1
University Of Amsterdam
Graduate School of Communication
Master in Corporate Communication- Master Thesis
Work carried out by: João Pedro Sousa Guerra
Supervised by Piet Verhoeven
The MH370 flight: Finding the right words when
you lose a plane
Message Framing in Crisis Communication: An automated semantic-network
analysis
2
Abstract
This study examines the underlying frames present in organizational, public and media messages during a crisis situation, more specifically the “MH370 Malaysian Airlines flight disappearance”. The purpose of this study is to extract and analyze the different meanings present in organizational crisis communication messages and how are they reflected in the media and public message framings. To highlight the underlying implicit frames present in the messages, an automated semantic-based network analysis is employed. This study comprised analysis of four different sets of messages: Press-releases/updates and official tweets by Malaysian Airline, newspaper articles and user Tweets. By examining the different underlying frames present in each set of messages it became apparent that the organizational messages were framed differently. Also, there was no apparent effect of the organizational message framing in the user Tweets, while the contrary occurred in the newspaper articles. The extensive study of crisis situations using case studies provides not only guidelines to future crisis management strategies, but also empirical knowledge about the approaches used and their consequent efficacy.
3
“The secret of crisis management is not good vs. bad, it’s preventing the bad from getting worse”
-Andy Gilman, Comm. Core Consulting Group
Introduction
Back in 2007, more than 1000 flights from the airline company JetBlue were canceled due to an ice storm that hit the east cost of the United States. An event of this magnitude is sure to have a strong negative impact on a company. It is up to the company to understand the problem and deal with it in such a fashion that more damage to the company is avoided. At the time, the CEO of the company David Neeleman adopted a position in which the blame was taken and assumed all by the company. A media campaign was initiated in order to reach a vaster public, a customer bill of rights was introduced and a detailed list of every action the company would take to compensate the affected passengers was made available. It is in situations of crisis that companies are put to the test, reinforcing the necessity to be ready to situations that could affect a company’s reputation and, therefore, profit capacity.
Over the years, crisis management and crisis handling strategies have gain territory in the fields of communication practice and research. With the rise of social media, crisis management has been going through a process of transformation by both professionals and researchers (Veil, Buehner & Palenchar, 2011). When looking at the current media spectrum, it is, without a doubt, clear that social media has risen to be one of the most important and used forms of media, shining amongst others due to its open and interactive characteristics (Mayfield, 2008). As Kaplan & Haenlein (2010) suggest, social media should be considered as a revolutionary trend in
4
which organizations should be highly interested in. Due to its popularity and massive usage, organizations realized and understood that social media can be used as an effective communication tool for both promotional and crisis management purposes.
According to Dimmick, Chen and Li (2004), there is a certain level of competition between the Internet and what is considered traditional media. Some scholars as Brown & Golsbee (2000) and Outing (1999) believe that the Internet, due to its dynamic and interactive characteristics, will at some point draw audiences away from traditional media. Nonetheless, and even though television has been losing audiences to the Internet, newspapers are still considered a reliable source of information (Dimmick, Chen and Li, 2004) and still play an important role on nowadays’ media spectrum.
When faced with a crisis situation, organizations react differently and use different media to reach the public in order to minimize the effects of the crisis. Organizations use different approaches and strategies to handle the communicative challenges that each media poses. A clear example of this is the use of press releases directed at traditional media, and the use of tweets (that contain a maximum of 160 characters) when the focus is social media (Taylor & Perry, 2005). There is, in a sense, differentiation in the strategies applied to each media, and this differentiation may lead to different communicative results. It is also essential to understand that each crisis is different (Siomkos & Kurzbard, 1994), but there is a necessity to thoroughly analyze and study each one, as there is knowledge that can be applied and can be used to better prepare companies for future crisis situations.
With that necessity in mind, the application of an automated content analysis to crisis communication-related messages has proven to be an efficient method to facilitate the
5
understanding of the meaning construction of a crisis (Leydesdorff & Hellsten, 2005). As Hellsten (2010) posits, identifying the meaning of the words presented in a message and the relative position of those words in word networks and repertoires, aids the process of meaning construction, and provides a better understanding of the overall crisis-management strategies. This co-word analysis plots the strength of the associations between key words in different messages, enabling the comparison of the implicit frames present in those messages (van der Meer, Verhoeven, Beentjes and Vliegenthart, 2014).
Framing is a well-established concept in which meaning is given to different clusters containing words that are interconnected. These words give meaning to the crisis (Van Der Meer & Verhoeven, 2013), and therefore, it is necessary to comprehend the different framing of the messages to finally understand their overall significance. Consequently, analyzing the different framing of the messages sent out by the organization and by social and traditional media, using an automated content analysis enables a clearer overview of the distinct strategies applied to a specific crisis and its consequences in terms of public and media output.
Finally, a good example of crisis where the different message framing can be observable is the one of the MH370 flight. The “Malaysia Airlines Flight MH370 disappearance” can be considered as one of the most impactful ones of the year of 2014. On the 8th of March of 2014, a scheduled international passenger’s flight heading from Kuala Lumpur’s international airport to Beijing, China disappeared. Little more is known about this case, hence making it one of the most intriguing and possibly one of the biggest crises to have ever hit the aviation sector. It is, therefore, necessary to acknowledge that the MH370 flight crisis is a very particular one. Due to this fact, this crisis reached vast proportions, and it was covered by different media all over the
6
globe, occupying the whole media spectrum for quite a while. The magnitude of this crisis both in traditional and social media coverage, justifies its use in this study. From a communicative and research point of view it is necessary to, and more than understanding what happened to the plane, analyze how the company behaved communication wise and what was the impact of these strategies on traditional media and the publics’ reaction.
Considering what was aforementioned, the following research questions were developed:
RQ1- Which meanings about an organizational crisis situation are visible in
press-releases/updates official tweets of the organization, newspaper articles and twitter messages from the public?
RQ2- Do these meanings differ and if so, how?
Theoretical Framework
The field of organizational crisis has been extensively studied, as it is crucial for organizations to be prepared for possible events that my influence their work and profit-making capacity (Coombs & Holladay, 1996).
According to Barton (1993), crises are negative events that will affect the image and reputation of an organization. As Falkheimer and Heide (2006) state, the field of crisis communication suffers from a lack of systematic knowledge and theoretical foundations, which again reinstates the necessity for, even more, extensive studies and research that do not only contemplate case studies but that aim for the creation of guidelines for crisis management. This
7
would aid both scholars and organizations to better understand how crises are generated and, therefore, how can they be resolved with minimal damage to organizations.
Nonetheless, this necessity is paired with the increased difficulty of finding cases that are large and broad enough for general rules and ideas to be synthetized (Seeger, 2006). Crisis communication studies should base their ground objectives on process improvement, using a systematic overview and an objective analysis of data and assessment of organizational processes, as suggested by Seeger (2006).
Crisis and Crisis Communication Strategies
It is relevant to have an overlook of what happens during a crisis and there is a necessity to separate a Crisis situation, and the communication tools used by an organization during a crisis in order to “repair” or minimize the damages to that organization (Coombs, 2007). As it was stated before, crisis can be defined as "a specific, unexpected, and non-routine event or series of events that create high levels of uncertainty and threaten or are perceived to threaten high priority goals," (Seeger, Sellnow, & Ulmer, 1998, p. 233), this can be translated to organizations as any event that will disrupt the normal functioning of an organization, and it is a situation that poses a threat to both the reputation and image of that said organization. There are many examples of crisis that affected companies differently. It is, therefore, very difficult to predict a crisis situation. Nonetheless, Crisis Communication plans and strategies, that can be defined as “actions that aim at “preventing or lessening the negative outcomes resulting from a crisis” (Spence, Lachlan & Griffin, 2007), are becoming a common practice on organizations.
8
In an increasingly demanding and challenging market spectrum, organizations have understood the importance of “being prepared” as image and reputation are becoming more and more essential to organizations (Benoit, 1997).
Nonetheless, crisis situations are normally difficult to prevent and most of the crisis communication strategies are still developed and used according to each situation, which again reiterates the necessity for empirical knowledge. It is necessary that both companies and researchers look at what has happened in past crisis situations to assess what “works and what does not”.
It is, then, necessary to consider that when faced with a crisis situation, organizations need to communicate with diverse audiences, and therefore, different strategies should be used. This is because different audiences and publics potentially have diverse interests, goals and concerns (Benoit, 1997).
As stated by Teece, Pisano, and Shuen (1997), organizations should have dynamic capabilities and should be able to adapt to both the internal and the external environment in which they are involved. This is especially true when crisis communication strategies are developed. There are several factors that influence organizations and different actors who should be addressed in order to minimize the effect of a crisis.
The necessity for a better understanding of a specific line of action and its results in an empirical case is contemplated in the sub-questions described below:
Sub-Question 1 – Are the Traditional and Social media organizational crisis messages
9
Sub-Question 2 – Does the public and news media frame the crisis in accordance with the
frames used by the organization?
Crisis and the tourism/aviation sectors
When talking about crisis, especially crisis related to disasters/accidents, one of the most affected commercial areas is the tourism one. It is plausible to state that tourism is one of the biggest, if not the biggest, industry in the world (Sausmarez, 2008). Also, and at the same time, it is one of the industries that most suffer from public opinion changes, and that is vulnerable to different aspects that cannot be controlled by the human hand, such as climate change, natural accidents, etc. (Sausmarez, 2008). With that in mind, it is possible to assume that companies that are closely related to tourism and traveling, need to be particularly cautious in order to avoid a crisis situation, and have to be prepared to deal with one in case it happens.
Media and crisis
In terms of crisis, different media channels play a key role in the reporting, perception and management of said crisis (Ma, 2014). Traditional media channels (such as newspapers, radio, and television) are, nowadays, suffering from the competition raised by social media. Even though traditional media used to be considered the most efficient means of communication, the innovative characteristics of the technology-based communication channels, such as its wide availability, and the possibility to share content freely and faster than on traditional media, are now posing a challenge to traditional media. Moreover, it is possible to state that a new media landscape is emerging (Kietzmann, Hermkens, McCarthy & Silvestre, 2011). These specific
10
social media characteristics result on Internet based communication channels to be disruptive and different from traditional media. While traditional media are often associated with a more unidirectional communication mean, social media at its core possesses characteristics linked with openness, conversation and community (Mayfield, 2007). Therefore, when it comes to crisis solving, traditional media and social media strategies are different. Even considering the rise of social media as a tool for crisis management, traditional media still play an important role in crisis management as it is still seen by many as a reliable source of information that depicts a “broader social spectrum of social reality constructions” (Schultz, Utz & Göritz. 2011). Traditional media strategies are based on one-way crisis communication tactics such as press releases, press conferences, etc. On the other hand, Internet-based communications have proven to use both traditional communication tactics, but also include new strategies that match the technology characteristics of social media, and these have as main ground intercommunication and open channels of discussion (Taylor & Perry, 2005). While it’s possible to assume that newspapers are one of the most important channels when it comes to traditional media crisis communication, Twitter on the other hand seems to establish itself as the main channel through which organizations communicate during a crisis communication as it leads to more positive outcomes for an organization (Schultz et. all, 2011).
To have a better understanding of an empirical case study, it is, again, important to study both the output given by a certain organization and the output given by the two different kinds of media (traditional and social). The strategies used for traditional media, more specifically press releases, and the tweets done over a period of two weeks after the incident can be seen as a reflection of an organization’s crisis management strategy and should be taken into account. This can be justified by the importance of “response time” associated with crisis increases, as there is
11
a necessity for reliable information and communication in order to solve the crisis (Hale, Dulek & Hale, 2005), this is especially true for crisis that involved deaths and disappearances as the familiars and the society in general seeks for fast and clear answers to what happened.
Message Framing
According to Borah (2011), when analyzing different messages there is a necessity to understand what the words in those messages mean. This process is usually called framing. Framing constitutes one of the most prominent roles in terms of attributing meaning to a message (Van Der Meer & Verhoeven, 2012). According to Entman (1993, p.52): “To frame is to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described”. Furthermore, according to Van der Meer, Verhoeven, Beentjes and Vliegenthart (2014), framing can be considered a “semantic-network approach” to a set of discourses that interact in a complex way, and that in which the distributions of words, their co-occurrence and variance are emphasized (Leydesdorff & Hellsten,2006).
When analyzing several messages, and based on the knowledge proposed by Hellsten, Dawson, & Leydesdorff, (2010) it is possible to make an analytical distinction between implicit and explicit frames. Explicit frames are observable in the word choices made by the authors, labeling issues in a certain way Hellsten et al. (2010). On the other hand, implicit frames are “patterns of words that co-occur throughout discourse and are not directly observable” (Van der
12
Meer et al., 2014). Implicit frames disclose the way a certain issue is conventionalized (Hellsten et al., 2010), in other words, implicit frames reveal the different communicative intentions present in a message by grouping words according to their co-occurrences.
The use of the implicit framing approach aids the understanding of the underlying communicative intentions of both the organization, social and traditional media network of messages. The analysis of the implicit frames present in the messages provides a better understanding of the overall communicative intention. Additionally, revealing the underlying implicit frames present in each network of messages offers a better understanding of each network and facilitates the association with the Crisis Type Matrix and the Situational Crisis Communication Theory.
The Crisis Type Matrix and the Situational Crisis Communication Theory
According to Coombs (1995), crisis response strategies are developed with 3 distinct objectives. These objectives comprise the necessity of controlling both the extent of the crisis, as well as shape the outcomes and perceptions of it by both external and internal stakeholders (Coombs, 1995). The use of two theories developed by Coombs served as a theoretical guideline to better understand what has happened communication wise, during a crisis.
Coombs (1995) suggests that there is a possible set of guideline actions to be taken considering a certain “type” of crisis. Likewise, Coombs (1995) uses the Attribution Theory, which deals with how the social actor uses information that has been given to him in order to arrive at casual explanations of events (Fiske, & Taylor, 1991), as the basis of these guidelines. Understanding how people process diverse types of information is a most valuable knowledge
13
that allows the creation of certain actions in order to reach different roles. In this study, the guidelines proposed by Coombs (1995) were adopted so that it is possible to paint a general image of what guidelines of action are used by an organization in order to deal with this specific crisis.
On a second stage of analysis, and in order to obtain a more grounded knowledge and understanding of this crisis, the Situational Crisis Communication theory (SCCT) (Coombs, 2007) was also applied and combined with the first. It is interesting to note that the Situational Crisis Communication theory (Coombs, 2007), was developed with experimental setups in mind. In its article Coombs (2007, p.163) states that
"Researching using SCCT relies on experimental methods rather than case studies”. As noticeable in figure 2, according to the SCCT theory, there are a number of variables that play an important role on the final “behavioral intentions” of an organization in regard to a specific crisis. This again, points out the
necessity for a manager to clear understand the crisis situation in order to take the appropriate set of crisis
responses (Coombs 2007). Nonetheless, and considering that the SCCT was designed to identify and create a set of guidelines for managers to use adequate crisis strategies in order to protect organizational reputation, it is appropriate to use these guidelines for the analysis of this specific empirical crisis situation.
14
Overall, combining supports more accurate and theoretically founded inferences about the strategies used by the organization in crisis.
On a final note, it was also necessary to take in account the output given by different stakeholders using the theories aforementioned, and therefore, create an overall view of the path: communicative strategy/empirical outcome.
Methodology
To have a better understanding of the meaning construction of the crisis communication strategies and outcomes, different sets of data that related to the “MH370 flight” were collected and analyzed using semantic networks to undercover the different implicit frames present in each set.
The MH370 flight
Specifically, the Malaysian airlines MH370 flight disappearance case was the one selected in this study according to what was mention previously. First of all, it is necessary to acknowledge that the MH370 flight crisis is a very particular one, since, even nowadays, what really happened is still not clear due to the absence of real evidence to where the plane is. Due to this fact, this crisis reached great proportions, and it was covered by different media all over the globe. In order to better understand this case, and in a very succinct way, the MH370 flight was a passenger flight that left the Kuala Lumpur International Airport on a Saturday, 8th of March
15
2014, with destination the Beijing Capital International Airport, in China. Eventually, during its flight path, all communications with the airplane were lost, and the plane was last spotted by military radars to never be seen again. This flight carried a total of 239 people, including 12 Malaysian crew members and 227 passengers from over 15 different nations. Looking at the data presented above, it is with no wonder that an event of these proportions shocked the world, and had extensive media coverage from all different kinds of media, generating a wave of solidarity and a massive search for the disappeared plane, which wasn’t, still nowadays, successful and what really happened is still unknown to the general public.
It is in the media magnitude that this case achieved that lays one of the main pillars for this study. As it was stated before, the tourism area, especially aviation companies are vulnerable to crisis and these can be devastating. The use of this case study can also be justified by the importance of having a closer look at a specific event or series of events in order to better understand the whole picture, as Gerring (2004, p.1) defines the use of a case study as a: “intensive study of a single unit for the purpose of understanding a larger class of (similar) units.”.
Data Collection
This study comprised several analyses that contribute to a better understanding of the communicative decisions made by the Malaysian airline, as well as the response by the public and media. It can be considered that the 4 different sets of data were divided into two different parts. The first part concerns the communicative actions made by the Malaysia Airlines company, and the second relates to the traditional and social media output.
16
Initially, and to have a better understanding of how the organization affected dealt with the crisis, all press-releases, media statements, press conferences reports and updates given by the Malaysian airlines were collected. As the focus of this study is to analyze the communicative actions performed right after the crisis, the press-releases/updates (N=29) ranged from the date of the incident until one month after it (8th of April, 2014). These were collected directly from the official “Media Statement & Information on Flight MH370” page (http://www.malaysiaairlines.com/mh370). It is, and although being available to the general public, possible to assume that this type of data is usually directed towards traditional media (newspapers, television), and it was, therefore, collected to help understand the traditional media oriented communicative actions taken by Malaysian airlines.
Also, to have a better understanding of the communicative actions taken by the Malaysia airline in terms of social media, all the tweets posted by the official Malaysia airline account (N=58) in a period of one month after the accident were collected. These were tweeted by the official account of the organization “@MAS”, and were accessed using the accounts’ official page (https://twitter.com/mas).
These first two sets of data comprised the initial part of the study in which the organization’s communicative actions directed at traditional and social media were analyzed.
Secondly, and in order to understand the subsequent media and user reaction to the incident, data from both traditional and social media was collected.
In terms of traditional media, a list of possible newspapers was created. Due to language restrictions and the necessity to access relatively old article archives, the Daily Mail (United
17
Kingdom) was the newspaper chosen. The Daily Mail’s 2014 news archive was accessed (http://www.dailymail.co.uk/home/sitemaparchive/year_2014.html) and all the articles relative to the MH370 flight disappearance were collected. Due to the great amount of articles, and to the fact that with time passage, the relevance of the articles decreases, the data collected only ranged from the date of the incident until two weeks after it.
Finally, a tweet search was conducted. As a result of the difficulty in collecting data from twitter, the hashtag #mh370 was chosen, due to being the most used one when users wanted to refer to the accident. Using the twitter search engine (https://twitter.com/search-advanced) a total of (N=350)Tweets from random users was collected from the date of the incident until up to two weeks after it. Being written in English and the use of the hash tag “#MH370” were the only conditions used for the collection of this specific data.
Analysis
Consequently, the process of analysis was performed according to the indications present on the “Pajek Manual” (Vlieger & Leydesdorff, 2010).
As the ultimate goal of this data collection procedure was to perform an automated content analysis to reveal the underlying organizational, public and media message framing, the data needed to be arranged. All the data was organized in different “WordPad” files where information that wasn’t relevant to the analysis (such as links, other hash tags and images) was removed so that only the pertinent text was present.
18
To create a matrix that would allow for an overview of the most-used words, the data was run through the program “FrqList.exe. Also, to facilitate the visual representation and overview of the data, a limit of 75 words per set of data was employed. To refine the data even further a list of “stop words” that should not be considered, was included.
Subsequently, the frequency list of words was run through both “Ti.exe” (for small text as tweets) and “FullText.exe” (for longer text). These programs generate several files that can be read by a statistical program such as SPSS, in which the different sets of messages were scrutinized for underlying framing components through the use of factorial analytical techniques.
Before any factorial analysis was performed, and considering that it is necessary that each variable has a (variance= <0) for a content analysis to be performed, a descriptive statistical test was executed.
Finally, with the different databases arranged and ready for analysis, a factor analysis was performed on each set of data. To each set of messages, a factorial analysis was employed. A total of 4 factorial analyses were then performed. Each factorial analysis was done in order to generate the different underlying implicit frames present in each set of messages. To facilitate the analysis and the visualization of the results the maximum number of factors was set to 6. It is important to consider that the factor that loads first has a higher portion of total explained variance (R2) and therefore. these may be considered the dominant frames in the texts while the other 5 are may be considered as sub-frames.
19
Furthermore, and for a better interpretation of the values obtained by the factorial analysis, a Varimax rotation was applied. The results of this rotation were used as the reference for further assumptions and discussions in this study.
After the factor analysis was performed, a Chronbach’s alpha check was performed to test for the reliability of each component. In this reliability check, the “Scale if item deleted” is selected, and if there was proof that the removal of an item would greatly improve the Chronbach’s alpha value of a component, the item would then be removed. As a rule of thumb, components with under (α = 0.60) were later dismissed and excluded from further analysis and discussion.
Results
To successfully answer the research and sub-questions, a number of analyses were performed. The method employed in this study, as stated before, is in accordance with the indications provided on the manual “Pajek Manual” elaborated by Vlieger & Leydesdorff (2010). Therefore, the results reported in this study are only the ones related to the SPSS and Pajek’s output. This facilitates the reading and the conclusions to be made of the results.
First set- Press releases/updates by Malaysia Airline
The first set of messages analyzed contained a total of (N=29) Press-releases/updates from the Malaysia Airlines company. According to the procedures aforementioned a total of 75 words
20
were extracted and combined into an SPSS data set. A frequency list of the words extracted can be seen on Appendix 1.
Furthermore, these words were compiled into an SPSS data base so that the analyses could be performed. Each word represents, then, a variable that was used to perform the analyses.
After the variables were compiled and labeled into an SPSS database a descriptive statistical test to these words was executed. It was then concluded that the variables/words selected were eligible to use, considering that all of them showed (variance= <0).
Furthermore, a factorial analysis with indications for a Varimax Rotation was performed.
A total of 6 components were extracted. Each component was tested for reliability using a Cronbach’s Alpha analysis. Also, each component was named according to the words/variables present in it.
An overview of the different components, the variables present and its corresponding Cronbach’s Alpha values can be seen on Table 1.:
Table 1. Results of Rotated Varimax Factorial Analysis and Chrobach’s Alpha reliability test on Press-releases/Updates by Malaysian Airlines
Factor Items Chronbach’s Alpha
1- General Info. Asset, Corridor, Operation, Search, Southern, Satellite, Air, Australia, Aircraft, Deployed, Investigation, Malaysian, International, Time, Rescue, Effort, Bring, Update, Statement, Information, Indonesia, Focus
0.95
2- Commitment Support, Day, Prime, Continue, Minister, Board, Families, Airline, Authorities, Malaysia, Family, Government, MH370, Ongoing, Arrangement
0.88
3- Company’s Intentions
Infant, Emergency, Crew, Passenger, Assist, Carrying, Contact, Including, Queries, Total, Public, Media
21
4- Flight details Kuala, Travel, Emotional, Lumpur, Care, China, Means, 0.85 5- Providing
Assistance
Assistance, Provide, Caregiver, Team, Confirm, Flight 0.69*
6- Organization’s future intentions and wishes
Lost, Situation, Hour, Currently, Centre, Regular, Beijing, Prayer
0.73**
*Chronbach’s Alpha value if item “Flight” was deleted *Chronbach’s Alpha value if item “Centre” was deleted
It is possible to assess through the table presented above that the Varimax rotated Factorial Analysis extracted 6 different factors. These factors are capable of explain a total of around 65.3% of all the variable variances.
It is important to consider that the Chronbach’s Alpha reliability test values for the factors “5- Providing Assistance" and “6- Organization’s future intentions and wishes” could be improved if one item was deleted. In the case of factor number 5, the Chronbach’s Alpha value of the component would be (α=0.69) if the item “Flight” was deleted. For factor number 6, the Chronbach’s Alpha value of the component would be (α=0.73) if the item “Centre” was deleted. Therefore, to improve the reliability of both components, these items were disregarded from further analysis.
In this set of messages, the first component, and also the dominant frame, relates to “General Info”, in which words like “Asset”, “Operation”, “Search”, “Investigation”, “Malaysian” and “International”, indicate that the Press-releases/updates were mainly used by the organization to send out important information about the flight disappearance. The second component, and first sub-frame, relates to the organization’s “Commitment” with words such as “Support”, “Continue”, “Ongoing”, and “Arrangement”. The other sub-frames align with the second, and demonstrate an overall crisis acceptance by the organization, and an intention of involvement
22
with the publics. This can be observed in the 3rd sub-frame that relates to the “Company’s Intentions”, that comprises words such as “Assist”, “Carrying”, “Contact”, and also on the 5th sub-frame, that relates to “Providing Assistance”, in which words like “Assistance, “Provide”, “Caregiver”, “Team” are used.
Using the program Pajek, it was possible to generate Figure 2, which provides another visual representation of the different factors loaded and the inter connectivity of the diverse words:
Figure 2. Pajek’s graphic representation of the 6 different factors, outliers and the words contained in each factor
Figure 2. also visually represents the group of words that didn’t load into any component due to the restrictions made (Max. 6 factors loaded), these words are presented in a Pink coloring and are not to be used in the rest of the analysis.
Second set- Official tweets by Malaysia Airlines
The second set of messages analyzed contained a total of (N=61) different tweets by Malaysia Airlines. According to the procedures aforementioned a total of 75 words were extracted and
23
combined into a SPSS dataset. A frequency list of the words extracted can be seen on Appendix 2.
After the variables were compiled and labeled into an SPSS database a descriptive statistical test to these words was executed. It was then concluded that the variables/words selected were eligible to use, considering that all of them showed (variance= <0).
Furthermore, a factorial analysis with indications for a Varimax Rotation was performed.
A total of 6 components were extracted. Each component was tested for reliability using a Cronbach’s Alpha analysis. Also, each component was named according to the words/variables present in it.
An overview of the different components, the variables present and its corresponding Cronbach’s Alpha values can be seen on Table 2.:
Table 2. Results of Rotated Varimax Factorial Analysis and Chrobach’s Alpha reliability test on official tweets by Malaysian Airlines
Factor Items Chronbach’s Alpha
1- Statement and Press-release info.
Statement, LT, Released, Yahya, Mar, CEO, Jauhari, Ahmad, Incident, Media
0.91
2- Commitment messages
Search, Rescue, Involved, Continue, Hour, Operation, Development
0.76
3- Company’s worries Relevant, Authorities, Crew, Locate, Concern, PAX, Aircraft 0.79
4- General Info. Southern, Multinational, Corridor, Lead, Effort 0.76
5- Providing Assistance
Response, Center, Established, Emergency 0.77
6- Empathy messages Thanks, Airline, Malaysia, Update, Support, Situation 0.55
It is possible to assess through the table presented above that the Varimax rotated Factorial Analysis extracted, as expected, 6 different factors. These factors are capable of explain a total of
24
around 34% of all the variable variances. Nonetheless, it is important to note that, when the Cronbach’s alpha reliability test was performed, the last component “Empathy messages” loaded (α = 0.55), which is a value under the threshold established of (α = 0.60) and was therefore disregarded due to not meeting the reliability values necessary for this analysis.
In this particular set of messages, the first component, and also the dominant frame, relates to “Statement and Press-release info.”, in which words like “Statement”, “Release”, “CEO”, “Incident” and “Media”, indicate that the Tweets sent out by the organization were mainly used to forward the users to the official Press-releases/updates. This suggests that the organization mainly used this platform as a “bridge” to the most important messages, the Press-releases/updates. This, again, suggests that the main tool used to communicate by the Malaysia Airlines during this crisis was indeed the Press-releases/updates. This may relate to the fact that Twitter only allows short messages in which elaborated messages are hard to construct. The second component, and first sub-frame, relates to “Commitment messages” with words such as “Search”, “Rescue”, “Involved” and “Continue”. The other sub-frames align with the second, and again demonstrate an overall crisis acceptance by the organization, and an intention of involvement with the publics. This can be observed in the 3rd sub-frame that relates to the “Company’s worries”, that comprises words such as Crew, “Locate”, “Concern”, and also on the 5th sub-frame, that relates to “Providing Assistance”, in which words like “Response, “Center” and “Established” are used.
Using the program Pajek, it was possible to generate Figure 3. It is important to reinstate that the figures provided by Pajek only represent a visual depiction of the components, and that is the reason why the last component “Empathy messages” is present.
25
Figure 3. Pajek’s graphic representation of the 6 different factors and the words contained in each factor
Observing the results of both sets of messages provides enough information to state that there are differences between the organizational message framing of the Press-Releases/Updates and the official Tweets.
Third set- Newspaper news related to the incident
The third set of messages analyzed contained a total of (N=46) different articles related to the MH370 disappearance from the newspaper “Daily Mail”. According to the procedures aforementioned a total of 75 words were extracted and combined into a SPSS dataset. A frequency list of the words extracted can be seen on Appendix 3.
After the variables were compiled and labeled into an SPSS database a descriptive statistical test to these words was executed. It was then concluded that the variables/words selected were eligible to use, considering that all of them showed (variance= <0). Furthermore, a factorial analysis with indications for a Varimax Rotation was performed.
26
A total of 6 components were extracted. Each component was tested for reliability using a Cronbach’s Alpha analysis. Also, each component was named according to the words/variables present in it.
An overview of the different components, the variables present and its corresponding Cronbach’s Alpha values can be seen on Table 3:
Table 3. Results of Rotated Varimax Factorial Analysis and Chrobach’s Alpha reliability test on official tweets by Malaysian Airlines
Factor Items Chronbach’s Alpha
1- General Info. Found, Crew, Government, Search, Traffic, Countries, Communications, Information, MH370, Ocean, Object, SHAH, Beijing, Kuala, March, Aviation, Investigation, Data, Vietnam, Hamid, Family, People,
0.95
2- Flight details Malaysian, Mile, South, Indian, GMT, Hotel, Malaysia, Flight, Official, Plane, Passenger, Captain, Missing, Contact, Zaharie
0.88
3- Flight Disappearance investigation
Ahmad, Police, Air, Airline, Fariq, Investigator, Lumpur, Time, West, Satelitte, Home, Relative
0.86
4- Information given out by Malaysia Airlines and Government
Jet, Minister, Expert, Boeing, Aboard, Force, Conference 0.85
5- Last
communications from MH370 flight
Sea, Southern, Board, Radar, Copilot, Aircraft 0.69*
6- Geographic plane’s search information
Chinese, System, China, Australian, Flying, Operation, Pilot, Hour
0.46**
*Chronbach’s Alpha value if item “Aircraft” was deleted **Chronbach’s Alpha value if item “Flying” was deleted
It is possible to assess through the table presented above that the Varimax rotated Factorial Analysis extracted 6 different factors. These factors are capable of explain a total of around 65.3% of all the variable variances. It is important to consider that the Chronbach’s Alpha
27
reliability test values for the factors “5- Last communications from MH370 flight " and “6- Geographic plane’s search information” could be improved if one item was deleted. In the case of factor number 5, the Chronbach’s Alpha value of the component would be (α=0.69) if the item “Aircraft” was deleted. For factor number 6, the Chronbach’s Alpha value of the component would be (α=0.71) if the item “Flying” was deleted. Therefore, to improve the reliability of both components, these items were disregarded from further analysis.
In this set of messages, the first component, and also the dominant frame, relates to “General Info”, in which words like “Found”, “Crew”, “Government”, “Communications”, “Information” and “Search”, indicating that the newspaper articles messages were mainly framed in terms of general information.
The second component, and first sub-frame, relates to the organization’s “Flight details” with words such as “Mile”, “Plane”, “Passenger”, and “Captain”.
The 4th component relates to “Information given out by Malaysia Airlines and Government” with words such as “Minister”, “Expert” and “Conference”, and provides evidence that the framing of the messages sent out by the organization is present in the framing of Traditional media messages.
Other sub-frames align with the second, and relate to information regarding the investigation as can be seen in the third sub-frame “Flight disappearance investigation” with words such as “Police”, “Investigator”, “Satellite”. This can be also observed in the 5th
and 6th sub-frames, “Last communications from MH370 flight” and “Geographic plane’s information”.
28
Using the program Pajek, it was possible to generate Figure 4, which provides another visual representation of the different factors loaded and the inter connectivity of the different words:
Figure 4. Pajek’s graphic representation of the 6 different factors and the words contained in each factor
Figure 4. Also visually represents the group of words that didn’t load into any component due to the restrictions made (Max. 6 factors loaded), these words are presented in a Pink coloring and are not to be used in the rest of the analysis.
Fourth set- Tweets by users using the Hashtag #MH370
The forth, and final, set of messages analyzed contained a total of (N=350) tweets by different users that contained the hashtag “#MH370”. According to the procedures aforementioned a total
29
of 75 words were extracted and combined into a SPSS dataset. A frequency list of the words extracted can be seen on Appendix 4.
Furthermore, these words were compiled into an SPSS database so that the analyses could be performed. Each word represents, then, a variable that was used to perform the analyses.
After the variables were compiled and labeled into an SPSS database a descriptive statistical test to these words was executed. It was then concluded that the variables/words selected were eligible to use, considering that all of them showed (variance= <0).
Furthermore, a factorial analysis with indications for a Varimax Rotation was performed.
A total of 6 components were extracted. Each component was tested for reliability using a Cronbach’s Alpha analysis. Also, each component was named according to the words/variables present in it.
An overview of the different components, the variables present and its corresponding Cronbach’s Alpha values can be seen on Table 4.:
Table 4. Results of Rotated Varimax Factorial Analysis and Chrobach’s Alpha reliability test on tweets by users using the hashtag “#MH370”
Factor Items Chronbach’s Alpha
1- Solidarity messages Launches, Evidence, Effort, Crowdsourcing, Look, Satelite 0.82
2- Flight
disappearance theories
Iranian, Holder, FakePassport, Mystery 0.79
3- Passengers’ cellphones info
Phone, Passenger, Ringing, Cell, Mobile 0.71
4- Possible plane’s location
30
5- General info. Malaysia, Airline, Missing, Jet 0.64
6- Stolen Passport Passengers info.
Stolen, Passport, Ali 0.75
It is possible to assess through the table presented above that the Varimax rotated Factorial Analysis extracted, as expected, 6 different factors. These factors are capable of explain a total of around 22.89% of all the variable variances.
In this set of messages, the first component, and also the dominant frame, relates to “Solidarity Messages”, in which words like “Effort”, “Crowdsourcing”, “Look” and “Sattelite are present, indicating that the dominant frame present in this set of messages relates to the personal wishes and solidarity of each user. The second component, and first sub-frame, relates to the “Flight disappearance theories” with words such as “Mystery”, “FakePassport”, “Iranian” and “Holder”. This specific sub-frame relates to the others, as there was a general concept of creating conspiracy theories about the incident as it is usual when there were no plausible explanations for what had happened. The 3rd “Passenger’s cellphones info” and 4th “Possible plane’s location” sub-frames are a good example of this wide-spread conspiracy theory framing, with words such as “Phone”, “Passenger”, “Ringing” for the 3rd
sub-frame and words such as “Jungle”, “Remote” and “Location” for the 4th
sub-frame.
The frames loaded provide enough evidence to conclude that there is no apparent presence of the information given out by the Malaysian Airlines in both the dominant and sub-frames extracted from the tweets sent out by users
31
Using the program Pajek, it was possible to generate Figure 5, which provides another visual representation of the different factors loaded and the inter connectivity of the different words:
Figure 5. Pajek’s graphic representation of the 6 different factors and the words contained in each factor
Discussion
The present study was set to analyze different sets of messages using an automated content analysis in order to extract the underlying factors of each of those set of messages. By unveiling these underlying factors present in each set of messages, it was possible to make several conclusions.
In this particular study, 4 different sets of messages were collected and used in the analysis. The first 2 sets (Official tweets and Press-releases/Updates) were used to understand the underlying structure of the implicit message framing present in the strategies used by the
32
organization responsible for the crisis. Accordingly, the other 2 (Tweets by users and newspaper articles) were collected and analyzed to unveil the implicit message framing present in those messages. The use of these particular sets of messages combined with an automated content analysis, supported an overview of both the organization’s communication strategies directed at Traditional media (Press-releases/updates) and Social media (tweets), with the Traditional media output (Newspaper articles) and Social media output (Tweets from random users).
When analyzing the results from the first set of messages, the Press-releases/Updates, it is possible to state that, the overall 6 factors extracted align with what is suggested in both the Crisis Type Matrix (Coombs, 1995) and the Situational Crisis Communication (Coombs, 2007) theories. As expected, the dominant frame extracted: “General Info”, relates to the information that the organization needs to send when faced with crisis-like situations. This frame is usually expected to be one of the most important as there is an obvious necessity for information. Nonetheless, the dominant frame extracted from this set of messages already suggests that the organization acknowledged the crisis and planned to create empathy with the public. Moreover, both Crisis Type Matrix (Coombs, 1995) and the Situational Crisis Communication (Coombs, 2007) theories suggest that when faced with an “Accident” type of crisis the organization should engage with the stakeholders, acknowledging the crisis and attempt to win forgiveness of the publics (Coombs, 1995). This is further supported when analyzing the other sub-frames extracted. These “mortification” actions (Coombs, 1995) can be observed more clearly in sub-frames such as the 2nd one: “Commitment”, which reinstates the organization’s intention to engage with the victims and with the public. Furthermore, this assumption can also be supported by the 5th sub- frame: ”Provide assistance”. The underlying frames extracted from this particular
33
set of messages reestablished that the organization’s assumed the responsibility for the crisis and tried to create a sense of involvement and, again, commitment with the victims.
Additionally, when looking at the results of the second set of messages, the official tweets sent by the organization, it is noticeable that, largely, the 6 factors extracted also align with what is suggested in both the Crisis Type Matrix (Coombs, 1995) and the Situational Crisis Communication (Coombs, 2007) theories. This can be better demonstrated by the sub-frames such as “Commitment messages”, “Company’s worries” and “Providing Assistance”. Nonetheless, and perhaps one of the most relevant conclusions to be drawn in relation with the present study, relies on the first component extracted. The first component, the dominant frame in the Tweets sent out by the organization relates to “Statement and Press-release info. It is important to understand that the dominant frame is indeed, the most significant frame existent in a set of messages, and also the one that explains the higher amount of variance. With that being said and considering that the dominant frame of this set of messages relates to the Press-releases/updates, it is possible to consider that main goal of these Tweets was to redirect the users to, what can be assumed as the most important messages sent from the organization, the Press-releases/updates. The reason for this can be the fact that Twitter limits the messages to a maximum of 140 characters, and due to this fact, the organization might have chosen to use the Press-releases/Updates as the main channel through which the organization is going to communicate the most important information.
Overall, it was possible to find underlying framing messages in the organizational messages developed by the organization for social and traditional media. The framing differences found in these messages subsequently provide an answer to Sub-question 1.
34
Moreover, this difference between message framing leads to the conclusion that the “traditional tactics” (Taylor & Perry, 2005) played a major role in the crisis communication strategies used by Malaysia Airlines, and while the social media messages complemented the overall strategy of crisis acceptance and empathy creation, those messages were also sent out to create involvement and raise awareness to the most important messages, the Press-releases/Updates.
With regards to the third set of messages, the newspaper’s articles, the dominant frame is “General Info”. This presents no surprise, as newspapers articles are often written to provide general information about a certain subject. It is, nonetheless, interesting to observe that, one of the sub-factors loaded in this set relates directly to the Malaysia Airline’s messages. The 4th component “Information given out by Malaysia Airlines and Government” consequently reinstates the importance of the messages sent out by the organization and the government. The report and use of these messages by traditional media, is complemented by the necessity for clear and concise information about the plane’s disappearance. This is especially important in this particular case due to the elevated amount of conspiracy theories and false information that arose. Finally, all the other factors loaded recount information related to the flight disappearance, aligning with the dominant frame.
In terms of the last set of messages, the user Tweets, it is necessary to acknowledge that the six factors loaded, relate mostly to solidarity messages and speculation theories. The dominant frame “Solidarity Messages” was expected as users tend to easily express sympathy and empathy feelings online when an incident like this occurs.
35
The essential need for an explanation of what happened, lead to great conjecture and suppositions. This assumption is supported by sub-frames such as “Flight disappearance theories” and “Possible plane’s location”.
It is interesting to discern that there are no significant components that relate to the framing present in messages sent out by Malaysia Airlines. Users tend to tweet more messages of support and possible theories about the flight, rather than comment on the actions undertaken by the organization. There seems to be no apparent mentioning of the messages sent out by the organization in the underlying frames present in this set of messages.
These facts provide enough evidence to answer the Sub-question 2, although there is a presence in terms of the organizational messages framing in the newspaper articles, this presence was not found in the user Tweets. Consequently, and due to the easy access to all kinds of information through the use of the Internet, conspiracy theories and speculation break out and the general public is usually fascinated by those theories and the popularity of such theories often presents a gradual growing rate and also become more elaborated over time (McHoskey, 1995). All these “distractions” may be the reason why it was impossible to identify a clear component that relates to the messages sent out by the Malaysian Airline.
Implications
It is imperative to understand that crisis situations present an organization with different challenges that are specific to each crisis. If one considers this argument, it might be difficult to understand why there is a need to study crisis that already occurred, extensively. Although there
36
is a great deal of particularity to each crisis, both organizations and scholars can learn from researches of this type in order to disclose what “worked vs. what didn’t” in a crisis strategy. More specifically, the analyses performed in this study allow the reader to have a better understanding of which underlying frames are present in the organization’s communication plans, and what could be the outcomes and implications of these frames in the output given by social and traditional media. One can assume that the strategy used by the Malaysia Airline is one that cherishes a sense of engagement and commitment with the victims and the community in general. By combining different approaches for traditional and social media, the organization was capable of, at least, direct the users to specific messages that were framed according to the organization’s intentions. Therefore, the results presented in this study, comply with the theories presented by Coombs (1995 and 2007), in which when faced with a crisis of this type (accidental), the organization should engage with the publics and be as transparent as possible in order to minimize the damage. These guidelines and specific message framing strategies may serve as a future reference to organizations when faced with similar crisis situation. On the other hand, it is necessary to consider that this study revealed that the messages sent out by the organizations in the period of time immediately after the crisis did not significantly impact the message framing used by tweet users. This can serve as a point of reference for scholars and even organizations, and raises awareness that the messages derived from Social media and the discussion matters present in this kind of media do not necessarily result in, or reflect the organization’s initial messages and intention. Social media and the messages produced on Social media are the result of several important factors that come into together, which may become an incredibly hard phenomenon to manage. With that being said, one of the most important results of this study suggests that both organizations and scholars should better understand, study and
37
test ways to be able to influence social media more directly and according to their previously established objectives.
Limitations
There are several limitations of this kind. One of the main limitations of these studies relates to data collection. Due to budget and time restraints, the overall data collected only represents a fraction of all the information flow generated after the crisis. The newspaper articles collected belonged only to one particular newspaper which can affect the generalization of “traditional media”. Likewise, language constrains come into play as the newspaper chosen had to be written in English. Furthermore, it is of great difficulty to collect data from social media platforms such as twitter, and the collection of this data had to be done manually. Considering the incredibly high amount of tweets sent out by users during the two weeks after the incident, the number of tweets used here only represents a fraction of the total tweets. Until there a better social media data collecting tool with widespread availability, studies like this may continue to suffer from a lack of total data and base their assumptions on analyzes performed in only a small fraction of the overall information available.
Finally, and one of the main limitations of this study relates to the time window from which the data was collected. This may be due to the difficulty to collect data both from newspapers and from social media. Data collected from a longer period of time would improve the reliability of assumptions and results presented in this study.
38
Conclusion
As Francis Marra (1998, p.461) states: “Crises, in almost all circumstances, immediately trigger a deluge of questions from an organization's many different publics.”.
Following the knowledge, and similarly to the study elaborated by van der Meer et al. (2014), by applying a semantic-network approach it was possible to build a more complex and structured understanding of crisis communication and the framing processes that are present in different sets of messages. This approach also endorses a more grounded understanding and visualization of the framing processes in terms of the interplay of several dominant domains (van der Meer et al. 2014).
Additionally, by unveiling the different underlying frames present in the sets of messages it was possible to answer Research Question 1, as the underlying frames extracted and mentioned in the results section of this study, represent the different meanings given to each set of messages. Correspondingly, by observing and concluding that from each set of messages, different underlying frames were extracted, it was possible to answer Research Question 2. The semantic networks present in each set of messages provided an overview of the existing meanings and their differences.
Furthermore, when taking a closer look at the results of the factorial analysis on both sets of data relative to the organization’s crisis communication strategy, it was possible to observe that the organization chose a posture of recognizing the crisis and tried to create involvement with the publics, offering support and regular updates on the search operation. This aligns with the theories and recommendations made by Coombs (1995, 2007), in which, and when faced with an
39
“accident type” of crisis” the organizations should adopt a humane posture and should engage with the publics in order to minimize the damages and create a sense of acceptance and empathy with the publics and media. It is interesting to observe that the loaded components from the sets of data relative to traditional and social media were mainly related to theories about the flight disappearance, prayer messages and general info about the incident. Nonetheless, when taking a closer look at the results presented in this study, it was possible to assess that the messages sent by the organization were largely disregarded by the social media users. No relevant evidence of the impact of the organization’s messages was found in the tweets sent out by users after the incident. While, and on the other hand, the messages sent out by the organization had a strong presence in the framing done by traditional media as proved by one of the factors loaded relating to these messages. This may lead to the conclusion that social media output fairly disregarded the messages sent out by the organization, being “conspiracy theories” and “support messages” the main frames used in this kind of messages. Similarly, it is possible to state that the communicative actions performed by the organization were successful in getting the traditional media.
However, in future research the data analyzed should comprise a larger period of time. This would allow for an improved understanding of the real consequences of the communicative actions performed by the organization, and it would also provide an improved overview of the underlying message framing present in both traditional and social media. Furthermore, other social media platforms such as Facebook should be considered. By doing so, it would be possible to create a summary of the underlying message frames present in social media more extensively. Finally, the language restraints were indeed a problem, and considering that the disaster occurred
40
and involved several Asian countries it could be interesting to also comprise the output given by social and traditional media that originates from these countries.
Overall, crisis management is a field that should be studied extensively. This study presented attention-grabbing results of a specific crisis. Although, and at first glance, it might be increasingly hard to relate two different crisis situations, the all-encompassing study of each crisis situation is a necessity so that the knowledge could be applied to better understand a future crisis and to create better crisis communication plans’ that might assist an organization in minimizing the damage and protect its’ image and reputation.
41
References
Barton, L.: 1993, Crisis in Organizations: Managing and Communicating in the Heat of Chaos (South-Western Publishing Company, Cincinnati, OH).
Benoit, W. L. (1997). Image repair discourse and crisis communication. Public relations
review, 23(2), 177-186.
Borah, P. (2011). Conceptual issues in framing theory: A systematic examination of a decade's literature. Journal of Communication, 61(2), 246-263.
Brown, J. R., & Goolsbee, A. (2000). Does the Internet make markets more competitive? (No. w7996). National Bureau of Economic Research.
Coombs, W. T. (1995). Choosing the Right Words The Development of Guidelines for the Selection of the “Appropriate” Crisis-Response Strategies.Management Communication
Quarterly, 8(4), 447-476.
Coombs, W. T., & Holladay, S. J. (1996). Communication and attributions in a crisis: An experimental study of crisis communication. Journal of Public Relations Research, 8, 279-295
Coombs, W. T. (2007). Protecting organization reputations during a crisis: The development and application of situational crisis communication theory. Corporate Reputation Review, 10(3), 163–177
Dimmick, J., Chen, Y., & Li, Z. (2004). Competition between the Internet and traditional news media: The gratification-opportunities niche dimension. The Journal of Media
Economics, 17(1), 19-33.
Entman, R. M. (1993). Framing: Toward clarification of a fractured paradigm. Journal of
Communication, 43(4), 51–58.
Falkheimer, J., & Heide, M. (2006). Multicultural crisis communication: Towards a social constructionist perspective. Journal of contingencies and crisis management, 14(4), 180-189.
Fiske, S. T., & Taylor, S. E. (1991). Social cognition, 1991. Social Cognition (2nd Ed.). Xviii,
717 Pp. New York, NY, England: Mcgraw-Hill Book Company.
Gerring, J. (2004). What is a case study and what is it good for?. American political science
review, 98(02), 341-354.
Hellsten, I., Dawson, J., & Leydesdorff, L. (2010). Implicit media frames: Automated analysis of public debate on artificial sweeteners. Public Understanding of Science, 19(5), 590-608.
