and allocation of resources for support to different affected groups. Based on the above, it appears that both crisis managers and citizens need to be sensitive to framing processes in media crisis reporting. Whereas frames may be used to reduce damage done, they may also form negative impressions towards individuals, organizations and society (An & Gower, 2009).
Global predictions warn of a dramatic increase in wildfire risk in a warming climate, by up to 20%-50% in disaster-prone landscapes, indicating that communities will face increased vulnerability to these risks (Bowman et al., 2017;Brenkert-Smith, Dickinson, Champ, & Flores, 2013;Calkin, Cohen, Finney, & Thompson, 2014;Calkin, Thompson, & Finney, 2015;Crow et al., 2017). Thus, the issue of how communities perceive and react to wildfire risk is important. There is evidence that news media are an important information source used by citizens to collectively define and respond to wildfire events (McCaffrey, 2004;Shindler, Brunson, & Stankey, 2002). However, according to Paveglio, Norton, and Carrol (2011) media as a component of public understanding or response to wildfire is not as well understood in comparison with other hazards. Looking at the literature, some scholars have studied how media frame the nature of wildfire.
For example, some work has emphasized wildfire in the context of climate change and more specifically, to what extent climate change may constitute an underlying cause to wildfire (Berglez & Lidskog, 2019). One conclusion is that the strong dramatization of wildfire in media may tend to push the climate change problem off the news agenda, although other studies argue that extreme events may in fact lead to a greater focus on climate change (see, e.g., Boykoff, 2012).
There are also studies that discuss natural disasters, including wildfires, in the public policy literature, suggesting that media attention may garner the consideration of the public and decision-makers and in consequence, open windows of opportunities for policy change to reduce community vulnerability to local risks (Crow et al., 2017). The literature gives evidence of an influence of media policy narratives on public opinion (Shanahan, McBeth, & Hathaway, 2011).
Based on the aforementioned, it appears that key issues regarding media framing of wildfire are concerned with the nature of wildfire in the context of climate change and framing effects on policy and public opinion. Empirical studies on actual media content, of authorities or the public's response separately or in interaction, hence what may be Electronic copy available at: https://ssrn.com/abstract=3605670 mediated through media to crisis managers and the public, seem to be largely missing. Given that the media represent the main sources of information for citizens, and that public support is often a necessary condition for policy implementation, media content is suggested to be a critical component for public and political interactions (Takahashi, 2011). Subsequently, to more fully understand media framing effects of wildfire events, it appears necessary to have a more thorough understanding of how media represent authorities and the public's responses. Scholars have indeed called for studies that examine "the underlying themes that guide broader representations of wildfire in the media including media coverage during and immediately after such events" (Paveglio et al., 2011, p. 43). Thus, the aim of this study was to identify key media frames relating to portrayals of public and authority responses during and after a wildfire crisis.
Previous research shows that media tend to cover what is happening now: the current impact on humans, the built environment and the natural environment. There is significantly more media coverage of wildfire in times of immediate risk rather than during the longer aftermath of anevent (Johnson, Bengtson, & Fan, 2009). This may be of significance, as frames tend to change, making different aspect of events more pronounced at different times, a process known as "frame changing." Based on this, it seems important to understand how frames may change over time and with regard to the different phases of crisis management. The psychological impact of a disaster can also be expected to differ depending on how much time has passed (Houston, Pfefferbaum, & Rozenholtz, 2012). Moreover, the dynamic process of media building up the salience of a topic on the media agenda has received little attention (Chyi & McCombs, 2004).
The focus of the present study is not only on the operational crisis, but also on the aftermath and the accountability phase in order to explore how frames may change over the course of the event.
This paper is organized as follows. Methodological considerations are described, followed by a presentation of the results. The Results section is structured in two parts, describing the first and second major themes identified. As the emphasis of the article is authorities' and the public's responses to wildfire, more attention has been given to the second major theme identified. The longitudinal approach is integrated into both sections. Lastly, the results are discussed and potential practical implications are outlined.

| ME THODS
This section comprises a description of the data collection process and the method for data analysis. The study is based on media articles relating to the previously described wildfire crisis that directly or indirectly affected inhabitants in a large area of Sweden.

| Data collection
Previous research on media coverage of wildfire has tended to focus upon the "prestige press," whereas few studies have emphasized analyses of popular mass-market newspapers. There is thus a need to investigate also mass-market newspapers, given their share of the market in relation to the "quality press." The same applies to regionally based newspapers, given their potentially significant role in influencing public perception (Anderson, 2009). In the present study, the source of literature comprised media articles from both major national and regional newspapers: Aftonbladet, Expressen, Dagens Nyheter, Svenska Dagbladet, Göteborgsposten, Vestmanlands Läns Tidning and Sveriges Radio. We selected the newspapers because of their dominance of socio-political topics in Sweden. They are part of the daily Swedish media supply and constitute mainstream news texts; thus, they do not direct their news to any specific group of individuals. As the aim of this paper was not to scrutinize media as such, but rather to study the overall media picture of authorities' and the public's responses during a specific type of crisis (read: wildfire), the analyses do not distinguish between different types of news media.
The newspapers were searched using the electronic Swedish media archive Mediearkivet, covering the above-presented papers with keywords relevant for the study topic, that is a variety of combinations including wildfire, "Västmanlandbranden". The searches took place during March to June 2016. As noted above, to identify potential frame-changing patterns or any reframing processes at work we searched the databases for two time periods: 1. The acute phase of the disaster as it unfolded and its imme-

| Data analysis
The data analysis comprised the examination and analysis of media articles of a wildfire event to identify portrayals of the public and authorities' responses (Entman, 1993). Media frames constituted the unit of analysis. As there is an apparent lack of studies that lay emphasis on media frames of authorities and the public's responses during wildfire events, we used a version of the thematic method of analysis that was developed (TA), [primarily] for use within the qualitative paradigm (Braun & Clarke, 2013;Burr, 1995), thus an inductive "bottom-up" approach to analysis (Frith & Gleeson, 2004).
Subsequently, the pattern/theme was not dependent on quantifiable measures, but instead the prevalence or the "keyness" of a theme rather depended "on whether it captured something important in relation to the overall research question" (Braun & Clarke, 2013:82).
TA is suggested to "provide accessible and systematic procedures for generating codes and themes from qualitative data" (Clarke & Braun, 2017:298). Codes are described as the building blocks for themes that are underpinned by a central organizing concept-a shared core idea. Themes provide a framework for organizing and reporting Electronic copy available at: https://ssrn.com/abstract=3605670 the researcher's analytic observations (Clarke & Braun, 2017:298). It is thus not only a matter of summarizing the data content, but also of identifying key features of data guided by the main questions. In our case, themes and sub-themes are the product of in-depth reading of media articles that are arranged in major themes. Subsequently, the first step of the analysis was made through so called "open coding," which meant that media articles were examined line by line to identify themes related to authorities' and the public's responses during the wildfire crisis. Next, codes were formulated into words describing the text content. To illustrate from an excerpt in a newspaper article: Many bore witness to non-existent or poor information from authorities.
The text section was coded as Non-existent or poor information.
Such codes were initially organized in tables with the name of the newspaper, name of the author, date of issue, article excerpts and the page of the article excerpts. The second step was to sort codes with similar content into different sub-themes. The previous example was sorted into the sub-theme Lack of information, underpinning the theme Absence, passivity and inadequacy of authorities and finally into the major theme How was the event managed. This process of analysis was conducted by making constant comparisons between the text sections, codes and themes. In practice, the steps of analysis were not strictly sequential, as we constantly moved back and forth in re-examining text sections, sub-themes, themes and major themes in an iterative working procedure. The main author was responsible for the initial coding of the data. Reliability was strengthened by a continuous dialogue between the authors on the coding procedure and on specific examples, particularly in ambiguous or uncertain cases.
As the process of analysis continued, the next step was to identify and organize all data that related to the previously classified patterns. Finally, an attempt was made to integrate the identified themes in order to gain a more holistic understanding of the subject matter.
A core theme/frame was identified as Responsibility/accountability, a concept that is suggested to capture and summarize the core of a coherent and meaningful pattern in the data. Subsequently, there was a progression from description (organizing to show patterns in semantic content and summarizing) to interpretation, attempting to theorize the significance of the patterns and their broader meanings and implications (Parker, 2004). Increasingly an interdependence between the major themes became visible. The overall structure of the core theme, the major themes, the themes and sub-themes is presented in Figure 1 below. It should be noted that the sub-themes are underpinned by numerous codes, which are not illustrated in the figure.

| RE SULTS
In the thematic analysis of media reports of authorities' and the public's response during and after the wildfire, the core theme Responsibility/accountability emerges. The process of framing appears to give meaning to the very nature of the crisis as well as constituting a kind of foundation for interpreting both authorities' and the public's response patterns. According to the thematic analysis, this core theme is underpinned by two major themes: 1. Why did the fire occur and the crisis escalate, and 2. How was the event managed?
From a media framing perspective, both themes are suggested to be outlined with regard to questions of accountability; who was accountable for the fire and who was accountable for managing the crisis? A shift in the major themes could be detected over time. The reason why the fire occurred and the crisis escalated is reframed, while more unfavourable aspects of how the event was managed tend to be reported after 1 year. The themes and sub-themes that underpin these major themes are elaborated below.

| Why did the fire occur and the crisis escalate?
The first major media theme that emerges from data concerns the reasons why the initially small fire developed into a full-scale crisis; "Why did the fire occur and the crisis escalate?" Answering this question appears to have been central in media at the time, while this also appears to have given meaning to the very nature of the event; who was to blame or who failed to prevent it. The three themes building up this major theme concerning the cause of the fire are as follows: (a) Extreme weather conditions due to climate change, (b) a workplace accident and (c) inadequate crisis management.

| Extreme weather conditions due to climate change
The sub-themes underpinning the theme Extreme weather conditions due to climate change are: • Politicians are to blame • The consumer society is to blame These sub-themes are further explained below.
The media initially frame the wildfire based on a discourse as regards accountability in relation to climate change and an increase in extreme weathers. Media discuss climate change in itself as the reason for the fire to occur and the crisis to escalate, but also indirectly by targeting politicians' actions (or inactions) and their failure to take the necessary measures that in turn are proposed to indirectly have brought about these consequences.
We need to reinforce our preparedness with regard to crisis events related to extreme weather. Unfortunately, | 73 NILSSON aNd ENaNdER the climate prognosis suggests that this kind of event will become more commonplace. (Dagens Nyheter, 2014-08-06) Sweden is not well enough equipped for climate change.
[…] Sweden needs to act to handle future wildfires, floods and other climate-related events. (Dagens Nyheter, 2014-08-23) The government is suggested to have been insufficiently attentive to warning signals from authorities and to have ignored an increase in extreme weathers, in spite of grim future scenarios. The crisis is additionally framed in media with regard to the consumer society, hence pointing to the individual consumers, and their negative impact on the environment.
Thus, it is implicitly suggested that the individual citizens have a responsibility for extreme weathers, which give rise to the wildfire.

| A workplace accident
The sub-themes that build up the second media theme defining the crisis as being a workplace accident are as follows: • The capitalist society is to blame • The forestry industry is to blame • The forestry firm is to blame • The single forester/entrepreneur is to blame This media frame of responsibility is complex, as it appears to involve several domains of accountability, ranging from the global to the individual level, which unravels in a kind of ripple effect. Taking its starting point in the forest industry being poorly managed for reasons of capitalist profit, shifting to a forestry firm engaging in soil scarification/preparation during warm and dry weather conditions, thus heightened risk of fire, and ultimately indicating a workplace accident as a forester failed to take proper safety measures. To illustrate a code building up this theme: The overall thematic structure of authorities' and the public's response during and after a wildfire crisis from a media framing perspective [Colour figure can be viewed at wileyonlinelibrary.com] The wildfire most likely started because of soil scarification. The driver of the forest machine called [emergency number, authors'' remark] to get help after having tried to put out the flames himself. (Aftonbladet, 2014-08-09) Climate change is of course not the villain here. It is the forestry policy and their profiteers. (Dagens Nyheter, 2014-08-10)

| Inadequate crisis management
When studying media frames 1 year after the fire, the thematic analysis shows how extreme weather or a workplace accident linger as explanations as to why the event occurred. However, a new media theme of responsibility emerges and seems to dominate the former two. Looking back 1 year after the crisis, the former underlying causes are no longer explicated to define the nature of the crisis, but are rather seen as consequences of inadequate crisis management. Hence, media emphasis has shifted to instead frame the inadequacy of the crisis management system as being the focal cause to the wildfire. Whereas the fire was reported to have started due to climate change or as a workplace accident in articles found in the first media search, in the second data set, it is the shortcomings of the response system that are instead put forward as the main reason for the fire developing into a large-scale crisis. The sub-themes underpinning the theme "Inadequate crisis management" include: • The international community is to blame • The state and municipalities are to blame • First responders (the emergency service, the firefighters, the police) are to blame The media expositions of crisis response within this frame of accountability are intertwined and complex and relate to different societal levels.
It is not until the catastrophe is a fact that our decision makers decide to appoint resources. Instead of being proactive, to take preventive actions, they act when it [the wildfire, authors' remark] has already happened. […] There is reluctance at all levels in society to assume costs that may not be necessary. (2014-08-08) Authorities are guilty of catastrophic work from the start, and that is the opinion of almost everyone we have talked to. (Expressen, 2014-08-06) If preparedness was good, the Government would have understood already the second day that the local emergency services had lost control and thus made sure that their crisis office was working and available. (Expressen, 2014-08-07) As the actors hold different work roles, underlying reasons for blame are linked to the respective actors falling short in performing their specific task assignments. Yet in parallel, media frames them as unable to perform as a well-functioning unified crisis management system. It could be noted that the tone of voice was considerably more gentle, understanding and accepting of potential flaws during the time of the crisis, as media conveyed the system to be subject to unrealistic demands regarding authority preparedness and immediate crisis response. In line with the initial framing of the wildfire (i.e. as a climate crisis and/or a work accident), actors within the crisis management system were rather suggested to be well prepared, but unfortunately having to deal with an unfeasibly exceptional and explosive concoction of heat, drought and wind.

| In sum
The analysis demonstrates how the media framing contributes to three completely different approaches to the wildfire crisis. There is also a shift over time.

| How was the event managed?
The second major theme that was identified during the media analyses, in terms of portrayals of authorities' and the public's response both during and in the aftermath of the wildfire crisis, refers to "How was the event managed?" This major theme is built up by two themes: (a) absence and inadequacy of authorities, and (b) the presence, activity and coping ability of citizens. The results indicate two contrasting frames of responsibility that are implicitly interrelated, as the public is increasingly suggested to be forced to act because of the inadequacy of authorities' (lack of) response. The sub-themes that underpin the themes describing the authorities' and the public's respective response patterns are further expanded below.

| Absence, passivity and inadequacy of authorities
The analyses of the media material show how the first theme "Absence, passivity and inadequacy of authorities," underpinning the second major theme "How was the crisis managed," is framed with regard to Responsibility/accountability of those handling the wildfire crisis. The following sub-themes were identified as follows: Electronic copy available at: https://ssrn.com/abstract=3605670 • Information, BUT lack of information • Collaboration, BUT deficient collaboration • Political responsibility, BUT lack of political responsibility The overall picture of media reports of crisis response is that authorities do provide information, they do collaborate and they do take political responsibility. However, favourable aspects of the authorities' response are continuously contrasted by a "but", followed by a number of negative counterparts that progressively seem to dominate the positive, ultimately pointing to the inadequacy in handling the wildfire crisis. The sub-themes are presented below.

Information, BUT lack of information
According to the thematic analyses of media reports, authorities responded somewhat well to the wildfire crisis by giving relevant and accurate information. Crisis information was conveyed in different kinds of fora and concerned issues on how to respond to the crisis event (i.e. to stay indoors until further notice, where to turn to in order to get up-to-date information or help, when and how to evacuate, to inform if cattle had been left behind). The analysis also identified exhortations to stay away from the fire area, to respect barriers and to report in the case of break-ins or looting.
The police caution people of going too close to the fire area, as there is an imminent danger of falling trees. (Göteborgsposten, One type of information appears aimed at de-emphasizing risks in order to calm individuals' anxieties and worries of the fire having burnt down, for example, one's own house. Another type referred to appeals for collaboration and common humanity by asking the public to help in any way possible by donating food and drink, giving shelter, etcetera.
The overall picture of crisis information in media, as rendered above, suggests that authorities acted in an adequate manner.
However, at the same time, that picture was contrasted with a more negative evaluation of authority response that progressively appeared to dominate the positive. Intense criticism was directed towards both municipalities and authorities as information to those affected was reported to have been insufficient, inferior or even almost non-existent, for example, regarding the fire whereabouts or alarming reports.
Many bore witness to non-existent or poor information from authorities. (Svenska Dagbladet, 2014-08-07) Communication has really failed. Sweden does not seem able to handle such big events.
(Svenska Dagbladet, 2014-08-06) Media gave an account of individuals even questioning if there was a fire at all, due to the lack of information. Those affected were said to have been frustrated and irate as they were described as having been left utterly clueless until flames appeared right on their doorstep, even though authorities had "promised" to evacuate with good foresight. Consequently, individuals were on the verge of becoming victims to the fire as they were overlooked and "forgotten" by authorities.  (Sveriges Radio, 2014-08-05) It feels as if they forgot our entire village. (Dagens Nyheter, 2014-08-08) Indeed, one fatality was attributed to the lack of information. In others cases, individuals were reported to be upset for having been evacuated unnecessarily or to lack information on when they could return home. There were technical difficulties on informational websites due to the pressure of many visitors, and the fact that information was only provided in the Swedish language was criticized. Apart from lack of information, there were additional reports of defective or contradictory facts while there appear to have been many rumours circulating.
The media reported how the public generally had to cope with great uncertainty.
In the analysis of the time 1 year after the fire, positive aspects of information sharing were no longer reported. It was rather a onesided matter of pointing to the lack and inadequacy of risk assessments, which at worst had put individuals in danger of their lives.
Not only had authorities missed individuals caught in the midst of fire, but they had also failed to hinder people from getting into the dangerous fire area. In other cases, authorities were accused of erroneously and inhumanly having hindered individuals from going into the fire area to engage in live-saving measures.

Collaboration, BUT deficient collaboration
At the first time point measured, favourable aspects of collaborative efforts were reported to have taken place already 1 week before the initial flames, as the county administrative board and the emergency service issued recommendations for the forestry on how to relate to dry weather and heightened risk of wildfires. Following the wildfire outbreak, there were initial reports of the crisis management system as being well prepared. The media analysis yielded numerous accounts of well-functioning collaboration.
We need to call attention to all the amazing human efforts made, everybody are important in handling this event; the firefighters, the helicopter pilots, the police, etc, everybody that contribute in one way or another.
Everybody should really know that they are important and that they have done an excellent job! (Vestmanlands Läns Tidning, 2014-08-12) As the crisis was reported to enter a second phase, the rescue operation was announced to almost double in reinforcement to ensure work around the clock.
In spite of reported favourable response patterns, there was also much media criticism directed towards collaboration in handling the wildfire crisis. In particular, disapproval was directed towards the passivity and inaction of authorities. Media's criticism grew continuously more serious, claiming nothing to have been managed properly from the crisis outset until its end. The state was accused of prioritizing in the wrong manner, while it was suggested to be difficult to catch up when preparation was insufficient. Media further reported on a lack of strategy and inability to apprehend and organize reinforcements. Whereas there were accounts of the public sector viewing citizens as amateurs in handling crises in general, media reports reflected disputes and harsh criticism towards the County Administrative Board as regards the overall set up of appeals made to the public to donate resources of various kinds. Crisis management cannot depend on donations! One year later, media still pointed to favourable aspects of collaborative measures taken; many had worked hard to save forests, buildings and lives. Nevertheless, the overall media picture still reported there to have been considerable shortcomings in the management structure or no crisis management at all and increasingly depicted the aforementioned deficiencies to be the very cause of the initial fire in fact having developed into a fullscale crisis.

Political responsibility, BUT lack of political responsibility
The issue of political responsibility is highly evident in media, both at the time of the crisis and during the aftermath.
Favourable aspects that were reported included the state declaring the crisis a national matter, thus relieving the affected municipalities. An additional positive aspect involved the exercise of symbolic leadership, by politicians showing presence, empathy and commitment.
Aspects of political responsibility initially accounted for in media held that Sweden did not find herself in a bad starting position. However, preparedness did reflect the Swedish risk profile, where wildfires were not considered a common threat. Thus, it was viewed as reasonable that the wildfire would cause great difficulties. In spite of such reservations, criticism grew steadily.
Some media reported that the Swedish Government Offices' crisis organization did not function at all. There were unclear messages regarding financial compensation, whereas municipalities initially were expected to pay the whole bill themselves. It was a common media theme that the municipalities finally forced the state to prioritize adequately and assume formal responsibility by leaving meetings in anger and in consequence enforcing instead of receiving economic support. There were also reports of the government as well as cabinet ministers responding late, and of the Minister of state being passive and slow. Politicians were accused of failing to take environmental issues seriously and to prepare for the hazards of climate change.
The government needs to look in the mirror. [ (Dagens Nyheter, 2014-08-08) The Government has closed their eyes to reality. The risk of extreme weathers should not have come as a surprise. (Aftonblandet, 2014-08-06). Electronic copy available at: https://ssrn.com/abstract=3605670 It is worrying that politicians decide to classify the wildfire as "beyond their control". This sends out a clear signal of their ambitions as regards the country's capability to handle greater emergencies, crises and threats to the sphere of foreign and security policy. (Svenska Dagbladet, 2015-08-03) Politicians were also accused of using the crisis in election campaigns, subtly burdening others with guilt while pursuing personal political gains. They were further blamed for not daring to face the reactions of the public, for withholding information that would be to their disadvantage (but which might have saved lives) and for not caring for those affected in rural areas. Even though media criticized politicians for not talking to people, politicians that did shoulder symbolic leadership by visiting those affected were accused of consuming too many resources. Overall, the media analysis demonstrates a general view that politicians were to blame for insufficiencies in both crisis preparedness and crisis management, forcing authorities and other societal functions to react in an ad hoc manner.
At the second time point studied, positive aspects in media of political responsibility refer solely to politicians taking corrective actions in the crisis aftermath, by compensating economically, apologizing and working to improve for the future. As noted above, a shift in comprehension of the nature of the crisis as being due to inadequate crisis management was conveyed. When politicians tended to hold on to an explanation of the crisis as caused by climate change, this was now stated to reflect their reluctance to learn and improve.

| The presence, activity and coping ability of citizens
The second theme within the major theme "How was the crisis managed" refers to the presence, activity and coping ability of citizens. This major theme is underpinned by the following sub-themes: • Responsible crisis management, • Well-meaning but misguided, • Inadequate behaviours, and • Criminal illegal behaviours.
The substantial content comprising these sub-themes is explicated below.

Responsible crisis management
A majority of the codes identified in media reports with regard to the public's responses suggest that voluntary workers had a decisive role in the overall handling of the wildfire, as it would simply not have worked without their participation. Swedish society was reported to show strength through these people. They were described as "unknown heroes" that did "extraordinary work" by providing information, coming up with resources (food, clothes, shelter, etc.), engaging in firefighting and so forth.

This is a milestone for how civilians can contribute to
Swedish crisis management. (Dagens Nyheter, 2014-08-08) The support of thousands of volunteers has played a decisive role. (Vestmanlands Läns Tidning, 2014-08-12) Volunteers represent a substantial part of the work in the fire area. […] The place is crawling with voluntary people that are helping within and outside the fire area. No one knows how many they are. After the event, one will not even know how many they have been. Volunteers are something of the great fire heroes. (Vestmanlands Läns Tidning, 2014-08-13) Voluntary efforts moreover included neighbourhood collaboration, looking out for and warning each other, giving support during evacuation and attending to those "left behind" by authorities and first responder personnel, etc. Media additionally recounted stories from citizens themselves; they found it natural to participate, to help and to do their best, as it would have been strange to impose all work on the state.
Besides showing solidarity and a great will to help, the media analyses suggest the general citizen to assume responsibility by taking proper action. There was a feverish activity concerning practical issues, still most people were reported to have remained calm and collected. Those affected stood up for themselves by questioning the general lack of information, the insufficient number of personnel, etc. Appeals made for the public came gradually to point to a lack of ability on the behalf of the County Administrative Board, suggesting that the public had to do the work of authorities.
In media reports 1 year after the crisis, media reports still accounted for solidarity among the public. The general idea was that voluntary workers did have a profoundly favourable effect on solving the crisis. A somewhat new aspect stemming from crisis collaboration concerned cohesion and deepened relationships between neighbours, reinforced by people doing good together.

Well-meaning but misguided
The second sub-theme of the public's responses identified in media reports is underpinned by codes illustrating, for example, an excess of donations and workers, but which had to be declined as authorities could not administer them logistically, nor ensure their safety.
Electronic copy available at: https://ssrn.com/abstract=3605670 Overall, the police are working hard to keep helpful people away from the dangers of the fire. People are calling from all over the country and that is nice, but they cannot just call and ask: "Where do you want me?" There is a need to know who they are. (Svenska Dagbladet, 2014-08-06) There was a perception that citizens at times even complicated the crisis work undertaken, as they lacked competence of what to expect in the fire area.
People have given so much that there is no need for more help now. Those that take in things announce that they have enough. (Expressen, 2014-08-07) Overall, the media painted a general view that authorities in charge unfortunately lacked competence and resources to make use of the voluntary help offered. For that reason, public responses that were motivated by good intentions became misguided. Another aspect underpinning this theme involved unintentionally unwise decisions as individuals cut fences in order to "help" wild animals escape the flames, but which created traffic risks.
The police exhort people not to cut wildlife fences to let stressed animals close to the wildfire area out. To cut the fences means that we let them out on the road, which will do more harm than good.

Inadequate behaviours
The third theme identified in the data refers to inadequate behaviours of the public, for example individuals refusing to leave their houses for fear of break-ins or deciding to "stay and fight the fire" neglecting predictable risks, calling helplines about trifles thus blocking for real emergencies and individuals not respecting occlusions to the wildfire area in spite of the dangers.
In spite of dangers, people continue to try to get into the wildfire area. The police have their hands full trying to keep people away. (Svenska Dagsbladet, 2014-08-06) The most distinct aspect of inadequate behaviours in media reports 1 year after the crisis concerns so called "fire tourists".
Fire tourists are described as intrusive, disturbing and acting disrespectfully by not leaving local inhabitants alone, resulting in outright conflicts and stress. Snooping individuals were further accused of not respecting barriers, creating traffic interruptions and forcing the police to pickets, thus appropriating resources that could have been of better use elsewhere. Also, biologists were described as having taken great risks when visiting the burnt forest for scientific reasons.

Criminal illegal behaviours
The fourth theme identified refers to criminal illegal behaviours. This is relatively one-sided, solely comprising negative aspects of the public's response at the first time point analysed. The media reported on actual break-ins and rumours of thieves taking advantage of empty houses due to people having fled the wildfire, while a pyromaniac was said to be somewhere close to the fire area.
As more people have been evacuated from their homes and cottages in the forest area, rumors of thieves operating within the barriers have begun to circulate. Moreover, during Saturday evening, [name of newspaper] published data on the police looking out for a pyromaniac. (Vestmanlands Läns Tidning, 2014-08-11) It should be noted that the theme Criminal illegal behaviours concerns only a small group of individuals.

| In sum
Regarding the second major theme "How was the crisis managed" which also includes the crisis aftermath, there is a pattern inherent in the data, illustrating how the negative aspects of crisis management in media reports increasingly come to dominate the positive. In contrast, the public was reported to gradually step in to compensate for the flaws of the management system. It can be noted that there are more unfavourable aspects of authorities' responses reported when looking back at the crisis 1 year after, as compared to the reports during the actual crisis. This pattern becomes more evident when examined from the backdrop of the results of the first major theme, that is the shift in the attributions regarding the nature of the crisis to mainly being due to deficiencies in crisis management.

| D ISCUSS I ON
The aim of this study was to identify key media frames relating to the portrayals of authorities' and the public's response during and after a wildfire crisis. The core frame identified is Responsibility/accountability, a concept that is suggested to capture and summarize the core of a coherent and meaningful pattern in the data. As such, the core media frame is underpinned by themes reflecting both authorities' and citizens' response, as the deficiencies of the former are implicitly suggested to have forced citizens to act to compensate for their absence, passivity and inadequacy. Here, media appears to play a crucial role in framing the social dimension of crisis management. Whereas previous research has considered media frames of authorities and of the public separately, the main theoretical contribution of this article is the identification of an interrelationship between the frames in relation to different groups of individuals responding to crisis. These findings raise questions of a theoretical nature. Previous research suggests that the use of different media Electronic copy available at: https://ssrn.com/abstract=3605670 frames depends on crisis type (An & Gower, 2009). Frames are thus suggested to exist independently and side-by-side. This appears to not always be the case as the accountability frame in this study was found to assign blame to some, while by doing so implicitly commending others, thus producing moral judgements (Entman, 2007).
Frames appear highly interrelated and intertwined. Some are construed in an explicit manner, while some appear more implicitly in the data, pointing to frames existing in parallel and at different levels of abstraction. These results may have implications for the way we think of framing effects, as multiple narratives may be at work at the same time.
The results suggest that authorities have failed at crisis management in spite of numerous accounts of proper by-the-book responses. These latter (favourable) media reports come through as descriptive and neutral by merely reproducing what are regarded as (adequate) crisis management actions. When taking the perspective of authorities, media report how they do give information, collaborate and take political responsibility. At the same time, the deficiencies of the crisis management system become increasingly evident.
Media give an account of improper response actions that, in comparison with the aforementioned, are no longer purely descriptive and neutral, but rather of a morally judgemental character from the vulnerable and affected individuals' point of view. This is not problematic in itself, as media should play the role of an active watchdog and be a sensor of potential social problems (von Dohnanyi & Möller, 2003). However, the same kind of responses that were previously framed in a positive manner from an authority perspective appears now to be implicitly evaluated against the backdrop of an ideal mirror, a criteria perspective of having an informed public, an efficient collaboration, characterized by national accountability and foresight.
From such a standpoint, authorities do not meet the criteria. The results explicate that complex and subtle judgemental media framing processes are at play. The very same behaviour has been shown to be both commended and criticized. For example, politicians acted poorly by not talking to people, whereas politicians shouldering symbolic leadership were accused of consuming too many resources.
Similarly, authorities acted wrongly in allowing people to drive straight into the flames, although they simultaneously hindered them from saving lives in the fire area. These results point to recognized difficulties in crisis management, namely to find clear-cut standards that can be applied every time. The media preference to report bad rather than good news (Dennis & Merrill, 1996) may serve as one explanation why proper response actions are framed in a merely descriptive and neutral fashion, whereas deficiencies appear framed with regard to so called emotionalized blame. As individuals tend to give more weight to negative information (Baumeister, Bratslavsky, Finkenauer, & Vohs, 2001), emotional news frames are likely to affect emotional response (Kim & Cameron, 2011) and emotionalized blame attributions have been shown to influence both blame perceptions and populist attitudes. This kind of media framing may thus create negative impressions among the public towards authorities that may ultimately aggravate efficient crisis management. This article contributes to an understanding of how media frame blame, thus how "blame games" may unfold, where the longitudinal approach demonstrates how the blame game is accentuated over time as favourable aspects of crisis management becomes increasingly overshadowed by criticism. It can be noted that the political context at the time might provide an explanation as to the heavy blame directed towards authorities, particularly in the crisis aftermath. Since the first media search, several assessments and evaluations of the crisis management system were made public (see, e.g., Regeringskansliet, 2015), directing sharp criticism to the crisis management system.
Main conclusions included lack of risk assessments of wildfire crises prior to the Västmanlandsbranden besides deficiencies in coordination, routines, roles, etc., which were reflected in media reports at the time.
Considering previous research on media framing of crisis and the literature on accountability and blame games, it is perhaps not surprising that the core theme of Responsibility/accountability emerged in the data. However, whereas current literature merely states that the responsibility issue or issues concerning accountability will be posed following crisis (see, e.g., Kuipers & 't Hart, 2014;Resodihardjo, Carroll, Eij, & Maris, 2016), this study provides empirical evidence of this process. The crisis appears to be subject to a kind of reconstruction process, which is illustrated throughout the article as a whole, but also within the smaller thematic units. For example, this is made particularly clear in the case of the sub-theme "A workplace accident," where media appear to struggle to "name" failures and assign blame, by targeting various actors at different hierarchical levels that are all framed as blameworthy in the quest of finding the truth and allocating responsibility. On the one hand, the incident appears to be subject to a politicization process by being made big enough to violate crucial public values (the capitalist society), thus arousing collective stress. On the other hand, the event is re-defined as due to the mistakes made by a single forester/entrepreneur. In that regard media rather seem to depoliticize the event. Brändström and Kuipers (2003) explicate that these kind of "blame games" may couple those being blamed with critical values that have the power to "unleash an expanding wave of criticism that may result in a thorough delegitimization" (p. 281) of those involved.
Another aspect that deserves attention refers to the idea of the public being forced to act as a form of compensation for authorities' failure. Media reports illustrate how citizens were abandoned, uncared for, etc., thus having to act themselves. It is portrayed as negative, on the verge of embarrassing, that citizens were even asked to contribute with food, shelter, etc. One explanation for the use of such media logic may be that emergency response depends greatly on a trained workforce (e.g. fire, rescue, medical, care and relief) in most OECD countries, including Sweden (McLennan & Birch, 2005), while there is increasing recognition that citizens need to take personal responsibility in countries that are particularly vulnerable to wildfire threats (Paveglio, 2010). As media frames have been shown to encourage individuals to think, feel and decide in particular ways (Entman, 2007), the subtle media messages or cultural norms embedded in news coverage may discourage private initiatives and undermine public acceptance, which actually constitutes a major factor in the ability of authorities [federal agencies] to effectively manage forest fires (Shindler et al., 2002). This is thus an illustration of a specific event that must be understood with regard to a wider socio-political context. Media have a role to critically examine the work of authorities. However, it appears to be a difficult balancing act to criticize power without risking patronizing the private citizen and promoting a paternalism rationale that holds that the task of crisis managers' work is to help and protect citizens who cannot manage on their own (Hobbins, 2017).
Another observation we believe is worth making is that the depictions in media of citizens' responses to the crisis illustrate mainly adequate behaviours, rather than those negative myths identified in many previous studies.

| Practical implications
This study points to several practical implications. First, this article sheds light on media's framing of crisis management responses, thus indirectly on performance, by seeking to answer what went wrong, what was (not) done to prevent and contain the crisis and who should be held responsible (Boin, McConnell, & 't Hart, 2008).
The construction of blame frames in the media is suggested to have produced three very different approaches to the nature of the same wildfire crisis. The longitudinal approach explicates how the event was reframed several times. As previous research has shown media frames to influence both crisis management and public opinion, exposure to several different frames may create confusion and inadequate expectations on authorities' responses. It may also have a negative effect on crisis management itself if leaders respond to mixed and misleading signals of the nature of the crisis or if authority representatives lack a common view of what needs to be done.
Second, media have a role to approach public actions such as crisis response in a critical manner. If investigative journalism evaluates crisis management on the basis of idealistic criteria, the outcome will inevitably be negative. Bearing in mind that a crisis involves a community of people, perceiving an urgent threat to core values or life-sustaining functions, which must be dealt with under conditions of uncertainty (Rosenthal, Boin, & Comfort, 2001), ambiguity and control loss are part of its nature. To use unrealistic standards risks creating unnecessary stress among leaders and undermining learning from experiences. The "media blame game" has been shown to constitute a pressing challenge for [political] leaders (Boin, 't Hart, McConnell, & Preston, 2010). If leaders' fear of being criticized and blamed during and in the crisis aftermath becomes a logic for crisis management actions, this will most likely [re]direct resources to the wrong places. Anderson, Chubb, and Djerf-Pierre (2018) report how the occurrence of media blames games near-silenced meaningful discussion of more relevant topics in relation to extreme weather disaster (Crow et al., 2017).
de Vreese (2005) states that current research has discussed media framing of different events and issues without addressing the implications for audiences. Previous research has shown that citizens tend to personalize media criticism of governmental responses after extended news media coverage, while biased media framing of the public during crisis may influence citizens' responses as well as impact negatively on crisis management (Nilsson et al., 2016;van der Meer & Verhoeven, 2013). This study illustrates another kind of subtle bias inherent in media reporting that may create unrealistic expectations on authority response besides risking to weaken a sense of individual responsibility among the public. Thus, not only leaders, but also the individual citizens need to be sensitive to media framing processes both before and during crisis events. This study may also be of value to media representatives in thinking about their ethical responsibilities as concerns crisis communication. Some scholars "urge, to radically re-think how we understand the media's responsibility towards vulnerable others" (Chouliaraki & Stolic, 2017:1162.
If properly handled, news frames can be used to reduce anger and blame in crisis , thereby strengthening crisis management structures.

| Limitations
This study has some limitations. In developing our analysis, we were limited to data obtained from one crisis event. Thus, we do not claim to have captured all the components of the processes at work in media framing of authorities' and the public's response to crisis. In addition, as the study takes an inductive approach, it may be accused of relying on a small sample and being difficult to replicate (Hertog & McLeod, 2001). The lack of representativeness is, however, inherent to the chosen qualitative method. Of course, there may be many contextual variations with regard to the same kind of crisis events, but also within the broad field of crisis management as a whole.
Another limitation may be that the persistence of specifically identifiable narratives may differ from one newspaper to another (Morehouse & Sonnet, 2010). Smith (1992) thus suggests that national, regional and local sources can provide complementary narratives, while recognizing that analysing these narratives requires close attention to framing practices that become evident over multiyear time spans. The authors have tried to address such apprehensions by including both local and national tabloids and quality papers and using a longitudinal approach to analyses. This study has been confined to a Swedish context, thus limiting the range of printed media examined. This limitation calls also for multicultural approaches, since research demonstrates national variations in media systems and media logic regarding the recognition of citizens own responsibility during crisis (McLennan & Birch, 2005;Paveglio, 2010;Strömbäck & Dimitrova, 2006). It should also be emphasized that the concepts derived from the data are of a sensitizing rather than a definitive character, as described by Blumer (1954). Even so, the results may serve as a starting point for crisis managers and citizens to reflect on the role of the media and to enhance awareness of and sensitivity to the complex Electronic copy available at: https://ssrn.com/abstract=3605670 logic and subtle messages of media reporting. The results may also be valuable to the media themselves in thinking about their way of framing crisis with regard to the potentially counterproductive outcomes that these frames may produce.

ACK N OWLED G EM ENTS
This work was supported by funding from the Swedish research council for sustainable development, Formas (grant no. 211-2014development, Formas (grant no. 211- -1885.