Framing terrorist attacks: A multi-proximity model

Abstract

Guided by the perspective of media framing theory and a newly proposed multi-proximity model, we analysed the content of news reports on five major terrorist attacks from four international news media. We identified three frame packages, each with four functional frames suggested by Entman. Regression analysis revealed the differential impact on media frames of multiple proximities in ideology, religion and bilateral relations. When the media host country and victim country are bound by close relations, the news frames applied to such attacks will be more negative. The same is true for religious proximity between the media host country and victim country. In contrast, close religious proximity between the media host country and the attacker group will lead to the adoption of a less negative news frame for an attack. Interestingly, the proximity of ideology between the media host country and victim country did not show the expected explanatory power for the media frames.

Keywords

Comparative studies international terrorism attack news frames proximity

Introduction

Global terrorism has become a prominent component of the daily agenda of the public and the media over the past decade. Because of the controversial nature and wide socio-cultural and political significance of terrorism, scholars have become increasingly concerned with how it is portrayed in mainstream media outlets since the September 11 and subsequent international terrorist attacks (Jackson and Sinclair, 2013).

Much of this scholarship has focused on the point that terrorism is a worldwide phenomenon, and there are enormous regional differences in how what are designated terrorist activities are presented, framed, understood and evaluated across different regions and, by extension, media outlets (Norris et al., 2003). Evidence suggests that an attack that is defined as a terrorist attack in one country may be described and categorized in a completely different way in another country or region (Li and Li, 2015; Medvedeva and Hinnant, 2011; Miao, 2014).

Previous research has tended to attribute media coverage of terrorism to cultural clashes and compared the differences between Arab and Western countries (Yarchi et al., 2013). While such comparisons have been useful to some extent, they generally have not dealt with the more general issues of how and why terrorism is identified, produced and represented in the news media. Our study attempts to provide a wider spectrum for terrorism report studies by using representative media and cases in an attempt to analyse to what extent international media frames are influenced by the interactions of multiple factors such as ideology, religion and international relations. By using terrorism attacks as a case study, this project intends to contribute to the knowledge of framing studies specifically with regard to international political communication within the background of across-media systems and across-country reporting of issues.

Entman's definition of media frames

In the 1980s, frames were described as ‘schemata of interpretation’ or ‘interpretive packages’ (Gamson and Modigliani, 1989; Goffman, 1974: 21). An interpretive package includes a central organizing idea (the frame) and a number of condensing symbols that often refer in shorthand to the frame and its position (Gamson and Modigliani, 1989).

Since the 1990s, Entman's definition of framing has been the one most frequently used (Chen and Yang, 2015; Matthes, 2005), and his work had a powerful catalytic effect on the field over the next 30 years (Scheufele and Iyengar, 2012). According to Entman (1993), framing is defined as selecting ‘some aspects of a perceived reality and making them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described’ (p. 51; Entman, 2003). In this definition, Entman suggests four sub-frame elements (issue definition, causation, moral judgement and remedy suggestions), which can be applied in empirical tests.

Previous empirical studies were operationalized mainly by some of these four elements. For example, Brownell et al. (2010) studied the frame of responsibilities and solutions in controversial issues. Kim and his colleagues emphasized the sub-frames of issue definition, causation and solutions for the issue of illegal immigration but omitted the sub-frame of moral judgement, thus rendering the frame package incomplete. This article attempts to operationalize Entman's four sub-frame elements into an ideally coded framing package (Kim et al., 2011).

Proximity, news value and media frames

Because of its controversial nature and its wide socio-cultural and political significance, international terrorism has attracted a considerable body of scholarly analysis. However, there is no consensus on the definition of terrorism; additionally, the process of news selection is also more complex than a simple relationship between attributes of events and coverage, and therefore, news media in different countries tend to cover the same news event from different viewpoints (Gerhards and Schäfer, 2014; Jackson and Sinclair, 2013; Weimann and Brosius, 1991).

Relating our research to the findings in existing studies, we found that news value theory might provide an explanation for this phenomenon. Particularly, in international terrorism coverage, the proximity to the event is one of the important factors of news value and a major determinant of what frame is preferred in the stories (Cotter, 1999; Gerhards and Schäfer, 2014; Schaefer, 2003).

Previous scholars of terrorism news coverage studies have discussed the importance of proximity by analysing the differences in coverage between local and foreign milieus and emphasizing country-specific or regional differences in media coverage that led to different media frames within a global media context (Medvedeva and Hinnant, 2011; Zhang et al., 2013). However, the proximity factor in news value theory does not necessarily limit to geographical closeness. Events in countries with which the home country media source has a particular bond or similarity can have the same effect on the selection of news frames (Cotter, 1999). In the case of the terrorism events, the proximity element is a factor that can be weighted differently according to where this event happens. For example, the proximities of ideology, bilateral relations and religions have vary depending on where the event occurs, and where it is reported (Yousaf, 2015). We will examine each of these proximity predictors separately in the following discussions.

Ideological proximity

Among the factors in predicting a news frame, ideology is the most influential. According to scholars like Campbell, Buzan and Raza, much of the analysis in this vein examines the representation of terrorism and counter-terrorism through the lens of specific ideologies, such as ‘orientalism’ or ‘cold war’, between the Western and Eastern world, discusses the unequal power between political opponents or analyses the dominant ideological influences on journalists (Buzan, 2006: 1101; Campbell, 2017: 11; Raza et al., 2012).

For example, the U.S.-backed Western news media portrayed the 9/11 events as ‘terrorist’ attacks against democracy and described the following ‘war on terror’ as ‘rational’ vs. ‘irrational’ and ‘modernity’ vs. ‘anti-modernity’ (Campbell, 2017: 11; Norris et al., 2003: 6; Nossek and Berkowitz, 2006). In contrast, they were reluctant to define an attack that occurred in a ‘rival’ country as terrorism (Li and Li, 2015: 139). For example, the railway station bombing in China was portrayed as oppression caused by the government's crackdown on Uighur separatists, which led to the persecution of innocent civilians and, in turn, fuelled the separatists’ cause (Miao, 2014). While the U.S. news media might decline to designate the Chinese incidents as terrorism and criticize the autocratic regime by using the fight against terrorism as a cover for their own internal crackdowns on political opponents, they quickly labelled the Paris shootings in 2015 as an ‘act of terrorism’ and expressed global solidarity in fighting terrorist attacks (Buckley and Fawn, 2003: 203; Fu, 2015: 46; Nail, 2016; Wilson and Piazza, 2013; Wilson, 2015).

Hence, some Eastern scholars argue that the Western media's coverage of terrorist attacks in non-democratic countries is ‘hypocritical’ and ‘irresponsible’ (Li and Li, 2015: 139). These scholars criticize the Western news media for applying one set of discourse to domestic attacks in their countries but another set when similar incidents occur in the territories of their political opponents; scholars such as Medvedeva and Hinnant refer to this phenomenon as a ‘double standard’ (Medvedeva and Hinnant, 2011; Miao, 2014: 29; Sun, 2014: 145).

The existing literature on terrorism coverage indicates that the root of this argument lies in the division between ‘democratic’ and ‘autocratic’ and the unequal power between Western and Eastern countries, with Western democratic countries regarding their counterpart governments as regimes with fewer human rights and fewer restrictions on administrative power and also as playing a formidable role in news media systems (Aksoy et al., 2012: 3; Wilson, 2015; Wilson and Piazza, 2013: 941). In contrast, Eastern countries regard the U.S.-backed rhetoric about democracy as an excuse for furthering Western hegemony and perceive attacks such as the events of 9/11 as ‘counter-hegemonic political violence’ against the West (Butko, 2005: 21).

Evidence indicates that although the world has undergone rapid changes, such ideological divides between different regimes remain essentially the same (Xin, 2010). When journalists cover stories, how they frame the ‘others’ will be congruent with the dominant ideology of their society (Lee and Wang, 2016: 1908; Norris et al., 2003). In previous studies, scholars have tested a few specific ideological indexes, such as political rights and civil liberty, to quantitatively measure this divide between Western and Eastern countries (Yarchi et al., 2013). In this way, they have provided necessary empirical support for testing the predictive power of ideological proximity in news frames.

Religious proximity

In addition to ideology, communications scholars have identified religion as a salient indicator that can affect media coverage, especially with regard to terrorism activities. Scholars have hypothesized that the intrinsic religiousness of journalists helps to organize the dominant frames in terrorism-related coverage (Powell, 2011). Because journalists use religion to distinguish between different ethnic groups and cultures, they particularly regard Muslims/Arabs/Islamists as possible participants in organized terrorist cells that act against other groups (Powell, 2011). For example, studies that focus on violent events indicate that journalists tend to equate Arabs with Muslims and to assume that Islam is a religion that sponsors terror and that terrorists oppose a Christian-dominated society such as the United States (Chuang and Robin, 2014; Nacos, 2016; Powell, 2011).

These ideas about groups, religions and populations are difficult to change, Western media can use these stereotypes to apportion blame across the whole of Islam and frame it as orthodox, barbaric, evil and opposed to any kind of modernity or freedom (Bhatia, 2008; Karim, 2003; Said, 1994). Based on previous studies, it is apparent that the Western news media routinely construct Muslims as the ‘others’, whereas Christians belong to ‘our’ group (Bhatia, 2008; Karim, 2003: 196; Said, 1994). Thus, the effect of religious proximity is prominent among journalists from Christian countries.

Meanwhile, many Islamic scholars have also analysed this issue (Nacos, 2016; Nacos and Torres-Reyna, 2005; Powell, 2011). Their studies tend to support the attackers and criticize the Western countries’ biased attitudes towards the issues of the Arab world. For example, scholars found that the Muslim news organization Al Jazeera tended to show pity for the victims in the attacks as innocent civilians and defined the 9/11 attacks with the frames ‘tragedy’, ‘crime’ and ‘humanitarian’ rather than ‘military’ or ‘war’ (Jasperson and El-Kikhia, 2003: 127). The following international ‘war on terror’ was a ‘hostile invasion’ that brought the battlefield into countries such as Iraq and was met with ‘massive resistance’ (Gerhards and Schäfer, 2014: 14).

Current studies show that Islamic attacks are more likely to be described as terrorism in the Western news media, especially after the 9/11 events; the designation ‘Arab’ or ‘brown’, ‘once the signifier of an exoticism’, now represents a negative ‘other’ (Semati, 2010: 253). Meanwhile, Islamic news media refuse to relate their religious group to ‘terrorists’ and place more emphasis on the ‘humanistic’ perspective of such stories, asserting that criminals and extremists cannot represent the Muslim population as a whole (Atran, 2003: 257).

Bilateral relations proximity

Beyond considering the influence of ideology and religion in portraying violence, scholars of framing effects note that it is equally important to consider the dynamic factor of bilateral relations in predicting trends in news coverage (Lecheler et al., 2009). For instance, Yousaf (2015) analysed terrorist attacks that occurred in Pakistan, which has had a strong bilateral relationship with China for 60 years, sharing economic ties and strategic alliances and maintaining close military relations. Thus, the news frames of Chinese news media tend to be more critical of terrorist attacks in Pakistan, representing Pakistan as a ‘victim’ and a ‘sufferer’ from terrorist attackers and positively supporting the counter-terrorism and security strategies of Pakistan's government. Not only were all news stories associated with Pakistan under the international relations frame positively reported in Chinese news media, but Pakistan was also praised as an ‘all-weather strategic cooperative partner’ of China, one of the strongest of its allies (Yousaf, 2015: 3058).

In comparison, Pakistan's relationship with the United States is based on interests rather than trust and is limited to cooperation and collaboration in the war against terrorism. News related to terrorism is often framed by the U.S. news media in a way that presents Pakistan as responsible for attacks in the region. For example, the frames ‘outside threat’, ‘Islamic militancy’, and ‘harbouring terrorism’ were adopted for the country of Pakistan in such coverage (Yousaf, 2015: 3058).

Additionally, it is worth noting that while the dominant ideology and religion are slow to change, bilateral relations between two countries or countries within regions are relatively dynamic. For example, the foreign policies of the United States have undergone a series of changes since 11 September 2001 (Nacos, 2016; Norris et al., 2003). When the U.S. media reported on the events of September 11 and the aftermath, President George W. Bush adopted the ‘war on terror’ frame as the primary standard for reinterpreting and understanding ‘friends’ and ‘enemies’ around the globe (Norris et al., 2003: 15).

In several respects, the 11 September 2001 terrorist attacks led to improvement in U.S.-China relations. As a result, the former ideological ‘rivals’ became economic ‘competitors’, and the frame used by the U.S. news media for China shifted from ‘illegality’ to ‘irresponsibility’ in the international environment (Wang, 2015: 281). Even though the U.S. news media still criticize China and present it as an authoritarian state, the frames used to describe it have become less overtly negative and have also taken on a different kind of emphasis.

These findings suggest that the process of news selection is more complex than a simple relationship between the attributes of events and coverage of them. Coverage of external or transnational issues, such as international terrorism, is often moulded by journalists with a more specific domestic view by considering the proximity of ideology, religion and bilateral relations.

The integration model of multi-factor proximity

As summarized above, there has been considerable scholarly investigation of fields that may influence media frames. However, some limitations still remain. For example, there is little empirical support for the often implied yet rarely tested influence of ideology on news frames. Among the research into how ideology influences terrorism coverage, Yarchi et al.'s (2013) study is an exception: they have contributed an empirical study of this factor by measuring ideological indexes, such as political rights and civil liberty, by applying data from Freedom House.

Their conceptualization of ideology provides useful reference points for us to consider in our study. Their findings are to some extent in need of re-contextualizing because their related variables were coded only with two points, ‘yes’ or ‘no’, which is not a suitable approach for measuring proximity. Additionally, two indicators might not be able to provide a wider ideological prism for different regimes and media affiliations. Hence, this article studies media host countries and media victim countries across a wider array of ideological positions by measuring this concept with more ideological indexes and proximity indicators, such as ruling patterns, electoral rights, press freedom and civil liberty.

The literature lacks a well-designed study that combines representative cases and news media from a media host country, a victim country and an attacker group. Additionally, the framing of controversial international events is not decided by a single predictor but is more cohesively based on an interplay of ideology, religion and bilateral relations. Certain studies have taken these scattered factors into consideration separately (Zhang et al., 2013), but a dynamic multi-factor model to integrate them is still lacking. Therefore, this study presents a dynamic model of multi-factor proximity, as shown in Figure 1.

Figure 1.

The integration model of multi-factor proximities.

The three parts have been analysed as a 3 × 3 (ideological/bilateral relations/other proximities × media host country/victim country/attacker group) pattern. For the attacker group, the current study includes as relevant only the proximity of religion between the media victim country and the attacker group. This article attempts to integrate those related predictors to test media frames with multiple factors and provide a slightly broader understanding of international news coverage.

Methodology

Sampling

Since international terrorism has been interpreted and structured primarily in terms of the ongoing conflict between autocratic and democratic regimes, or between the Arab and Western worlds, we wish to examine news stories from all sides of this alleged conflict. Therefore, we chose serious news media outlets that represent different countries and regions of the world and are relevant due to their scope and impact.

This study proceeded from a quantitative content analysis of the frames used in three newspapers: the New York Times, an influential and elite newspaper in the United States; the People's Daily, the official flagship Chinese newspaper; Ming Pao (Hong Kong) a prestigious Chinese-language newspaper; and one television outlet, Al Jazeera (Qatar), a pan-Arab broadcaster. We chose the three newspapers and one television broadcaster because representative television programmes in China and Hong Kong are not available and the most popular news medium in the Arab world is television. According to Semetko and Valkenburg (2000), the news frames of political conflicts might differ in type of news outlet, e.g., sensationalist or serious news, but do not differ significantly between medium (television vs. the print press). Additionally, the general aim of this project was to glean a general picture of the way major news media systems cover terror, and Al Jazeera is the only channel that describes occasions where Muslims are victims of discrimination in the West, thus providing an alternative viewpoint for controversial international issues (Gerhards and Schäfer, 2014). Hence, we include Al Jazeera as a balancing voice among the news frames.

The U.S. New York Times and the Chinese People's Daily were selected to represent two ideological ‘rivals’ between Western and Eastern regimes; the Hong Kong newspaper Ming Pao was chosen to present another balancing power among these news media systems. Hong Kong is a former British colony that is now a special administrative region of China governed by a policy of ‘one country, two systems’; its ideological and international positions are therefore jointly influenced by both mainland China and Western countries (Ho et al., 2003). By examining these news media systems, we extend the scope of our analysis to a broader context.

Similar differences can be ascribed to differences in the respective case selections. According to Schmid (2012), terrorist attack is a violent struggle, employed by individual, group or state actors, for idiosyncratic or political reasons, which purposely utilizes or threatens to utilize violence against civilians to create fear and anxiety. Guided by this most quoted definition, five terrorism cases were examined in this article: the Xinjiang Urumqi attack (2009, China), the Orlando nightclub shooting (2016, the United States), the Beslan hostage crisis (2005, Russia), the London subway bombings (2004, the United Kingdom), and the Ankara bombings (2015, Turkey). Each case selected for this study was a terrorist attack that generated the largest casualties in the victim country in two decades. Additionally, the media host countries and victim countries represent diverse cultural backgrounds.

Among the cases, the victim country of the London subway bombings is the United Kingdom, which is an ally of the United States and the former sovereign state of Hong Kong, and the Ankara bombing occurred in Turkey, which is a Muslim victim country. Russia, the victim country of the Beslan hostage crisis, is an ideological ‘rival’ of the Western countries that shares closer relations with China.

For the People's Daily (China) and Ming Pao (Hong Kong, officially a territory of China), the Xinjiang attack is a domestic case, as is the Orlando shooting for the New York Times (United States.). However, for Al Jazeera (Qatar), all of the selected cases are foreign events; therefore, we analysed only its international English news stories. For the New York Times, People's Daily and Ming Pao, we analysed the news reports in their native languages (English or Chinese). By including cases from Western countries, a Muslim country, China and Russia, we are able to provide comparative analyses that show clear differences between the wide range of regimes and news media systems (Table 1).

Table 1.

Cases information.

	Time	Location	Attackers	Deaths
Beslan hostage crisis	09/01/2004	Russia	North Ossetia separatists	326
Orlando nightclub shooting	12/06/2016	United States	Chechen American	50
Urumqi attack	05/07/2009	China	Xinjiang separatists	197
London subway bombings	07/07/2005	Britain	European jihad organization	56
Ankara bombings	10/10/2015	Turkey	Kurdish separatists	96

The period for news sample collection covered 2 weeks after the terrorist attacks occurred. Keywords such as ‘terrorism’, ‘bombings’ and ‘violence’ were used to search news media databases from the established time period. After the results were screened for relevancy, a total of 588 news stories were considered relevant to our research topic: 147 from the New York Times (24.9%), 154 from the People's Daily (26.1%), 123 from Al Jazeera (20.9%) and 165 from Ming Pao (28.0%).

Measurement of independent variables

In the proposed model, we considered the relevant proximities among the media host country, victim country and attacker group in ideology, bilateral relations and other factors as dependent variables. After preliminary analysis, we focused on the proximity of ideology, bilateral relations and religion (including the victim country and attackers). There were 20 proximity pairs for each indicator (4 media sources examined over 5 cases), each of which was measured on a three-point Likert scale (0–2). Because of uncertainty regarding other characteristics of the attacker groups (ideology, bilateral relations), we measured only the proximity of religion between the media host country and the attackers.

Ideological index

Based on data from Freedom House, the scores for four indicators (ruling patterns, election rights, press freedom and civil liberty) for each of the seven media host and victim countries were converted to a three-point scale of 0 to 2. The four indicator scores were combined to generate an ideology index ranging from 0 to 8.

Taking the coding process of ruling patterns as an example, China, Russia and Qatar are categorized as authoritarian regimes; Turkey is regarded as a hybrid regime; and the United States and Hong Kong are regarded as flawed democratic countries. The United Kingdom is the only regime considered a full democracy. The original full score assigned by Freedom House is 10; the scores were re-coded on a scale of 0–2. China, Russia and Qatar were coded as 0, Turkey and Hong Kong as 1, and the United States and United Kingdom as 2. Based on the same coding procedure, the four ideological indexes were coded.

Ideological proximity

For each compared pair, the ideological index score of the media host country was subtracted from the ideology index score of the victim country. Using absolute values, the result was then converted to another 0–2 scale.

For example, the proximity of the ruling pattern between China/Russia and the United States/United Kingdom was measured as 2. The proximity between China, Russia and Qatar was measured as 0. The proximity between the United States and the United Kingdom was also measured as 0. Hong Kong was measured as 1 with all of the other countries except Turkey. Other ideological proximities were coded in the same way.

Religious proximity 1 (between media host country and victim country)

This measurement was based primarily on the percentage of the population within each country that identifies as the dominant religion, as measured by official data from each country at the time that the attacks occurred. China (85.6%) and Hong Kong (70%) are made up mainly of atheists. The dominant religion in the United States and the United Kingdom is Christianity, representing 78% of the population in the United States and 64% in the United Kingdom. The dominant religion in Qatar (91.5%) and Turkey (99%) is Islam. The dominant religion in Russia is Russian Orthodox Christianity, representing 78% of the population. Proximity was coded as same religion vs. same religion = 0; one religion vs. another religion = 1; atheism vs. other religion = 2.

Religious proximity 2 (between media host country and attackers)

The same procedure was employed as in the Religious proximity 1 section.

Bilateral relations

Bilateral relations is a dynamic variable based primarily on the political relationship between two countries at the moment of the specific attack. By checking documents and contracts, statements made by leaders and other official materials, the bilateral relations between each pair of countries was coded as friend, neutral or enemy (friend = 0; neutral = 1; enemy = 2). Materials from governments’ partnership statements, foreign policy decisions and bilateral political cooperation were coded. For example, China has 82 official bilateral partners, categorized on 16 levels (Wuhan Gov., 2017), and Russia is in the category of the strongest allies. Thus, the relation between these two countries was coded as 0.

For the detailed coding results for ideological index and multiple proximities across regions and the mean and standard deviance of each content indicator, see Tables 8 to 10 in the Appendix.

Measurement of dependent variables

Three media frame packages with 12 sub-frames drawn from the coverage of terrorist attacks were used as dependent variables (Table 2). After a close reading of approximately 20% of the sampled news, 12 sub-frames with relevant keywords were identified that matched the 4 functions in Entman's definition of frame. These sub-frames were used in a multi-frame coding procedure to code each news story (Table 3). Instead of marking only the dominant frame, this study gave each sub-frame a coding score on a three-point scale so that every news story had a set of 12 scores. With factor analysis, these 12 sub-frames could be grouped into three frame packages, each with 4 sub-frames that matched the respective relevant functions.

Table 2.

The 12 sub-frames.

	Negative	Neutral	Positive
Issue definition	Anti-human	Crime	Resistance
Assigned causation	Global terror organizations	Appeal	Oppression
Moral judgement	Evil	Paranoid	Unfairness
Remedy suggestion	Strikes	Precautions	Reallocate interests
Cronbach's alpha	0.985	0.946	0.921

Table 3.

Guiding questions and keyword information for frames coding.

Issue definition
(1)	Does the story suggest some defining message of the event?
	Keywords: crime, terror, war, protest, revenge, ethnic conflict, deadly clashes, riots, rampage, etc.
(2)	Are quotation marks used in the story to imply disagreement?
	Keywords: ‘terrorism’, ‘double standard’, the ‘three evils’, etc.
(3)	Does the story suggest labels for the attacker?
	Keywords: separatists, freedom fighters, demon, criminal, murderer, stabber, etc.
Assigned causation
(1)	Does the story suggest that some level of the government/organization/person is responsible for the event?
	Keywords: global terror organization, government of the victim country, individual attacker, foreign power, etc.
(2)	Does the story suggest that ideology is responsible for the event?
	Keywords: clash of cultures, the West vs. Islam, socialism vs. capitalism, etc.
Moral judgement
(1)	Does the story contain any moral judgement message?
	Keywords: inhuman, evil, autocracy, hatred, unfair, colonize the minority, extreme, etc.
(2)	Does the story refer to attacks that occurred in other countries?
	Keywords: Russia's 9/11, etc.
(3)	Does the story justify its evaluations?
	Keywords: justice vs. evil, civilize the barbarians, ensure national security, human rights, etc.
Remedy suggestion
(1)	Does the story suggest certain solutions?
	Keywords: global cooperation, boom economy, strikes, flexible governance, education, self-determination, allow independence, redistribute benefits, etc.
(2)	Does the story suggest that some level of government has the ability to alleviate the problem?
	Keywords: high efficiency, confidence, solidarity, quick response, have experiences, be alert, etc.

Inter-coder reliability checking

Two coders were trained to measure the news using a Likert three-point scale. First, the coders randomly selected 60 news items to perform a reliability test using a Krippendorff's alpha value of 0.82 for dependent variables and 0.91 for independent variables. The coders then finished coding independently. The dominant frame was coded as 2 and the marginal frame as 1; if no frame was identified, then it was coded as 0. The unit of analysis is every piece of a news story, and multiple frames were allowed in the coding process.

Data processing

Except for report frequencies and correlations, the SPSS functions of factor analysis and multiple regressions were also applied in the data analysis.

Factor analysis

Because the four indicators of ideology were coded based on the available indexes from Freedom House, the data are more precise for these indicators than for other proximity variables. We reduced these four indicators into one factor of ideology through SPSS factor analysis and obtained a Cronbach's alpha value of 0.87 (Table 4).

Table 4.

Factor analysis of the four indicators of ideological proximity.

Indicators	Scores
Medium/victim ruling pattern proximity	0.971
Medium/victim electoral rights proximity	0.707
Medium/victim press freedom proximity	0.978
Medium/victim civil liberty proximity	0.858

Also using SPSS factor analysis, the 12 sub-frames were merged into three factors. The sub-frames ‘resistance’, ‘unfairness’, ‘oppression’ and ‘relocate interests’ became the ‘resistance’ factor, with a Cronbach's alpha value of 0.921. The sub-frames ‘anti-human’, ‘global terror organization’, ‘evil’ and ‘strikes’ became the ‘anti-human’ factor, with a Cronbach's alpha value of 0.985. In addition, the sub-frames ‘crime’, ‘appeal’, paranoid’ and ‘precautions’ became the ‘crime’ factor, with a Cronbach's alpha value of 0.946 (Table 5).

Table 5.

Factor analysis of the 12 sub-frames.

Function	Sub-frame	Scores
Issue definition	Resistance	0.953	−0.212	−0.428
Assigned causation	Oppression	0.982	−0.204	−0.373
Moral judgement	Unfair	0.987	−0.201	−0.381
Remedy suggestion	Relocate interests	0.974	−0.200	−0.374
Issue definition	Anti-human	−0.512	−0.459	0.837
Assigned causation	International terror organization	−0.375	−0.330	0.944
Moral judgement	Evil	−0.416	−0.352	0.946
Remedy suggestion	Strikes	−0.372	−0.322	0.958
Issue definition	Crime	−0.268	0.855	−0.461
Assigned causation	Appeal	−0.190	0.947	−0.318

Research hypotheses

Based on the above reviews and analysis, the researchers proposed the following four hypotheses:

Hypothesis 1. The higher the religious proximity between the media host country and the attackers, the more likely the media will be to adopt a positive frame in reporting the attack.

Hypothesis 2. The higher the religious proximity between the media host country and the victim country, the more likely the media will be to adopt a negative frame in reporting the attack.

Hypothesis 3. The closer the bilateral relations between the media host country and the victim country, the more likely the media will be to adopt a negative frame in reporting the attack.

Hypothesis 4. The higher the ideological proximity between the media host country and the victim country, the more likely the media will be to adopt a negative frame in reporting the attack.

Results of data analysis

Descriptive results

Among the four news organizations studied, the People's Daily tended to use the ‘anti-human’ factor (85.2%) most frequently when covering the selected terrorist attacks, with the ‘resistance’ factor rarely appearing. As the People's Daily reported, ‘The Urumqi attack is a serious terrorism attack organized by the “three evils.” Terrorism should be struck hard internationally, and anti-terrorism cooperation within countries should be established’.

Al Jazeera tended to report such on the attacks using the ‘resistance’ factor (58.7%) and ‘crime’ factor (36.7%) and seldom used the ‘anti-human’ factor.

President Putin has ordered sweeping changes to Russia's political system to help combat ‘terrorism’, immediately drawing accusations of exploiting this month's bloody school siege in Beslan to boost his power. Ferazouli concluded: ‘Stalin may have been brutal, but at least he did not kill us like the 150,000 Russian troops currently stationed in Chechnya’.

Among the four news organizations studied, the People's Daily tended to use the ‘anti-human’ factor (85.2%) most frequently when covering the selected terrorist attacks, with the ‘resistance’ factor rarely appearing. Al Jazeera tended to report on the attacks using the frames ‘resistance’ (58.7%) and ‘crime’ (36.7%) but seldom used ‘anti-human’. The New York Times mainly used ‘anti-human’ (58.3%) to report on the attacks, with ‘resistance’ (35.4%) and ‘crime’ (18.1%) also being used. Ming Pao's pattern of coverage was similar to that of the New York Times; it mainly used ‘anti-human’ (54.9%), followed by ‘resistance’ (28.4%) and ‘crime’ (23.5%).

Analysing framing patterns on a case-by-case basis for each attack reveals a more insightful profile for each medium. In covering the Urumqi attack, both the People's Daily and Al Jazeera took an extremely one-sided position, but in opposite directions. Ninety-nine percent of the People's Daily's frame fell into the ‘anti-human’ category, while all reports on the same attack from Al Jazeera described it as ‘resistance’. The New York Times took a stance similar to that of Al Jazeera, with 86% of its stories using the frame ‘resistance’.

China and Qatar media followed a similar pattern in covering the Beslan hostage crisis. The People's Daily overwhelmingly (98%) used the ‘anti-human factor’, while Al Jazeera leaned towards the ‘resistance’ factor (79%). In covering the Beslan story, both the New York Times and Ming Pao took moderate positions.

Coverage of the London subway bombings was the only incident among those studied for which media from China, Hong Kong and the United States all used ‘anti-human’ as their dominant news frame. Al Jazeera also adopted the mainly neutral frame ‘crime’ rather than the relatively more supportive ‘resistance’ frame.

In covering the Orlando nightclub shooting, the New York Times was the only medium to use ‘anti-human’ as the main frame, while the Chinese and Hong Kong newspapers both adopted the ‘crime’ frame.

The Ankara bombings demonstrated a relatively weaker media agenda, as they made up only 42 of the 588 news reports we studied. All four media sources used their own framing pattern in this case (Table 6).

Table 6.

Frequency and percentage of frame factors of terrorism cases in media.

Terrorism case	Frame factors	People's Daily	Al Jazeera	New York Times	Ming Pao
Urumqi attack	Resistance	0 (0)	35.9 (28)	30.8 (24)	33.3 (26)
	Crime	5.3 (1)	0 (0)	21.1 (4)	73.7 (14)
	Anti-human	98.5 (68)	0 (0)	0 (0)	43.7 (31)
Orlando shooting	Resistance	9.5 (2)	71.4 (15)	4.8 (1)	14.3 (3)
	Crime	15.7 (8)	25.5 (13)	27.5 (14)	31.4 (16)
	Anti-human	6.7 (2)	0 (0)	56.7 (17)	36.7 (11)
Beslan hostage crisis	Resistance	0 (0)	(11)	(20)	(8)
	Crime	8.3 (1)	25.0 (3)	41.7 (5)	25.0 (3)
	Anti-human	40.6 (41)	0 (0)	43.6 (44)	15.8 (16)
London bombings	Resistance	8.3 (3)	27.8 (10)	50.0 (18)	13.9 (5)
	Crime	8.6 (6)	34.3 (24)	50.0 (35)	7.1 (5)
	Anti-human	14.0 (17)	5.0 (6)	55.4 (67)	25.6 (31)
Ankara bombings	Resistance	0 (0)	57.1 (4)	28.6 (2)	14.3 (1)
	Crime	9.5 (2)	38.1 (8)	47.6 (10)	4.8 (1)
	Anti-human	20.0 (3)	13.3 (2)	60.0 (9)	6.7 (1)

Explanation of findings

Using the three media frame factors as dependent variables, multiple regression models were estimated. The predictors included ideological proximity, ideology index, bilateral relations, religious proximity 1 and religious proximity 2. These variables were entered as three groups, and all interactions among the independent variables were controlled for (Table 7).

Table 7.

Predicting the news frames with proximity variables (β).

Variable	Resistance frame	Crime frame	Anti-human frame
Medium/attacker religious proximity	−0.394***	−0.202***	0.405***
Medium/victim religious proximity	0.045	0.059	−0.114**
R sq. change	0.176	0.062	0.259
Medium/victim bilateral relations	0.370***	−0.110*	−0.190***
Medium/victim ideological proximity^a	0.070	0.084	−0.086
R sq. change	0.053	0.018	0.038
Medium/victim ideological proximity	−0.340***	0.112	0.046
R sq. change	0.053	0.004	0.001
Total R sq.	.263	.083	.298
Uncertainty explained	51.3%	28.9%	54.6%

Factor.

*p < .05; **p < .01; ***p < .001.

The model to predict the anti-human frame showed a significant regression equation (F(5, 580) = 49.218, p < .000), with an R square of 0.298. The predicted anti-human frame is equal to −0.215 + 0.590 (medium/attacker religion) – 0.115 (medium/victim religion) – 0.334 (bilateral relations). These results indicate that the less religious proximity there is between the media host country and the country of the attackers, the more religious proximity there is between the media host country and the victim country, and the closer the bilateral relations between the media host country and the victim country, the more likely the media will be to apply the most negative news frame, ‘anti-human’, in their reports.

A significant regression equation is also found in the resistance model, with an equation of F(5, 580) = 30.789, p < .000 and an R square of 0.263. The predicted resistance frame is equal to 0.415 – 0.574 (medium/attacker religion) + 0.651 (bilateral relations) – 0.104 (ideology index). In the crime model, the equation is F(5, 580) = 10.544, p < .000 and the R square is 0.083. The predicted resistance frame is equal to 0.164 – 0.295 (medium/attacker religion) – 0.193 (bilateral relations). These results indicate that the more religious proximity there is between the media host country and the country of the attackers, the more hostile the bilateral relations are between the media host country and the victim country, and the higher the democracy index score of the media host country is, the more likely the media will be to apply the most positive news frame, ‘resistance’, when reporting on attacks.

The multi-proximity model showed less (but still significant) power when used to predict the neutral frame ‘crime’. Attacker-related religious proximity and bilateral relations demonstrated predictive power: the more religious proximity there is between the media host country and the country of the attackers and the more friendly the bilateral relations are between the media host country and the victim country, the more likely the media will be to use the ‘crime’ frame to define the attack.

Taken together, the score of adjusted R square shows that the current predictive model can explain large amounts of uncertainty in the two dependent variables. The adjusted R square values 0.257, 0.292 and 0.075 mean that approximately 51.3% of the variance has been explained for the resistance factor, 54.6% of the variance has been explained for the anti-human factor and 28.9% of the variance has been explained for the crime factor. This finding demonstrates the quality of the research design for this study.

Testing of research hypotheses

Hypothesis 1 is accepted. The outcome of the regression analysis shows that higher religious proximity between the media host country and the attackers is very strongly related to news coverage adopting the positive frame ‘resistance’.

Hypothesis 2 is largely accepted. The result of multiple regression analysis shows a moderate positive correlation between the religious proximity of the media host country and the victim country and news coverage applying a negative ‘anti-human’ frame in reporting on the attack.

Hypothesis 3 is also accepted. The result of the data analysis shows that bilateral relations between the media host country and the victim country are a very stable and powerful predictor for which news frame will be applied. It is positively related to the frame ‘anti-human’ and negatively associated with the frame ‘resistance’.

Unfortunately, Hypothesis 4 is rejected. The ideological proximity between the media host country and the victim country showed no significant relation with any frame. The only notable finding regarding ideology was that media located in a country with a higher democracy index score are more likely to adopt a positive frame when reporting on a terrorist attack.

Conclusion and discussion

This study provided a systematic comparative analysis of news frames used by representative global news media when covering terrorism attacks that occurred on various continents. Both the media sources and the selected terrorist incidents crossed ideological, ethnic, religious, linguistic and geographic lines. A newly proposed multi-proximity model shows strong predictive power for different factors on news frames. Additionally, three consistent frame packages were identified that are a good match for the functions of media frames. Limitations and suggestions for further study will also be discussed.

First, the broad range in the media sources and attacks studied largely increased the generalizability of this article. The scope covered exceeded that of other studies that have been conducted to date on similar topics, especially in the inclusion of a medium based in and an attack that occurred in an Arab country. Four media organizations (based in China, Hong Kong, the United States and Qatar) and five attacks (occurring in China, the United States, the United Kingdom, Russia and Turkey) represent a wide spectrum that crosses different political systems, religious beliefs, ideologies, cultures and geography and supports the core goal of our research.

Second, to identify possible reasons for divergent news frames in the coverage of terrorism attacks, this study proposes an innovative integrated model of multi-factor proximities. By measuring proximities in ideology, dominant religion and bilateral relations between media host countries and victim countries and calculating the association of these proximities with frame packages, this model can explain a large amount of uncertainty in media frame selection. Religious proximity (between the media host country and both victim countries and attackers) and bilateral relations are strongly predictively correlated with frames.

However, ideology, the variable that many news content analyses conducted across countries or media systems have concluded is the most influential, did not show the expected impact on media frames in our study. One possible reason lies in the definition of ideology and the power structure in global information flow. According to the ideological categories of Freedom House, China and Qatar are grouped together at the lower end, Hong Kong is in the middle and the United States is at the higher end. However, in the coverage of the terrorist attack that occurred in China, Al Jazeera (with the same religious beliefs as those of the attackers) often adopted less negative news frames in opposition to those used by the People's Daily, China's official news medium. In addition, because the United States controls the lion's share of the information flow in global communications, most other countries, to a certain extent, will follow its frames in covering terrorism that occurs in the United States. For these reasons, the boundary that was expected to be created by ideological differences may have been blurred.

Another contribution of this study is demonstrating a convincing application of Entman's four functions perspective in an empirical frame analysis study. In their analysis of media frame studies across decades, Matthes and Chen suggested two key criteria for a theory-oriented frame study: solid conceptualization and including antecedents and consequences (Matthes, 2009). In this article, factor analysis categorized 12 news frames into three differentiated frame packages (anti-human, crime and resistance). Within each frame package, the relevant sub-frames (issue definition, responsibility attribution, moral judgement and suggested remedies) were grouped together with high reliability scores. Not satisfied with identifying frame patterns in different media, this study went further in systematically analysing the relationship between the contextual characteristics of the media, and the data showed that a large amount of uncertainty in media frames could be explained.

Limitations

The realized limitations of this study should be addressed. If not restricted by time and the size of the research team, more representative media and attacks across time should be included in the study. Currently, there is only one media source to represent each category, but ideally, there should be two or more news organizations for each category of political/news media system and cultural/religion system. Similarly, only one attack is included for each victimized country, which restricts the generalizability of the findings and prevents testing cross-time analysis within the same type of countries. Most importantly, Al Jazeera, the most popular news medium in Arab countries, is a television broadcaster, and only its English news reports are analysed, which is a serious limitation. Additionally, by broadening the range of attacks examined, the dynamic nature of bilateral relations across time between countries can be tested. Another limitation is that this study did not factor in the position of each sampled country in the international communications landscape or its ability to control information flow. The news frame used by a superpower country has an agenda-setting impact on other countries. Last but not least, the varying details of each terrorist attack, such as casualties, means, circumstances and salience of field location, may also influence the reporting frames adopted by media. Hopefully, the team will have the chance to address these issues in a follow-up study.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This article is supported by the Chongqing University “Fundamental Research Funds for the Central Universities” 2018 CDXYXW0043 and 106112017 CDJXY070002.

Appendix

References

Aksoy

Carter

Wright

(2012) Terrorism in dictatorships. The Journal of Politics 74(3): 810–826.

Atran

(2003) Genesis of suicide terrorism. Science 299(5612): 1534–1539.

Bhatia

(2008) Discursive illusions in the American national strategy for combating terrorism. Journal of Language and Politics 7(2): 201–227.

Brownell

Kersh

Ludwig

et al. (2010) Personal responsibility and obesity: A constructive approach to a controversial issue. Health Affairs 29(3): 379–387.

Buckley

Fawn

(2003) Global Responses to Terrorism: 9/11, Afghanistan, and Beyond, London: Psychology Press.

Butko

(2005) Terrorism redefined: Terrorism as “counter-hegemonic political violence”. Peace Review: A Journal of Social Justice 1(18): 145–151.

Buzan

(2006) Will the ‘global war on terrorism’ be the new Cold War?. International Affairs 82(6): 1101–1118.

Campbell A (2017) Framing terrorism. Oxford Research Encyclopedia of Criminology 00: 1–34. Available at: https://doi.org/10.1093/acrefore/9780190264079.013.148 (accessed August 2017).

Chen

Yang

(2015) An empirical analysis of impact from applied symbols and interpreting frames on attitude and public opinion-in case of “mainland visitor waves” in Macao. Journalism and Communication. 25(7)35–50.

10.

Chuang

Robin

(2014) Shifting signifiers of otherness: The 2002 “DC Snipers” in the US press. Communication, Culture & Critique 7(4): 541–558.

11.

Cotter

(1999) From folklore to “News at 6”: Managing language and reframing identity through the media. In: Bucholtz

Liang

Sutton

(eds) Reinventing Identities, Oxford: Oxford University Press, pp. 369–387.

12.

Entman

(1993) Framing: Toward clarification of a fractured paradigm. Journal of communication 43(4): 51–58.

13.

Entman

(2003) Cascading activation: Contesting the White House's frame after 9/11. Political Communication 20(4): 415–432.

14.

(2015) International anti-terrorism should abandon Double Standard. China Report 13(12): 46–47.

15.

Gamson

Modigliani

(1989) Media discourse and public opinion on nuclear power: A constructionist approach. American Journal of Sociology 95(1): 1–37.

16.

Gerhards

Schäfer

(2014) International terrorism, domestic coverage? How terrorist attacks are presented in the news of CNN, Al Jazeera, the BBC, and ARD. International Communication Gazette 76(1): 3–26.

17.

Goffman

(1974) Frame Analysis: An Essay on the Organization of Experience, Cambridge, MA: Harvard University Press.

18.

Chau

Chiu

et al. (2003) Ideological orientation and political transition in Hong Kong: Confidence in the future. Political Psychology 24(2): 403–413.

19.

Jackson

Sinclair

(2013) Contemporary Debates on Terrorism, London: Routledge.

20.

Jasperson

El-Kikhia

(2003) CNN and Al-Jazeera's media coverage of America's war in Afghanistan. In: Norris

Kern

Just

(eds) Framing Terrorism: The News Media, the Government, and the Public, London: Psychology Press, pp. 113–132.

21.

Karim

(2003) Islamic Peril: Media and Global Violence, Montreal: Black Rose Books Limited.

22.

Kim

Carvalho

Davis

(2011) The view of the border: News framing of the definition, causes, and solutions to illegal immigration. Mass Communication and Society 14(3): 292–314.

23.

Lecheler

de Vreese

Slothuus

(2009) Issue importance as a moderator of framing effects. Communication Research 36(3): 400–425.

24.

Lee

Wang

(2016) A comparative investigation into press–state relations: Comparing source structures in three news agencies' coverage of the North Korean missile crisis. International Journal of Communication 10(22): 1907–1928.

25.

(2015) The “Double standards” issues in the Western mainstream media: The case study on Chinese terrorism attacks. Journal of Yun Meng 36(5): 139–145.

26.

Matthes

(2009) What's in a frame? A content analysis of media framing studies in the world's leading communication journals, 1990-2005. Journalism & Mass Communication Quarterly 86(2): 349–367.

27.

Medvedeva

Hinnant

(2011) Terrorism coverage in the U.S. and Russian Newsweek. International Communication Research Journal 1(46): 29–55.

28.

Miao

(2014) Response to the Western media: “double standard” in Kunming violent terrorist event. Journal of external communication. 26(4)29–30.

29.

Nacos

(2016) Terrorism and Counterterrorism, London: Routledge.

30.

Nacos

Torres-Reyna

(2005) Framing Muslim-Americans before and after 9/11. In: Norris

Kern

Just

(eds) Framing Terrorism: The News Media, the Government, and the Public, London: Psychology Press, pp. 133–158.

31.

Nail

(2016) A tale of two crises: Migration and terrorism after the Paris attacks. Studies in Ethnicity and Nationalism 16(1): 158–167.

32.

Norris

Kern

Just

(2003) Framing Terrorism: The News Media, the Government, and the Public, London: Psychology Press.

33.

Nossek

Berkowitz

(2006) Telling our story through news of terrorism: Mythical newswork as journalistic practice in crisis. Journalism Studies 7(5): 691–707.

34.

Powell

(2011) Framing Islam: An analysis of US media coverage of terrorism since 9/11. Communication Studies 62(1): 90–112.

35.

Raza

Jan

Sultan

et al. (2012) Portrayal of war on terrorism in Pakistani print media exploring peace framing in daily nation and business recorder. Asian Journal of Social Sciences & Humanities 1(4): 97–108.

36.

Said

(1994) Orientalism, New York: Vintage.

37.

Schaefer

(2003) Framing the US embassy bombings and September 11 attacks in African and US newspapers. In: Norris

Kern

Just

(eds) Framing Terrorism: The News Media, the Government, and the Public, London: Psychology Press, pp. 93–112.

38.

Scheufele

Iyengar

(2012) The state of framing research: A call for new directions. The Oxford Handbook of Political Communication Theories, New York: Oxford University Press.

39.

Schmid

Price

(2012) The revised academic consensus definition of terrorism. Perspectives on Terrorism 6(2): 158–159.

40.

Semati

(2010) Islamophobia, culture and race in the age of empire. Cultural Studies 24(2): 256–275.

41.

Semetko

Valkenburg

(2000) Framing European politics: A content analysis of press and television news. Journal of communication 50(2): 93–109.

42.

Wuhan G (2017) Inventory of the Partnership of China. Available at: http://www.whfao.gov.cn/html/guide/knowledge/201602/t20160215_46744.shtml (accessed 2 October 2018).

43.

Sun

(2014) American mainstream media using “Double standards” to definition terrorism attacks: The case of Wall Street journal. New Media and Society 21(4): 26.

44.

Wang

(2015) A frame analysis of the Times on the image of China (1992-2008). In: Chen

(ed.) Analysis of the News Frames and National Images in Major Events, Macau: University of Macau, pp. 281–303.

45.

Weimann

Brosius

(1991) The newsworthiness of international terrorism. Communication Research 18(3): 333–354.

46.

Wilson

(2015) Strategic Partners: Russian-Chinese Relations in the Post-Soviet Era, London: Routledge.

47.

Wilson

Piazza

(2013) Autocracies and terrorism: Conditioning effects of authoritarian regime type on terrorist attacks. American Journal of Political Science 57(4): 941–955.

48.

Xin

(2010) Chindia's challenge to global communication: A perspective from China. Global Media and Communication 6(3): 296–301.

49.

Yarchi

Wolfsfeld

Sheafer

et al. (2013) Promoting stories about terrorism to the international news media: A study of public diplomacy. Media, War & Conflict 6(3): 263–278.

50.

Yousaf

(2015) Representations of Pakistan: A framing analysis of coverage in the US and Chinese news media surrounding operation Zarb-e-Azb. International Journal of Communication 9(23): 3042–3064.

51.

Zhang

Shoemaker

Wang

(2013) Reality and newsworthiness: Press coverage of International Terrorism by China and the United States. Asian Journal of Communication 23(5): 449–471.