Framing in Crisis Communication:
Investigating the alignment of frames across the media, the public, and the
organization in the discourse on the GermanWings airplane crash in March 2015
Master’s Thesis
Graduate School of Communication
Masters’ Programme Communication Science
Author: Vanessa Girnth Student Number: 10830030 Thesis Supervisor: Theo Araujo 28th January 2016
Preface
Dear reader,
Before you lies the thesis “Framing in Crisis Communication: Investigating the alignment of frames across the media, the public, and the organization in the discourse on the
GermanWings airplane crash in March 2015”, which is based on a content analysis of messages on the GermanWings airplane crash. This thesis is the last part of my Masters’ programme Corporate Communication at the University of Amsterdam. I was engaged with working on my thesis for three month full-time, starting at the end of October 2015.
I worked with much pleasure on my thesis together with my supervisor, who supported me during this process. I would like to thank Theo Araujo for being always available to answer questions and his useful feedback on my work.
Furthermore, I would like to thank my friends who made my studentship at the University of Amsterdam a wonderful, instructive and eventful time. Finally, I would like to acknowledge my mother, sister and boyfriend for always understanding my decisions and supporting me during my study so far.
Vanessa Girnth 27th January 2016
Abstract
The present study investigates whether and if so how, the media, the organization and the public align on the same frames in their messages on the GermanWings airplane crash in March 2015. The data in this study consists of 2062 newspaper articles, press releases and Facebook posts from the three actors. Analyses examined whether frame alignment of emotional and responsibility frames occurred in one or more of the three phases of the crisis. Frame alignment in this study is about the usage of similar specific frames in the same phase of the crisis. The news value of proximity towards an event, how geographically close people are to it, is applied in this study as a determining factor of alignment as well as the source of the message. Furthermore, expectations of frame alignment depended on the type of frame investigated. While emotional frames aligned immediately after the crisis, no frame alignment was found for responsibility frames. In Germany more framing of emotions existed compared to the United States and the Netherlands. The countries did not differ in terms of frames of responsibility. Within this study, the angle taken to investigate frame alignment differed from earlier studies. The current study helps organizations to understand the different actors involved in the communication on a crisis. Besides, it gives insights into differences between frame types and their alignment across crisis phases. The results indicate that different alignment rules might apply for emotional and responsibility frames compared to other frame types. Future research needs to investigate how the time phases of a crisis can be identified and to what extent organizations are able to influence the framing process of a crisis situation to reduce the damage a crisis can cause the organization.
Keywords. Crisis communication, framing, frame alignment, attribution of responsibility,
Introduction
Almost every organization has to face a crisis during the time it operates. A crisis for an organization is the: “perception of an unpredictable event that threatens important expectations of stakeholders and can seriously impact an organization’s performance and generate negative outcomes” (Coombs, 2007, p.19). Therefore, it is necessary for an organization to understand the crisis and develop an appropriate crisis response strategy to make sure its potential reputation damage is limited (Coombs, 2007; Fediuk, Coombs & Botero, 2010). Such responses of an organization are made up of frames, which play a major role in communication (Holladay, 2010). By framing a message in a specific way, perceptions of stakeholders can be influenced (Entman, 1993) and negative outcomes of the crisis situation might be minimized (Canel & Sanders, 2010; Coombs, 2007; McDonald, Sparks & Glendon, 2010). Framing within this context, is: “to select some aspects of a perceived reality and make them more salient in a text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation […] for the item described” (Entman, 1993, p. 52), which leads to the processing of a message in a specific way. Not only the organization uses frames in a crisis. Everybody who is communicating uses frames, as it is impossible to engage in communication without framing (Entman, 1993). Therefore, also stakeholder groups are framing the crisis, such as the media or the public.
The communication, or discourse (Weick, 2011), through which the frames develop during a crisis can be categorized in different phases. In the first phase of the discourse there is little information about the crisis and speculations, different ideas and different frames are present (Van der Meer et al., 2014). In the second phase the alignment comes into play due to more information provision and interchange between the different parties (Schultz, Kleinnijenhuis, Oegema, Utz & Atteveldt, 2012). This alignment is about all actors using the same frames on the crisis. In the third phase the frames de-align. Reasons for the crisis are clearer and the crisis is not recent anymore. The complex situation is understood by the different stakeholders and there is no need for alignment anymore (Van der Meer et al., 2014). Within these three phases all social actors will apply frames and try to make sense of the situation. At some point of the discourse, the parties will come to reach a collective sense, the frame alignment (Van der Meer, Verhoeven, Beentjes & Vliegenhart, 2014).
Earlier, this process of alignment was considered to take place only between the media and the organization and the public were considered less important (Van der Meer et al.,
2014). In today’s world, they gained influence through to the so-called ‘empowerment of publics’ given to them by the Internet (Castells, 2007). This empowerment reduces the organizations’ impact on messages about a crisis (Stephens & Malone, 2009; Castells, 2007) and makes it more important to understand the influence of all three actors simultaneously. The frames of the three actors can be about different issues that are of importance in a crisis. Five important types of frames and issues have been identified by An and Gower (2009). Which of those five types occur in a specific crisis depends on the situation itself (An & Gower, 2009).
This study will investigate the crisis of GermanWings, which started in March 2015. On the 24th March an airplane of GermanWings crashed in the French Alps and all of its passengers died. The airplane crash had a huge impact on society as a lot of people were involved and died. This resulted in a severe crisis for GermanWings. The organizations’ reputation of a trustworthy and well performing organization was at stake. With regard to the airplane crash of GermanWings, two of the five types of frames are of specific interest here: responsibility and human-interest frames (An & Gower, 2009). These will be used to investigate the frame alignment process in the communication on the crisis.
The current research contributes to a better understanding of how organizations can protect their reputation in times of crisis and avoid reputation damage (Gilpin & Murphy, 2010; Cornellissen, Carroll & Elving, 2009). It also enlarges the field of knowledge on the relative power of the three different actors. Understanding the active role of the public in today’s world helps organization to react appropriately upon stakeholders’ requests and opinions and with how to use social media in crisis (Schultz, Utz & Goeritz, 2011). The results of earlier studies on frame alignment and the results of Van der Meer et al. (2014) will be extended. A manual content analysis in contrast to a computer-assisted one will investigate specific frames to understand whether frames of responsibility and emotional frames align in one or more phases of the crisis. This makes it possible to not only look at if the same word pairs are used by different actors, but whether it is actually the same content that is communicated. The following research question will be answered by using a number of hypotheses. These will give answers to questions in the domain of crisis communication that are not yet fully understood.
RQ1: To what extent do emotional and responsibility frames by the media, the public and the organization on the airplane crash of GermanWings align across different phases of the crisis?
Theoretical Background
All aspects of a crisis will be framed to some extent. Which frames will be used and who are the important actors framing the crisis depends on the impact of the crisis on stakeholders living in the area where the crisis happened, on who can be held accountable for the crisis (Heath, 2010), and the media attention given to the crisis (Coombs, 2007; Seletzky & Lehman-Wilzig, 2010; Cornelissen, Carroll & Elving, 2009). At some point of the crisis, all different stakeholder groups are expected to use the same frames on the crisis and to align on frames (Van der Meer et al., 2014).
Framing Process
All actors involved in the communication processes engage in framing, as it allows depicting a situation in a less complex and more comprehensible way (Hallahan, 1999). In the communication on a crisis, framing can be the selection of specific aspects of an event or the interpretation from a specific angle (Entman, 1993). Framing will always take place in communicative actions, because it is an inherent process to human communication (Van Gorp, 2007). In a situation like a crisis where there is a lot of communication, frames play a major role. Framing helps to make sense of the social reality. This sense making process needs to be understood by the organization, because it can lead to different levels of harm for an organization (Weick, 1988). Therefore, within the current research it will be investigated how frames come into play and to what extent the different actors in the framing process influence each other (Vliegenhart & van Zoonen, 2011).
Which frames people use depends on numerous factors such as their culture (Van Gorp, 2007). People from different cultures understand the world differently and therefore also construct different frames. Individual perceptions, involvement in the crisis and the opinion towards the organization involved can also influence the framing process of an individual (Entman, 1993). Those different frames of people and groups can interact with each other through interchange (Van der Meer et al., 2014). This can happen for example on communication platforms like social media where people exchange opinions and communicate.
Identifying Frames
There are numerous different types of frames and manners of how framing can be used (Thaler & Sunstein, 2008). Within the current research, the focus lies on specific frame types used in crisis situations based on the more general framing definition by Entman (1993).
An and Gower (2009) found out that there are five types of frames. The human-interest frame, the conflict frame, the morality frame, the economic frame and the attribution of responsibility are used dominantly in crisis situations. It depends on the type of crisis situation whether all five types are present in the communication on the crisis or only some of them (An & Gower, 2009). The human-interest frame, which depicts the issue from an emotional angle and the attribution of responsibility are taken into closer consideration within this research. These are two types of frames that are expected to play a crucial role in an airplane crash where 150 people lost their life. Based on the two basic frame categories, responsibility and emotions, it can be searched for specific frames that fall into one of the two categories (An & Gower, 2009). Factors of the crisis, such as the consequences and the impact on people’s lives can have an influence on the framing processes as well as other external factors (Lee, 2009).
Frame Alignment
Different researchers investigated how the different frames come into play on communication platforms (Entman, 1993; Miller & Riecher, 2001; Van der Meer et al., 2014). To be able to analyse this process, Heath (2010) divided a crisis in three different stages, wherein the framing occurs. These stages are the pre-crisis, the crisis and the post-crisis. The alignment occurs in the second phase and is crucial to solve the organizational crisis and to reach the outcome of an understandable situation for all stakeholders (Weick, 1988). Within this alignment stage, the different parties communicate about the issue in more or less the same way. They agree upon certain ways of looking at the issue (Weick, 1988). After the crisis has been solved frame alignment becomes needless.
Researchers differ in terms of how many stages of a crisis they identify; some describe more stages than Heath (2010) does. Furthermore, instead of stage, terms as phase or cycle are also used (Van der Meer et al., 2014; Miller & Riecher, 2001). Van der Meer et al. (2014) identified three stages, like Heath (2010). All the researches on framing of a crisis situation have in common that at some point an alignment of frames is recognized (Snow, Burke
Rochford, Worden & Benford, 1986; Miller & Riechert, 2001; Snow, Vliegenhart & Corrigal-Brown, 2007; Van der Meer et al., 2014). This alignment might be due to relations between the different actors and their interchange by using: “collective processes of negotiation over meaning” (Vliegenhart & van Zoonen, 2011, p. 111).
This sociological view on sense making of the actors (Weick, 1988) fits to the idea of frame alignment to make sense of a complex situation. The different actors are expected to take part in the communication and framing process on the crisis(Castells, 2007). This leads to the expectations of finding frame alignment in the communication of GermanWings in the second phase of the crisis between the already known actors the organization and the media as well as the public as empowered through the Internet (Castells, 2007). The following hypothesis arises:
H1: The frames of the organization, the media and the public on the airplane crash of GermanWings will align in the second phase of the crisis.
Proximity of the Crisis
An external factor that influences how stakeholders frame a crisis is the proximity (Lee, 2009). People who are living close to the area where a crisis happened are likely to be more affected by the consequences. Therefore, they are more involved and interested in information on the crisis. They might also suffer more under the outcomes or damages of the crisis. This leads to the fact that people close to the crisis, might evaluate the event as of more impact and as worse compared to people living further away, thus influencing the way how they frame the crisis.
The proximity factor is closely related to news media values. These values influence whether a crisis gets high amount of attention and thus to what extent it influences the public (Lee, 2009). One of the values is next to actuality of an event, whether it happened in the area of the target group of the newspaper (Seletzky & Lehman-Wilzig, 2010). An event is likely to be perceived as more newsworthy by journalists when it is fulfilling the requirements of being in close proximity to his or her target group (Vliegenhart & van Zoonen, 2011; Miller & Riechert, 2001). Based on this knowledge it can be expected that events will be reported and discussed differently at different places in the world as the proximity influences the media coverage. In different countries different proximity to the crisis comes into play (Van Gorp, 2007; Vliegenhart & van Zoonen, 2011).
This knowledge leads to the assumption that a crisis such as the one of the airplane crash of GermanWings will get differing media attention depending on the target group of the media organization and the proximity of the event. Furthermore, public living in close proximity may frame a crisis differently. These differences in framing might be in terms of how bad the consequences are for stakeholders or in terms of emotional involvement of people (Lee, 2009; An & Gower, 2009).
It can be expected that Germany will elaborate more on the crisis due to a German organization being involved. Furthermore, 72 of the victims were people carrying the German identity, and thus also have relatives mainly in Germany (Business Insider, 2015). This means that the crisis of the airplane crash will have more impact on German citizens and the country as a whole compared to the Netherlands. The neighbouring country is likely to be less involved as the organization is not a Dutch one and only one Dutch passenger was on board of the airplane (Business Insider, 2015). With regard to an area that is even further away than the Netherlands, it can be expected that the United States will be even less involved than the Netherlands. This should be the case due to not being part of the geographical region wherein the airplane crash has happened and to the fact that only two victims carried an American identity (Business Insider, 2015). The different degrees of involvement can be expressed by emotional frames. Besides, German citizens might be more interested and concerned about who is responsible for the crisis. The following hypothesis emerges based on earlier research applied to the theory of framing in a crisis situation:
H2: Nations in close proximity to a crisis rely on different responsibility (A) and emotional (B) frames compared to nations further away during the first phase of a crisis.
Applying the knowledge on sense making from Weick (1988), it can actually be expected that people from different countries will come to a collective sense making at some point to understand the complex situation. This can be reached nowadays through the interchange of information via the Internet. Media organizations are interacting with each other and provide newspaper articles online and the public engage in dialogue through for example social media platforms (Bucher, 2002; Castells, 2007). Interaction of the different nations thus is inevitable and frame alignment can be expected across nations. According to earlier research this will take place in the second phase of the crisis (Snow, Burke Rochford, Worden & Benford, 1986; Miller & Riechert, 2001; Snow, Vliegenhart & Corrigal-Brown, 2007; van der Meer et aal., 2014).
H3: The different frames used by the different nations will align in the second phase of the crisis.
Framing of Emotions
Another factor that can have influence on how stakeholders frame a crisis is emotions (An & Gower, 2009). Emotions often come into play in crises as such situations bring danger or fear with them. How a message is presented can influence what kind of emotions people will feel after having heard or read about the crisis (Hoeken, 2003). This can be explained with the example of a situation that is framed as a catastrophe instead of an accident. Even the different usage of words, such as catastrophe versus accident might lead to arousal of different emotions. Nabi (2003) found out that emotions are not only responses to frames, but can also function as a frame themselves. Up until today, the framing of emotions has not been defined or analysed in detail. Emotional frames are related to human-interest frames that have been defined and investigated earlier (Semetko & Valkenburg, 2000). This type of frame is one of the five dominant frame types found in crisis communication (An & Gower, 2009). It is a frame that describes a situation from a human angle and also considers or evokes emotions. Nabi (2003) takes the human-interest frame one step further. She suggests that perceiving and describing a situation in terms of emotions leads to the existence of frames of emotions.
It can be expected that the presence of framing of emotions will play an important role in a crisis situation. Emotions that are used as frames in messages can arouse for example negative feeling in the receiver (Coombs & Holladay, 2005; Kim & Cameron, 2011; Jin, 2009). Jin (2009) has also explained that: “individuals’ perception of a crisis is not strictly a function of the crisis stimulus itself, but involves their own interpretation of the crisis” (p. 310). Such own interpretations lead to an interplay of media and organizational frames and individual framing of a situation in the communication processes. The evolving individual frames of emotions can for example spread via the Internet through social media. Based on the empowerment of stakeholders (Castells, 2007) it can be expected that frames of emotions are dominant in a crisis and play a main role in terms of framing processes and frame alignment on an issue. Whether newspapers also use frames of emotions might also be interesting to find out, based on different kinds of newspapers, as they might differ in terms of their usage of emotional frames (Vincent, Imwold, Masemann & Johnson, 2002).
Up until recently, there was a gap with regard to research on framing of emotions in crisis situations (Jin & Pang, 2010). To get further insights on the topic of framing of emotions a sub-research question will be applied within the current research.
RQ2: To what extent is framing of emotions used by the different parties when communicating about the GermanWings airplane crash?
Which and to what extent emotions are induced by a crisis depends on the characteristics of the crisis situation (Jin, 2009). How strong the emotions are might be due to the proximity of stakeholders to the crisis (Seletzky & Lehmann-Wilzig, 2010). Differences in the emotional frames used in Germany compared to the Netherlands and the United States (Martin, 1988) can be expected. The news coverage is likely to be less emotional in the United States and in the Netherlands due to less involvement in the crisis and therefore also due to less relevance in terms of news factors for the media (Martin, 1988).
H4a: There will be more framing of emotions used in Germany compared to the United States.
H4b: There will be more framing of emotions used in Germany compared to the Netherlands. Furthermore, it is interesting to find out whether the frame alignment process is applicable to the concept of framing of emotions (Miller & Riechert, 2001; Van der Meer et al., 2014). Frame alignment has very generally been investigated in terms of word occurrences. By investigating the phenomenon in more detail, further insights on frame alignment can be gained. It might actually be the case that frame alignment does not apply to all types of frames when investigating frame content in more detail. Jin (2009) found out that which emotions are experienced depends on the specific crisis situations, but not on the individual. Almost the same emotions are experienced by different people in the same crisis situations (Jin, 2009). Therefore, framing of emotions is expected to not vary as much as other types of frames in the first phase as the public, the media and the organization all feel negative emotions due to the crisis. In terms of the framing of emotions, it is more difficult to change perceptions dramatically due to the fact that the crisis is a negative situation and can for example not be depicted as a positive one. This leads to the expectation that frame alignment will already come into place in the first phase of the crisis when it comes to emotional frames and not only in the second phase.
H5: The framing of emotions about the crisis situation is already in alignment between the different actors in the first phase of the crisis.
Framing of Responsibility
As mentioned earlier, another important frame that has been identified in research is the framing of responsibility (Coombs, 2007). It is also one of the five dominant types of frames used in crisis communication (An & Gower, 2009). The framing of responsibility is closely related to framing of emotions but is more about peoples’ ideas and reasoning on what causes have led to the crisis (Coombs, 2007). Such causes are determined due to the accessible information on the crisis. Coombs (2007) explained that there are different kinds of crisis types. These crisis types are the victim cluster, the accidental cluster and the intentional cluster (Coombs, 2007). In contrast to emotional frames and human-interest frames, the perceived responsibility that an organization has for a crisis is the main focus within the responsibility frames (An & Gower, 2009). In the victim cluster the organization has no responsibility and in the intentional cluster the organization has full responsibility. The accidental cluster lies in between in terms of responsibility (Coombs, 2007). Other parties than the organization might also be incorporated in the responsibility attribution, namely the media and the public.
It can be expected that in each organizational crisis attribution of responsibility will take place and therefore, responsibility frames are important in crisis situations (Coombs, 2007; An & Gower, 2009). In contrast to emotional frames, the responsibility frames are expected to vary much due to different parties’ interests (Brown & White, 2010; Kanso, Levitt & Nelson, 2010). This leads to a second sub-research question within this research:
RQ3: To what extent do different parties influence the framing of responsibility of the crisis of GermanWings?
Up until now, it has not been found out who the most influential actors or frame developers within this frame alignment process are. Van der Meer et al., (2014) have shown that the public adapted the frames of the organization and it has been proven by many researchers that the media influence the public through for example agenda-setting (Carroll & McCoombs, 2003; McCoombs & Shaw, 1972). The media report about an issue and therefore the public read about it in for example newspapers. Journalists apply frames and make a topic more salient compared to others and by framing have influence on how the public perceive an
issue and think about it (Hunter, van Wassenhove, Besiou & van Halderen, 2013). Though apart from the news media setting the agenda, there has not been investigated how the three actors, including the organization, relate upon each other in terms of framing and whether organization and public make use of agenda-setting as well.
As the attribution of responsibility can be influenced by information on the crisis it might be the case that news media and the organization are more likely to set up frames that include responsibility issues (Kanso et al., 2010; Coombs, 2007). The organization will try not to be depicted as having caused the crisis intentionally and might try to influence the responsibility frames of other stakeholders (Coombs, 2007), such as the media to reduce the damage of the crisis on the organizational reputation (Kanso et al., 2010; Liu, 2010). Being perceived as a victim of the crisis or having caused it accidentally will lead to less damage for the organization (Coombs, 2007). Therefore, the organization is expected to make use of agenda-setting theory as well. It can try to influence the media in terms of responsibility attribution and to be depicted as having only little responsibility. As the organization has inside information on the crisis and is likely to communicate after a crisis with the media it can try to influence the framing of responsibility of the media to get its’ own frames accepted. If this happens, influential relationships might be found in terms of the usage of responsibility frames across the three crisis phases.
H6a: The organization will influence the framing of responsibility used by the news media. Furthermore, the media might play an important role in the framing of responsibility as journalists sometimes even do have more knowledge and insights on an organizational crisis than the organization itself (Stephens & Malone, 2009). Therefore, frames of responsibility are also expected to be made up by the news organization and accepted by the public. Furthermore, it might be the case that due to differences in perceptions, the news media frame the responsibility of the organization differently. Theories like the agenda-setting theory (Carroll & McCoombs, 2003) have already proven that the media influences the public. This is also expected in terms of framing of responsibility (Entman, 1993). Within their articles, journalists will apply frames as a result of the inability to communicate without using frames (Entman, 1993). The public is confronted with the applied frames of responsibility of the journalists while reading the articles and are likely to adapt these frames (Hoeken, 2003). Therefore, also in this case influential relationships might be possible to identify across crisi phases.
H6b: The news media will influence the framing of responsibility of the public.
Whether the public’ empowerment (Castells, 2007) is strong enough to influence other actors in a crisis is also interesting to investigate. Journalists might read through social media content and other communication on the Internet such as blogs when writing about a crisis. Through this process they can come across frames different from the ones applied by the general media (Castells, 2007). Whether journalists accept these frames is interesting to investigate.
H6c: The public will influence the framing of responsibility of the news media.
The public requires providing fast and much information on a crisis (Coombs, 2007). While reading or listening to these, they are influenced by the applied frames of journalists. Journalists are the producers of the news media, from which the public receive their information. It might be the case that the news media and its journalists are only a step between the organization and its other stakeholders. If the organization is able to influence the framing of responsibility by the news media, it might indirectly also be able to influence the public through the media.
H6d: The public framing of responsibility will be influenced by the organization through the news media.
Figure 1 illustrates the relationship between the different social actors in terms of influencing their framing of responsibility.
Figure 1.
The relationship between the different social actors in terms of framing of responsibility.
In terms of closeness of the event, the same arguments as for the emotional frames are expected to apply. German citizens feel more impact of the crisis and might therefore also be
more engaged in figuring out the responsibility of the organization for the crisis (Seletzky & Wilzig, 2010).
H7a: There will be more framing of responsibility used by the different parties in Germany compared to the Netherlands.
H7b: There will be more framing of responsibility used by the different parties in Germany compared to the United States.
With regard to frame alignment it can be expected that all frames will align after the first period of the crisis (van der Meer et al., 2014). Due to the different interests of the parties involved, the theory of van der Meer et al. (2014) is expected to be applicable in terms of framing of responsibility. In the first phase there is not enough information on what caused the crisis to make up strong frames of attribution of responsibility and the different parties all perceive the situation differently. After the first phase, the sense making process is expected to come into play (Weick, 1988). For understanding the complex crisis situation the alignment of frames is necessary to bring the crisis to an end in terms of understanding it (Van der Meer et al., 2014).
H8: Framing of responsibility by the different parties will align in the second phase of the crisis.
Method Material
Within this research, text messages dealing with the crisis of the GermanWings airplane crash on the 24th March 2015 were analysed. The researcher collected a number of different text messages to get insights on the frames used referring to the airplane crash.
Media Depiction. First of all, newspaper articles were collected through LexisNexis
Academics. These are articles from highly circulated serious and popular newspapers from the countries under investigation. Information on the newspapers can be found in table 1. It shows which newspapers where chosen and how many articles from each newspaper have been found that dealt with the crisis under investigation. Serious newspapers are expected to look at an issue from a less sensation seeking perspective than popular newspapers (Vincent et al,
2002) and the mix of both types of newspapers included will give a more detailed overview of the media depiction.
For the collection of the newspaper articles the researcher used the search term ‘GermanWings’ in the period between 24th
March 2015 and 24th September 2015. By applying this search term, 253 articles have been found from all newspapers selected. In the case of the German newspaper Bild the articles were collected through the newspapers’ website. Based on the review 19 articles have been deleted from the sample being identified as not dealing with the crisis under investigation. In total, 16% of the total sample of text messages is newspaper articles. These articles represent the media depiction on the crisis of the airplane crash.
Table 1.
Newspaper articles coded for from the three different countries (n= 329).
Newspaper Number of articles Kind of newspaper
U.S. newspapers
New York Times (3) 64 Serious
USA Today (4) 21 Popular
German newspapers Handelsblatt (2) 66 Serious Bild (1) 95 Popular Dutch newspapers Volkskrant (6) 35 Serious Telegraaf (5) 48 Popular Total n= 329
Note. Between brackets (*) the number coded for in data-set.
Organizational Depiction. The messages of the organization are press releases, speeches and
Facebook posts of GermanWings as well as Lufthansa. Posts of Lufthansa are also considered organizational messages as GermanWings is a subsidiary of it. In total, 41 organizational messages have been collected. Press releases have been used from the organizational
webpage1. 11 German and 10 English press releases have been found. Next to this, the researcher collected two speeches and transcribed them. The speeches are from the CEO of GermanWings and Lufthansa. From the Facebook page, the researcher collected 12 other posts of GermanWings. These were posted on the page itself either as a response to the airplane crash or as comments on posts of stakeholders on the page.
Public Depiction. The Facebook page was used within this research as an issue arena
(Luoma-aho & Vos, 2010). This means that the social media platforms can be seen as places where public discussion on issues takes place. The issue of the airplane crash of GermanWings has been discussed at different places and stakeholders started up own dialogues on the social media platform. On such issue arenas, the public also engages in framing and people influence each other. Facebook is expected to represent a broad variety of stakeholders from different countries within this research and is therefore a good representation of the different issue arenas. In total, the researcher collected 3769 public posts through the GermanWings Facebook page, which represent the public messages. Due to the high amount of posts only every second post was coded. Furthermore, posts without content, posts that were unreadable and not representing words, and posts that only included smileys were deleted from the sample (385 posts). All in all, within this research 1692 messages (82.1%) represent the media depiction.
Procedure
Within this research a manual content analysis has been conducted. The researcher started with an open coding to investigate specific frames that were striking within the topic of the GermanWings airplane crash. It has been looked for the categorical frames of An and Gower (2009) using the instructions of van Gorp (2010) to identify specific frames.
The researcher searched for, and identified specific frames that included frames of responsibility and emotional frames. Two parties that were often assigned responsibility were found. Besides, the researcher found four specific emotional frames that occurred often. Responsibility of the organization and the co-pilot was found to be discussed often. Furthermore, emotions were directed towards the victims, showing empathy and pity for them and their families. Negative emotions towards the co-pilot and the organization were also often articulated in messages as well as positive emotions towards the organization
1
mentioning that customers still trust in Lufthansa/ GermanWings and still want to fly with the airline.
Development of Codebook. After the open coding, a codebook was developed making it
possible for the researcher to identify the frames found in open coding through all messages. For setting up the codebook, the responsibility frame from research by Semetko and Valkenburg (2000) has been used in a refined manner. The emotional framing aspect has been based on the study by An and Gower (2009) and taken one step further from only identifying human-interest frames to looking more specific at emotions which often come into play in crisis and which can be used as frames in itself.
Based on the research of Semetko and Valkenburg (2000) the methodology for answering the codebook questions has been used. The answer categories Yes (1) and No (0) for the coding of the emotional frames and for one of the questions coding for responsibility have been taken over. The specific responsibility frames needed to be answered based on four answer categories, (0) no responsibility of organization/ co-pilot, (1) victim cluster, (2) accidental cluster, (3) intentional cluster. Differentiation between attributions of responsibility and framing of emotions towards different actors in the crisis, gives more specific knowledge on the content of frames articulated within the crisis. An example of a content specific frame is the attribution towards the co-pilot of intentionally crashing the airplane. In this case the researcher coded for whether there was attribution of responsibility given to the co-pilot and if so which of the three types. The specific emotional frames were about positive/ negative emotions towards the organization and negative emotions towards the co-pilot and sympathy for the victims. These needed to be answered with either Yes or No.
Text messages often included anger towards the co-pilot and accusations or questions about his intentional actions that led to the crash of the airplane. Such messages were often coded as messages that included both, responsibility and emotional frames. The divide in presence of attribution of responsibility and emotional frames and the coding for the specific parties involved makes the frame alignment analyses based on content possible. An overview of the specific frames coded for can be found in table 2. This table shows that in the total sample there are many emotional frames present, 61.3 % of all text messages include an emotional statements. In contrast to that, only 20.7 % of all messages include responsibility frames.
Table 2.
Occurrence of frames coded for (N= 2062).
Frame Coding Descriptive Percentages
General frames measures Any attribution of
responsibility present
427 (Yes) 20.7%
Any emotional frames present 1264 (Yes) 61.3% Frame alignment measures
Organizational responsibility 7 (Victim cluster) 46 (Accidental cluster) 58 (Intentional cluster)
0.3% 2.2% 2.8% Co-pilot responsibility 2 (Victim cluster)
15 (Accidental cluster) 261 (Intentional cluster)
0.1% 0.7% 12.7% Negative emotions towards
organization
48 (Yes) 2.3%
Positive emotions towards organization
155 (Yes) 7.5%
Negative emotions for victims 972 (Yes) 47.1% Negative emotions towards
co-pilot
54 (Yes) 2.6%
Note. N= 2062.
Furthermore, codes for the proximity towards the crisis, the time phases of the crisis and the publishing source have been developed. Whether a message is in proximity to the event was identified based on the country of origin. 61% (n= 1257) of the text messages came from Germany and only between 4.4 and 4.8% of the messages came from the Netherlands and the United States and were considered as more far away from the crisis. Besides, a number of messages with an unclear country of origin (n= 579) was found.
Altogether the codebook consists of 14 items of which two items identify presence of emotional and responsibility frames and seven identify content-specific frames. The whole codebook can be found in Appendix A.
The Crisis Phases
The messages have been grouped in three phases according to the publishing date. These three groups represent the three time phases of the crisis identified in earlier research (Miller & Riecher, 2001). Messages between the 24th and the 25th March 2015 have been considered to belong to the first phase of the crisis, messages between the 26th and 31st March to the second phase. Messages from the 1st April 2016 onwards belong to the third phase.
The categorization of the different dates was done subsequently to the closed coding. After having read all messages of the sample, the researcher was able to recognize content specific difference. Based on these differences the categorization has been made. Due to the fact that there are no clear theoretical instructions on how to divide the crisis into phases, categorization based on content was the only possible method. The first phase of the crisis included 1008 messages (48.9%). Messages in this phase informed readers about the crash of the airplane and the victims as well as speculations about what caused the crash. The second phase included 662 (32.1%) messages and started when the attribution of responsibility became clearer. Messages in this phase included clearer attribution of responsibility and were less about providing only the information that the airplane crash has happened. In the third phase, the attribution of responsibility is very clear and emotional statements were more about anger and hate towards the co-pilot who crashed the airplane. There are 392 (19%) messages that belong to the third phase.
Intercoder Reliability
After completion of the codebook, a second coder applied the codebook to find out whether the instructions are clear and if they can be understood by other people than the researcher. This coder coded 10% of the 2062 messages, namely 207 messages. Intercoder reliability analyses have been conducted using Krippendorf’s alpha. The reliability was tested for twelve of the 14 items of the codebook. Eleven of the twelve items had a Kalpha of .86 or higher which was very sufficient. The item positive emotions towards the organization had a Kalpha of .66, which was of lower reliability but still accepted regarding the norms expecting a Kalpha of .60 or higher. All reliability values can be found in Appendix B.
Ethical Considerations
Within this research, all information and posts of individuals on Facebook have been treated anonymously and private information, as for example the name of the author, were not taken into consideration.
Results
To answer the eight hypotheses within this research a number of statistical analyses were conducted using SPSS. The hypotheses were answered using primarily logistic regression analysis and descriptive statistics.
Figure 2 and 3 give an overview of the usage of all frames of responsibility and emotional frames across the three actors, the media, organization and public.
Figure 2.
Presence of Frames of Responsibility across Actors (n= 427).
0,00% 20,00% 40,00% 60,00% 80,00% 100,00% P re sence o f re spo ns ibi lity f ra m es
Presence of Frames of Responsibility across Actors
Media Public Organization
Figure 3.
Presence of Emotional Frames across actors (n= 1264).
Based on the data and the figures, it becomes clear that most responsibility frames are used in the second and third phase of the crisis (37.2% and 37.5%). Figure 2 shows for example that the organization did not use any frames of responsibility and less emotional frames than the other two parties. Furthermore, it becomes clear that the media used most emotional and responsibility frames. In Appendix C, figures 4 and 5 can be found. They represent the usage of general frames across nations.
H1 suggested alignment of emotional and responsibility frames by the media, the organization and the public in the second phase of the crisis. After this, Chi-square analyses were conducted to investigate the relationship between the publishing source and the different frames. For the two specific responsibility frames the results showed a significant relationship between organizational responsibility and media, organization and public χ2(6) = 23.144, p ≤ 0.001, as well as for co-pilot responsibility type in relation to publishing source χ2(6) = 150.74, p < 0.001. There were not enough expected frequencies in each answer category (Field, 2009) of the variables. One of the assumptions of logistic regression is thus violated. Still, logistic regression analyses for every specific frame were conducted. For emotional frames binominal regression models were used as the variable was dichotomous. For the responsibility frames, multinomial logistic regression was conducted because these variables included four answer categories.
0,00% 20,00% 40,00% 60,00% 80,00% 100,00% Presence of em ot ional f ram es
Presence of Emotional Frames across Actors
Media Public Organization
The first analysis used the media, organization and public to predict organizational responsibility frames (χ2= 19.87, p < .005 with df= 6). The model was significant, meaning that one or more of the three actors used more frames of responsibility than the others. Therefore, this analysis already showed opposite results than what was expected for H1. For the responsibility frames on the co-pilot, logistic regression was not possible to be conducted due to too little values in each category.
To test H1 with regard to the emotional frames, binary logistic regression showed that all models were significant, in contrast to what has been expected in H1. Within the analyses shown in table 3, the organization has been used as a reference group. Model 1 and 4 do not show where the differences between the groups lie. By using a different group as a reference, namely the media, it becomes clear that the media used more negative emotions towards the organization (χ2= 35.93, p < .001 with df= 2). In model 4 the difference was made by the public who used much more negative emotions towards the co-pilot (χ2= 100.04, p < .001 with df= 2). All the results of the different logistic regression analyses lead to the rejection of H1.
Table 3.
Logistic regression analyses for alignment of different emotional frames in the second phase of the crisis (n= 662). Model 1: Neg. emotions twd. organization Model 2: Pos. emotions twd. organization Model 3: Neg. emotions for victims Model 4: Neg. emotions twd. co-pilot
Eβ SE β Eβ SE β Eβ SE β Eβ SE β
Media .00 12710.18 69234638 4.6 3580.6 8 4.06* 0.71 .00 12709.55 Public .00 12710.18 6.202 0.71* 9.70** 0.78 .00 12709.55 Model χ2 0.81** 21.12** 25.53** 48.22** Nagelkerke’s R² .01 .09 .06 .20 n= 662 662 662 662
H2 expected that different emotional and responsibility frames are used in the first phase by nations who are more far away from the crisis. To test H2, the same analyses as above were used, but with the variable country of origin (Germany and Netherlands, controlling for Unites States) as independent variable. Germany as country close to the crisis versus the Netherlands and the United States as most far away from the crisis were used, and the first phase was selected. Furthermore, all answer categories other than the three countries under investigation were excluded from the analyses through the select cases function.
For the responsibility frames, the variable co-pilot responsibility was excluded from the analysis as there were too little values in each category (Field, 2009). The logistic regression was not able to give any results. The model for the relationship between organizational responsibility and country of origin was found not to be significant (χ2= 1.15,
p =.766 with df= 6). In contrast to what has been expected, there were found no significant
different results and the model was unable to predict differences in terms of framing of responsibility in the first phase of the crisis based on the country of origin of the text message. For the emotional frames, results of the different models can be found in table 4. Models have been found to be significant, but there are no significant relationships between the country of origin and the occurrence of specific emotional frames.
The only significant relationship found was for the emotional frame relationship between sympathy for the victims based on the country of origin of the text message (χ2= 4.07, p < .001 with df= 1). Nagelkerke’s R² of .01 indicated a very weak relationship between prediction and grouping. The odds of finding an emotional statement that indicates sympathy towards the victims in a German text messages increases by 1 (= 2.18*100-100). The odds of finding an emotional statements with sympathy towards the victims in a text message of the other countries was found to be not significant with p= .091. For the variable positive emotions towards the organization, there was found a significant difference but this relationship only showed an effect for the usage of no frames showing sympathy for the victims as can be seen in table 4. For the variable negative emotions toward the co-pilot, the analysis could not be conducted due to the fact that there were no negative emotions towards the co-pilot present in the first phase of the crisis. Based on the analysis it can be assumed that the frames are not differing as much as was expected and that overall H2 needs to be rejected.
Table 4.
Logistic regression analyses for variation of different emotional frames in the first phase of the crisis (n= 734). Model 1: Neg. emotions twd. organization Model 2: Pos. emotions twd. Organization Model 3: Neg. emotions for victims
Eβ SE β Eβ SE β Eβ SE β
Germany 6953263.5 6 6893.04 .82 .58 2.18 .37* Netherlands .00 6893.04 .13 .53** 1.83* .36 United States Model χ2 0.29** 0.13** 4.07** Nagelkerke’s R² .01 .00 .01 n= 734
Note. *= significant at < .05; **= significant at < .001.
H3 suggested that the responsibility and emotional frames of the different nations, Germany, the Netherlands and the United States align in the second phase of the crisis. Therefore, it was expected to find no significant difference between the frames of the countries. The same analyses as for H2 have been conducted, selecting the second phase of the crisis and only the countries under investigation. The first analysis showed that there was no significant relationship between presence of organizational responsibility in text messages and country of origin (χ2= 5.83, p = .443 with df= 6). In contrary to that, the second analysis showed a significant relationship between presence of co-pilot responsibility in text messages and the country of origin (χ2= 73.06, p < .001 with df= 6). Nagelkerke’s R² of .22 indicated a weak relationship between prediction and grouping. Unfortunately only one relationship had predictive power. The odds of finding no responsibility frame in a German text message decreases by 1 (= -2.16.72*100-100). Based on these results it can already be stated that in terms of co-pilot responsibility no frame alignment has been found.
For the specific emotional frames, the results can be found in table 5. The first emotional statement about negative emotions towards the organization was not significant,
but the other three variables all showed a significant relationship. As model 1 was not significant, another logistic regression has been conducted with a different reference category than the United States. With Netherlands as a reference category, the model was significant (χ2= 5.02, p < .001 with df= 2). Nagelkerke’s R² of .01 indicated a very weak relationship between prediction and grouping. The relationship predicting the finding of negative emotions towards the organization in a German text message was not significant, p= .200. For the United States it was significant. The odds of finding no responsibility frame in a German text message decreased by 1(=-.543*100-100). The other three models showed little differences between the two countries when using the United States as a reference group. The Netherlands as a reference group showed increased differences. German text messages were more positive about GermanWings, they included more negative emotions towards the victims and less negative emotions towards the co-pilot compared to the United States. These analyses showed that there are differences between the usage of frames by the different nations in the second phase of the crisis. Therefore H3 needs to be rejected.
Table 5.
Logistic regression analyses for variation of different emotional frames in the second phase of the crisis (n= 405). Model 1: Neg. emotions twd. organization Model 2: Pos. emotions twd. Organization Model 3: Neg. emotions for victims Model 4: Neg. emotions twd. co-pilot
Eβ SE β Eβ SE β Eβ SE β Eβ SE β
Germany .00 5991.62 .00 5991.62 1.49 .34 2.33 .60 Netherlands .00 5991.62 1.00 8919.69 1.04 .47 .23 .64* Model χ2 3.28 9.31** 2.00** 22.60** Nagelkerke’s R² .03 .07 .01 .14 n= 405
Note. *= significant at < .05; **= significant at < .001.
The sub research question about the extent of emotional framing used by the different parties was answered using the results of the three hypotheses on emotional framing and descriptive statistics. There are much more emotional frames used (1246= Yes, N= 2062) compared to responsibility frames (427= Yes, N= 2062). Furthermore most emotional frames
are used in the first phase of the crisis (82% of emotional frames). H4a and H4b were about the amount of emotional framing used by the different nations. Therefore, the whole data from all phases and the general variable about presence of emotional frames was used in the analysis. To answer hypotheses H4a and H4b, first Chi-square was conducted. Chi-square for country of origin of the text message and whether there was any framing of emotions present in the message χ2(5) = 152.73, p < 0.001 was found to be significant. Most emotional statements were found in German text messages and in messages from other countries than the ones under investigation. This leads to the assumption that H4a and H4b can be accepted. Logistic regression analyses have been conducted to find confirmation for the results found by Chi-square. For these analyses only the three nations have been selected, and unidentified countries as well as other nations have been excluded from the analysis through selecting cases. The logistic regression showed that the model was significant (χ2= 39.02, p < .001 with
df= 1). Nagelkerke’s R² of .04 indicated a very weak relationship between grouping and
prediction. Germany as the country of origin of the text message was significantly related to finding an emotional frame in the message. The odds of finding an emotional frame in a German text messages increased by 1(=2.69*100-100), for the other two nations the odds of finding an emotional frame decreased. It was expected that Germany would have the most emotional statements in comparison to the United States and the Netherlands. This was found to be true. Based on the Chi-square analysis and the logistic regression H4a and H4b can be accepted.
H5 also investigated the emotional frames. It suggested that the emotional frames of the different actors are in alignment. In contrast to H3, this hypothesis investigated only emotional frames and the alignment across all three phases. Therefore, results might differ from the results found for H3, for which only emotional frames in the second phase were compared across nations. To answer H5, logistic regression analyses have been conducted for all specific frames of emotions as independent variables. Cases have been deselected again, as all countries are compared. To accept the hypothesis, the results of the analyses needed to show no significant differences as the frames are expected to align. Negative emotions towards the organization (χ2= 28.90, p = 1.000 with df= 3), positive emotions towards the organization (χ2= 38.53, p = 1.00 with df= 3), negative emotions for the victims (χ2= 32.22, p = 1.000 with df= 3) and negative emotions towards the co-pilot (χ2= 103.10, p < 1.000 with
specific emotional frames used are the same across the different actors and H5 can be accepted.
The following results of H6a, b, and c help answering the second sub research question. This question is about the extent of influence of the different parties on each other’s use of responsibility frames. Hypotheses H6a, b and c are all about the influence of the three actors, the public, the media and the organization upon each other and their usage of frames. They are all answered using descriptive statistics. Cross-tabulations are conducted to investigate which frames are used by which actors. Therefore, the time phases under consideration were selected. The organizational responsibility was categorized under the accidental cluster three times by the media in the first phase. The public used all three types of responsibility, the victim cluster one time, the accidental cluster and the intentional cluster both six times. The organization did not use any attribution of responsibility in the first phase. All three actors did not use any framing of the responsibility of the co-pilot in this phase.
The usage of the specific responsibility frames in the first and second time phase can be found in table 6 and table 7 below. The usage of frames in the third phase can be found in Appendix D. Based on the descriptive statistics it can be seen that the co-pilot responsibility frame was used most often.
Table 6.
Frames of organizational and co-pilot responsibility used by three actors in first phase (n=1008).
Organizational Responsibility
Victim Accidental Intentional
Media 0 3 0
Public 1 6 6
Organization 0 0 0
Table 7.
Frames of organizational and co-pilot responsibility used by the three actors in the second phase of the crisis (n=662).
Organizational Responsibility Co-Pilot Responsibility
Victim Accidental Intentional Victim Accidental Intentional
Media 0 8 4 0 7 73
Public 5 4 33 2 5 72
Organization 0 1 0 0 2 0
Total 5 13 37 2 14 145
Note. n= 662 applies for both organizational and co-pilot responsibility attribution frames.
The cross-tabulations for the relationship between organizational responsibility and media, organization and public was found to be significant χ2(6) = 20.23, p < 0.005 in the first phase. Of the 1008 text messages that were analysed in the first phase, only 16 included organizational responsibility. The relationship between organizational responsibility and media, organization and public was also found to be significant in the second phase of the crisis χ2(6) = 23.14, p < 0.001. The organization only used the accidental cluster, which had been used by the news media in the first phase of the crisis already. TH6a needs to be rejected as no influence of the organization on the media can be expected based on the data.
Considering the other specific frame of responsibility, no attribution of responsibility was given to the co-pilot in the first phase of the crisis. Therefore, it will be looked at the second and third phase of the crisis for this variable. The relationship between co-pilot responsibility and publishing source was found to be significant in the second phase of the crisis (χ2(6) = 150.74, p < 0.001). The media and the public both used the intentional cluster most. The relationship between co-pilot responsibility and the three actors in the third phase was also found to be significant with χ2(4) = 138.90, p < 0.001. In this phase, most of the framing of responsibility was found to be intentional and done by the news media. Based on these two phases nothing can be said about the relationship examined in H6a. The hypothesis was rejected based on the results of the variable of framing organizational responsibility.
H6b expected an influence by the news media on the public. H6c expected this influence the other way around. Due to the results shown in table 6 and 7, it can be stated that most attribution of responsibility towards the organization was done by the publics in the first phase of the crisis. In the second phase of the crisis this attribution of responsibility by the
public increased even more. The news media also attributed responsibility and used the accidental cluster most in the second phase. This frame might have been taken over from the public. Based on the visual analysis it is difficult to examine an influential relationship between both actors and due to the fact that there are no logistic analyses in the different phases of the crisis possible because of too little cases to calculate most models, H6b and H6c were rejected. There was too little framing of attribution of responsibility done by the media in the first phase of the crisis to investigate an influential relationship.
H6d expected a mediating relationship of the media on the framing of responsibility by the public influenced by the organization. Looking at table 6 and 7 it becomes clear that H6d needs to be rejected. The organization itself used very little frames of responsibility. The only frame that was used was the accidental cluster which was not the main frame used by the other two parties. Chi-square for the relationship between organizational attribution of responsibility (χ2(3) = 1.37, p = .714) was found to be not significant. Co-pilot responsibility (χ2(3) = 15.60, p ≤ 0.001) was found to be significant, but only predicting no occurrence of responsibility frames. No further analyses were conducted as visual analyses presented in table 6 and 7 already showed that the organization did not influence the public. Therefore, also no mediating relationship can exist and hypothesis H6d was also rejected. Referring back to the sub-research question it can be stated that there seems to be little or no influence of the different actors on each other’s framing of attribution of responsibility.
H7a and H7b expected more framing of responsibility by German text messages compared to Dutch and American text messages. First Chi-square was conducted. The results showed that there was a significant relationship between country of origin of the text message and whether there was any attribution of responsibility in the message χ2(5) = 249.15, p < 0.001. Most of the attribution of responsibility was present in American and Dutch text messages. This already showed that H7a and H7b needed to be rejected as Germany was expected to use more framing of responsibility than the other two countries. A logistic regression was also conducted, to confirm the results that lead to the rejection of H7a and H7b. Again, as for H4a and H4b only the text messages from Germany, the Netherlands and the United States were included, all other cases were deselected. In total, 1447 messages were analysed. The model was significant with χ2= 192.61, p < 0.001 with df= 1. Nagelkerke’s R² of .18 indicated a weak relationship between grouping and prediction. There was a significant relationship between Germany as the country of origin of the text message and finding no frames of responsibility in the message. The odds of finding a frame of responsibility in a
German text messages decreased by 1(=.99*100-100). Therefore, H7a and H7b needed to be rejected as had already been assumed. German text messages used the least amount of responsibility frames compared to the United States and the Netherlands.
H8 assumed frame alignment of responsibility frames in the second phase of the crisis. To analyse H8, the data from the second phase of the crisis was used. It was expected to find no significant difference between the different sources, news media, public and organization in terms of responsibility frames. Therefore, only the two specific frames were used in the analyses. The regression for the relationship of organizational responsibility and the origin of the text messages being from the public, the news media or the organization was significant (χ2= 19.87, p < .005 with df= 6). Nagelkerke’s R² of .058 indicated a very weak relationship between prediction and grouping. The multinomial regression analysis was not possible to be conducted for the co-pilot responsibility due to the fact that there were too little values for a number of categories (Field, 2009). The analysis showed that H8 needed to be rejected as there were significant differences found between the sources.
An overview of all the accepted and rejected hypotheses can be found in table 8. Table 8.
Overview of acceptance/ rejection of hypotheses tested within this research.
Hypotheses Analysis Accepted/
rejected H1 Frame alignment of organization,
news media and public in 2nd phase
Binary and multinomial logistic regression
Rejected
H2 No frame alignment of different nations in 1st phase of crisis
Binary and multinomial logistic regression
Rejected
H3 Frame alignment of different nations in 2nd phase
Binary and multinomial logistic regression
Rejected
H4a/b More framing of emotions in
Germany compared to United States/ the Netherlands
Binary logistic regression
Both Accepted
H5 Emotional frame alignment across news media, organization and public
Binary logistic regression
Accepted
H6a/b/c Influence of organization/ news media/ public on framing of
Chi-square and visual analysis
responsibility of news media/ public/ news media
H6d Public framing influenced by organization through news media
Chi-square and visual analysis
Rejected
H7a/b More framing of responsibility in Germany compared to Netherlands/ United States
Binary logistic regression
Both Rejected
H8 Framing of responsibility in alignment by news media,
organization and public in 2nd phase
Multinomial logistic regression
Rejected
Within this research, the crisis was distinguished in three time phases. It was investigated whether alignment of frames took place in the first or second phase of the crisis. The following table gives a summary of all the results that were found based on the three time phases. It shows whether frames of the different nations and the different actors aligned and if so in which phase this happened.
Table 9.
Summary of all results showing the occurrence of frame alignment across the phases.
Phase 1 Phase 2 Phase 3
Emotional frames Frame alignment across nations/ No frame alignment across actors No alignment across actors/ nations Frame alignment across actors/ nations* Responsibility frames Frame alignment across nations/ No frame alignment across actors No alignment across actors/ nations No frame alignment across actors/ nations*
Note. *analyses have been done with basic frames (any attribution of responsibility and any emotional frames) to give a summary of frame alignment.
