Student: Gigi Levens
Student number: 5897807
Master’s Thesis
Master’s programme Communication Sciences
Supervisor: dr. P.H.J. Schafraad
Date of completion: 29th of January 2016
Reporting on your own mess
while cleaning it up
Frame usage of news media in crisis vs. not in crisis
Mary Mapes - CBS News Dan Rather - CBS NewsUSE OF FRAMES: NEWS MEDIA IN CRISIS
Abstract
Abstract
Crisis management is a relevant topic for organisations. The ability to downplay an event that
might be a negative influence is part of maintaining the reputation of the organisation. One of
the important aspects of crisis management is dealing with the media. Because they have
the power to downplay or amplify a crisis, it is only understandable that a part of the overall
crisis management of an organisation is to deal with a crisis going public through the media.
What gives the media so much power over determined incidents is that they have the can to
narrate the news in a specific way, and with the use of particular language, develop a
concept of the issue in the public’s mind. That phenomenon is called framing, and it is the
organisation’s job to try to make sure the frame the media will use works in their benefit. But
what happens when the organisation going through a crisis is a media organisation itself?
This thesis analyses three case studies of media organisation that face a reputation crisis
caused by the actions of their employee(s). Through content analysis of news transcripts
from the affected and other relevant networks, the study tries to state a difference in the use
of frames between the media outlet in crisis and the ones not in crisis, also looking at the
behaviour of the “familial” media outlets (i.e. independent media outlets that are ultimately
owned by the same media conglomerate). Results show that there is indeed a difference
between the different groups of media when reporting on a crisis of a media organisation.
Keywords: Crisis communication, framing, victim frame, blame frame, intent frame,
remedy frame, judgement, frame alignment
USE OF FRAMES: NEWS MEDIA IN CRISIS
Table of Contents
3
Table of Contents
1. Introduction
5
2. Theoretical Framework
7
2.1. Crisis communication
7
2.1.1. Crisis communication theories
7
2.2. Frames
8
2.2.1. Why and how frames work
8
2.2.2. Frames in the media
10
2.2.3. Frame alignment theory
12
2.2.3.1.
After frame alignment
12
2.2.4. Frame analysis
13
3. Methods
16
3.1. Research design
16
3.2. Sampling procedure
16
3.3. Coding procedure
17
3.4. Measures
18
3.4.1. Independent variables
18
3.4.1.1.
Outlet variables
18
3.4.1.2.
Time variables
18
3.4.2. Dependent variables: frames
18
3.4.2.1.
Judgement
19
3.4.3. Inter-coder reliability
20
4. Results
22
4.1. Downplaying and amplifying crisis with frames
22
4.1.1. “Master” frames
22
4.1.2. Sub-frames
23
4.2. The conglomerate
23
4.2.1. “Sister” company vs. media outlet in crisis
24
4.2.2. “Sister” non-crisis vs. other media outlets not in crisis
24
4.2.2.1.
“Master” frames
24
4.2.2.2.
Sub-frames
24
4.2.3. Conglomerate vs. non-conglomerate non-crisis
25
4.2.3.1.
“Master” frames
25
4.2.3.2.
Sub-frames
25
4.2.4. Implications
26
USE OF FRAMES: NEWS MEDIA IN CRISIS
Table of Contents
4
4.3.1. CBS with Mary Mapes and Dan Rather
27
4.3.2. FOX News with Bill O’Reilly
27
4.3.3. NBC News with Brain Williams
27
4.3.4. Implications
28
5. Discussion and Limitations
31
5.1. Findings
31
5.2. Conclusion
32
5.3. Limitations
32
5.4. Future Research
33
References
34
Appendix A: Cases
A1
Appendix B: Codebook
B1
Appendix C: Syntax used
C1
Appendix D: Inter-coder reliability
D1
USE OF FRAMES: NEWS MEDIA IN CRISIS
Introduction
5
1. Introduction
Organisational crises are inherently public phenomena. Given their nature as media events,
“most stakeholders will learn about a crisis from news reports.” (Coombs, 2007, p. 164) In
fact, significant events with the potential to damage an organisation’s reputation only turn into
crises once they enter the public sphere. Crisis communication scholars have derived a
range of principles and propositions that enhance our understanding about the genesis,
dynamics, and implications of crises as media events (i.e. McQuail, 2012). One of the main
aspects to consider is the ability of the media to prioritise coverage of specific crisis events,
which as a result will amplify public perception and relevance of the issue. For this reason,
organisations are usually advised to form partnerships with news media before crises erupt;
not doing so can result in drawing negative attention to the crises, and gives the media a
certain lead over the situation (Pidgeon, Kasperson, & Slovic, 2003).
Framing is one of the media communication theories applied to explain the role of
news media in distributing information, and it concerns “the process by which people develop
a particular conceptualisation of an issue or reorient their thinking about an issue” (Chong &
Druckman, 2007, p. 104). The way a situation is explained or framed could have a direct
effect on how the audience will perceive it. In relation to crisis communication, framing theory
concentrates mainly on the way the news media following the crisis portray the organisations
involved. As an organisation, the goal is to attempt to frame the media’s response to the
crisis in a favourable way, and they can either accept or even reframe the intended message
(Sellnow & Seeger, 2013). Research papers and literature make reference to organisations
working alongside with the media in an attempt to control the angle and amount of
information available to the audience. In their book Theorizing Crisis Communication,
Sellnow & Seeger (2013) analyse the framing in the April 2001 collision of a Chinese
interceptor jet and a US intelligence-gathering plane off the coast of China. The results
showed that the coverage in both countries contrasted severely, given that the news media
in each country framed the news in a way that was favourable to its government. The
governments of each country played the role of the organisations that succeeded in working
alongside with the media in order to present their angle to the crisis.
But what happens when the organisation affected by a crisis is itself part of the news
media landscape; what if the mediators are also the main characters in a crisis event? Do
most crisis communication propositions still apply, or do they need to be modified? Especially
since the media do have to adhere to truthfulness, ethical norms, accountability and
standards of objectivity (McQuail. 2012, chapter 8). Can we derive new and insightful
findings from these types of crisis? Motivated by such questions, this thesis will investigate
precisely a few of such cases. Specifically, CBS News and their former anchor Dan Rather
USE OF FRAMES: NEWS MEDIA IN CRISIS
Introduction
6
and former producer Mary Mapes; Fox News Channel with their anchor Bill O’Reily; and NBC
News with their former anchor Brian Williams. All are cases in which employees either made
a mistake or were caught being untruthful or “misremembering” events.
The research question for this thesis is as follows: “Does a media organisation that is
suffering a reputation crisis use frames differently when reporting on the crisis than a media
organisation that is not in crisis, and does the theory of frame alignment apply?” Given the
fact that the media play such an important role in crisis communication, the main goal is to
determine if there are any specific effects to an organisational crisis when the organisation
itself is a media institution.
The first part of the research will include a literature review analysing the main
concepts and theories related to crisis communication, framing, development of the use of
frames over time, and how the frames will be analysed. Also, it will look for previous
researches that refer to media institutions as the organisation facing the crisis. During the
second phase, a content analysis will be conducted of transcripts of news programmes
reporting about the three crisis cases. The data obtained will be analysed and it’s relevance
for the research question and its hypothesis will be assessed in the Results-chapter. Last but
not least, there will be a discussion of the obtained results and of which a conclusion will
emerge. The limitations of the research will be recognised and suggestions for future
research will be given in the last paragraph of this theses,
USE OF FRAMES: NEWS MEDIA IN CRISIS
Theoretical Framework
7
2. Theoretical Framework
The main focus of this literature review is to examine to what extent the literature is
applicable to the case studies and to determine whether there are any knowledge gaps that
need to be filled.
2.1. Crisis communication
A crisis is an event that suddenly occurs. It demands a quick reaction and it interferes with
the organisational performance (Millar, 2004). This is because the crisis brings, or has the
potential of bringing, reputational damage to an organisation, which can influence its future
profitability, growth, and even its survival in a negative way (Lerbinger, 1997). During crises,
people want to find order in chaos: they try to make sense of what is happening (Gray,
2003). Crisis communication plays a central role in effective crisis management, since the
more the media report about an organisation’s crisis, the more the public will perceive it as
important (Pidgeon, Kasperson & Slovic, 2003).
Crisis communication can be defined in different ways. The current consensus states
that crisis communication should be considered as an integrated part, and as a critical
element, of the overall crisis management process (Coombs, 1999). Effective crisis
communication is also essential for maintaining a positive relationship with stakeholders
during the crisis (Fishman, 1999). Crisis communication affects the perception of the public
of the organisation during and after the crisis (Williams & Treadway, 1992), thus influencing
the organisation’s reputation.
2.1.1. Crisis communication theories.
There are many theoretical models for an effective crisis communication. The Theory
of Image Restoration (Benoit, 1995) as well as the Situational Crisis Communication Theory
(SCCT) developed by Coombs and Holladay (2002) are chosen for this thesis, since these
theories fit the purpose of the current research best. The reason of this is because they are
both useful for crisis communication (De Wolf & Meijri, 2013) and for distinguishing different
frames. More on this in the paragraph “Frame analysis”.
The Theory of Image Restoration was developed to help managers to preserve the
organisation’s positive image and reputation through communication (Benoit, 1995). This
theory is actually not intended for use within crisis communication, but it could be useful for
this type of communication since crises often generate negative perceptions about the
organisation, and these affect its image and reputation.
USE OF FRAMES: NEWS MEDIA IN CRISIS
Theoretical Framework
8
The SCCT is among the most used theories in the field of crisis communication (De
Wolf & Mejri, 2013). It notes important variations in crises. Because of these variations the
authors suggested a broader typology of crisis situations and ten crisis-response strategies
to deal with these types crises. The SCCT also suggests that each crisis generates specific
attributions of responsibility for the crisis (Coombs, 2007), and it is further concerned with the
historic behaviour of the organisation indicated by, but not limited to, the organisation’s crisis
history. The SCCT states that the negative impact of a crisis on an organisation’s reputation
is intensified if it has a negative performance history (Coombs, 2007).
2.2 Framing
The origins of framing lie in the fields of cognitive psychology (Bartlett, 1932) and
anthropology (Bateson, 1972). It was adopted by other disciplines, usually with a slightly
different meaning and was applicable in a variety of ways. Research approaches that
analyse message content to assess how the media represent a topic are often referred to as
frame analyses, even if they distinguish no frame at all. In recent framing research the
attention has shifted to either very specific issue-frames that are only applicable to certain
topics, or to broadly defined generic frames, that can be present everywhere (De Vreese,
Peter, & Semetko, 2001).
2.2.1. Why and how frames work.
Literature on framing usually gives the impression that frames can be found in several
places within the communication process (Entman, 1993). Goffman (1981) considers frames
independent from the individual, with their own logic and meaning, and emphasizes their
connection with culture. Culture refers to an organised set of beliefs, codes, stereotypes,
values, norms, frames, etc. that are shared in the collective memory of society (Zald, 1996),
which an individual cannot change. Because of this frames are conceptually situated largely
externally of the individual (Goffman, 1981). Individuals can make use of these cultural
phenomena; media workers apply and magnify them in media content and present them to
their audiences (Shoemaker & Reese, 1996, p. 60).
Although frames in culture are hard to define, they can be reconstructed. They are
embedded in media content when journalists construct the news message in such a way that
elements refer to a frame. Each frame that a journalist applies in a text can be represented
as a frame package, which is a cluster of logical organised devices that function as an
identity kit for a frame. Because of this an important part of frame analysis is the
reconstruction of frame packages. A frame package is made up of three elements: the
USE OF FRAMES: NEWS MEDIA IN CRISIS
Theoretical Framework
9
manifested framing devices; the manifested or latent reasoning devices; and an implicit
cultural phenomenon that displays the package as a whole (Gamson & Lasch, 1983;
Gamson & Modigliani, 1989).
Media use various framing devices (such as word choice, metaphors,exemplars,
descriptions, arguments and visual images) to manifest a frame (Gamson & Lasch, 1983;
Pan & Kosicki, 1993). All possible framing devices that indicate the same core idea,
constitute the manifested part of a frame package. These devices are held together by a
central organising theme. This theme is the actual frame, which provides the frame package
with a structure (Donati, 1992). In this research the focus is on frame packages in which the
cultural phenomenon archetype (e.g. victim; Berns, 2004) functions as a central theme.
Therefore, essential to a frame package are reasoning devices. These are explicit and
implicit statements that deal with justifications, causes and consequences in a temporal
order, and complete the frame package (Gamson & Lasch, 1983; Gamson & Modigliani,
1989). The reasoning devices are related to the four framing functions that Entman (1993)
distinguishes: the promotion of a particular problem definition; causal interpretation; moral
evaluation; and treatment recommendation. In his 2004 book Projections of power, Entman
argues that the application of a frame should promote an interpretation, problem definition,
and causal relationships.
The connection between framing devices. reasoning devices and the actual frame
occurs during the interpretation phase of the message. This process of interpretation ensures
that the complexity of the event is reduced to a more understandable and plausible whole.
The frame can suggest a definition, an explanation, a problematisation, and an evaluation of
an event and can ultimately result in a number of conclusions. Because of this the media
provide the public with both information on the event itself and how it should be interpreted.
This means that framing is a form of meta-communication (Van Gorp, 2005); the frame
specifies the relationship between a number of elements in a text on the basis of which a
topic may be defined and understood (Bateson, 1972). This type of communication helps the
receiver to structure and define reality. The notion of meta-communication implies that the
meaning readers assign to a text is not determined merely by the explicit information that the
text contains, but also by implicit information between the lines (Gurevitch & Levy, 1986).
Frames don’t need to occupy a central position in the structure of a text; it can fleetingly be
present in a number of devices. Therefore, an important aspect of the framing on the receiver
side is that it provides a context in which the news message can be interpreted (Cappella &
Jamieson, 1997).
A frame is not linked to any particular topic; it is possible to identify frames that define
an identical situation in a different way. Which means that a topic can be framed in several
ways, and one frame can be applied to various topics. It’s also important to note that a frame
USE OF FRAMES: NEWS MEDIA IN CRISIS
Theoretical Framework
10
is characterized by some level of abstraction, because of this it can be applied to vastly
different issues. Even an issue-specific frame is in fact is linkable to another, more abstract
“master”-frame (Van Gorp, 2005).
2.2.2. Frames in the media.
In framing theory it is assumed that media makers deliberately make use of frames.
In this sense, the “selection” of a frame is a conscious decision on the part of the journalist.
The frame that eventually gets embedded in the news message is not determined in advance
by the reported item. If this were the case it would mean that there would be a “correct”
frame that corresponds with the event. The task of the journalist would then consist of
representing this correct frame as accurately as possible (Van Gorp, 2007). The elements
belonging to this structure stem from the occurrence, whereas this is not the case for the
connection between these elements in the news story (Hackett,1984).
A key event, such as an organisational crisis, can lead to the activation of alternative
frames in the media (Scheufele, 2004). In these specific situations certain sponsors of a
particular frame, such as interest groups, spin doctors, advertisers, etc., may strategically try
to convince the media to cover the crisis in accordance with “their” frame. This frame came to
be by prior strategic decision making about the manner in which their viewpoints should be
announced (Entman, 2004). Frame sponsors are concerned with directing the perception
and the frame selection of journalists as they report on an event (Pan & Kosicki, 1993).
It can be assumed that the sponsors deliberately choose which frames to present to
convince the receiver of their viewpoints and inform them about the situation (Van Gorp,
2007). In these cases it is still possible for the journalist to report on the event with a different
frame (a counterframe) to the one presented, and can even ignore the proposed frame
altogether (Benford, 1993; Callaghan & Schnell, 2001; Gamson & Modigliani, 1989).
Several studies show that the influence from sponsors on frames seems to be the
strongest for issues in which journalists and other players in the policy arena can find shared
narratives with these sponsors. Also, organisations in crisis tend to downplay and be more
positive about their crisis in comparison to their competitors that are not in crisis (Benson,
1988; Veil et al., 2012). This might explain why organisations in crisis frame the crisis
differently than organisations that are not in crisis. The organisation in crisis needs to protect
or repair its reputation and thus chooses a narrative with specific frames to appease it
sponsor, in this case itself. They do this because framing processes are very important for
the shaping of organisational reputation (Coombs, 2007), the prevention of crisis escalation
(Seeger, 2002), and the avoidance of public confusion and even panic (Liu & Kim, 2011; Van
der Meer & Verhoeven, 2013). The other organisation that is not in crisis, does not have the
USE OF FRAMES: NEWS MEDIA IN CRISIS
Theoretical Framework
11
same goal as the organisation in crisis, hence it could choose a different narrative. Which
could lead to different narratives about the same crisis.
These studies, although interesting, are not necessarily applicable to situations in
which the organisation in crisis is part of the media landscape, since they were based on
situations in which the organisation in crisis was not a media organisation. Also, bearing in
mind that media organisations are bound to a stricter set of rules (McQuail, 2012), the
following hypothesis should be interesting to test: “The media outlet in reputation crisis uses
frames to downplay the crisis, while the media outlets that are in crisis uses frames to amplify
the crisis.”
News media organisations, just like any other organisation, are for-profit businesses.
This means that they have to produce news and news programmes that have to be
consumed by as many viewers as possible. The more viewers a program has, the higher the
ratings of this program will be. The higher the ratings are, the more advertisers will pay the
media organisation to advertise in that specific timeslot. The bigger the company gets, the
more dominant it becomes on the market, the more viewers it can reach, the more profits it
can make (McQuail, 2012, chapter 9). This fight for ratings, market share and profits has lead
to a higher concentration, sometimes in quite extreme forms. In 1983 90% of the media in
the United States of America (USA), the country in which the crises studied took place, was
owned by 50 companies. In 2011 this number went down to 6 conglomerates (Business
Insiter, 2012). According to McQuail (2012, p. 232) less media companies, ergo a high media
concentration, could lead to a loss of diversity, maybe even objectivity. Since ownership of
media inevitably influences content of the media, which in turn is in line with the concept of
Freedom of the Press – the right that any entity can publish without censorship (McQuail,
2012, p. 228 & p. 557). This means that each company has the legal right to decide for itself
how to report on a crisis. This could lead to a variety of frames between companies or
conglomerates, since they all could be catering to different sponsors due different interests.
But within the conglomerate the use of frames could be similar, since the companies
compromising of the conglomerate ultimately have the same goal and cater to the same
frame sponsors. This could lead to a conflict between what a news media outlet ought to
from the conglomerate’s perspective and what McQuail’s (2012, chapter 8) principles and
standards of news media. Particularly if “one of their own” is in crisis: will media outlets within
a conglomerate turn a blind eye to their “familial” bond – thus treating the in crisis outlet as if
it were an actual competitor from a “rival” conglomerate –, or will they ignore the principles
and standards and stand up for their “sister” company?
USE OF FRAMES: NEWS MEDIA IN CRISIS
Theoretical Framework
12
2.2.3. Frame alignment theory.
In organizational-crisis situations, sense-making activities rapidly emerge, resulting in
frames that help to comprehend what is going on (Cornelissen et al., 2009; Schultz & Raupp,
2010; Weick, 1988). Various interested parties are likely to differ in their initial production and
use of frames in a crisis situation. Therefore frame variation is the norm (Leydesdorff &
Hellsten, 2005; McCammon, 2012).
After the party-specific frame-building processes, the frames will meet on platforms
provided by the media. In order to solve organizational crises and to avoid uncontrollable
crisis magnification, it is crucial to arrive at collective meaning of the complex situation
(Seeger, 2002; Weick, 1988). Therefore, in a complex nexus of competing frames, there will
be a need to provide coherence to the issue (Hellsten et al., 2010; Snow et al., 1986) and
reduce ambiguity and uncertainty (Leydesdorff & Ivanova, in press). This results in mutual
borrowings and frame alignment among the parties. The need for meaning coherence may
urge party-specific discourses to open up, move toward one another, and eventually overlap,
instead of remaining isolated. This party rapprochement does not imply total fusion of the
parties’ perspectives. However, the interplay between the various parties’ meaning provision
and framing may lead to frame alignment regarding the meaning of the crisis among the
domains of PR, news media, and the public.
The notion of frame alignment in organizational crises relates to several empirical
findings, such as the cases of the French riots (Snow et al., 2007), H1N1 flu-pandemic (Liu &
Kim, 2011), financial crisis (Schultz & Raupp, 2010), the BP oil spill crisis (Schultz et al.,
2012), Max Havelaar skepticism (Van der Meer, 2014), and the explosion of a chemical plant
(Van der Meer & Verhoeven, 2013). Despite their differences, these cases point clearly in the
same direction: framing among domains aligns over time after initial differences in framing.
2.2.3.1. After frame alignment.
Crisis communication literature acknowledges the dimension of time in crises and that
crises evolve in several phases (Coombs, 2012). It is natural to assume that these phases
influence the way news media use frames in their narratives of the crisis. When mutual
sense making of the crisis is completed, frame alignment is no longer a necessity. Therefore,
frame alignment is a temporal state that, once achieved, cannot be taken for granted as it is
subject to reassessment. The alignment is periodically necessary for “ameliorating” the
prospect of misframing or interpretative errors (Snow et al., 1986). This was confirmed in Van
der Meer, 2013, in which they analyse the Moerdijk chemical plant disaster.
The impermanence of frame alignment emphasizes that frame development is
characterized by a dynamic process (Scheufele, 2004) and that frames are inherently
USE OF FRAMES: NEWS MEDIA IN CRISIS
Theoretical Framework
13
unstable (Steinberg, 1998). In this regard, a crisis-specific pattern in the framing among the
domains is assumed.
In short, the routine narratives and frames of news media are disrupted by a crisis
situation (Schultz et al., 2012), producing crisis-specific frames and frame alignment to
provide clarity for the public (Snow et al., 1986). When those circumstances become less
pressing, in terms of a decrease of crisis emergency and a mutual understanding of the
crisis, news media have again space to select specific parts of the crisis situation framing,
creating a de-alignment. Essentially following the development of the crisis, as
acknowledged by Coombs (2012). Van der Meer et al. (2014) confirm this theory of frame
alignment in their research. However, all research mentioned is based on situations in which
the organisation in crisis is not a part of the media landscape. Assuming that the media
coverage of a crisis of a media organisation is different than the coverage of a crisis of a
non-media organisation, and keeping in mind that the non-media has to adhere to stricter rules
(McQuail, 2012). Will the development of the use of frames follow the path proposed by the
frame alignment theory, even if the organisation in crisis is part of the media?
2.2.4. Frame analysis.
Van Gorp (2007) proposes a reliable method to analyse frames in media content. He
based his methodology on the constructionist approach, and therefor integrates several
aspects of the communication process in the analysis. This is done by using the idea that
frames are part of culture as a tool to understand the underlying framing processes and to
guide research on framing, Essential for the construction of frames for this research are the
frame package elements of Gamson and Lansch (1983) and Gamson and Modigliani (1989).
The first element, the “manifested and/or latent” reasoning devices, deal with
justifications, causes and consequences. Its function is quite similar to the functions of
frames proposed by Entman (1993): the promotion of a particular problem definition, causal
interpretation, moral evaluation and treatment recommendation. Thus, the reasoning devices
of the frames used in this research will fulfil these functions.
The second element of which a frame package consists of is the “manifested framing
devices”. These indicate what the core idea is of the frame. Since this research is done
within the scope of crisis communication, theories on crisis communication might have some
interesting insights. Coombs’ SCCT and Benoit’s Image Restoration Theory provide
typologies on crisis situations and crisis-response strategies. These often centre on
attribution of responsibility, intentions, possible actions of the organisation in crisis and
judgements. By analysing a portion of the sample used for this research, the following
framing devices were either chosen or derived from the literature: offensiveness of the crisis;
who is pointing out the wrong doing; who is the victim of the crisis; who is held accountable
USE OF FRAMES: NEWS MEDIA IN CRISIS
Theoretical Framework
14
for the crisis; what are the actions of the responsible; was the crisis caused intentionally; how
can the crisis be solved; and how is the crisis judgement. These framing devices will each
have their own reasoning device and therefore fulfil a framing function. Coombs and Benoit
also specify how some of these frame devises can be further specified. These specifications
will also be used.
The last element that Gamson and Lansch (1983) and Gamson and Modigliani (1989)
specify is the “implicit cultural phenomenon” and is closely related to the framing devices.
This element is important, since it gives the frame a meaning that is embedded in culture.
This in turn can facilitate the interpretation of the frame. Based on the frame devices
mentioned above, the cultural phenomenons used for this research are offensiveness,
archetypes (accuser, victim, responsible), actions, intent and judgement (i.e. Bern, 2004;
Rowan, 2015).
To summarise, the frames that will be used in this research will contain one of the
cultural phenomena, will contain a core idea and will fulfil a specific function. For a visual
overview, see figure 1 on the next page. How specific frames will be defined will be
discussed in the Methods-chapter.
USE OF FRAMES: NEWS MEDIA IN CRISIS
Theoretical Framework
15
Identifying causes
Archetypes
Actions
Intent
Accuser
Victim
Denial
Evasion of responsibility
Mortification
Circumstantial
Accidental
Responsible
No denial
Simple denial
Shifting blame
Provocation
Defeasibility
Accident
Good intentions
Preventable
Intentional
Is the situation problematic?
Endorsing remedy
Passing judgement
Offensiveness
Remedy
Judgement
Reducing
Increasing
Bolstering
Minimisation
Transcendence
Attacking accuser
Compensation
Figure 1. Sketch of the, still incomplete, frame analysis matrix for this research based on Van Gorp (2007),
Entman (2004), Benoit (1995) and Coombs (2007).
USE OF FRAMES: NEWS MEDIA IN CRISIS
Methods
16
3. Methods
In this chapter the sampling procedure and measures of this research will be presented.
Also, some attention will be given to the coding procedure and the inter-coder reliability.
3.1. Research design
This study used a mixed design: a comparative, longitudinal, two-part quantitative content
analysis of a multiple embedded case study (Huttner, et al., 1995). This study was
comparative because, by means of a hypothesis and sub-questions (addressed in the
Theoretical Framework, and will be repeated in the Results), this thesis answered the
research question “Does a media organisation that is suffering a reputation crisis use frames
differently when reporting on the crisis than a media organisation that is not in crisis, and
does the theory of frame alignment apply?” It compared the usage of frames during the crisis
coverage of media outlets in crisis versus media outlets not in crisis. This study was also
longitudinal, because this thesis analysed how the usage of frames developed over time (e.g.
for the duration of the crisis), in order to see whether there is frame alignment. Since the
study used a sample made up of three cases and the only focus of the analysis will be
frames, discarding all other aspects of the cases, the study could be called a multiple
embedded case study.
3.2 Sampling procedure
The sample was focused on the coverage of three reputation crisis cases caused by an
employee of a TV network based in the USA. The first case involved Mary Mapes (a
producer) and Dan Rather (an anchor) both employed at the time at CBS. Both got fired after
it seemed that they didn’t do their job properly: they failed to authenticate documents that
would smear George W. Bush’s reputation during his re-election. The second case involved
Bill O’Reily (an anchor) at Fox News Channel. He was caught being untruthful about his
reporting of the Falkland Wars: he wasn’t present on the islands in the warzone, he was
about 1900 kilometres further in Buenos Aires reporting a minor riot. This opened the door to
more of his past exaggerations. The last case involved Brian Williams (an anchor) at NBC
who was also untruthful about his reporting. This particular crisis was regarding his
embellishment of what occurred in Iraq. He’s been demoted to MSNBC. For a more detailed
description of the cases, see Appendix A.
USE OF FRAMES: NEWS MEDIA IN CRISIS
Methods
17
These cases were selected according to a number of criteria. First the crisis had to be
caused by an employee. This way the frames can be used in the most diverse way, i.e. one
could blame or victimise the employee and/or the organisation employing him or her.
Secondly, the crisis needed to be fairly recent, since this would facilitate the gathering of
media coverage about the crisis. Then the crisis had to be old enough for it to be, somewhat,
concluded. This made it possible to study the development of the crisis over time, making
studying frame alignment possible. The selected cases that survived the initial selection were
again filtered with one more criteria: they all had to take place in the same country. This was
important because different countries have different communication/media cultures. This
could lead to different reporting styles of the same type of crisis. The results of such research
would be hard to interpret, and for an accurate interpretation each country involved would
need a decent sample, so that the results can be explained per country as well. The
timeframe in which this thesis has to be completed is, unfortunately, too short for this bigger
research.
The coverage on these cases was taken from the online database LexisNexis, using
specific search keywords, specified in Appendix A in the case description. This yielded a
convenience sample consisting of 459 transcripts from a number of TV networks (ABC
News, Bloomberg News, CBS News, CNBC News, CNN, Fox Business Network, Fox News
Channel, MSNBC, NBC News and PBS), and all coverage of every crisis from its starting
date up until 31
stof October 2015.
3.3. Coding procedure
As mentioned before, the quantitative content analysis consisted of two parts, the codebook
(see Appendix B) was divided in the same way. The first part was done with the help of a
computer; making it a computer assisted content analysis, and was used to code the
“objective” information from the sample. This was information about date, length, which TV
network the item comes from, etc. The syntax used was an adaptation of the SPSS-code
used for prominence analysis (see Appendix C). This is reflected in the “A. General
Information”-section of the codebook. There are a number of reasons why only this section
will be coded with a computer. These reason have to do with the advantages and
disadvantages of computer assisted content analysis.
One of the advantages of computer analysis is that a computer can process a large
sample in little time, which could save some time. Computers can also process textual
material reliably. This advantage, however, is also a disadvantage. Where humans can read
a text, attribute meaning to words, therefore interpreting what they read and make decisions
on how to code the text; computers can only code strings of characters. The researcher has
USE OF FRAMES: NEWS MEDIA IN CRISIS
Methods
18
to insert these strings in syntax before the coding starts and anything that might be relevant
that isn’t coded will be left out (Krippendorff, 2013, chapter 11.2). Because of this only the
variables with a set amount of values were coded with a computer.
The second part of the content analysis was largely done manually and is reflected in
“B. Item Content”. Some variables were calculated by SPSS with the help of the manually
coded variables, code also available in Appendix C. In this section the variables focused on
the content regarding the use of frames in the items. The variables used to code the frames
were based on the literature and the coding results of a small portion of the sample. How
these variables came about and how these were coded will be discussed in the next
paragraph, “Measures”.
3.4. Measures
3.4.1. Independent variables.
3.4.1.1. Outlet variables.
For each item the news outlet that published it was coded in the variable Outlet.
These were coded in a numeric way. This facilitated the coding of the other news outlet
related variables such as Crisis (whether a news outlet is in crisis or not), Umbrella crisis (to
group the outlet in crisis with it “sister” company, suitable for comparing the conglomerate to
the rest of the non-crisis media), Sister Umbrella (to code if an outlet is the “sister” of the
outlet in crisis, suitable for comparing the two) and Sister VS (to code if an outlet is the
“sister” of the outlet in crisis, suitable for comparing the “sister” against the other non-crisis
outlets).
3.4.1.2. Time variables.
By coding the date that the items were published, it was possible to analyse how the
crises develops over time. It also facilitated the grouping of the coverage in bigger time
periods, such as weeks. Grouping the coverage in larger periods of time helped made
evaluating the development of frames easier, since it eliminates the day-to-day “noise”.
3.4.2. Dependent variables: frames.
In the “Theoretical Framework” a sketch of the frame matrix used in this research was
presented (figure 1). This was constructed according to Van Gorp (2007) constructionist
approach, Entman (2004)’s views on framing and the crisis communication theories of
USE OF FRAMES: NEWS MEDIA IN CRISIS
Methods
19
mentioned: frames that could be applicable to a diverse range of situations. Hence it was not
enough to properly guide the analysis of frames in this research, which compromises of three
quite specific cases. To make the scheme more appropriate the small portion of the sample
was coded to provide the much needed sub-frames: frames that are specific to these three
cases.
The sub-frames found for the Responsible “master” frame were “Network” (the
network in crisis is responsible for the crisis), “Employee” (the employee of the network in
crisis is responsible for the crisis), “Competitor” (a competitor of the network in crisis is
responsible for the crisis), “Audience” (a competitor of the network in crisis is responsible for
the crisis), “Colleague” (a competitor of the network in crisis is responsible for the crisis) and
“Source” (a competitor of the network in crisis is responsible for the crisis). And
“Other”-category was added for good measure, in case a new responsible was pointed out that was
not defined. These sub-frames were binary variables (i.e. 1 = yes, it is present, 2 = no, it is
not present). The reason why each sub-frames had it’s own variable instead of giving the
“master” frame several options, was because there were several items that used several
sub-frames. By giving each sub-frame its own variable, the use of multiple sub-frames could
be coded more easily. The same process was repeated for the “master” frames Accuser,
Victim, Intent, Reducing and Increasing Offensiveness and Remedy. The Mortification frame
(in which the responsible expresses disappointment, guilt and/or apologizes) will one be
coded for presence and will not contain any sub-frames..
Due the high amount of frames present in the analysis, the “master” frames
themselves were converted into overview variables, in which was coded if its sub-frames
were present or not. This was done with SPSS syntax (present in Appendix C).
3.4.2.1. Judgment.
The Judgement “master” frames was different type of frame than the ones previously
mentioned; it didn’t code the presence of something it coded the sentiment of the crisis
coverage; therefore a different style of coding was required. For this variable a sentiment
analysis was executed. This required a text-file with the relevant portions of the transcripts,
wordlists containing words with a positive or negative connotations, and syntax with which
Python can code the sentiments present (also present in Appendix C). Preparing the
text-files was done at the same time while coding the frames to prevent re-visitation of the
transcripts. The wordlists used were developed by Hu and Lui (2004) and Lui, Hu and Cheng
(2005). Python coded positivity and negativity scores (amount of positive or negative words
divided by the total amount of words) according to these wordlist. These scores were used to
calculate the “master” frame variable Judgement, which was a continuous scale variable
ranging van -1 to 1.
USE OF FRAMES: NEWS MEDIA IN CRISIS
Methods
20
Figure 2, on the next page, showcases the complete set of frames that were used
and briefly describes who they were coded. Each frame is coded in its own variable, which
carries the same name as the frame.
3.4.3. Inter-coder reliability.
Although there was just one coder that coded all the items for this research, is
important to assess if several coders will come to the same conclusions when given the
proper training with the codebook. To assess the inter-coder reliability a fellow student was
asked to code 10% of the sample. Using the ReCal2 online tool for calculating reliabilities,
the 59 variables that needed to be coded manually were tested.
The overall conclusion is that the reliability is acceptable, but not ideal. Even though
all alpha scores were either very low or undefined, the percentage of agreement was high
(73.2% to 100%). This result isn’t as strange as it seems, when you keep in mind that all
variables were binary (which leaves little room for covariance) and factoring the coincidence
when there’s 100% agreement is not possible (Krippendorff, 2004). For a complete overview,
see Appendix D.
USE OF FRAMES: NEWS MEDIA IN CRISIS
Methods
21
Identifying causes
Archetype
Intent
(binary)
Accuser
(binary)
(binary)
Victim
Responsible
(binary)
Accidental
Circumstantial
Preventable
Intentional
Network
Employee
Competitor
Other people involved
Other
Network
Employee
Competitor
Profession
Audience
Colleague
Other
Network
Employee
Competitor
Audience
Colleague
Source
Other
Actions
Denial
(multiple)
Evasion of responsibility
(multiple)
Mortification
(binary)
No denial
Simple denial
Shifting blame
Provocation
Defeasibility
Accident
Good intentions
Is the situation problematic?
Endorsing remedy
Passing judgement
Offensiveness
(binary)
Remedy
(binary)
Judgement
(sentiment analysis)
Reducing
Increasing
Nothing
Leave with pay
Positive score
Negative score
Judgement
Bolstering
Minimisation
Transcendence
Attacking accuser
Compensation
Other
Suspension
Demotion
Fire the person
Retire
Other
Figure 2. The frame analysis matrix for this research based on Van Gorp (2007), Entman (2004), Benoit (1995),
Coombs (2007), Hu and Lui (2004) and Lui, Hu and Cheng (2005). In parentheses is described how the variables
were coded, and apply to all the variables present in the column. The exception to this, is the variables with
“multiple”. In these cases the frames written in the column will be used to define the “higher” frame.
USE OF FRAMES: NEWS MEDIA IN CRISIS
Results
22
4. Results
In this chapter the results of each hypothesis will be discussed. It is important to note that if
the frames and other variables presented in the methods are not presented during the
discussion of the results, it means that the results of these variables were not significant.
Also note that some frames have deviating sample sizes. The reason for this is that they
were coded a bit differently than the other frames. This has to do with the instructions given
in the codebook (available in Appendix B).
4.1 Downplaying and amplifying crisis with frames
The first hypothesis stated that the media outlet in crisis (N = 27) would use frames to
downplay the crisis, while the media outlets that are not in crisis (N = 432) would use frames
to amplify the crisis. Whether a frame is considered as to downplay or amplify the crisis, is
previously presented in the Theoretical Framework and Methods.
Most frame variables were analysed with a Fisher’s exact test. This statistical method
was chosen because most frames were coded in binary variables and the expected counts of
these variables were likely to be less than 5. The Denial frame and the Evasion of
Responsibility frame were both analysed with a Chi-square test, since these variables
provided more than 2 options for the coder. The Judgement frame, which was a scale
variable, was analysed with an Independent Samples T-Test to be able to compare the
means of the two groups. Table 1 gives a quick overview of the significant results.
4.1.1. “Master” frames.
The first “master”-frame that is significantly used differently by both groups of outlets
is the Problematic frame. Outlets in crisis tend to use this frame less in their coverage than
outlets that are not in crisis. This association, however significant, is quite weak (p < .01, Phi
= –.23, N = 209). The next “master” frame with a significant, but with a weak association,
result is the Increasing Offensiveness frame. The outlet in crisis will use this frame less in
their coverage of their own crisis than outlets that are not in crisis reporting on the same
crisis (p < .01, Phi = –.23, N = 209). The Victim “master” frame is the third frame in which
there is a significant difference in use. Although the association is very weak, it is possible to
state that media outlets in crisis are more inclined to use this frame in their coverage of the
crisis than their not-in-crisis counterparts (p < .05, Phi = .11, N = 459). The fourth and last
“master” frame with a significant difference in use in the coverage of a crisis is the Intent
frame. Outlets that are in crisis use this frame less while covering their own crisis than the
USE OF FRAMES: NEWS MEDIA IN CRISIS
Results
23
outlets that don’t have a crisis. This association is, also, very weak (p < .01, Phi = –.14, N =
459).
4.1.2. Sub-frames.
There are also 4 sub-frames in which these two groups differ from each other. One
such frame is the Network sub-frame of the Victim “master” frame. Outlets in crisis tend to
use this frame more in their coverage than outlets that are not in crisis (p < .01, Phi = .14, N
= 459). The next sub-frame with a significant, also belonging the Victim “master” frame, is the
Employee sub-frame. It is possible to state that media outlets in crisis are more inclined to
use this frame in their coverage of the crisis than their not-in-crisis counterparts (p < .01, Phi
= .14, N = 459). The third sub-frame that media outlets in crisis use more than the other
media, is the Source sub-frame belonging to the Responsible sub-frame (p < .05, Phi = .15,
N = 459). The last frame is the Preventable sub-frame belonging to the Intent “master” frame.
The news media in crisis also use this more than their non-crisis counterparts (p < .05, Phi =
.11, N = 459). All associations mentioned were quite weak.
Table 1
Difference in the use of frames between in-crisis outlets and not-in-crisis outlets
Frame
Level of significance p
Value of Phi
Sample size N
Problematic
< .01
-.23
209
Increasing Offensiveness
< .01
-.23
209
Victim
< .05
.11
459
Network
< .01
.14
459
Employee
< .01
.14
459
Responsible
Not significant
Source
< .05
.15
459
Intent
< .01
-.14
209
Preventable
< .05
.11
459
Note. The difference in sample size is due the difference in the way of coding the variables. These
instructions were given in the codebook.
4.1.3. Implications.
The analysis for hypotheses yielded a slim set of significant results for 4 out of 10
“master” frames and 4 out of 34 sub-frames (18.2% of the frame variables) present in the
sample. All were accompanied by very weak to weak associations. Both groups used frames
that downplay and amplify the crisis. But it is clear that the outlets in crisis use the downplay
frames more than the crisis free outlets. They use the Victim-related frames the most, usually
calling themselves and their employee the victims. They hold a third party responsible (their
source) and state that the crisis could be prevented, thus implicitly stating that it’s not a
structural problem. It is also clear that frames that amplify the crisis are used less by the
outlet in crisis than the outlets not in crisis. The in-crisis outlets avoid calling the crisis
problematic and increasing its offensiveness. They also mention the intention behind the
USE OF FRAMES: NEWS MEDIA IN CRISIS
Results
24
crisis less. Because of these results the first hypothesis “The media outlet in reputation crisis
uses frames to downplay the crisis, while the media outlets that are not in crisis uses frames
to amplify the crisis” can be supported even though the differences are very large.
4.2. The conglomerate
Since that ownership influences content and reduces diversity, therefore
compromising objectivity, and that media conglomerates dominate the media landscape in
the USA (Business Insider, 2012), it could be assumed that this is also reflected in the way
media reports on their own and their competition’s crisis. To research this assumption only
second and third case of the sample was used. This due the fact that CBS News, the outlet
in crisis in the first case, is the only outlet used of the conglomerate of which it is part of (CBS
Corporation). The remaining sample (N = 301) was further divided and 4 groups: in crisis
outlets (N = 11), “sister” companies (N = 9), the conglomerate in crisis (N = 20), and the rest
of the non-crisis media (N = 281). The same statistical tests mentioned in paragraph 4.1
where used.
Table 2 and 3 give a quick overview of the significant results.
4.2.1. “Sister” company vs. media outlet in crisis.
There were no significant differences in the use of frames between the media outlet in
crisis and it’s “sister” company.
4.2.2. “Sister” company vs. other media outlets not in crisis.
4.2.2.1. “Master” frames.
There are no “master” frames in which the “sister” company of the media outlet in
crisis differs significantly with a media outlet not in crisis, that doesn’t belong to the same
conglomerate.
4.2.2.2. Sub-frames.
There are several sub-frames in which the defined groups significantly differ from
each other. One such frame is the Bolstering sub-frame of the Reducing Offensiveness
“master” frame. The “sister” companies use this frame more than the other non-crisis media
outlets (p < .01, Phi = .42, N = 135). The same is true for the Network (p < .05, Phi = .19, N =
290) and Employee (p < .01, Phi = .42, N = 290) sub-frames of the Victim “master” frame.
These associations are weak to moderate.
USE OF FRAMES: NEWS MEDIA IN CRISIS
Results
25
Table 2
Difference in the use of frames between “sister” outlets and the other not-in-crisis outlets
Frame
Level of significance p
Value of Phi
Sample size N
Reducing Offensiveness
Not significant
-.23
209
Bolstering
< .01
-.23
209
Victim
Not significant
.11
459
Network
< .01
.14
459
Employee
< .01
.14
459
Note. The difference in sample size is due the difference in the way of coding the variables. These
instructions were given in the codebook.
4.2.3. Conglomerate in crisis vs. non-conglomerate not in crisis.
4.2.3.1. “Master” frames.
The first “master”-frame that is significantly used differently by both groups of outlets
is the Problematic frame. The conglomerate in crisis tend to use this frame less in their
coverage than outlets that are not in crisis. This association, however significant, is weak (p
< .05, Phi = –.27, N = 139). The next “master” frame with a significant, but with a weak
association, result is the Reducing Offensiveness frame. The conglomerate in crisis will use
this frame more in their coverage of their own crisis than outlets that are not in crisis
reporting on the same crisis (p < .05, Phi = .21, N = 139). The Increasing Offensiveness
“master” frame is the third frame in which there is a significant difference in use. Although the
association is very weak, it is possible to state that media outlets in crisis are less inclined to
use this frame in their coverage of the crisis than their not-in-crisis counterparts (p = .01, Phi
= -.28, N = 139). The fourth and last “master” frame with a significant difference in use in the
coverage of a crisis is the Victim frame. Outlets that are in crisis use this frame more while
covering their own crisis than the outlets that don’t have a crisis. This association is, also,
very weak (p < .01, Phi = .18, N = 301)
4.2.3.2. Sub-frames.
There are also six sub-frames in which these two groups differ from each other. One
such frame is the Bolstering sub-frame of the Reducing Offensiveness “master” frame.
Outlets in crisis tend to use this frame more in their coverage than outlets that are not in
crisis (p < .05, Phi = .25, N = 139). The same is true for the Network (p < .01, Phi = .23, N =
301) and Employee (p < .01, Phi = .41, N = 301) sub-frames of the Victim “master” frame.
These associations are weak to moderate. The next three sub-frames with a significant
difference belong to the Responsible “master” frames. The conglomerate in crisis tends to
use the Employee sub-frame (p < .05, Phi = -.15, N = 301) less while using the Competitor
sub-frame (p < .05, Phi = .15, N = 301) and Audience sub-frame (p < .05, Phi = .24, N = 301)
more than the other media outlets. These associations are quite weak.
USE OF FRAMES: NEWS MEDIA IN CRISIS
Results
26
Table 3
Difference in the use of frames between the conglomerate and the other not-in-crisis outlets
Frame
Level of significance p
Value of Phi
Sample size N
Problematic
< .01
-.23
209
Increasing Offensiveness
< .01
-.23
209
Victim
< .05
.11
459
Network
< .01
.14
459
Employee
< .01
.14
459
Responsible
Not significant
Source
< .05
.15
459
Intent
< .01
-.14
209
Preventable
< .05
.11
459
Note. The difference in sample size is due the difference in the way of coding the
variables. These instructions were given in the codebook.
4.2.4. Implications.
The first set of analysis show that there is no significant difference between the
coverage of the outlet in crisis and its “sister” outlet that is not in crisis when it comes to the
use of frames. When comparing the “sister” outlet with the other outlets that are not in crisis,
the analysis shows that there is a quite modest difference between the two groups. 4 out of
28 frames present in sample reported a significant difference between the two groups. This
difference is a bit smaller, and has a stronger measure of association, than the difference
discussed in the first paragraph: 8 out of 44 (18.2%) variables with very weak to weak
association vs. 3 out of 28 (10.7%) variables with weak to moderate association. The last
analysis, in which the conglomerate in crisis is compared to the outlets that are not in crisis
and are not part of the conglomerate, shows that there is quite some difference between the
two groups. 10 out of 28 variables (35.7%) of the frame and judgement variables present in
the sample report a significant difference between the conglomerate in crisis and the rest,
with very weak to moderate measures of association.
These results suggest that “sister” outlets of the outlet in crisis seem to have a similar
view on the crisis as the crisis outlet has, but also seems to differ little in their coverage when
comparing them to the rest of the non-crisis media. When the coverage of the conglomerate
as a whole is compared to the other media outlets that are not in crisis, the difference seems
to be amplified. Therefore it could indeed be argued that ownership of media has an
influence on news media during a crisis, making the conglomerate as a whole stand-up for
their “family”-member in crisis, but the individual “familial” outlets do try to maintain a certain
amount of objectivity.
4.3. Frame alignment theory
The last question inquires whether the theory of frame alignment, as proposed by Van der
Meer, Verhoeven, Beentjes & Vliegenthart, 2014, is also applicable when the crisis only
USE OF FRAMES: NEWS MEDIA IN CRISIS
Results
27
involves media outlets. Contrary to previous hypotheses and question, this hypothesis will be
tested not with the sample as a whole, but for each individual case. This because the three
crises studied all occur at different times and have a different duration and coverage rhythm.
Due lack of time, these analysis will only focus on the frame alignment between in-crisis
media outlets and non-crisis media outlets.
The individual data files of cases were first split for the time variable “Week”. This way
SPSS will first divide the sample into weeks and than will compare the in-crisis and
not-in-crisis groups with each other within these weeks. The same tests described in 4.1 were
used.
4.3.1. Case 1: CBS News with Dan Rather and Mary Mapes.
The first crisis discussed, “belongs” to CBS. It spans a period of 30 weeks, with
sporadic mentioning up to 565 weeks. CBS reports a total of 15 times, while the other media
outlets do this 143 times.
In the first week of the crisis there was just 1 item by CBS is the only one covering the
crisis. This isn’t strange since the crisis originated on the Internet and blew over to TV later
on. Within the confines of this sample it means that CBS could start of with its own frames
and therefore could set the tone for the coverage of this crisis on TV. This week was followed
by a period of de-alignment in frames with an occasional alignment of certain frames when
CBS reports on the crisis. For a rough overview see Table 4 on page 29.
4.3.2. Case 2: FOX News Channel with Bill O’Reilly.
The second crisis discussed spans a period of 3 weeks. Fox News Channel only
reports three times on the crisis in the first week of the crisis. During this week the
Responsible frame was the only frame with perfect alignment. The other frames were used
differed from each other in a non-significant way.
4.3.3. Case 3: NBC News with Brian Williams.
The last crisis discussed spans a period of 39 weeks. Just like with the last case,
NBC did not only report not much on the crisis: 8 times. The crisis starts with a mix of aligned
and de-aligned frames. Followed by a week of quite some de-alignment. NBC then goes a
few weeks quiet to remerge with similar coverage as the other news media, causing frame
alignment in the frames used. See Table 5 on page 30 for a rough overview of the
USE OF FRAMES: NEWS MEDIA IN CRISIS