When the myth of safety collapses : the interplay between TEPCO, media, government, and the public awareness in the case of Fukushima Daiichi disaster

When the myth of safety collapses

The interplay between TEPCO, media, government, and the pubic awareness in

the case of Fukushima Daiichi disaster

Reina Kurobori

University of Amsterdam Graduate School of Communication

Master's Thesis

Student number: 10700900

Program: Master's Program in Communication Science Track: Corporate Communication

Supervisor: Dr. Friederike Schultz Date of submission: 26 June 2015

Abstract

Studies regarding how organizations minimize their reputational damages via communication in crises receive much attention today. However, they are also criticized for neglecting the constructivist approaches that actors rather make sense of each other during crises in order to maintain their legitimacy. The present study therefore concerns how different actors interact with each other through communication during the crisis. Taking Fukushima Daiichi disaster as an example case, the study content analyzed a total of 2134 documents issued between March and June 2011 by Tokyo Electric Power Company (TEPCO), media, and the Japanese government using free open source called Amsterdam Content Analysis Toolkit (AmCAT). The study also included data extracted from Google Trends in order to examine

agenda-setting effect between the media and the publics. Frequency analysis reveals the results which overall contradict to the classic agenda-setting, and -building effects except for the partial confirmation recognized in the results of intermedia agenda-setting. Semantic network analysis highlights different associations of concepts for each medium rather than similarities between these actors. Another set of associative frames demonstrates how the linkage of concepts changes over time. It is notable that three months after the outbreak, the only issue agenda strongly appears between TEPCO, media, and the government was TEPCO's responsibility. The fact that all actors failed to address other issues might have contributed to the criticism regarding the crisis management in this case.

Keywords: Semantic network analysis, crisis communication, agenda-setting,

When the myth of safety collapses: The interplay between TEPCO, media, government, and the pubic awareness in the case of Fukushima Daiichi disaster

"At the level of social life, what is called the adjustment of man to his environment

takes place through the medium of fictions" (Lippmann, 1922 p. 15). Do we live in a fictitious world? Not necessarily. However, this statement made by the prominent author Walter Lippmann certainly has a point. As he argues, the reality is so enormous, manifold, and short-lived that we judge and select a part to create our own map, which helps make our way across the world filled with ocean of information (Lippmann, 1922).

Especially in crises where ambiguity clouds the vision to create the map, people are in search of information. At the same time, those who caused the problems attempt to distribute appropriate information that can maintain people's positive perception towards them (Sohn & Lariscy, 2015). Living in an economic driven world today, research concerning the way in which minimizes the financial and reputational damage of corporations has been much discussed, particularly through the application of Situational Crisis Communication Theory (e.g. Coombs, 2007; Coombs & Holladay, 2002). Through the notion of agenda-setting and -building, researchers also have been examining the relationships how information from communicators influences receivers. Nevertheless, these approaches lack constructivist perspective that actors negotiate and make sense of each other in crises via communication so that they find the ways to clarify and justify themselves (e.g. Patoriotta, Gond, & Schultz, 2011). Recently, a new concept of associative frames has been proposed (e.g. Schultz, Kleinnijenhuis, Oegema, Utz, & Van Attenveldt, 2012) to overcome such deficits, but it still remains unclear how interaction between various domains in crises produces the agendas, which affects the reconstruction of the "reality" due to the limited number of studies. Moreover, little research has been conducted on the governmental crisis communication (Schultz & Raupp, 2010).

The current study therefore attempts to fill these gaps in research by focusing on one of the recent and worst nuclear crises that occurred at Fukushima Daiichi nuclear power plant in Japan in 2011. The main research question is "How did different actors

communicatively interact with each other during Fukushima Daiichi disaster?" The

three sub research questions were also formulated to specify the interest of this study.

Sub-question 1: How did the media's interaction in the case of Fukushima Daiichi disaster

associate with attention and network of other domains? Sub-question 2: How did the interplay between Tokyo Electric Power Company (hereinafter TEPCO), media, and the government change over the period of three months? Sub-question 3: How did the Japanese and the British newspapers report the crisis differently? To answer these questions, the current study content analyzes documents issued by three main communication media with the help of free software called Amsterdam Content Analysis Toolkit (hereinafter AmCAT). The study also includes data from Google Trends, which can numerically present any terms searched online, to examine how the media reporting influenced the public awareness.

The next section addresses relevant literature concerning sensemaking and different levels of agenda-setting and -building. The method includes overall research design, data collection, and analysis plan. Results are also analyzed in the paper, following which the theoretical and practical implications are discussed.

Theoretical background

Making sense of crises via corporate, political, and medial communication

Coombs (2007) defined a crisis as "a sudden and unexpected event" (p.164) that disturbs the daily operation of organizations and moreover threats their financial performance and reputation. Besides physical and emotional damage causing to people, reputational damage due to crises could be fatal to the business of organizations. Coombs (2007) proposed the SCCT (Situational Crisis Communication Theory), which empirically demonstrated how

organizations could minimize reputational damages. Accordingly, there are three types of crises (i.e. victim cluster, accidental cluster, and preventable cluster) and organizations are able to avoid the worst-case scenario by taking appropriate responses (e.g. denial, apology) that fits into the situation.

Although SCCT is a very practical approach, it is criticized to be too simple and pay far too much attention to organizations, especially in the private sectors (Schultz & Raupp,

2010; Schultz, Utz, & Göritz, 2011). It is also considered that SCCT lacks the interactive

nature of communication. Some scholars addressed the notion of sensemaking in crises, which referred to acts by organizations, seeking answers for the questions such as "what is going on?" "why it happened?" and "who is responsible?" through the interaction with others

(Maitlis & Sonenshein, 2010; Weick, Sutcliffe, & Obstfeld, 2005). During the crisis, uncertainty is one of the main components and hence organizations delivering the proper messages and the publics understanding what happened enable the society to move forward and proceed to the next step of taking actions (Gioia & Chittipeddi, 1991; Weick et al., 2005). In other words, sensemaking is considered as to bring a sense of control to the crisis, and consequently reduces the damage and reputation at stake.

The following part illustrates how different domains interact, make sense of each other, and affects to one another by introducing the concept of agenda-setting and - building.

The interplay between the media and the public: Agenda-setting

Agenda-setting concerns how agendas in the media are transferred to the public. This derives from the question that some media agendas receive great amount of attentions from the publics while others do not (Dearing & Rogers, 1996). The notable Chapel Hill study by McCombs and Shaw (1972) is considered as a classic case, which analyzed the relationship between the media and the publics, and confirmed how media coverage positively affected public agendas. This implies that when media provide news, people would be able to identify

and judge what are the crucial issues for them (Carroll & McCombs, 2003). Press "may not be successful much of the time in telling people what to think, but it is stunningly successful in telling its readers what to think about (Cohen, 1963, p.13, italics added). When the

agenda-setting process is continuously practiced, public opinions would be formed (Carroll & McCombs, 2003). The study started in the political context, yet later studies (e.g. Carroll & McCombs, 2003; Kim, Kiousis, & Xiang, 2015) included the corporate context. A study that further confirmed agenda-setting between the media salience and financial performance of companies to some extent as well as corporate reputation perceived by the publics (Kiousis, Popescu, & Mitrook, 2007).

First level of agenda-setting (Salience of objects). The idea of agenda-setting is

usually discussed at different levels. The first level concerns the salience of objects, as Kim et al. (2015) called it visibility. Salience is comprised of two aspects, attention, and prominence (Kiousis, 2004). Attention is externally based and a commonly used perspective that refers to cues such as the frequency of the story covered (Kiousis, 2004; McCombs & Shaw, 1972). Some studies found the positive correlation between news media and the public (e.g. Iyenger & Simon, 1993; Kiousis, 2004; Kiousis & McCombs 2004). Prominence concerns

"positioning of a story within a media" (Kiousis, 2004, p.74). Pages allocated to stories in the newspapers, a size of the headline, and the articles are considered as contributing to the public agendas (Carroll & McCombs, 2003). The study forms the next question using the attention approach:

RQ1: How do the Japanese media interact with the publics in terms of attention?

Third level of agenda-setting (Associative frames). In addition to the first level,

there is a second-level of agenda-setting, which relates to attributes. Attribute implies "a property, quality, or characteristics that describes an object (Kiousis, Mitrook, Wu, & Seltzer, 2006, p. 269). Unlike the first level, the second level offers more information and explanation

on what the object is about (Carroll & McCombs, 2003). Attribution has two aspects: substantive and evaluative (Carroll & McCombs, 2003; Kim et al., 2015). A typical substantive aspect is described in relation to the political campaign. The example includes ideology, opinions, and personality of political candidates (McCombs, Lopez-Escobar, & Llamas, 2000). On the other hand, an evaluative aspect includes the tone of the article as in positive, negative or neutral (Wu & Coleman, 2009).

In addition to the first level of agenda-setting (attention) and the second level of agenda-setting (attributes), the latest research started to focus on agenda-setting at the third level, also known as associative frames or network agenda-setting. This rather new concept concerns "the multiple, non-hierarchical" (Schultz et al., 2012, p.99) linkage between objects, attributes, and even the combination of both (McCombs, Shaw, & Weaver, 2014).

Researchers attempt to understand more precise picture that one builds about an event in one's head. The idea is based upon the framing theory, which Entman (1993) simply noted as "selection and salience" (p.52). This implies that there are a lot of different versions of the reality to an event, and communicators intentionally select and emphasize certain

perspectives that reflect their viewpoints for their own benefits (Birkland & Lawrence, 2009; Entman, 2003; Guo, Vu, & McCombs, 2012; Reese, 2007; Vargo, Guo, McCombs, & Shaw, 2014). When one aspect or concept is more noticeable than others, it is more likely that information receivers would be influenced by its logic. Daily piece of information may be only a part of a big picture concerning the event, but it would accumulate and affect the public gestalt over time (Vu, Guo, & McCombs, 2014).

The interplay between TEPCO and the media: Corporate agenda-building

As explained above, agenda-setting concerns the impacts of media on the publics. Agenda-building, an extending concept of agenda-setting, deals with "sources' interactions with gatekeepers" (p. 91), meaning how organizations influence on the news agendas (Ohl,

Pincus, Rimme, & Harrison, 1995). In the current study, a corporate organization's impact on the media agenda is understood as corporate agenda-building while a political organization's influence on the media agenda is referred to as political agenda-building. The influence could be various since the idea is based on different organizations carry their own unique identity, philosophy, goals and opinions so as the media do (Cornelissen, Carroll, & Elving, 2009; Walters & Walters, 1992). Therefore, when media take some information for their news coverage from certain organizations, this means the media have already embeded a certain point of view that helps shape their stories (Walters & Walters, 1992).

Corporate agenda-building has evolved from the notion of political agenda-building that is to be explained in the next part. Based on past research (e.g. Turk, 1985; Kaid, 1976), it has been assumed and verified that information subsidies (e.g., press releases, statements, PR Newswire) had positive associations with the media agendas also in the context of corporate communication (Kiousis et al., 2007; Ohl et al., 1995). Ohl et al. (1995) verified agenda-building effect in the attention of news reporting and additionally found that the same points of views in the press releases were reflected in the news articles at the time of the takeover of a company. Kiousis et al. (2007) also confirmed the same effect by comparing the public relations materials with the US news reporting. In contrast, there is also a study, which did not find the transfer of salience at the first level (Kim et al., 2015). Therefore the next research question is submitted:

RQ2: How does TEPCO interact with the media in terms of attention?

As previously mentioned, the third level of agenda-setting refers to the transfer of network issue agendas in the media to those in the publics. The third level of either corporate or political agenda-building thus deals with the transfer of network issue agendas from the organizations to the media. The study by Nijkrake, Gosselt, and Gutteling (2015) made a comparison between frames introduced by a Dutch hospital during its crisis and frames of

news coverage. The results showed the difference that the media used various frames to describe the crisis while the hospital was attached to a perspective concerning personal stories. Schultz et al. (2012) provided visualized networked graphs of both press releases by British Petroleum (BP) and news media (US and UK news) with regard to BP's oil spillage crisis. As similarities as well as differences were revealed, the study concluded that the third level agenda-building effected only to a certain extent. Accordingly, the following research question was formulated:

RQ3: To what extent do the Japanese media mirror TEPCO's associative frames? The interplay between the government and the media: Political agenda-building

The notion of agenda-building was firstly studied in the context of political communication, especially regarding the political electoral campaigns (e.g. Kaid, 1976; Kiousis et al., 2006, Lancendorfer & Lee, 2010; Walters, Walters, & Gray, 1996) Voters have been hugely depending upon how the media describe the candidates in the elections. The rapid news cycle and demand for cost-efficiency in production made the media rely on the political candidates and vice versa (Curtin, 1999; Tedesco, 2005). Research also examined political agenda-building effects between the information subsidies concerning local

legislation (e.g. Weaver & Elliott, 1985), public information offices (Turk, 1985;1986; Walter & Walter, 1992) and the news media, and the results demonstrated positive associations between them. However, the effect would vary depending upon the issues. For instance, Kaid (1976) confirmed the positive relationship between press releases of political candidates and the media coverage, but the personal information of candidates was more frequently observed than other issues. It is still difficult to grasp the definite tendency in attention. Therefore the following question is proposed:

RQ4: How does the government interact with the media in terms of attention?

Hänggli & Kriesi, 2010, Lee & Basnyat, 2013, Tedesco, 2001) provided mixed results. For example, Lee and Basnyat (2013) found both similarities and differences in the governmental documents and news coverage during the crisis. They presumed that journalists wrote news articles by using press releases as guidance. Since the relationship between the political and media narratives at the third level remains uncertain, the next question was formulated:

RQ5: To what extent do the Japanese media mirror associative frames of the Japanese

government?

The interplay between different media: intermedia agenda-setting

The notion of agenda-setting is extended to the association between different media. This is the question of how the salience of issues or associative frames can be transferred from one media outlet to others. This is called intermedia agenda-setting. Especially, the present study pays particular interest to the effect emerging between media from different countries: international intermedia agenda-setting.

Research in the area of this subject in terms of attention started with investigating relationships between different news media, for instance, the effect of major newspapers on regional newspapers and broadcast media (Denham, 2014; Golan, 2006). Recent studies examined how the Internet bulletin boards (Lee, Lancendorfer, & Lee, 2005), blogs (Sweetser, Golan, & Wanta, 2008), and Twitter (Broersma & Graham, 2012) positively correlated to news media mainly in the political concept. The most recent study by Kleinnijenhuis, Schultz, Utz, and Oegema (2015) attempted to clarify the intermedia agenda-setting in an international context. They examined whether an increase in the amount of the US news coverage resulted in the increase in the case of the BP's oil spill crisis. The results indicated that there was an international intermedia agenda-setting effect of the US news on the UK news, and not the reverse effect (Kleinnijenhuis et al., 2015).

non-Western news media is yet to be uncertain. The location of the outbreak was in Japan. In addition, more documents are usually accessible in Japanese, so this might lead overseas journalists to depend upon the Japanese media reports. Secondly, Japan is ahead of time compared to the UK. It can be presumed that the Japanese media set the agendas for the day before the British media report about the topic. On the other hand, the press club (kisha kurabu) system that the Japanese major news agencies belong to is distinctive. That is, only the members are allowed to access to certain information, through for example, attending informal briefings (Gamble & Watanabe, 2004). These authors also pointed out the

uniqueness of the Japanese media's editorial policy (fuhen futo, which means to be fair and objective). Even though this sounds positive, Gamble and Watanabe (2004) mentioned rather negatively that the policy was used "as a justification for not criticizing anything except the safest targets" (p.42). This study formulated the following question to ascertain whether the intermedia agenda-setting effect between the Japanese and the Western media exists:

RQ6: How does the Japanese media interact with the UK media in terms of attention?

At the third level, strong relationships between different media including newspapers, television news, online media, and radio was found (Vu et al., 2014). When comparing the associative frames of the American and the British media on the BP oil spill crisis, however, the evidence suggested that there were differences as well as similarities (Schultz et al., 2012). For instance, strong associations between the cause and the problem, the court and the

consequence were found in the both media. However, more concepts were linked in the US coverage. The present study aims to compare the associative frames of the Japanese and the UK news media and identify how similar or different they actually are. Accordingly, the next research question is formed:

Conceptual model

According to the aforementioned research questions and hypotheses, the study developed a conceptual model shown below (Figure 1).

Method Research design

This is a case study, which contains a longitudinal element. The present study content analyzes materials of various actors involved in the Fukushima Daiichi disaster, one of the most severe crises that happened to the post second world war era in Japan.

Case selection

The earthquake with magnitude 9.0 and the tsunami, which hit on Tohoku region on 11 March 2011, caused a series of unexpectedly critical problems to the site including hydrogen explosions and meltdowns. Consequently, the disaster changed the lives of a large number of people. Central government designated evacuation zones, and especially, the 20km radius from the plant was forbidden to entry. Radioactive releases brought concerns

about health issues as well as water and food contamination. Major events of the crisis are listed in Appendix A.

The main reasons behind the selection of this crisis are: the impact and the ambiguity of the crisis, the media criticism, and strong involvement of the political actor. Firstly, the incident was simply enormous and internationally influential. Secondly, the crisis brought a great deal of ambiguity. Media reporting from the site was physically restricted and thus the media faced the situation where they had to heavily rely on the information from the operator of the plant, TEPCO and the government. The crisis strongly required scientific knowledge, which also became a burden to all actors when delivering information that all can understand. Thirdly, both TEPCO and the government have been bitterly criticized in the media for their responses. Abraham (2011) noted in an American magazine The Atrantic,"I become more concerned about the country's fate because of the government's inability to convey accurate information". Another article noted on 13 March 2011 states, "most frightening, though, was the utter lack of transparency from the authorities" (The Japan Times, 2011a). Criticism continues even today for TEPCO's lack of preparedness (The Guardian, 2015). Fourthly, the involvement of the government is distinctive in this case. As promoting nuclear power as new source of energy was a part of its policy, the Japanese government worked hand in hand with TEPCO ever since the disaster emerged.

Data collection

Different actors involved in the Fukushima Daiichi disaster, yet this study particularly paid attention to the following four predominant actors: TEPCO, the media, the Japanese government, and the publics. The study collected and analyzed relevant English data from above-mentioned actors: press releases, news articles, governmental statements and data presented on Google Trends. The time frame for data collection was three months: 11 March to 11 June 2011. Data was retrieved during last 11 days of April 2015. The types and numbers

of documents collected for analysis are presented in Table 1.

Source Detail Documents

TEPCO Press Releases 649

Japanese media The Japan Times AJW Asahi

Total

406 288

694

British media The Guardian

The Daily Telegraph The Times The Independent Total 168 182 245 80 675 Government Statements by PM Statements by CCS Others Total Total sum 18 94 4 116 2134 Table 1.

Overview of the data collected for analysis.

Press releases by TEPCO. TEPCO is one of the most crucial actors in this crisis as

an operator of the Fukushima Daiichi nuclear power plant. This study employed press releases because TEPCO managed to disseminate them ever since the outbreak. TEPCO's social media use (i.e. Twitter, Facebook) started after May 2011, so they were judged as inappropriate in this study. English version of press releases (n=649) was retrieved from the official homepage of TEPCO where archived documents are listed chronologically under "Releases•Announcements". Except for clearly unrelated titles, all the other press releases were collected. As regards the documents on a daily plant status update, this study only used the latest plant update of the day after the first week for a practical reason. Moreover,

appendices attached to press releases, were used only when they were in the form of documents instead of tables and graphs.

News coverage. The Japanese and the UK media were selected as news media. As the

disaster occurred in Japan, the Japanese media have to take part in this research as key actors. The British media were chosen as the Western media. Nuclear disasters are closely linked to

the environmental issues, and the UK is considered as 1) keeping "the best environmental record" (p. 642) and 2) the publics and the government in the UK tend to act in favor of the environment (O'Donnell & Rice, 2008). This study thus assumed that the news coverage about the crisis would appear more in the British media than the US newspapers.

Two Japanese English-text newspapers are regarded as the Japanese media in this study: The Japan Times (political orientation: centrist, http://www.japantimes.co.jp) and Asia

and Japan Watch (AJW) Asahi (political orientation: left, https://ajw.asahi.com). The main

reason for the choice of newspapers is practicality, whether they are accessible through their webpages with monthly subscription fees but without the mandatory option of paper

subscription in Japan. Their political orientations are not strongly affected for the selection.

The Japan Times is one of the most well known English newspapers distributed in Japan.

Articles (n=406) were retrieved from the archive on the webpage. Since there was no search engine, articles were selected according to the title. When it was difficult to judge from the title, the contents were checked with the search term "Fukushima" and "nuclear". AJW Asahi is an English digital version of The Asahi Shimbun, the second biggest national paper (the circulation is nearly eight billion copies) in Japan (The Asahi Shimbun, 2015). It is known to have been writing articles against the nuclear power. Articles (n=288) were retrieved from the website with the search terms including both "Fukushima" and "nuclear."

The British newspapers in this study merely include quality papers corresponding with the selection of English-text Japanese broadsheet papers. The UK media refer to: The

Guardian (political orientation: center-left), The Daily Telegraph (right), The Times

(center-right), and The Independent (center-left). These media were selected based on their political positions and the number of available articles. All the articles (n=675) were retrieved via LexisNexis with the search terms containing both "Fukushima" and "nuclear".

statements by the Prime Minister (PM) and by the Chief Cabinet Secretary. PM is a head of the government in Japan, selected by the Diet and appointed by the Emperor. PM's official statements reflect the policies and opinions on events and issues. The Chief Cabinet Secretary serves as the voice of the cabinet. He faces the journalists daily, usually twice a day from Mondays to Fridays at the press conferences to report or make comments on relevant events (Prime Minister's Office, 2015). From the very beginning of the outbreak, the officials made statements and held press conferences concerning the crisis. Compared to PM statements (n=18), Statements by the Chief Cabinet Secretary (n=94) contain more detailed information concerning the crisis, and Q&A sessions with journalists. Four statements by the Deputy Chief Cabinet Secretary were also included in this dataset. All the statements were retrieved from the official website of the PM's Office where all the documents were chronologically archived under the section of "Reconstruction following the Great East Japan Earthquake".

Public awareness. Awareness is opposed to the term "ignorance", and if a topic or an

event does not appear on media, it implies that it does not exist as people's concern

(McCombs, 1977). The term "public awareness" is understood as "the level of attention the public devotes to objects", which could be measured by public polls (Carroll, 2004, pp. 13). Instead of public polls, this study introduced Google Trends. Although the results are limited to the Internet users, Google Trends indicates the behavior of those interested in seeking more information about an event (Kleinnijenhuis et al., 2015). Any terms could be searched

together with the location and the date (month and year). Categories of topics and search areas (e.g. web search, image search, YouTube search) can also be selected. The current study conducted a web search of all categories in Japan between March and June 2011. It also has to be noted that the numbers of search terms shown as a result range from 0 to 100. This is not an absolute number but the highest frequency is treated as 100 (Google, 2015).

Data analysis strategy

In order to analyze the collected data, the study introduced an automated content analysis using AmCAT. This open source software functions as data analysis as well as database. In addition to the efficiency of handling big datasets, AmCAT serves to identify "patterns between actors and issues" (Van Attenveldt, 2008, p. 166).

Prior to the actual analysis, the study newly created a project on AmCAT, uploaded and stored all the materials mentioned at the data collection. Then, key actors and issues within the Fukushima Daiichi crisis were identified through reading a small sample (about 10%) of aforementioned materials. Based on the exploratory analysis, Lucene search strings were built and refined. When a keyword was found in the right context, it was considered as a "hit". The study recognized eight actors and 11 issues. Eight actors are: TEPCO, the Japanese government, residents, the US, experts, international organizations, the Japanese media, and reactor makers (manufacturing companies). Issues contain six main concepts: the problem, the cause, the consequence, the responsibility, the comparison, and the solution. The consequence has four subcategories: electricity, environment, health, and new energy. The solution also consists of two subcategories: evacuation, and Fukushima plant. Since the crisis has been regarded as the world's worst nuclear disaster since Chernobyl, negative attributes were also added as a concept. See Appendix B for the finalized list of search strings.

This study conducted two types of analyses: frequency analysis, and semantic network analysis. RQ 1, 2, 4, 6 were formulated to investigate whether the attention is transferred from one medium to another. For RQ1, the study conducts linear regression to ascertain the relationship between the media and the publics. For RQ 2, 4, and 6, attention is compared between two domains in question based on the table, which demonstrates overall articles and hits. The study also compares the salience of objects during three days at four different points of time in order to observe how the increase in attention of one medium on

one day would predict another medium's reports on the next day. For instance, when corporate agenda-building effect is analyzed between 11 - 14 March, this means that

TEPCO's press releases on 11 - 13 March and news articles on 12- 14 March are compared.

Semantic network analysis. For RQ 3, 5 and 7 concerning the associative frames,

the study carries out a semantic network analysis using AmCAT, which provides networked graphics in relation to the selected concepts (Van Attenveldt, 2008). Associative frames are built on the idea of similarity judgments introduced by Tversky (1977) that a statement "a is like b" is not necessarily equivalent to "b is like a", meaning the association of two objects are asymmetrical (Schultz et al., 2012; Tversky 1977). For example, one might think about "nuclear" when the term "Fukushima" appears. However, when one firstly encounters the term "nuclear", one may not think about "Fukushima" since there are so many other aspects or words relating to nuclear. Association of a with b is called a conditional probability and written as p(b|a) = 0.86. In this case, the result tells that 86% of a probability that concept b appears when the concept a is mentioned in a randomly selected document.

AmCAT does not simply find associations of different concepts, but also visualize the strength of these associations. The strength varies from zero to one. When the conditional probability is low, an arrow between two concepts is narrow, whereas when it is strong, the arrow would be thick. To avoid the complication in the graphs, only strong conditional probabilities are visualized. The optimum threshold was selected for each medium.

Results Preliminary analysis

An overview on attention paid by different actors at Fukushima Daiichi disaster between 11 March and 11 June 2011 is provided in Figure 2. Since the last week includes the data of two days, the graph showed a drastic decrease. The study firstly observed the graphs of Japanese actors. TEPCO and the government had higher attention in the first week

compared to the other time frames (TEPCO: n=63, the government: n=14). Afterwards, both continued to be flat, especially, the government remained at a very low level. The highest peak in attention was found in the third week (n=102) for the media. After the considerable decrease in the fourth week, the attention gradually dropped and stayed flat. In spite of the long holiday from 29 April until 8 May in Japan, there was no significant effect found among Japanese actors. With regard to the British media, the attention of the first week after the outbreak is the highest as well as TEPCO and the Japanese government, even though the volume of the UK news (n=204) is far more than that of others. After the first week, a significant decrease was recognized and by the end of April, to only 11 articles a week.

0 50 100 150 200 250 2011 /3/1 1 2011 /3/1 8 2011 /3/2 5 2011 /4/1 2011 /4/8 2011 /4/1 5 2011 /4/2 2 2011 /4/2 9 2011 /5/6 2011 /5/1 3 2011 /5/2 0 2011 /5/2 7 2011 /6/3

Figure 2. Attention of each media TEPCO Japanese Media UK Media Government “Troubles occurred” “Progresses made” “More issues surfaced”

Time frames. Using the Japanese media, which showed more visible tendencies at

different point in time than other actors, the study divided three months into three different time periods for the analysis. The first time frame starts from 11 until 31 March. This period was selected because there was a decrease in the attention after its highest peak. The present study calls this period "troubles occurred". After the earthquake and tsunami, numerous

problems occurred in chains such as the hydrogen explosion, release of radioactive materials, smokes. The second time frame is between 1 April and 5 May when the attention hit the bottom in the Japanese media reporting. The study sets this period as "progresses made". During this period, TEPCO demonstrated the roadmap for the problem solving (17 April), and the area within 20 km radius from the plant was closed down (22 April). The third phase is after 6 May (until 11 June) when "more issues surfaced". Water leakage and radioactive materials in the soil at the site were detected. The president of TEPCO resigned to take responsibility for the financial loss (20 May).

Agenda-setting

RQ 1 was formulated based on the classic agenda-setting that an increase in the amount of the Japanese news reports would increase the public awareness. Since the news articles were selected based on the words "Fukushima" and "nuclear", the same words were searched on Google Trends in Japan between March and June 2011. Due to the technical reason, data from 6-10 March 2011 was included in Google Trends. Figure 3 shows the attention of the Japanese media and the term searched on Google Trends.

As both numbers are at the interval/ratio level in terms of statistics and the data is normally distributed, linear regression was conducted to ascertain the relationship of the number of news articles (independent variable) on the volume of search terms (dependent variable). Accordingly, no significant association was found between them. The study also conducted linear regression by switching the independent and dependent variables if there was any relation although this means there is an opposite agenda-setting. No significant relationship was confirmed (See Table 2). These suggest that there is no notable interaction between the media and the publics in terms of the transfer of attention.

0 20 40 60 80 100 120 2011 /3/6 2011 /3/1 3 2011 /3/2 0 2011 /3/2 7 2011 /4/3 2011 /4/1 0 2011 /4/1 7 2011 /4/2 4 2011 /5/1 2011 /5/8 2011 /5/1 5 2011 /5/2 2 2011 /5/2 9 2011 /6/5

Figure 3. Attention of the Japanese Media and Public Awareness

Japanese Media

GoogleTrends "Fukushima nuclear"

Table 2.

Interaction between the Japanese media and the public awareness (attention) Independent variable R-squared

(Model)

Regression (p)

(ANOVA) Beta (Effect) P (Significance)

Number of articles (media) .11 .25 .37 .25

Volume of search terms (public

awareness) .11 .25 .30 .25

Corporate agenda-buildin: Attention and associative frames

Attention. RQ2 concerns the first level of agenda-building, assuming that the

attention would be transferred from TEPCO to the Japanese media. According to the Table 3, which demonstrates the overview of actors and issues, the highest attention found in TEPCO was the problem (82%), followed by the government (59%), the solution for the plant (43%), TEPCO (36%), and residents (26%). On the contrary, the most frequently mentioned concept in the Japanese media is the problem (76%), which is still lower than that in TEPCO. When comparing these data, every actors and issues appear in the media while four actors and issues (international organizations, the Japanese media, reactor makers, and the consequence regarding new energy) were not mentioned at all in TEPCO. There are are nine actors and issues more frequently mentioned in the media than TEPCO: TEPCO, residents, the cause,

and the solution concerning evacuation. Only the problem was found higher in TEPCO.

TEPCO Japanese Government Japanese Media UK Media

Articles Hits % Articles Hits % Articles Hits % Articles Hits %

Actors TEPCO 235 1027 36 91 700 78 425 2399 61 220 692 33 Japanese Government 364 1251 59 115 2586 99 406 1769 59 231 565 34 Residents 166 167 26 75 117 65 258 363 37 180 247 27 US 29 54 4 34 70 29 115 185 17 66 103 10 Experts 19 39 3 53 187 46 219 527 32 181 278 27 International Organizations 0 0 0 26 52 22 48 103 7 87 158 13 Japanese Media 0 0 0 7 9 6 14 16 2 46 59 7 Reactor Makers 0 0 0 1 1 0 38 137 5 28 72 4 Issues Problem Fukushima Plant 532 4960 82 94 333 81 524 2717 76 479 2156 71 Negative Attributes 2 2 0 9 10 8 104 131 15 190 220 28 Cause 13 21 2 39 56 34 130 208 19 110 170 16 Consequence Electricity 48 88 7 16 49 14 54 116 8 54 68 8 Consequence Environment 1 1 0 37 149 32 104 422 15 83 239 12 Consequence Health 22 37 3 22 72 19 102 294 15 131 326 19 Consequence New Energy 0 0 0 21 58 18 55 131 8 94 212 14 Responsibility 117 117 18 42 76 36 57 69 8 52 52 8 Comparison 0 0 0 3 15 3 131 329 19 187 488 28 Solution Evacuation 116 119 18 95 224 82 257 355 37 254 449 38 Solution Fukushima Plant 279 1001 43 100 277 86 344 706 50 281 483 42 Table 3.

Overview of actors and issues

The present study further investigated how TEPCO's information on one day would predict the report by the media on the following day. On the basis of the overall data, the following four periods of time was chosen: 1) 20-23 March, 2) 4-7 April, 3) 14-17 May, and 4) 6-9 June. Results merely revealed that there was no strong evidence that the salience of object was transferred from TEPCO to the Japanese media on the daily basis (Appendix D). There maybe a transfer of salience of certain objects for the overall period of time, but it is not clear whether one day's press releases would affect another day's media coverage.

Associative frames. In order to answer the RQ3, semantic network analysis was

conducted on AmCAT, and consequently, associative frames of TEPCO (Figure 4) and the Japanese media (Figure 5) were created. Threshold for TEPCO is 0.75 while the Japanese media is 0.7. Conditional probabilities of both actors are presented in AppendixC.

The results provide very different network graphs for each medium. One of the

and issues related strongly at the level of p > 0.75. Surprisingly, the strongest associations found in TEPCO are related to the United States (p(TEPCO|US) = 0.99, p(Problem Fukushima Plant|US) = 0.98, p(Solution Fukushima Plant|US) = 0.97, p(Japanese

Government|US) = .93)). For example, the press release on 22 March mentioned one of the

problem-soling actions conducted with the help of the US as follows: "At 2:00 pm, March

18th, spraying water by fire engines was started with the cooperation of Self-Defense Forces and the United States Armed Forces" (TEPCO, 2011a). For some documents, TEPCO added new information to the old ones, so this sentence repeatedly appeared. On 11 April, TEPCO

also referred to the support from the States, "we have received a wide range of support and

assistance from our national government and related Ministries, [...] foreign governments and companies from various countries including United States of America [...]" (TEPCO, 2011b).

Secondly, it is worth noting that the responsibility is connected strongly only to TEPCO (p(TEPCO|Responsibility) = 0.80) such as "we would like to make our deep apologies for concern and nuisance about the incident [...]" found in the press release on 18 March (TEPCO, 2011c). However, the association of the responsibility with the government is rather low (p(Japanese Government|Responsibility) = 0.28).

The third point to note is in relation to the solution. The solution regarding evacuation connects strongly with TEPCO and with the problem (p(TEPCO|Solution Evacuation) = 0.79,

SolutionEvacuation ConsequenceHealth Consequence Electricity US ProblemFukushimaPlant TEPCO Responsibility Residents SolutionFukushima Plant JapaneseGovernment 0.76 0.76 0.78 0.87 0.82 0.89 0.80 0.85 0.79 0.97 0.93 0.99 0.98 0.85 0.91

p(Problem Fukushima Plant|Solution Evacuation) = 0.85), yet not with the government at the

level of p > 0.75. As regard to the problem concerning the Fukushima plant is strongly linked neither to TEPCO nor the government. It is rather connected to the US, residents and the problem.

In comparison with TEPCO's associative frames, the Japanese media's associative frames (Figure 5) are a lot less complex. They are composed of two clusters above p = 0.70. In the media's reports, the only strong association regarding the solution is the one with TEPCO (p = 0.71). The problem was found only in the context of comparison with other similar disasters (p = 0.70). The difference was found in terms of the responsibility. Both TEPCO and the government are connected to the responsibility strongly (p(Japanese Government|Responsibility) = 0.75, p(TEPCO|Responsibility) = 0.75). The quote from the crisis, national policy minister Genba was reported on 30 March: "'Since the government has been promoting nuclear energy as official policy, it is necessary for it to [...] take

responsibility,' Genba said, indicating the government will step in [...]" (The Japan Times, 2011b). On another day (14 April), an article referred to the apology made by TEPCO: "TEPCO President Masataka Shimizu offered apologies [...]" (The Japan Times, 2011c).

Political agenda-building: Attention and associative frames

Attention. RQ3 was proposed based on the idea that the salience of objects in the

Figure 5. Associative frames in the Japanese media (threshold .70)

Responsibility Comparison ProblemFukushimaPlant JapaneseGovernment TEPCO SolutionFukushima Plant 0.70 0.71 0.75 0.75

government would be transferred to the Japanese media. According to Table 3, all the actors and issues were at least once covered in the governmental documents. The highly frequently appeared concepts are: the government (99%), the solution (Fukushima plant: 86%,

evacuation: 82%), the problem (81%), and TEPCO (78%). However, none of the

above-mentioned concepts emerged more in the Japanese media reporting. The only concepts mentioned more in the Japanese media are following two: negative attributes (the Japanese media 15%, the government 8%), and the comparison (the Japanese media 19%, the government 3%). Political agenda-building is rather recognized as a reverse effect.

Observation of the daily tendencies in attention did not provide sufficient evidence in favor of political agenda-building effects. The study looked into the following four time periods: 1) 23-26 March, 2) 4-8 April, 3) 10-13 May, and 4) 23-26 May. Although objects mentioned in the governmental statements mostly appeared in the media, there was no indication that the governmental documents intensified the agendas in the media. See

Appendix D that demonstrates transfer of attention between the government and the media.

Associative frames. Semantic network analysis was conducted on AmCAT in order

to answer the research question 3. Threshold for the Japanese media is 0.70 while that for the government is 0.97. See Appendix C for the conditional probabilities. The results

Figure 6. Associative frames in the Japanese government (threshold .97)

Japanese Government Comparison ConsequenceEnv JapaneseMedia InternationalOrg Responsibility ReactorMakers US 1.00 1.00 1.00 1.00 1.00 0.99 0.99 0.99 0.98 0.97 Problem Fukushima Plant Solution Evacuation TEPCO

demonstrate dissimilar network graphs as illustrated in Figure 4 and 6. Associative frames in the Japanese government demonstrate 1) complexity, as they include 11 actors and issues, and 2) overall strong associations between concepts. Five associations are perfectly related, the Japanese government with the media, comparison, international organizations, and reactor makers as well as the problem with reactor makers. At one of the press conferences on 17 May, for instance, the Chief Cabinet Secretary stated concerning the fact-finding mission by IAEA (International Atomic Energy Agency): "As the IAEA is an internationally authorized body on nuclear power, I believe that the Cabinet will seek its involvement in the

investigation [...]. However, this will be subject to further consultation with the IAEA once the Government incident investigation committee has been launched (Prime Minister's Office, 2011a). All the concepts connect to the government at the level of p > 0.92 in accordance with the conditional probability table (Appendix C).

The third point is that the responsibility is strongly associated with the government (p(Japanese Government|Responsibility) = 0.97). For example, the government repeatedly referred to its accountability on 30 May, "it is natural that Government takes responsibility for all areas of governance. While it is the Government that has responsibility, [...] then it is the responsibility of the Government to correct and reform [...]"(Prime Minister's Office, 2011b). The conditional probability table shows that the concept is also linked with TEPCO,

but much lower in the present narrative (p(TEPCO|Responsibility)= 0.85).

When it is compared to the associative frames in the Japanese media, the differences stand out. The graphic itself is less complicated and overall strength in association is lower than the political narrative. Unlike the governmental statements, the strength of associations between actors and issues and the government are weaker, ranging from p = 0.39-0.75. With regard to the comparison, it was very strongly connected to the government in the political narrative (p = 1.00). On the other hand, the comparison was strongly related to the problem (p

= 0.70) in the media. The same link also appeared in the government, almost at the same level (p = 0.68). As previously mentioned at the corporate agenda-building, the difference found in the media is that the responsibility was strongly linked with both TEPCO and the government at the same level of strength (p = 0.75).

Intermedia agenda-setting: Attention and associative frames

Attention. RQ6 concerns the intermedia agenda-setting. In accordance the Table 3,

the Japanese media report actors and issues extensively. The highest percentage of the concepts covered is the problem (76%) followed by TEPCO (61%), the government (59%), the solution for the plant (50%), and residents and the solution for evacuation (both 37%). On the other hand, the overall attention in the UK media is lower than the Japanese media. The highest attention was found in the problem (71%), and the following actors and issues a lot lower than the problem: the solution for the plant (42%) and for evacuation (38%), the government (34%) and TEPCO (33%). Comparison revealed that there are nine actors and issues found more in the UK media, including international organizations, the Japanese media, negative attributes, and the consequence regarding health and new energy.

Daily tendencies in attention also support the results deducted from the Table 3 to some extent. Although there was no apparent trend in the light of salience of actors, but two issues (negative attributes and the consequence regarding new energy) were mentioned more in the UK media compared to the Japanese. Appendix D demonstrates all the tables of the following period 1) 15-18 March, 2) 28-31 March, 3) 24-27 May, and 4) 6-9 June.

Associative frames. For RQ 4, semantic network analysis was conducted on AmCAT.

Accordingly, Figure 4 and 7 demonstrate the network graphs of the Japanese (threshold 0.7) and the UK media (threshold 0.75) respectively. The conditional probabilities are listed in the Appendix C. As previously mentioned, the Japanese media contains double clusters in the associative frames. On the contrary, the complexity of associative frames in the UK media is

prominent. Another distinctive feature is that the associative frames are unidirectional, meaning all the concepts (nine actors and issues) at p > 0.75 direct to the problem. The strongest association is observed between the US and the problem (p = 0.81). Almost three months after the outbreak (8 June), one of the British news articles still reports the concern over the problem as follows: "The amount of radiation released by the Fukushima Daiichi

nuclear power plant [...] could have been more than double that originally estimated [...],

Japan's nuclear safety agency said. It is not clear whether the revised account concerning the Fukushima meltdown, the world's worst nuclear accident since Chernobyl in 1986 [...]" (The Guardian, 2011).

As seen in the Japanese media narrative, the comparison was linked with the problem and it was also seen almost at the same level in the UK media. In contrast, the association regarding the responsibility shows the dissimilarity between the Japanese media and the UK media. Although the responsibility is associated with both TEPCO and the government in the Japanese media (p(TEPCO|Responsibility)= 0.75, p(Japanese Government|Responsibility)= 0.75), the issue does not link strongly to neither TEPCO nor the government in the UK news articles (p(TEPCO|Responsibility)= 0.08, p(Japanese Government|Responsibility)= 0.07).

Additional analysis: The overall interaction among Japanese actors over time

One of the main research questions established in the beginning of this study was to ascertain the change of interaction over time. Since the concept of time was not included in

Figure 7. Associative frames in the UK media (threshold .75)

US ProblemFukushimaPlant Comparison JapaneseMedia ConsequenceEnv InternationalOrg Experts SolutionEvacuation Consequence Health Japanese Government 0.81 0.75 0.75 0.77 0.77 0.77 0.79 0.79

the research questions formed in the theoretical background, the additional analysis was conducted. All the Japanese actors (i.e. TEPCO, the Japanese media, and the government) were treated as one domain that set agendas and a semantic network analysis was conducted to examine how the interaction leads to frame agendas in different time periods classified in the preliminary results. Threshold for the associative frames in the first period is 0.8, and the rest are 0.7. Conditional probabilities for each period are presented in the Appendix C.

"Troubles occurred" period. Associative frames of the first period (11 - 31 March)

are presented in Figure 8. The frames are unidirectional as all the actors and issues above the threshold connected to the problem. Especially, the solutions concerning evacuation,

international organizations, and experts, have the same strong association with the problem (p=0.85) For instance, the press release on 13 March stated: "Re-activation of Reactor Core Isolation Cooling System was attempted but failed, [...] unable to confirm the level of water injection to the reactor [...]" (TEPCO, 2011d). An article on 16 March also described the situations at the nuclear power plant: "Officials of TEPCO [...] said the white smoke from the No.3 reactor was probably evaporating water from the storage pool for nuclear fuel rods" (AJW Asahi, 2011a).

Progresses made" period. The second time frame is between 1 April and 5 May.

Figure 8. Associative frames during “troubles occurred” period” (threshold: 0.8)

ProblemFukushimaPlant Responsibility ConsequenceEnv Comparison JapaneseMedia US Experts SolutionFukushimaPlant ConsequenceHealth TEPCO InternationalOrg SolutionEvacuation Residents 0.83 0.84 0.84 0.84 0.81 0.83 0.83 _0.85 0.85 0.85 0.83 0.83

During this period, Figure 9 shows simpler associative frames in comparison to the first phase, containing three clusters of associative frames. The first cluster concerns the responsibility: when responsibility is mentioned, there is strong association to TEPCO

(p(TEPCO|responsibility) = 0.73). The second one shows that the probability of appearing the problem is high when the solution regarding the plant and reactor makers are mentioned (p(Problem Fukushima Plant|Reactor Makers) = 0.70, p(Problem Fukushima Plant|Solution Fukushima Plant) = 0.72). The political concepts emerged in the third cluster, with the association of the government and the US p(Japanese Government|US) = 0.71).

"More issues surfaced" period. The third period starts 6 May and ends 11 June.

Figure 10 shows a single associative frame, which demonstrates the strong link between the issue of responsibility and the actor TEPCO (p(TEPCO|Responsibility) = 0.82). The

association was, for example, observed in the context of the head of TEPCO's resignation on 20 May: "TEPCO President Masataka Shimizu, 66, plans to resign to take responsibility for the utility's handling of the accident" (AJW Asahi, 2011b).

Figure 9. Associative frames during “progresses made” period (threshold: 0.7)

US JapaneseMedia JapaneseGovernment 0.71 0.72 Responsibility TEPCO 0.73 ReactorMakers ProblemFukushimaPlant Solution FukushimaPlant 0.72 0.70

Figure 10. Associative frames during “more issues surfaced” period (threshold 0.7)

Responsibility

TEPCO 0.82

Additional findings

The present study takes particular note of the responsibility that showed interesting results. Categorized as one of the issues, responsibility demonstrated strong associations with TEPCO and the government among Japanese domains. Within TEPCO's narrative, 18% of the press releases mentioned about responsibility, and it was strongly connected to TEPCO (p=0.80) as shown in Table 3 and Figure 5. On the other hand, according to Table 3 and Figure 6, the government mentioned 36% of its documents about responsibility, which strongly linked to the government itself (p =0.97). The Japanese media did not mention responsibility as frequently as other Japanese actors, but it was linked to both TEPCO and the government. Moreover, the UK media, which remarked on responsibility as much as

Japanese media did (8%), showed association of the responsibility with TEPCO (p =0.55), and the government (p =0.46)

Discussion Summary of findings and theoretical implication

The present study investigated how actors interact with each other via communication in Fukushima Daiichi disaster, and three sub research questions were formulated. The first sub question concerned the Japanese media's interaction in relation to attention and

associative frames. RQ1 dealt with the agenda-setting between the media and the public awareness, but the statistical analysis did not confirm the transfer of attention from the media to the publics. The reverse effect was not found either. RQ2 and 3 were related to corporate agenda-building. There seemed to have the media's influence on the first level, but limited to few concepts. Daily tendency was not clearly found. Associative frames demonstrated how different concepts were stressed in these two media outlets, since the same concepts found in the graphs were related to other concepts in a different degree of strength. RQ4 and 5

was not detected, and it rather seemed to have a reverse effect. Associative frames in the government hugely differ from those in the media, especially in terms of their strength.

Surprisingly, these results did not confirm clear agenda-setting and -building effects in terms of attention compared to previous studies, except for Kim et al. (2015) which did not find corporate agenda-building effect at the first level. As for agenda-setting, this maybe the implication that media audience are not simply affected by the media reports but also from other flow of information such as bloggers, television hosts, or someone who could bring information which fits to their logic (Vargo, et al., 2014) At least, for agenda-building effects, it was overall recognized that the media mentioned the topics that both TEPCO and the government covered. It can be estimated that journalists build stories on the basis of the information and the data they were distributed from the organizations. This is in line with the reasoning suggested by Lee and Basnyat (2013). In terms of the relationship between the media and the government, a tendency of a reverse effect was observed. One reason could be considered that the government's statements contained Q&A sessions. Journalists may well ask questions that relate to their own ideas for articles, and the responses from the officials appear more likely in their final outputs.

The second sub research question concerned the agenda-building effects over time. AmCAT results demonstrated dissimilar associative frames for each period that this study classified. Various actors and issues were strongly connected to the problem during the first period. However, in the end of the third period, the only strong association recognized was the one between TEPCO and the responsibility.

The third sub question dealt with the intermedia agenda-setting between the Japanese and the British media at both the first and the third level. The issue regarding the attention was analyzed in RQ6. The overall table indicated the intermedia agenda-setting effect in attention was only found to few actors and issues. Transfer of attention was partially

recognized when looking into the daily tendencies. One issue was regarded as attributes, which is agenda-setting effect at the second level. In terms of associative frames (RQ7), the findings suggest the differences more than similarities. While the UK media contain many concepts which all directs to the problem, associative frames of the Japanese media were limited to the responsibility, the solution, and the problem. As previous studies pointed out, both the first and the third level of intermedia agenda-setting effects were only found to a certain extent (Kleinnijenhuis et al., 2015, Schultz, et al., 2012). At the third level, as argued in the theoretical background, the difference of the media culture between two countries may have contributed to form different issue networks. Otherwise, news values may be still greatly influenced by the geographical proximity to an event (Ruigrok & Van Attenveldt, 2007).

All in all, the present study displayed the unique communicative interaction between different actors in the case of Fukushima Daiichi disaster. From the point of classic

agenda-setting and -building effect, the study did not recognize the obvious transfer of objects both in frequency analysis and simple day-to-day comparison between two actors. However, research results differentiated contextual information that each medium provides and how the interplay changes overall agendas over the period of three months.

Limitations and suggestions for the future research

Despite the contribution made by the present study, limitations also have mentioned. Firstly, research sorely examined English materials in order to use AmCAT instead of manual coding which restricts the amount of data analyzed. Secondly, the selection of news media was limited owning to the limited choice of English-text Japanese articles. The problem embedded in this point is that it is not clear whether English contents are the true reflection of what was discussed in Japanese, as Tollefson (2014) also questioned in his paper. Thirdly, AmCAT itself may have a deficit in presenting associative frames. For example, a strong

connection between reactor makers and few concepts in the governmental documents, yet reactor makers were mentioned only in one article. As AmCAT is able to weigh the number of hits, it is difficult to explain why this emerged.

In order to overcome these limitations, the present study suggests the following change and inclusion to the future research. Firstly, Japanese-text materials as well as wider variety of news media should be included for the analysis. For instance, the public narratives in the social media such as Twitter and Facebook in Japanese should also be analyzed. In terms of news media, tabloids, regional and economic newspapers, magazines and even television news programs would contribute to grasp more balanced view on agendas in the media. By using above-mentioned documents, the combination of computer-assisted and manual content analyses would be a beneficial method. In terms of investigating the salience of objects, conducting time series analysis is encouraged to strengthen findings in the present study, which mostly depended upon Table 3. Lastly, longer-term observation of this crisis is suggested, since it is ongoing and has a possibility of prpovididng more agendas when the reactor decommissioning process progresses. This would be a beneficial remark for

organizations both in private and public sectors to ascertain how their responses would meld in to become agendas and reach to the publics.

Practical implications

Fukushima Daiichi disaster was an event that the society never foresaw. Failure to provide solutions due to technological deficits and poor preparedness itself contains a lot of problems. This study suggests the problem deriving from a communication perspective. One of the distinguished findings in this study was related to the responsibility frame that how strongly it appeared together with TEPCO and the government. The fact that all the actors resulted in providing sorely this frame, especially within three months after the outbreak, may have contributed to blind people with everything else that happened around the disaster.

References

Abraham, D. (2011, March). How bad communication is undermining Japan's crisis response.

The Atrantic, Retrieved 17 June 2015 from

http://www.theatlantic.com/international/archive/2011/03/how-bad-communication-is -undermining-japans-crisis-response/72573/

AJW Asahi (2011a). High levels of radiation at stricken reactor, 16 March. AJW Asahi (2011b) Tepco downing in red ink, loses 1.25 trillion yen, 20 May. Birkland, T. A., & Lawrence, R. G. (2009). Media framing and policy change after

Columbine. American Behavioral Scientist, 52(10), 1405-1425.

Broersma, M., & Graham, T. (2012). Social media as beat: Tweets as a news source during the 2010 British and Dutch elections. Journalism Practice, 6(3), 403-419.

Carroll, C. E. (2004). How the mass media influence perceptions of corporate reputation:

Exploring agenda-setting effects within business news coverage (Doctoral

dissertation). University of Texas, Austin.

Carroll, C. E., & McCombs, M. (2003). Agenda-setting effects of business news on the public's images and opinions about major corporations. Corporate Reputation Review,

6(1), 36-46.

Cohen, B. C. (1963). The press and foreign policy. Princeton, New Jersey: Princeton University Press.

Coombs, W. T. (2007). Protecting organization reputations during a crisis: The development and application of situational crisis communication theory. Corporate Reputation

Review, 10(3), 163-176.

Coombs, W. T., & Holladay, S. J. (2002). Helping crisis managers protect reputational assets: Initial tests of the Situational Crisis Communication Theory. Management

Communication Quarterly, 16(2), 165-186.

Cornelissen, J. P., Carroll, C., & Elving, W. J. L. (2009). Making sense of a crucial interface: corporate communication and the news media. In C. Chouliaraki & M. Morsing (Eds.), Media, organisation and identity (pp. 1-22). Hampshire: Palgrave McMillan.

Curtin, P. A. (1999). Reevaluating public relations information subsidies: Market-driven journalism and agenda-building theory and practice. Journal of Public Relations

Research, 11(1), 53-90.

Dearing, J. W., & Rogers, E. M. (1996). Agenda-setting. Thousand Oaks; California: Sage Publication.

Denham, B. E. (2014). Intermedia attribute agenda setting in the New York Times: The case of animal abuse in U.S. horse racing. Journalism & Mass Communication Quarterly,

91 (1), 17-37

Du, Y. R. (2012). Intermedia agenda-setting in the age of globalization: A multinational agenda-setting test. Global Media and Communication, 9(1), 19-36.

Entman, R.M. (1993). Framing: Toward clarification of a fractured paradigm. Journal of

Communication, 43(4), 51-58.

Entman, R.M. (2003). Cascading activation: Contesting the White House's frame after 9/11.

Political Communication, 20(4). 415-432.

Gamble, A., & Watanabe, T. (2004). A public betrayed: an inside look at Japanese media atrocities and their warnings to the West. Washington DC: Regnery Publishing. Gioia, D. A., Chittipeddi, K. (1991). Sensemaking and sensegiving in strategic change

initiation. Strategic Management Journal, 12, 433-448.

Golan, G. (2006). Inter-media agenda setting and global news coverage. Journalism Studies,

Google, About Trends graphs (2015) retrieved 14 June from

https://support.google.com/trends/answer/4355164?hl=en&rd=1

Greenpeace (2012). Lessons from Fukushima. Retrieved on 24 June 2015 from

http://www.greenpeace.org/international/Global/international/publications/nuclear/20 12/Fukushima/Lessons-from-Fukushima.pdf

Guo, L., Vu, H. T., & McCombs, M. (2012). An expanded perspective on agenda-setting effects: Exploring the third level of agenda-setting. Revista de Comunicación, 11, 51 68. Retrieved 15 March 2015 from

http://udep.edu.pe/comunicacion/rcom/pdf/2012/Art051-068.pdf

Hänggli, R., & Kriesi, H. (2010). Political framing strategies and their impact on media framing in a Swiss direct-democratic campaign. Political Communication, 27(2), 141-157.

Independent Investigation Commission on the Fukushima Nuclear Accident (2012). Chosa

Kenshou Houkokusho [The Fukushima Daiichi nuclear power station disaster:

Investigating the myth and reality]. Tokyo: Discover 21.

Iyengar, S., & Simon, A. (1993). News coverage of the Gulf crisis and public opinion: A study of agenda-setting, priming, and framing. Communication Research, 20(3), 365-383.

Kaid, L. L (1976). Newspaper treatment of a candidate's news releases. Journalism Quarterly,

53(1), 135-137.

Kim, J. Y., Kiousis, S., & Xiang, Z. (2015). Agenda building and agenda setting in business: Corporate reputation attributes. Corporate Reputation Review, 18(1), 25-36.

Kiousis, S. (2004). Explicating media salience: A factor analysis of New York Times issue coverage during the 2000 US presidential election. Journal of Communication, 54(1), 71-87.

Kiousis, S., McCombs, M. (2004). Agenda-setting effects and attitude strength - Political figures during the 1996 presidential election. Communication Research, 31(1), 36-57. Kiousis, S., Mitrook, M., Wu, X., & Seltzer, T. (2006). First- and second-level

agenda-building and agenda-setting effects: Exploring the linkages among candidate news releases, media coverage, and public opinion during the 2002 Florida

gubernational election. Journal of Public Relations Research, 18(3), 265-285. Kiousis, S., Popescu, C., & Mitrook, M. (2007). Understanding influence on corporate

reputation: An examination of public relations efforts, media coverage, public opinion, and financial performance from an agenda-building and agenda-setting perspective.

Journal of Public Relations Research, 19(2), 147-165.

Kleinnijenhuis, J., Schultz, F., Utz, S. & Oegema, D. (2015). The mediating role of the news in the BP oil spill crisis 2010. Communication Research, 42(3), 408-428.

Lancendorfer, K., & Lee, B. (2010). Who influences whom? The agenda-building relationship between political candidates and the media in the 2002 Michigan Governor's race. Journal of Political Marketing, 9(3), 186-206.

Lee, S. T., & Basnyat, I. (2013). From press release to news: Mapping the framing of the 2009 H1N1 A influenza pandemic. Health Communication, 28(2), 119-132. Lee, B., Lancendorfer, K. M., & Lee, K. J. (2005). Agenda-setting and the internet: The

intermedia influence of internet bulletin boards on newspaper coverage of the 2000 general election in South Korea. Asian Journal of Communication, 15(1), 57-71. Lippmann, W. (1922). Public Opinion. New York: The Macmillan Company.

Maitlis, S., & Sconenshein, S. (2010). Sensemaking in crisis and change: Inspiration and insights from Weick (1998). Journal of Management Studies, 47(3), 551-580.

McCombs, M. (1977). Agenda setting function of mass media. Public Relations Review, 3(4), 89-95.