Food fraud : the horsemeat scandal : a case study of different presentations in a shared crisis

(1)

Food Fraud: the Horsemeat Scandal

A Case Study of Different Presentations in a Shared Crisis

Master’s Thesis

University of Amsterdam’s Graduate School of Communication

Master’s programme Communication Science

Supervised by Dr. Piet Verhoeven

[27-06-2014]

(10627359)

(2)

Abstract

In this research, the influence of multiple organizations’ crisis communication in a shared crisis on news media reporting is explored. This is done with a case study on the organizations Tesco, Iceland, IKEA and ASDA, who were involved in the UK horsemeat scandal. Through quantitative content analysis, it is found that regardless of the crisis cluster, the organizations involved in the horsemeat scandal mostly use rebuild strategies as their crisis response. In their crisis communication, the organizations take their responsibility, making attribution framing the most common frame in the press releases. Even if an organization perceives itself as innocent, it may accept the responsibility and become a villain in order to restore their image. In this, the media play a great role. Public relations strategies are considered to be successful when the content and framing of the press releases are taken over as intended by the media. With that said, the media take over the frames of each organizations’ press release, except for attribution framing which is more present in the press releases than in the news media reports. In terms of quoting, it can be stated that Iceland and IKEA failed horribly while Tesco and ASDA’s press releases were taken over for, on average, 10%. In terms of success in influencing the media reporting, Tesco stands out by being the most present in news media reports and having the highest quoting percentage. Results show that the media’s presentation of the organizations involved in the horsemeat scandal does not solely depend on the crisis communication strategies; there is more to it. Remarkable insights on the way different parties handle, and are presented in, a shared crisis are gained, which provide us with new perspectives on crisis communication and interesting topics for future research.

Key words: crisis communication, public relations, press releases, media, framing,

(3)

Acknowledgements

In memory of Wim, who always believed in me.

During the process of writing a thesis, the author needs to find inspiration and support when needed. Looking back on this period, I feel blessed for having been surrounded by people who were always there to help me.

First and foremost, I would like to show my gratitude towards my supervisor, Dr. Piet Verhoeven, for his encouragement during this journey. During the period leading up to the completion of this thesis, you have been a tremendous help. Thank you for always having me leave your office with a lot of valuable insights and recommendations.

Mom and dad, for taking me to, and leaving me in, the library to spend endless of hours reading books and catering to my continuous eager to learn everything about anything. The ground of the knowledge I have today, comes from you.

My boyfriend, you have shown remarkable patience despite my occasional burst outs of stress and you have always been there to talk some sense back into me. Somehow, you always know exactly what I need and for that, I thank you.

Also, I am grateful for having a loving family and good friends who have listened, given their support and encouraged me during this period.

Last but not least, here is to caffeine and sugar, my companions through many long nights of writing.

Thank you all for your loving support, I will always carry it with me in my heart.

Amsterdam, June 2014

(4)

Introduction

Thousands of jokes and numerous of memes are made about it but the horsemeat scandal is a serious case. This scandal shed a light on the questionable practices of retailers and good companies; companies that are expected to cautiously follow the strict health and safety standards. Major UK retailers - including Tesco, Iceland, Aldi, Lidl, Ikea, ASDA and the Co-op – were exposed and it was revealed that they sold beef products that contained horsemeat. Shortly after this revelation, the scandal bombarded all media outlets and subsequently, affected the entire European beef supply chain, with abattoirs, suppliers, manufacturers and retailers all involved (Holmes & Sudhaman, 2014).

The horsemeat scandal includes the key elements of a crisis. First, non-labelled use of horsemeat (instead of beef) results into an involuntarily exposure of consumers; given that the fraudulent composition of the product is not communicated to the public. Second and third, the scandal leads to much uncertainty since the source of the problem was initially hard to identify and thus, it was difficult to answer the questions raised after the revelations. Fourth, the fraud is not limited to the UK alone, it is universal; several European countries are involved and the problem has expanded to a pan-European scale. Fifth, it concerns a combination of a technological hazard (fraudulent mixing of ingredients during food processing) and a possible bioactive risk (products of animal origin), which is a highly sensitive risk-combination with substantial potential to evolve into a crisis. Finally, there are some media-triggers such as the presence of suspects and a link with crime (Verbeke, 2013).

This sudden and unexpected event disrupts the organizations’ operations and comes with both financial and reputation threats. In this crisis, the stakeholders are physically, emotionally and financially harmed. As a result, people might think badly of the organization, causing the crisis to negatively affect the organizational reputation. In this, the news media play a critical role since most stakeholder will learn and gather

(6)

their information from news reports (Coombs, 2007). Since the average person’s real contact with events in society is limited, it is the media that construct the public’s knowledge of organizations, politicians, companies and situations (McQuail, 2000). It is therefore suggested that the media’s coverage of an organization during a crisis is directly related to particular communicative behaviours and practices of the organization (Romenti & Valentini, 2010).

In almost all successful crisis management efforts, communication is key. Communicating poorly during a crisis often makes bad situations even worse. In times of crisis, an organization’s publics are full with questions. Reporters, employees, stockholders, government officials, and local residents all want, and need, to know: What happened? Who did it happen to? When? Where? How? Why? Organizations that wait, or fail, to answer these questions are likely to suffer unnecessary financial, emotional and perceptual damage. The ability to communicate quickly and effectively is absolutely one of the most important parts of effective crisis management (Marra, 1999).

What is striking in the media reports concerning the horsemeat scandal, is that it is mainly focusing on food fraud rather than food safety, in terms of concerns about the intake of horse mediatisation through consuming the meat, or moral ‘disgust’ regarding the fact that many people could have eaten horsemeat. The scandal was merely related to the failure of performance by the retailers in not supplying the consumer with the product they promised. The retailers that are involved in this scandal lost the trust of their consumers and the need for crisis management and recovery of this scandal is high (Holmes & Sudhaman, 2014).

Scientific and Societal Relevance

There is a research gap in empirical research about crises in which multiple parties are involved. While the effectiveness of different response strategies is extensively examined in previous research, the involvement of multiple parties in one and the same

(7)

crisis is neglected. This research aims to explore this ignored matter. Subsequently, this thesis fills a research gap in the field of public relations and crisis communications research by analyzing associations between content and strategic framing in public relations and the news.

Examining this phenomenon also has societal relevance. It can serve as an advice for companies as it will teach them how to influence media reporting in times of crisis. Since there are multiple organizations involved, a reflection on the most successful crisis communication strategy can be made. Additionally, it will provide them information on how to deal with multiple parties involved in a shared crisis. Finally, by using a recent crisis as a case study, the results are as up to date as possible and thus, are applicable to crises today.

Doing a case study on the horsemeat scandal leads us to be able to answer the central research question for this thesis:

RQ: How do the different involved parties and the media handle the same crisis?

The central research question needs to be broken down into sub questions. This is done to guide the direction of this study, to narrow down the necessary literature and to create a method suitable to answer these questions. This comes down to this research being divided into three parts, presented below.

Part One – Press Release Analyses

Research Question 1: What crisis communication strategies are used by the different parties involved in the Horsemeat Scandal?

Sub-Research Question 1.1: Are there differences in the crisis type applicable for each party? If yes, in what respect?

Sub-Research Question 1.2: Are there differences in the type of crisis response between the different parties? If yes, in what respect?

(8)

Sub-Research Question 1.3: Are there differences in image restoration strategies used by the different parties? If yes, in what respect?

Part Two – News Media Reports Analyses

Research Question 2: How are differing parties presented in the news media reports during a shared crisis?

Sub-Research Question 2.1: Are there differences in the levels of tone of the news media towards the parties (positive, negative, and neutral) when covering crises? If yes, in what respect?

Sub-Research Question 2.2: Are the different parties framed differently by the news media (attribution of responsibility, human interest, conflict, morality, and economic)? If yes, in what respect?

Part Three – Press Releases vs. News Media Reports

Research Question 3: Which party’s crisis communication strategy was the most successful with regard to influencing the news media reporting with their press releases?

Sub-Research Question 3.1: To what extent do the news media take over the content of the press releases of the different parties?

Sub-Research Question 3.2: To what extent does the associative frame in the parties’ press releases associate differ from the associative frame in the UK media?

Theoretical Background

From a communications standpoint, a crisis is a business or organizational problem that is exposed to public attention that threatens a company’s reputation and its ability to conduct business. Coombs (2007) describes a crisis as a “sudden and unexpected event that threatens to disrupt an organization’s operations and poses both a financial

(9)

and a reputational threat” (p. 164). In a crisis, the organization’s competence or honesty is under attack because some or all of the communication about the threat is outside the organization’s control (Lattimore, Baskin, Heiman, & Toth, 2007; Smith, 2009 as cited in Swanson, 2012).

As with the case of the horsemeat scandal, a crisis initially involving a single organization can draw the attention to the practices of the entire industry and/or market segment and thus, drawing other organizations in it as well (Fortunato, 2011 as cited in Swanson, 2012). Therefore, it is fundamental that organizations in crisis offers a timely, appropriate, consistent and effective response (Coombs, 2007; Greer & Moreland, 2003; Jin, Pang & Cameron, 2006; Stephens, Malone & Bailey, 2005).

By using the correct public relations strategy, an organization aims to influence the public’s perception of the organization and to maintain and/or restore a positive image (Coombs, 2004; Coombs, 2007; Coombs & Holladay, 2002; Stephens, Malone & Bailey, 2005; Utz, Schultz & Glocka, 2013). Especially in a controversial situation like this, surrounded by potentially dangerous scientific or medical matters, the ability to frame an issue is especially important (Andsager & Smiley, 1998; Stephens & Malone, 2009).

There are three domains which are essential for the consequences and escalation of organizational crises due to their role in crisis communication and framing: public relations, news media and the public (Cornelissen, Carroll & Elving, 2009; Schultz, Kleinnijenhuis, Oegema, Utz, & Van Atteveldt, 2012; van der Meer & Verhoeven, 2013). The information disseminated by an organization in crisis and the media channel(s) used for that dissemination need to have a strong strategic match in order for the communication to be effective (Coombs, 2007). This is because the final judgment on resolution of a crisis is likely to come through the media – and experience has shown that many members of the public will adopt the media’s

(10)

perspective (Coombs, 2007). The role of the organizations’ public relations strategies are analyzed with the following research question:

Research Question 1: What crisis communication strategies are used by the different parties involved in the horsemeat scandal?

Over the past decades, Benoit’s (1997) image restoration theory and Coombs’ (2002) situational crisis communication theory have provided the dominant paradigms for organizations’ strategy development and crisis communication research (Kim, Avery & Lariscy, 2009; Utz, Schultz & Glocka, 2013). Therefore, these are the theories discussed and applied in this research.

Situational Crisis Communication Theory

The situational crisis communication theory (SCCT) offers a framework for understanding how crisis communication can be used to protect an organization’s reputation during a crisis. Guidelines for crisis communication, including a mechanism for anticipating how stakeholders will react to a crisis in terms of the reputational threat posed by the crisis, is projected by the SCCT (Coombs, 2007).

To protect the organization from reputational threats during a crisis, the SCCT posits that the crisis situation needs to be understood in order to decide on the crisis response strategy. In this, the crisis type determines the reputational threat. Each crisis type features certain aspects of the crisis and indicate how stakeholders should interpret the crisis (Coombs & Holladay, 2002). Crisis types can be seen as a certain frame and allows the crisis manager to establish or shape the crisis frame. This frame then determines how much stakeholders attribute responsibility for the crisis to the organization.

In conclusion, we cannot assume that every organization involved is likely to take the blame and show responsibility for the crisis, and thus, present it as a preventable crisis; expected is that they are also likely to play the victim role. Moreover,

(11)

the media are able to shape the crisis type for each organization in the way that they present them (Sellnow & Seeger, 2013). Therefore, it is expected that the organizations involved are differing in crisis type relating to the horsemeat scandal. From this, the following sub-hypothesis is formed:

Hypothesis 1.1: Fewer organizations will fall into the victim crisis type, more of the involved organizations will place themselves in the preventable crisis type cluster.

Subsequently, the crisis type assigned influences the crisis response strategy chosen. A crisis response strategy can be described as what the management says and does after a crisis occurs. It is used in order to repair the reputation, reduce negative effects and to prevent negative behavioural intentions (Coombs, 2007). The SCCT’s list of crisis response strategies is built around the perceived acceptance of responsibility for a crisis. These strategies can be divided into three groups: denial, diminish and rebuild strategies (Coombs, 2006)1.

Claeys and Cauberghe (2014) state that the crisis communication will only be successful when the crisis response strategy matches the crisis type, a finding based on the SCCT by Coombs. Especially in the case of a preventable crisis, rebuild strategies seem to be most appropriate whereas denial strategies are common when the organization takes the position of a victim. However, since we expect differences in crisis type, differences in crisis response strategies are to be expected as well. This comes down to the following sub-hypothesis:

Hypothesis 1.2: Organizations placing themselves in the victim crisis type are likely to use denial response strategies, while organizations in the preventable crisis type cluster will be using rebuild strategies.

1_{More detailed information on the SCCT’s crisis response strategies can be found in the}

(12)

Image Restoration Theory

The SCCT draws upon the crisis response strategies presented in the image restoration theory by Benoit (1997) and thus, in order to create a holistic image of crisis communication, the image restoration theory needs to be included as well (Coombs, 2007).

Benoit (1997) puts ‘image’ forward as being essential to organizations and a key concept in public relations. As a result, the image restoration theory is created as a viable approach for use in developing and understanding messages that respond to organizations’ image crises. The image restoration strategies take on different approaches: denial, evasion of responsibility, reducing offensiveness of event, corrective action and mortification2. The choice of strategy depends on the level of offensiveness and attributed responsibility, which are likely to vary for each organization. From this, the following hypothesis is formed:

Hypothesis 1.3: Organizations placing themselves in the victim crisis type are likely to use denial image restoration strategies, while organizations in the preventable crisis type cluster will be merely using mortification and corrective action image restoration strategies.

Based on the previously mentioned literature and proposed sub-hypotheses, we can propose the following overall hypothesis to answer the main question of the first part of this research:

Hypothesis 1: The different parties involved in the horsemeat scandal will use differing crisis communication strategies.

Media Framing

As has come forward from the review of the crisis communication literature, the media seem to play an important role. The way an organization deals with their crisis and especially, the way in which it is reported publicly, is able to make or break their

2

More detailed information on the Image Restoration Strategies can be found in the Method section.

(13)

reputation. This can be accounted to the fact that the main source of information is the news media, making media coverage a key feature of reputation management (Carroll, 2004; Carroll & McCombs, 2003; Coombs, 2007; Dutton & Dukerich, 1991Meijer, 2004; Romenti & Valentini, 2010).

Hence, the framing of the crisis through the news media is extremely important (Carroll, 2004; Carroll and McCombs, 2003; Coombs, 2007; Meijer, 2004; Romenti & Valentini, 2010). Particularly when a crisis includes controversy, involving potentially dangerous scientific or medical matters, framing is essential (Andsager & Smiley, 1998). These media frames serve as working routines for journalists, enabling them to rapidly identify and classify information and to ‘package’ it for their audiences (Scheufele, 2000). Thus, to describe the term ‘media frames’, we can point to the following definition of Gamson & Modigliani (1987, p. 143 as cited in Scheufele, 2000):

“A central organizing idea or story line that provides meaning to an unfolding strip of events. The frame suggests what the controversy is about, the essence of the issue.”

The framing of the different organizations involved in the horsemeat scandal in the media is analyzed with the following research question in mind:

Research Question 2: How are differing parties presented in the news media reports during a shared crisis?

The news media’s tone. Romenti and Valentini (2010) found that most of the

newspapers’ coverage on crises framed the crisis responses with a negative tone. The media seldom reported the actions and information providing strategies of the organizations and when doing so, this was done with negative associations. Schultz, Kleinnijenhuis, Oegema, Utz and van Atteveldt (2012) confirm this finding: they argue that since negativism and conflict lead to more publicity, the media focus more on causes and consequences (negative aspects) than the organizations’ press releases.

(14)

This can be expected for the horsemeat scandal as well. It could be, for example, that the organizations provide solutions in their press releases but the media will give more attention to the consequences of the horsemeat in beef products. The good news about bad news is that it sells. Therefore, we propose the following sub-research question and hypothesis:

Hypothesis 2.1: The general tone of the news media towards the organizations when covering the crisis will be negative.

The news media’s frame. Semetko and Valkenburg (2000) present five crisis

news frames: attribution of responsibility, human interest, conflict, morality, and economic. Research has shown that crisis news stories used news frames in the order of predominance: attribution of responsibility, economic, conflict, human interest, and morality (An & Gower, 2009; Semetko & Valkenburg, 2006). Yet, findings also show that in crisis reporting, the framing differs among the three crisis types (victim, accidental and preventable). Moreover, in crisis reporting, the media tend to assign specific blame to the individual or organization, especially in preventable crises. This makes the attribution of responsibility the most used frame when there is strong controllability and intentionality on the part of the actor (An & Gower, 2009).

The second most frequently used frame in crisis news is the economic frame. Recall crises dealt mainly with economic consequences on the organization itself and consumers (An & Gower, 2009). This is also likely to be applicable in the case of the horsemeat scandal, since this crisis involved the recalling of several products of a few organizations as well. Compared to these frames, the other three frames are not used as frequently in crisis news. However, with the moral aspect in mind, the morality frame may be visible in the media reporting as well. Since the organizations are expected to handle the crisis differently, the frames presented in the media may differ as well. With reference to the frames used, the sub-research question and hypothesis are as follows:

(15)

Hypothesis 2.2: The attribution of responsibility frame will be the most used frame by the news media, while the economic and morality frames will be used less and the human interest and conflict the least.

Regarding the second part of this research, the evaluation of the media reporting during the horsemeat scandal, the overarching hypothesis is as follows:

Hypothesis 2: The differing parties will all be negatively presented in the news media reports, but, within different frames varying from attribution of responsibility to economic framing and morality frames.

Success Factors of Press Releases

In the field of public relations research, the main focus lies on the analysis of the interplay between PR, news and publics (Seletzky & Lehman-Wilzig, 2012; Van der Meer & Verhoeven, 2013). Public relations strategies are considered to be successful when the content and framing of the press releases are taken over as intended by the media and consequently, influences the stakeholders’ perceptions of the organizations (Seletzky & Lehman-Wilzig, 2012). To evaluate the rate of success of the responses used by the organizations involved in the horsemeat scandal, this research aims to answer the following question:

Research Question 3: Which party’s crisis communication strategy was the most successful with regard to influencing the news media reporting with their press releases?

Publishing prepared news. With the cutbacks on newspapers and the

subsequent drawback in news sources in mind, many journalism and public relations researchers come to the conclusion that the press release is still the most common form of information for a journalist (Seletzky & Lehman-Wilzig, 2012). It is estimated that between 25% and 50% (at times even above 80%) of the news items originate in some form of public relations (Lee & Solomon, 1990; Stauber & Rampton, 1995; Curtin & Rhodenbough, 2001 as cited in Seletzky & Lehman-Wilzig, 2012). The willingness to publish ‘prepared’ news is seemingly growing (Seletzky & Lehman-Wilzig, 2012). Still,

(16)

‘pre-packaged’ news is, in general, not fully copy-pasted but rather supplemented with context and gathered reactions on the news (Hijmans, Schafraad & Buijs, 2009). Never before has research made a concrete comparison between the offer of press releases of multiple organizations involved in a crisis and the subsequent media reporting. Therefore, it is unclear to what extent organizations will be successful in their attempt to influence the media. This makes us unable to propose a hypothesis and thus, this leads to the explorative research question:

Sub-Research Question 3.1: To what extent do the news media reports take over the content of the press releases of the different parties?

Associative frames. Overall, news media tend to rely on the frames provided

by the most influential policy actors. By possessing a monopoly, or near-monopoly, on information provided through press releases, policy actors are able to perform a certain degree of influence on the dissemination of the news stories (Andsager & Smiley, 1998). Journalists tend to rely on the expertise of others and their influence in framing controversial issues is visible. Yet, the media are also seen to be using other frames than the ones used in the published press releases of the organizations involved in a particular crisis. The tone of the news media continues to be negative and accusative, regardless of the solutions proposed by the involved organizations (Romenti & Valentini, 2010).

In relation to the horsemeat scandal, we expect the organizations to use different crisis response strategies and thus, different frames. Subsequently, the content and associative frame in the media will not always be similar to the one presented in an organization’s press release. This leads to the following sub- hypothesis:

Hypothesis 3.2: The associative frame in the organizations’ press releases differs from the associative frame in the news media reports.

(17)

A crisis response can be seen as a failure when the news media reject the crisis manager’s frame and continue by using another frame. Stakeholders will then receive competing frames and will select the one provided by the source they perceive as most credible (Coombs, 2007; Coombs & Holladay, 2002). Organizations using a consistent crisis response that matches the crisis type, are expected to be the most successful in handling the horsemeat scandal. This comes down to the overall hypothesis, answering the main question of the final part of this research, presented below:

Hypothesis 3: The most influential organization, using a matching and consistent crisis response, will be more successful in influencing the media reporting than the other organizations.

Method

The aforementioned research question shows that the aim of this research is to determine the relationship between the strategies used in the organization’s crisis communication and the following media frames in the UK news reports. A time of crisis is chosen because a crisis is able to threaten an organization’s high priority goals (Seeger, Sellnow, & Ulmer, 1998) and often results in high media attention and corporate attempts to communicatively repair organizational legitimacy (Pattriotta, Gond & Schultz, 2011).

The case study “the horsemeat scandal in the United Kingdom” is used as an example because it led, and still leads, to a large amount of media coverage and a spread of the crisis discourse all over Europe. Moreover, it dramatically affected the foundations of the entire beef-supply chain. Subsequently, this gives us the ability to make a comparison between different parties involved. This aim, to determine a relationship between the crisis response strategy and the media framing, fits well with quantitative research (Hopkins, 2008). Since this research is descriptive and we want to establish the associations between the variables, a quantitative content analysis is

(18)

conducted. Quantitative content analysis utilizes a variety of tools and methods to study media content. This technique allows for an objective, systematic, and quantitative description of the manifest content of communication (Berelson, 1952). It enables the making of inferences by systematically and objectively identifying specified characteristics of messages (Holsti, 1969).

Research Method

In order to be able to describe the crisis communications strategies used by the involved parties and their reflection in the media reporting, a quantitative content analysis is conducted on the press releases and news media reports of each party. The case is analyzed with the help of the SCCT by Coombs (2007), image restoration theory by Benoit (1997) and the media frames presented by Semetko and Valkenburg (2000). This crisis involves the following different parties: Tesco, Iceland, Aldi, Lidl, Ikea, ASDA and the Co-op. This is a shared crisis, through which the parties are all affected differently and thus, cope with differently. That is why, this research aspires to compare the differences in their crisis response strategies and reflects on how these strategies influence the media reporting.

This research examines if competing parties unite in order to tackle the shared crisis and portray consensus to the outside world or if they save their own businesses and reputations, even if it is at the expense of other parties. Analyzing this provides us with an in-depth understanding of how different organizations handle the same crisis. First, the organizations’ press releases are analyzed and compared to see whether they differ in terms of crisis communication strategies. Second, their presentation in the media is evaluated. At last, the news media reports are checked for the frames they apply to each party and whether they differ from one another. This enables the drawing of conclusions on how the parties are able to influence the way the media frames them. Additionally, the news media reports are analyzed to see how much the news media take over from the published press releases of the UK retailers to see which party has

(19)

the most ‘power’ over the media reporting. Thus, the processes of crisis communication, strategic framing and application of associative frames are examined by doing a comparative content analysis of the UK news articles concerning the horsemeat scandal and the press releases published by the UK parties involved.

Research Units

In this research, the population that is researched are press releases and news media reports. To maximize the validity and reliability, all press releases of the involved parties concerning the horsemeat scandal are included in this research. For the analysis of the news media, a subset of the largest newspapers of the UK is used. The news coverage on the horsemeat scandal of the newspapers, Daily Mail and The Guardian are selected for content analysis. These newspapers are selected because of their dominant status in printed media coverage in the UK. They are chosen from the most popular newspapers (The Sun, Daily Mail, The Daily Telegraph, The Times, The Guardian, The Financial Times and Daily Mirror as cited in Newspapers, 2014) because they provide the largest amount of articles on the horsemeat scandal. Finally, this sample includes both a quality and a sensationalist newspaper, making the analysis more suitable for generalization.

Sample.

Press releases. Acquired through the official websites of the organizations

involved, a total of seventeen press releases, published in a time frame of 15-01-2013 until 15-04-2013, are analyzed. This sample consists of press releases of Tesco, Iceland, IKEA and ASDA. The remaining organizations, Aldi, Lidl and Co-op, either did not have an official statement on their website, or, in the case of Aldi, made them unavailable for public access3.

3

(20)

News media reports. The news media reports are acquired through LexisNexis

and will include the reports of the following newspapers: Daily Mail and The Guardian. The sampling method used for the selection of newspapers is a stratified random sampling strategy. Only the largest newspapers are included in the analysis, simultaneously excluding smaller UK newspapers as they are not considered to be relevant enough for this research. This diverse sample includes left- and right alignment, quality journalism and sensationalism, business-oriented and a mix of these characteristics (Newspapers, 2014).

The time frame of the selected news media reports are the published news between the day of the first press release about the horsemeat scandal, and a week after the last one. This comes down to a time frame from 15-01-2013 until 16-04-2013. This three-month period is taken from the research done by Andsager and Smiley (1998), who also did a content analysis on the shaping of the news coverage, and happens to cover all published press releases regarding the horsemeat scandal with the first being released on 15-01-2013 and the last one on 09-04-2013.

Data collection. The data collected includes a total of seventeen press

releases, divided among the following four companies: Tesco (5 press releases), Iceland (1), IKEA (5) and ASDA (6). Through LexisNexis, a total of 251 news media reports from Daily Mail and The Guardian was acquired. After removing the double articles, the ones that included columns, comments or feature stories, articles that did not have their main focus on the horsemeat scandal and the ones that did not involve a crisis response of any of the involved organizations, there are 28 valid ones left. The remaining articles focus on the horsemeat scandal and one or more of the organizations involved. Thus, the data collected from seventeen press releases and 28 news media reports is used and analyzed4.

4

An overview of the news media reports and the press releases they are matched with can be found in Appendix 2.

(21)

Operationalization

5

Situational crisis communication theory.

Crisis type. The dependent variable crisis type shows the three crisis clusters

presented in the SCCT by Coombs (2004). Based on the study of An and Gower (2009), crisis type is measured by ascribing the applicable crisis type from the thirteen options: rumour, natural disaster, malevolence/product tampering, workplace violence, challenge, megadamage (technical error accident), technical breakdown accidents, technical breakdown recalls, human breakdown accidents, human breakdown recalls, organizational misdeed with no injuries, organizational misdeed management misconduct (violate a law), organizational misdeed with injuries (stakeholders). From this, the crisis cluster follows in a second variable: victim (crisis type 1-4), accidental and preventable (9-13).

Crisis response strategy. Also following Coombs (2004) and his list for responding to crises according to the SCCT, the crisis response strategies include the following, ascribed as different items, in one dependent variable: full apology, corrective action, ingratiation, justification, excuse, denial, attack the accuser. Based on the codebook of An and Gower (2009), the crisis response is divided into three groups: deny (crisis response 6 and 7), diminish (3, 4 and 5) and rebuild (1 and 2).

Image restoration theory. The image restoration theory by Benoit (1997)

describes strategies that can be deployed to minimize the damage in a crisis where the organization’s reputation is likely to be under scrutiny. Benoit (1997) outlines the following strategies, which are translated into five variables: denial, evasion of

responsibility, reducing offensiveness of event, corrective action and mortification.

These strategies are all measured by answering multiple questions regarding whether the strategy is used ‘yes’ or ‘no’, which is based on the study of An and Gower (2009).

5

The full operationalization process can be found in Appendix 3 and the resulting codebooks can be found in Appendix 4 – Codebook Press Release and Appendix 5 – Codebook News Media Report.

(22)

Media framing. The media frames described by Semetko and Valkenburg

(2000) are also analyzed using the codebook of An and Gower (2009). The presence of the five frames/variables, attribution frame, human interest frame, conflict frame,

morality frame and economic frame, is measured by asking several questions

answered with ‘yes’ and ‘no’.

Inter-coder Reliability

To calculate the inter-coder reliability in this content analysis, we make use of Krippendorff’s Alpha. A second coder is called for to analyze two of the seventeen press releases (11.76%) and nine news media reports (33.33% of the total news media reports and 10.59% of the total analyses on the news media reports). The results show that Kalpha = 0.88, indicating a high inter-coder reliability. In addition, the probability (q) of failure to achieve an alpha of at least 0.80, when the entire sample would be tested, is only 3.49%.

Plan Of Analysis

Data reduction.

Image restoration strategies. The concepts within the image restoration

strategies consist of denial, evasion of responsibility, reducing offensiveness of event,

corrective action and mortification. Except for corrective action and mortification, the

concepts are measured using several items. These items consist of questions answered with ‘yes’ and ‘no’. All strategies are computed by adding up the related items and dividing this sum by the number of items. Finally, the newly computed variables are recoded back to ‘yes’ and ‘no’.

Media frames. The media frames included in this analysis are attribution

framing, human interest framing, conflict framing, morality framing and economic framing. These concepts are measured by the use of several items consisting of

questions answered with ‘yes’ and ‘no’. All frames are computed by adding up the related items and dividing this sum by the number of items. The newly computed

(23)

variables measure a frame consisting of a scale from ‘frame not present at all’ and ‘frame fully present’.

Hypothesis testing.

Cross tabulations and chi-square tests. A cross tabulation in combination

with a chi-square test is the most used test to analyze categorical data, showing the relationship between two categorical variables. This is done for hypothesis 1.1, 1.2, 1.3 and 2.1 because these hypotheses all ask if the residuals for one level of a variable differ as a function of another variable.

One-way analysis of variance. Hypothesis 2.2 analyzes whether there is any

difference between groups (organizations) on some variable (media frames). Here, an interval variable (media frames) is involved and thus, it is appropriate to use a one-way analysis of variance (ANOVA). This is used to determine the presence of significant differences between the means of three or more independent (unrelated) groups.

Multivariate analysis of variance. The one-way multivariate analysis of

variance (MANOVA) is used to determine whether there are any differences between independent groups on more than one continuous dependent variable. This is exactly what is tested in hypothesis 3.1, where it is analyzed whether the organizations differ in terms of quoting percentages in their press releases and news media reports. Here, several correlated dependent variables come into play and a MANOVA offers a single, overall statistical test on this set of variables instead of performing multiple individual tests.

Independent samples t-tests. An independent samples t-test compares the

means between two unrelated groups on the same continuous, dependent variable. For hypothesis 3.2, the means of the presence of a certain frame (dependent variable) in the press releases (group 1) and news media reports (group 2) are compared. An independent samples t-test is run per organization to see whether the frames used in the press releases and news media reports differ from each other.

(24)

Results

This research analyzes a total of seventeen press releases, with an average of 261 words, and 27 news media reports containing on average 493 words. The 27 news media reports are matched to one or more press releases, depending on the date of publishing from both. Subsequently, the news media reports are often analyzed multiple times, depending on the number of press releases it is matched with; each time to code it with another organization in mind. Examining the news media reports for each organization separately, leaves us with a total of 85 news media report analyses. In these analyses, Tesco’s response was covered seventeen times, Iceland’s three times, IKEA’s two times and lastly, the crisis response of ASDA was included in the news media reports six times.

Hypotheses Testing

Hypothesis 1. Hypothesis 1.1 assumes that there is a difference in the matter

of the type of crisis cluster between the organizations involved. More specifically, it expects that fewer organizations present themselves as a victim, the larger part of the involved organizations will place themselves in the preventable crisis cluster. To see whether we can accept, or need to reject, this hypothesis, a crosstab is created and a chi-square test is used to see if there is a relationship between the two categorical variables.

As we can see from Table 1, the largest part of the organizations chose to fall within the preventable crisis cluster, where a few also chose to play the victim. However, this relationship is not significant (N = 17, χ²(6, 17) = 8.716, p = 0.190). Since

the crisis cluster and the organization involved are not related, it is safe to say that the

differences are due to chance variation, which implies that each company falls in the same crisis cluster. Hence, hypothesis 1.1 can be rejected.

(25)

Table 1.

Crosstab Organization × Crisis Cluster

Organization Victim Accident Preventable N

Tesco 1 (20,0%) 0 (0,0%) 4 (80,0%) 5

Iceland 1 (100,0%) 0 (0.0%) 0 (0,0%) 1

IKEA 0 (0,0%) 0 (0,0%) 5 (100,0%) 5

ASDA 3 (50,0%) 1 (16,7%) 2 (33,3%) 6

Total 5 (29,4%) 1 (5,9%) 11 (64,7%) 17

Note. The value of Pearson Chi-Square is 8.716 with a significance level of 0.190 (p > 0.05). For hypothesis 1.2, it is assumed that organizations use differing response strategies in a shared crisis (H1.2a) and the response strategy used depends on the crisis type (H1.2b). Here, an organization putting themselves in the victim cluster is expected to use denial response strategies whereas those in a preventable cluster are assumed to use rebuild strategies. To find out if this is actually the case, a crosstab is created to first analyze whether there indeed is a relationship between the response strategy used and the organization involved.

Table 2 shows us that corrective action seems to be the most used crisis response (82,4%), after which comes denial (11,8%), followed by full apology (5,9%).

Corrective action and full apology together form the rebuild strategy, making this crisis

response appear as the most used in the horsemeat scandal. The other responses included in the content analysis appear to not have been used by the organizations involved at all. The results indicate that there is no statistically significant relationship between the type of crisis response and the organization involved (N = 17, χ²(6, 17) = 11.333, p = 0.079). However, when using a p-value of 0.10, we can state that this relationship is marginal significant. Thus, the first part of the hypothesis can be cautiously accepted; there are indeed differences in the crisis responses used by the different organizations involved. Finally, there appears to be a high association between the type of crisis response and the organizations (V = 0.577).

(26)

Table 2.

Crosstab Organization × Type of Crisis Response

Organization Full Apology Corrective Action Denial N Tesco 1 (20,0%) 4 (80,0%) 0 (0,0%) 5 Iceland 0 (0,0%) 0 (0.0%) 1 (100,0%) 1 IKEA 0 (0,0%) 5 (100,0%) 0 (100,0%) 5 ASDA 0 (0,0%) 5 (83,3%) 1 (16,7%) 6 Total 1 (5,9%) 14 (82,4%) 2 (11,8%) 17

Note. The value of Pearson Chi-Square is 11.333 with a significance level of 0.079 (p > 0.05 but p < 0.1). Cramer’s V has a value of 0.577, indicating a high association between the variables involved (p = 0.079).

For the latter part of hypothesis 1.2, we are looking for a significant association between crisis cluster and crisis response strategy in Table 3. With a p-value higher than 0.05, we can state that there is no association between crisis cluster and crisis

response (N = 17, χ²(4, 17) = 5.784, p = 0.216), meaning we have to reject the latter

part of hypothesis 1.2.

Table 3.

Crosstab Crisis Cluster × Type of Crisis Response

Crisis Cluster Full Apology Corrective Action Denial N Victim 0 (0,0%) 3 (60,0%) 2 (40,0%) 5 Accidental 0 (0,0%) 1 (100.0%) 0 (0,0%) 1 Preventable 1 (9.1%) 10 (90.9%) 0 (0,0%) 11 Total 1 (5,9%) 14 (82,4%) 2 (11,8%) 17

Note. The value of Pearson Chi-Square is 5.784 with a significance level of 0.216 (p > 0.05). Hypothesis 1.3 poses the assumption that organizations use differing image restoration strategies in a shared crisis (H1.3a) and the image restoration strategy used depends on the crisis cluster (H1.3b). To be more specific, it expects that organizations placing themselves among the victim crisis cluster are likely to use denial image restoration strategies, while organizations in the preventable crisis cluster will be expected to use mortification and corrective action image restoration strategies.

(27)

To test the first part of the hypothesis, a crosstab for each image restoration strategy is created to first analyze whether there indeed is a relationship between the

image restoration strategy used and the organization involved. For the final part, the

same procedure is followed with the variables crisis cluster and image restoration

strategy. The image restoration strategies denial and evasion of responsibility are

excluded from the analysis since these are constant variables. None of the organizations used this strategy in their press releases which means that there could not be a chi-square test performed.

Reducing offensiveness of event. There is no association between image

restoration strategy reducing offensiveness of event and the organizations (N = 17, χ²(3, 17) = 1.948, p = 0.583) as is shown in Table 4.

Table 4.

Crosstab Organization × Reducing Offensiveness of Event

Organization No Yes N Tesco 0 (0,0%) 5 (100,0%) 5 Iceland 1 (100,0%) 0 (0,0%) 1 IKEA 0 (0,0%) 1 (100,0%) 5 ASDA 5 (83,3%) 1 (16,7%) 6 Total 6 (%) 7 (%) 17

Note. The value of Pearson Chi-Square is 1.948 with a significance level of 0.583 (p > 0.05). Nevertheless, there is a significant, perfectly positive association between image restoration strategy reducing offensiveness of event and the crisis cluster (N = 17, V = 1, χ²(2, 17) = 17.000, p = 0.000). As can be seen in Table 5, the strategy is never used in the victim or preventable cluster, only once in the accidental cluster.

(28)

Table 5.

Crosstab Crisis Cluster × Reducing Offensiveness of Event

Crisis Cluster No Yes N

Victim 5 (100,0%) 0 (0,0%) 5

Accidental 0 (0,0%) 1 (100,0%) 1

Preventable 11 (100,0%) 0 (0,0%) 11

Total 16 (94,1%) 1 (5,9%) 17

Note. The value of Pearson Chi-Square is 17.000 with a significance level of 0.000 (p < 0.05). Cramer’s V has a value of 1.000, indicating a perfect association between the variables involved (p = 0.000).

Corrective action. As shown in Table 6, there is a significant, perfectly positive

association between image restoration strategy corrective action and the organizations (N = 17, V = 1, χ²(3, 17) = 17.000, p = 0.001). Except for Iceland, the organizations involved always use corrective action in their press releases as a strategy to restore their image.

Table 6.

Crosstab Organization × Corrective Action

Organization No Yes N Tesco 0 (0,0%) 5 (100,0%) 5 Iceland 1 (100,0%) 0 (0,0%) 1 IKEA 0 (0,0%) 5 (100,0%) 5 ASDA 0 (0,0%) 6 (100,0%) 6 Total 1 (5.9%) 16 (94.1%) 17

Note. The value of Pearson Chi-Square is 17.000 with a significance level of 0.001 (p < 0.05). Cramer’s V has a value of 1.000, indicating a perfect association between the variables involved (p = 0.001).

(29)

Yet, there is no association between image restoration strategy corrective

action and the crisis cluster (N = 17, V = 0.387, χ²(2, 17) = 2.550, p = 0.279) as can be

seen in Table 7.

Table 7.

Crosstab Crisis Cluster × Corrective Action

Victim 1 (20,0%) 4 (80,0%) 5

Accidental 0 (0,0%) 1 (100,0%) 1

Preventable 0 (0,0%) 11 (100,0%) 11

Total 1 (5,9%) 16 (94,1%) 17

Note. The value of Pearson Chi-Square is 2.550 with a significance level of 0.279 (p > 0.05).

Mortification. Table 8 shows that there is no association between image

restoration strategy mortification and the organizations (N = 17, V = 0.421, χ²(3, 17) = 3.009, p = 0.390).

Table 8.

Crosstab Organization × Mortification

Organization No Yes N Tesco 3 (60,0%) 2 (40,0%) 5 Iceland 1 (100,0%) 0 (0,0%) 1 IKEA 5 (100,0%) 0 (0,0%) 5 ASDA 5 (0,0%) 1 (100,0%) 6 Total 14 (82.4%) 3 (17.6%) 17

Note. The value of Pearson Chi-Square is 3.009 with a significance level of 0.390 (p > 0.05). Additionally, there is no association between image restoration strategy

mortification and the crisis cluster (N = 17, V = 0.118, χ²(2, 17) = 0.235, p = 0.889)

(30)

Table 9.

Crosstab Crisis Cluster × Mortification

Victim 4 (80,0%) 1 (20,0%) 5

Accidental 1 (100,0%) 0 (0,0%) 1

Preventable 9 (81,8%) 2 (18,2%) 11

Total 14 (82,4%) 3 (17,6%) 17

Note. The value of Pearson Chi-Square is 0.235 with a significance level of 0.889 (p > 0.05). The aforementioned results allow us to partly accept hypothesis 1.3: pertaining to the image restoration strategies, there are differences in use between the organizations. However, the image restoration strategy used does not depend on the crisis cluster.

Hypothesis 1: The different parties involved in the horsemeat scandal will use differing crisis communication strategies.

Based on the conclusions drawn from the tests performed on the sub-hypotheses above, we can state that hypothesis 1 can be accepted: the different parties involved in the horsemeat scandal do use differing communication strategies.

Hypothesis 2. Hypothesis 2.1 assumes that there will be no differences

between the tone towards the organizations presented in the news media reports; they are all negatively presented. To see whether we can accept, or need to reject, this hypothesis, a crosstab is created and a chi-square test is used to see if there is a relationship between the two categorical variables.

As we can see from Table 10, the media reporting generally holds a neutral tone towards the organizations (81,2%). Additionally, 14.1% of the news media reporting is positive towards the organizations and only 4.7% appears to be negative. The chi-Square test presented in Table 10 shows that there is a significant weak association between the organizations involved and the tone of the media (N = 85, V = 0.292, χ² = 14.488, df = 6, p = 0.025). Hence, hypothesis 1.2 can be rejected and we

(31)

can state that there is a difference in tone of the news media towards the organizations.

The differences in tone show that Tesco is presented most negatively (14,3%), whereas the other organizations are not presented negatively at all. Tesco also has the lowest score with regard to the media being neutral in their reporting on them; 60.3% against Iceland (90,0%), IKEA (100,0%) and (89,3%). However, they are also presented the most positively of all; 25.0% against Iceland (10,0%), IKEA (0,0%) and (10,7%).

Table 10.

Crosstab Organization × Tone

Organization Negative Neutral Positive N

Tesco 4 (14,3%) 17 (60,7%) 7 (25,0%) 28

Iceland 0 (0,0%) 18 (90.0%) 2 (10,0%) 20

IKEA 0 (0,0%) 9 (100,0%) 0 (0,0%) 9

ASDA 0 (0,0%) 25 (89,3%) 3 (10,7%) 28

Total 4 (4,7%) 69 (81,2%) 12 (14,1%) 85

Note. The value of Pearson Chi-Square is 14.488 with a significance level of 0.025 (p < 0.05). Cramer’s V has a value of 0.292, indicating a weak association between the variables involved (p = 0.025).

Hypothesis 2.2 examines whether there are differences concerning the framing in the media between the organizations involved. Here, it is expected that the attribution of responsibility frame will be the most used frame, while the economic- and morality frames will be used less and the human interest- and conflict frame the least. To test this assumption, an ANOVA is run in SPSS. Here, the groups are not of equal size and thus, we may not assume equal variances in the population on the basis of Levene's F-test.

(32)

Table 11 shows that the significance level of attribution framing is 0.002 (p = .002), which is below 0.05. and, therefore, there is a statistically significant difference in the mean of attribution framing between the organizations involved. This also goes for

human interest framing (p = 0.047) and conflict framing (p = 0.005).

Table 11.

One-Way Analysis of Variance of Media Frames by Organizations Involved

Media Frame Sum of Squares Df Mean Square F P Attribution Framing 1.744 3 0.581 5.601 0.002 Human Interest Framing 0.048 3 0.016 2.775 0.047 Conflict Framing 0.421 3 0.140 4.578 0.005 Morality Framing 0.007 3 0.002 0.146 0.932 Economic Framing 0.499 3 0.166 2.431 0.071

Thus, there are significant differences between the groups as a whole, with regard to attribution-, human interest- and conflict framing. The Scheffé post-hoc test shows that there is a significant difference in attribution framing between Tesco (M = 0.4114)and Iceland (M = 0.0900, p = 0.009) as well as between Tesco and ASDA (M = 0.1333, p = 0.016). A significant difference in conflict framing between Tesco (M = 0.1786) and ASDA (M = 0.357, p = 0.031) is found as well. However, no significant differences within human interest framing among the organizations was found (p > 0.05).

These findings leave us to partly accept hypothesis 2.2 since the framing does differ between the organizations, however, not between all organizations and it does not account for every frame used. Moreover, the expected order of use is almost

(33)

correct, except for conflict- and morality framing switching places: attribution framing (M = 0.2190, SD = 0.34796), economic framing (M = 0.1059, SD = 0.26831), conflict

framing (M = 0.0794, SD = 0.18585), human interest framing (M = 0.0235, SD =

0.07814) and morality framing (M = 0.0235, SD = 0.12374).

Hypothesis 2: The differing parties will all be negatively presented in the news media reports, but, within different frames varying from attribution of responsibility to economic framing and morality frames.

Consequently, based on the conclusions drawn from the tests performed on the sub-hypotheses above, we can state that hypothesis 2 can be partly accepted: the differing parties are not all negatively presented in the news media reports, but, to a certain extent, they are presented within different frames.

Hypothesis 3. For sub-question 3.1, the level of quoting from press releases in

the media is explored. By using an automated content analysis software6, the matched news media reports and press releases are compared to see how many words from the press release are taken over by the media. In this analysis, the shortest phrase needed to call it a match is three words and the fewest matches needed to report are ten. Hence, the press releases and news media reports not included in this report did not fulfil the aforementioned requirements.

In Table 12 below, the output of this comparison analysis is shown. Here, the column ‘Match’ refers to the amount of quoted words from the press release in the news media report. The percentages included describe the matched words as a percentage of the total amount of words of the press release (=L) and the news media reports (=R).

As can be seen in here, Tesco is quoted in the news media reports eleven times while ASDA is quoted ten times. In addition, the organizations do not seem to

6_{The automated content analysis software used is ‘WCopyfind’: an open source}

(34)

differ on the percentages that much since almost all values are around 10%. However, there is one outlier; Tesco’s first press release is taken over for 44% in one of the news media reports.

Table 12.

Press Release vs. News Media Comparison Report

Match Press Release (=L) News Media Report (=R)

14 (16% L, 2% R) ASDA 1 The Guardian 11

103 (44% L, 16% R) Tesco 1 The Guardian 11

17 (8% L, 2% R) ASDA 3 Daily Mail 1

14 (5% L, 2% R) ASDA 4 Daily Mail 1

To determine whether the differences between the organizations on their match between press release and news media report are significant, a one-way MANOVA is run in SPSS. This test is solely performed on the companies Tesco and ASDA since the other two, Iceland and IKEA, were not quoted in the news media reports at all.

(35)

In this analysis, the scores of the dependent variables (% of Quoting from Press

Release and % of Quoting from News Media Report) are combined in order to create a

more coherent picture of the influence of the independent variable (the organizations

involved). Thus, the MANOVA enables us to analyze whether Tesco has a statistically

significant larger combined quoting percentage than ASDA.

At first sight, in Table 13, it appears that Tesco scores relatively higher on being quoted. However, we can see that Wilks' Lambda has a significance value of 0.062 and from this, we can conclude that there is no statistically significant difference in quoting percentage based on the organization involved (F(2, 18) = 3.268, p > 0.0005; Wilk’s Λ = 0.062, partial η2 = 0.266). Nevertheless, we can still state that the level of quoting differs between the organizations involved since Iceland and IKEA are not quoted at all whilst Tesco and ASDA are.

Table 13.

Summary of the One-Way Multivariate Analysis of Variance of Quoting Percentages from Press Release and –News Media Report by Organizations Involved

Organization Mean SD N

% of Quoting from Press Release

Tesco 12.455 11,51836 11

ASDA 9.400 5,37897 10

% of Quoting from News Media Report

Tesco 4.455 4,20389 11

ASDA 2.000 1,15470 10

Note. The significance value of Wilks’ Lamba is 0.062 indicating that the one-way MANOVA is not statistically significant (p > .0005).

For hypothesis 3.2, the assumption holds that the frames used in the press release differ from the ones used in the news media reports. Again, it is analyzed whether this varies between the different organizations. Thus, there are seven variables involved in this analysis, namely: type of content (press release or news media report),

(36)

organization involved and the five frames. To test these differences, an independent

samples t-test is performed for each organization separately. Here, the groups are not of equal size and thus we may not assume equal variances in the population on the basis of Levene's F-test.

Tesco. This analysis found that with reference to attribution framing, Tesco’s

press releases contain an average presence of 0.88 (SD = 0.10954) which is a statistically significant higher presence than the news media reports (M = 0.4400, SD = 0.40143, T(17,838) = 3.838, p = 0.001). In addition, with regard to conflict framing, it appears that the news media reports (M = 0.1833, SD = 0.25820) have a significantly higher presence of the frame than Tesco’s press releases (M = 0.0000, SD = 0.0000,

T(14.000) = 2.750, p = 0.016). This means that Tesco’s attribution and conflict frames

differ from the ones presented in the news media reports. Regarding the other frames,

human interest-, morality- and economic framing, the news media did take over the

frames presented in Tesco’s press releases.

Iceland. In regard to the frames used in Iceland’s press release and the

presence of these frames in the news media reporting, no comparison could be made. The t-value could not be computed since the standard deviations of both groups are 0.00000. Therefore, we can assume that the frames present in the press release and in the news media reporting on Iceland do not differ from each other.

IKEA. The results shows us that the conflict- and morality frame are not

included in the analysis; this can be explained by the fact that IKEA did not use these frames in their press releases and simultaneously, these frames were not used in the news media reports covering IKEA.

Moreover, it tells us that with reference to attribution framing, IKEA’s press releases (M = 0.9600, SD = 0.08944) contain a statistically significant higher presence than the news media reports (M = 0.1000, SD = 0.24495, T(17,838) = 3.838, p =

(37)

0.000). This means that IKEA’s attribution framing differs from that presented in the news media reports. Relating to the other frames, human interest- and economic

framing, the news media did take over the frames presented in IKEA’s press releases

(p = 0.099 and p = 0.596). However, it must be said that both frames are barely used.

ASDA. From the results of the t-test performed on ASDA, we can conclude that

with regard to attribution framing, ASDA’s press releases (M = 0.8000, SD = 0.12649) contain a statistically significantly higher presence than the news media reports (M = 0.1200, SD = 0.24842, T(17,475) = 3.838, p = 0.000). This means that ASDA’s attribution framing differs from that presented in the news media reports. Concerning the other frames, the news media did take over the frames presented in ASDA’s press releases (p > 0.05).

With regard to hypothesis 3.2, we can state that the news media reports take over the frames in the press releases of every organization in the cases of human

interest-, conflict-, morality- and economic framing. Only Tesco was presented in the

news media reports in a conflict frame while it did not use this in its press releases. Additionally, all organizations had a higher presence of attribution framing than the level that was visible in the news media reports. Thus, hypothesis 3.2 can be partly accepted: while not all frames in the organizations’ press releases associate with the frames in the news media reports, these disassociations do not seem to differ between the organizations.

Hypothesis 3: The most influential organization, using a matching and consistent crisis response, will be the most successful in influencing the media reporting.

Based on the conclusions drawn on the sub-hypotheses above, we can state that hypothesis 3 can be partly accepted. The matching and consistent crisis response is not the determent factor of success in influencing the media reporting. Rather, it is the media who covers the largest information providing organization most extensively.

Food fraud : the horsemeat scandal : a case study of different presentations in a shared crisis