The Story of Two Big Chimneys: A Frame Analysis of Climate Change in US and Chinese Newspapers

Abstract

Despite mounting research on how the news media portrayed climate change in developed countries, disproportionately little attention was paid to one of the top greenhouse gas emitters in the world—China. This study systematically examined and compared how the US and Chinese newspapers portrayed climate change. An analytical framework, grounded in framing research traditions, took into account both theoretical and practical challenges in frame analysis by creating a hybrid structure of interlocked generic and issue-specific frames. The results showed both convergent and divergent pictures from the two countries’ newspapers shaped distinctly by the sociopolitical realities in both countries. The research also cautioned against making simplistic explanations of the results and offered culturally informed interpretations. A discussion on better research focus and design was offered for future research in this avenue.

Keywords:

• Climate Change
• Global Warming
• Framing
• Frame Analysis
• Science Communication
• Cross-culture

Introduction

Climate change has been the epitome of how scientific and political discourse clash. According to the Intergovernmental Panel on Climate Change (IPCC, 2007), mounting evidence ranging from widespread snow and ice melting to rising average sea level has made global warming “unequivocal.” Public perception of climate change, however, has been far less than unanimous in the US. While 40% of Americans believe that the climate is changing, one in four Americans remain skeptics (Gallup, 2014), a difference most effectively explained by partisan divide (Gallup, 2010). On a global scale, Americans are less likely to see global climate change as a major threat to their country than most people elsewhere (Pew Research Center, 2013).

Due to the chronic and accumulative nature of climate (Beamish, 2002), the public perception of the threats posed by climate change relies heavily on how news media portray it (Corbett & Durfee, 2004; Lacy, Rife, & Varouhakis, 2007). Over the past 20 years, media scholars have been exploring the news media’s representation of climate change in the US and other countries (see summary in Schäfer & Schlichting, 2014).

Most of the research, however, centers on Western developed countries (Anderson, 2009). China, currently, the largest greenhouse gas (GHG) emitter exceeded the second largest—the US—by nearly 50% (Boden & Andres, 2013) in terms of total CO2 emission as of 2010, yet China was studied in only six out of 273 (or 2.2%) published research projects on media representation of climate change (Schäfer & Schlichting, 2014). Comparison of the major stakeholders, especially those from non-Western sociopolitical backgrounds, “should play an important role in improving communication and help to identify the main barriers towards change” (Anderson, 2009, p. 176).

This study is, in part, a response to the disproportionately little attention that has been paid to China. It first reviews the social construction of climate change in both the US and China and demonstrates the strengths and weaknesses in the literature; then it examines climate change coverage in five major newspapers in the two countries by employing an interlocked analytical framework composed of four research traditions; finally, the results are compared with earlier studies, shedding new light on (1) how the framing of the issue in the two countries differed, (2) how the sociopolitical realities and media systems of the two countries shaped the coverage, (3) what the convergence and divergence in the representations suggest in terms of national identity and cultural differences, and (4) directions into theoretical and practical advancement for similar cross-cultural media studies in the future.

A timeframe from 2005 to 2008 was selected for two reasons. First, not until 2005 did a considerable number of news stories on climate change appear in the Chinese media (Schmidt, Ivanova, & Schäfer, 2013). Second, the time frame constitutes a global media attention cycle (Center for Science & Technology Policy Research, 2013), which accompanied the release of Al Gore’s documentary “An Inconvenient Truth” in 2006 and ended before the “Climategate” controversy and the Copenhagen Summit in 2009.

The US Media’s Portrayal of Climate Change in an International Perspective

Climate change was mostly discussed with other environmental issues by media scholars in the 1980s. Greenberg, Sandman, Sachsman, and Salomone (1989) examined environmental stories presented by three major news broadcasters from 1984 to 1986 and found that most of the stories were driven by human drama while offering little scientific background. Research on media coverage of climate change started to emerge in the 1990s, about the time of the Rio Earth Summit and the conclusion of the Kyoto Protocol (Schäfer & Schlichting, 2014).

Scholarship on climate change coverage often reacted to socio political realities. Wilkins (1993) looked at the US print media coverage of greenhouse effect for underlying news values between 1987 and 1990, and found that, contrary to the earlier finding that dramatization permeated environmental news, most GHG coverage was confined to scientific debate.

In contrast to Wilkins’ concern with the lack of collective political actions and ethical public discourse in the coverage, Trumbo (1996) analyzed a decade of climate change coverage in four major US newspapers from 1985 to 1994 and found a shift later in the period. In the early 1990s, politicians and interest groups significantly outweighed scientists as information sources; meanwhile, judgments and remedies became the focal point, suggesting “a politicization of the issue, an increase in its level of controversy, and a shift toward judgments and solutions” (p. 281). This shift, unsurprisingly, happened concurrently with major political powers’ involvement in mitigation, signified by US President George H. W. Bush’s commitment to the UN’s Framework Convention on Climate Change at the Rio Earth Summit in 1992.

Probably due to the politicalization of climate change, scholars in the 2000s began to worry about how accurately climate science was translated into media narratives. Drawing 928 climate change research abstracts from the ISI (Institute for Scientific Information) database, Oreskes (2004) found almost no rejection of the anthropogenic climate change consensus in the published climate science literature. However, the media portrayal in the early 2000s was far from consensus. The issue was framed primarily as a scientific and political controversy replete with rhetoric emphasizing scientific uncertainty and skepticism. The tendency was reinforced by the internal structure of the US news media, where the “collective newswire/news service community,” instead of the science community, served as “a dominant source of climate science news” (Antilla, 2005, p. 350).

A more systematic scrutiny of the US news media’s role in climate change communication was led by Maxwell Boykoff and his colleagues. By contrasting the scientific consensus on anthropogenic climate change with considerably uncertainty in media reports, Boykoff and Boykoff (2004) criticized the US news media of misusing “journalistic objectivity.” They argued that apparently “balanced” reporting in fact formed an “informational bias” that painted an elusive picture of scientific research on climate change, confused the public, and caused political inaction. The discursive mistranslation, as the authors believed, “is systematic and occurs for perfectly logical reasons rooted in journalistic norms and values” (p. 134).

Due to climate change’s deterritorialized nature, some studies centered on cross-national media coverage on climate change and cross-cultural perceptions of it. Media coverage in the US has been compared with that in the UK (Boykoff, 2007), France (Brossard, Shanahan, & McComas, 2004; Dispensa & Brulle, 2003), New Zealand (Dispensa & Brulle, 2003), and other developed and developing countries (Billett, 2010; Jones, 2006).

In addition to confirming the contentious framing of climate change in the US news media, Jones (2006) found more use of the “episodic frame” than the “thematic frames” (see Iyengar, 1989, 1991) in the US than in eight other countries. Brossard et al. (2004) concerned primarily with how journalistic norms and practices affected French and American journalists’ framing of climate change. Their analysis of climate coverage in Le Monde, a French newspaper, and the New York Times (1987–1997) showed that the contention in the French media came not from scientific uncertainty and domestic political struggles, but instead from an “international relations” frame, which dealt with the culpability of the US as both a cause and a solution to the problem. The conflicts between the US and the European Union regarding a climate change solution dominated other topics, such as potential consequences and domestic politics. Nonetheless, the international diplomacy perspective was largely absent in the New York Times’ coverage. To explain the findings, the authors resorted to differences in journalistic norms: the focus on consequences and business sources situated well with the American journalistic norm, whereas the focus on diplomatic tension was in line with the traditionally more opinionated news writing in France.

Extending his line of research and the “informational bias” criticism, Boykoff (2007) compared British newspapers’ coverage with their American counterparts from 2003 to 2006, and discovered an “evolutionary shift” (p. 475) in the US press, from considerable skepticism to more scientific and political consensus, indicating a convergence in tone between the two countries’ portrayals of the issue. He attributed the shift to various political movements, scientific activities, and meteorological events between 2003 and 2006.

But the good news was less reassuring than it sounded. A wider comparison (Painter & Ashe, 2012) involving print media in the US, Brazil, China, France, India, and the UK showed that skepticism toward climate change was mostly limited to the US and the UK. An even more inclusive comparison of climate change coverage in 27 countries by Schmidt et al. (2013) is probably the most culturally diverse of its kind to date. With a timeframe of 15 years (1996–2010), the study traced cross-nation differences in media attention and attempted to explain attention cycles based on extreme weather, climate change-related events, and other contributing factors with moderate success in certain countries.

Overall, despite the abundance of research on climate change communication, two challenging trends have emerged in recent years. First, most research focused on developed countries in the West with little attention to developing countries (Schäfer & Schlichting, 2014), who are either top contributors of GHG, such as China, or victims most severely affected by climate change, such as Indonesia. Even when they were included in cross-nation comparisons, they often represented a sample so small that culturally situated discussions on how climate change was framed by their own news media were almost impossible (Schmidt et al., 2013). Second, despite repeated calls for cross-cultural comparisons of climate change representation (Anderson, 2009; Brossard et al., 2004; Trumbo & Shanahan, 2000), the number of cross-sectional studies has been declining in the past decade, and research combining cross-sectional and longitudinal perspectives was even more scarce, accounting for only about 11% of all the climate change communication studies published in the 2000s (Schäfer & Schlichting, 2014).

Climate Change Reporting in China

It is well known that China has been plagued by a diversity of environmental problems, including water and air pollution, sand storms, and forest deterioration, due largely to its rapid economic growth (Yang, 2004). In 2010, China became the top total CO² emitter, followed by the US and India (Boden & Andres, 2013).

Growing concerns about the adverse effects of climate change were raised by Chinese and international intellectuals alike. After looking at years of climate change research in China, Lin, Xu, Wu, Ju, and Shiming (2007) concluded that climate change’s adverse impacts significantly outweighed positive results, leading to environmental challenges in “ecosystems, agriculture, water resources, and coastal zones in China” (p. 6). These issues would then lead to social issues, such as food insecurity, heavy disaster relief, and even political tension in Tibet (Lewis, 2009).

The Chinese Government’s responses to these concerns, nonetheless, seemed mixed. On the international stage, the government “has used its dual status as a developing country (with rights to and needs for development) and its growing role as a major contributor to global environmental problems (such as GHG emissions) to acquire substantial influence in international environmental negotiation” (Yu, 2008, p. 55). The government also showed a strong commitment to mitigating the environmental impacts of the nation’s industrialization by (1) developing a “comprehensive corpus of environmental policies and laws,” (2) increasing “environmental capacity through the steady buildup of environmental technologies”, and (3) cultivating a “relatively positive stance toward emerging green activism” (Ho, 2006, p. 24). Domestically, China has taken steps in recent years to reduce GHG emission. China has become a leader in renewable energy technology and has been aggressive in developing low- or zero-emission coal technology (Lewis, 2009).

On the other hand, the Chinese Government has been criticized for not being cooperative enough in multilateral negotiations to fight climate change, due in part to its caution against eco-colonialism and unwillingness to commit to concrete mitigation responsibility (e.g. mandatory GHG emission limits) that might hamper economic growth (Kobayashi, 2003). This criticism peaked during the Copenhagen Climate Summit in 2009, when China vetoed proposals for 50% reductions in global emissions by 2050 or 80% reductions by developed countries. Critics including Ed Miliband (2009), then British Environment Minister, accused China of hijacking the Summit and rescinding an opportunity that could usher international cooperation to a new era.

These voices were hardly heard in the Chinese news media. The rapid commercialization and technological advancement of the media system since the 1990s have complicated, rather than weakened, state censorship. Instead of publicly criticizing censored content, authorities now exert “passive censorship” by which content is taken down “quickly and quietly” without offering free publicity (Zhao, 2004, pp. 181–182). The degree of state control is often topic dependent. “Politically safe” topics, such as entertainment and business news, enjoy the least censorship. Local protests and deviant voices that may challenge political stability, on the other hand, are largely marginalized or muted either by government-issued censorship directives (Stockmann, 2012; Zhao, 2008) or the news media’s self-censorship (King, Pan, & Roberts, 2013). The changing media landscape, as Stockmann (2012) observed, has given rise to a more sophisticated censorship system and a more covert propaganda machine.

Albeit limited in number, several studies have provided sketches of how climate change was portrayed in China. In a study of four Chinese newspapers in 2005, Jia (2007) found that “the lack of details and controversies” in the news coverage “lowers the public interest” (p. 39) in the issue, a finding echoed by an experiment involving US and Chinese college students (Yao, Haggard, & Cai, 2014). The researchers found that Chinese students knew less about climate science, had a less positive attitude toward climate change, and a weaker behavior intention about the issue. Mirroring the government’s diplomatic stance on climate change, news stories blamed developed countries, especially the US, for contributing to the issue and for not taking adequate actions to mitigate the threat (Tolan, 2007). The stories made no mention that China was the second largest GHG emitter before 2007 (Jia, 2007).

de Burgh and Rong’s (2012) interviews with Chinese environmental journalists documented both opportunities and challenges in climate change reporting in the country. Acknowledging that economic sustainability is unattainable without environmental sustainability, the Chinese Government permitted more open environmental discussion in the news media, even though the media system still runs largely under an authoritarian paradigm. With an increasing sense of pride and responsibility among Chinese environmental journalists, they found climate change to be a particularly difficult topic to cover because of their self-acknowledged lack of scientific training. This weakness was especially worrisome before 2006, when most climate change stories heavily relied on recycled reports from Western science. The weakness became even more evident after 2006 with the increasing practice of attributing extreme weather in China to climate change (Tolan, 2007). Interestingly, the journalists considered climate change “controversial,” not in terms of its scientific uncertainty, but in light of the tension created when investigative reporting puts the government’s mitigation and adaptation efforts under public scrutiny (de Burgh & Rong, 2012).

Analytical Framework, Research Questions, and Hypothesis

Concerted endeavors in the late twentieth century, including Gamson and Modigliani’s (1989) study of the public discourse on nuclear power and Entman’s (1993) cornerstone work on framing theory, were devoted to developing theoretically and methodologically sound approaches to frame analysis. Framing research has been criticized persistently for lacking a systematic structure (Benford, 1997; Reese, 2007; Vreese, 2005). Part of the challenge derives from the overuse of issue-specific frames, such as those developed a priori for specific issues and research purposes and considered almost unusable in broader contexts. These frames, despite their ability to sometime reveal distinct characteristics of framing, fragment the system of frame typology, provide little value to theory-building over time (Scheufele, 1999, 2000), and often make it too convenient for researchers to find the evidence they were hoping to find (Hertog & McLeod, 2001). Even though scholars repeatedly recommended using generic frames, which transcend topical constraints and are identifiable across different topics or cultural contexts (Vreese, 2005), issue-specific frames still dominated in most frame analyses published between 1997 and 2007 (Borah, 2011).

Another challenge has to do with how frames are sometimes developed in a way that disconnects them from audience cognition. Cappella and Jamieson (1997) argued that frames must be commonly observed in journalistic practice in that aberrations beyond professional and organizational routines usually contribute little to the construction of social reality and invariably result in poor external validity. In other words, frames should be socially shared organizing principles that are readily recognized by the audience rather than those manufactured in the researcher’s imagination (Reese, 2001). The analytical framework of this study addressed both challenges (see Figure 1). This study employs four traditions in climate change communication and framing research at large—skepticism toward climate change, micro-issue salience, audience-based frames, and attribution of responsibility. Rather than a simple compilation of existing frames, the framework of this study identified the conceptual relationship of the four traditions and then connected them in an interlocked, progressive structure, providing an integrated analytical tool to examine climate change representation in a geopolitically sensitive light. The framework also avoided the fragmentation challenge by employing a hybrid structure comprised of both generic and issue-specific frames.

Figure 1 A hybrid analytical framework.

The first research question started with earlier work on skepticism toward climate change (Antilla, 2005; Boykoff, 2006; Boykoff & Roberts, 2007; Painter & Ashe, 2012), a significant issue-specific frame that largely reflected the public opinion and discourse in the US (Gallup, 2010, 2014). To compare the skepticism in both countries’ newspapers on a longitudinal basis, we asked.

RQ1: To what extend does the skepticism toward climate change vary by country and year?

This study borrowed Entman’s (1993) influential definition of framing, which used progressive stages of social issues as a frame capturing device. He said, “to frame is to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and treatment recommendation” (Entman, 1993, p. 52). The “definition—cause—morality—remedy” dichotomy provides researchers with an analytical tool to measure micro-issue selection and salience. For climate change coverage, the selection and salience of the four stages show the cues through which the media try to either guide or reflect public discourse.

RQ2: To what extent are the variances in the use of micro-issue salience (definition, cause, morality, and remedy) explained by country?

To further look into the fourth micro-issue, remedy, we were interested in how mitigation and adaptation achievements of different geopolitical scales (domestic, foreign, and international) were reported in both countries. So, we used these issue-specific frames and asked.

RQ3: How do the geopolitical boundaries of mitigation and adaptation achievement (domestic, foreign, and international) vary by country?

This study handled the second challenge in framing research—the disconnect between researchers’ unilateral imagination and audiences’ cognition—by adapting earlier research on audience-based frames in news consumption. Semetko and Valkenburg’s (2000) development of audience-based generic frames (conflict, human interest, economic, morality, and responsibility) was one of the successful endeavors to address the issue. In addition to using the five frames, we added “non-human interest” because climate change stories sometimes cover ecological impacts with no humans apparently involved.

RQ4: To what extent does the use of audience-based frames (conflict, human interest, non-human interest; economic, morality, and responsibility) vary by country and year?

Next, we focused on the responsibility frame in RQ4 and explored how remedial responsibility was attributed to different social institutions. An array of sociological works has demonstrated that individuals tend to simplify complex issues “by reducing them to questions of responsibility, and their opinion on issues flow from their answers to these questions” (Iyengar, 1991, p. 8). Based on the attribution of responsibility thesis developed mainly by Heider and Simmel (Heider, 1958; Heider & Simmel, 1944), this study drew upon Semetko and Valkenburg’s (2000) analysis of responsibility in media messages and examined the attribution of “treatment responsibility” to parties including individuals, governments, scientists, non-government organizations (NGOs), industry, and all humans in general. The following research question compared the attribution of responsibility in both countries.

RQ5: To what extent does attribution of responsibility (individual, government, industry, organizations, scientist, and all humans) vary by country and year?

Similar to the geopolitical examination of mitigation boundaries in RQ3, here, we further distinguished among attribution to domestic, foreign, and international governments. According to the Chinese Government’s tendency to downplay China’s total GHG emission (Tolan, 2007) and underscore developed countries’ mitigation obligation (Kobayashi, 2003), we hypothesized as follows:

H1: Chinese news media are more likely than their US counterpart to attribute mitigation responsibility to non-domestic governments.

As seen in Figure 1, the analytical framework of this study bears several theoretical and practical virtues. It integrates analytical tools used in earlier research into a networked structure, taking advantage of their conceptual connections. Vertically from the top, skepticism toward climate change may permeate every level of micro-issue salience; on the second and third level, the frames of conflict, human and nonhuman interest, and economic interest provide more detail of how problem definition and cause are portrayed; on the third and fourth level, attribution of responsibility was dissected in a microscopic view of various social institutions. Horizontally, two related dimensions—remedy and responsibility—were examined on three geopolitical levels, indicating how the media in each culture define the “us vs. them” approach to mitigation and adaptation, given that territorial thinking still dominates climate change narratives in media messages (Olausson & Berglez, 2014).

Furthermore, the hybrid model of generic and issue-specific frames responds to the long-lasting fragmentation challenge by situating issue-specific frames within the framework of generic frames, a “convergence” practice that represents “an attractive aspect of framing research” (Borah, 2011, p. 256). While some scholars have criticized the excess of issue-specific frames that are too discriminatory for the complexity of climate change (Olausson, 2013), others have underscored the necessity to offer culturally situated and diversified picture of climate change communication (Olausson & Berglez, 2014), which are often imperceptible to generic frames. The framework of this study is an attempt to address both concerns in a progressive structure where generality cascades to specificity and cultural universality migrates to cultural sensitivity.

Method

The unit of analysis of this study was the individual news story, including hard news, editorials, or opinion pieces that center on climate change. The inclusion of all three types of stories was based on the rationale that the social construction of climate change has to be examined under the full spectrum of media representation rather than a selected sub-genre of news writing.

Sampling

This study opted for three major newspapers in the US (the New York Times, USA Today, and the Washington Post), and two major national newspapers in China (the People’s Daily and the China Daily). They were selected due largely to their large circulations and considerable social influence in their native countries. The three US newspapers’ circulations have stayed consistently in the top five league in the past decade (Alliance for Audited Media, 2014), and a great number of studies used them for analyzing climate change coverage and other environmental issues (Boykoff & Boykoff, 2004; Jones, 2006; Trumbo, 1996; Wilkins, 1993). The People’s Daily, with a daily circulation of more than one million worldwide (People’s Daily, 2008) during the sampled timeframe, is known as the official voice of the Central Committee of the Communist Party in China. The China Daily, with a daily circulation of more than 300,000 worldwide (ChinaDaily.com, 2007), is the largest English-language newspaper published in China that targets primarily English speakers.

Except for the People’s Daily, all the other newspapers were systematically archived by LexisNexis. The sampling began with using two key phrases—“climate change” and “global warming”—to search for stories published from 2005 through 2008 on climate change in the four newspapers. The searches were stratified by year and publication in order to increase sampling representativeness and to minimize over- or under-representation of particular segments in the sample. For example, using the advanced search options in LexisNexis, a search was performed to list all relevant stories published in USA Today in 2005, followed by a screening that picked only stories centered on climate change. These articles were recorded and numbered in sequence. Sampling for USA Today’s coverage in the following years followed a similar method, and the same approach was used to record articles in the other newspapers. After all, the articles on climate change were collected and systematic sampling was then executed by (1) evenly dividing the search results in each publication/year strata in N-story segments, depending on the number of recorded articles and a workable sample size, (2) randomly determining a number n from 1 to N, and (3) sampling every nth story. The use of stratified and systematic sampling techniques integrates the beauty of both methods: stratified sampling’s precision in proportion and systematic sampling’s capacity to deal with large sampling frames. Stories in the People’s Daily came from the newspaper’s online portal (www.people.com.cn).¹ Each sampled People’s Daily story indicated that it also appeared in the print edition.

The final sample consisted of 444 stories in total, with 242 from the US and 202 from China. Each publication contributed to the sample as such: 91 stories from the New York Times, 69 from the USA today, 82 from the Washington Post, 104 from the People’s Daily, and 98 from the China Daily.

Key Variables

Skepticism.

Skepticism was defined as an attitude of doubt toward any of the following areas: (a) the existence of climate change, (b) the anthropogenic nature of climate change, (c) validity of the scientific research on climate change, (d) individual, corporate, governmental, or organizational interests to promote the seriousness of climate change. The variable was measured at the ordinal level: (1) little to no skepticism, (2) skepticism as one side of a relatively balanced account, and (3) strong skepticism. For instance, an article introducing or promoting “An Inconvenient Truth” would fall under the first category; a story quoting climate scientists’ agreement and disagreement with Gore’s claims would be rated as “balanced”; an article questioning the IPCC’s scientific authority or Gore’s political agenda would be considered “skeptical.” In the pilot study, disagreement arose regarding whether the mention of skepticism in an article should automatically exclude it from the first category. After discussion among the coders, opinion articles that mentioned opposite views should be coded according to the authors’ views, rather than the view of the “straw man.”

Micro-issue salience.

Micro-issue salience variables include definition, cause, morality, and remedy. Coders answered binary “yes or no” questions that aimed to identify the presence of the frames. For example, the question “Does the story describe or discuss what climate change is and/or how it manifests itself?” detects whether problem definition is present in the story. Note that these variables may coexist in a story even though they are mutually exclusive at the conceptual level.

The variable mitigation boundary further distinguished mitigation means geographically at three levels: mitigation that is accomplished, being executed, or being developed by (1) the domestic country, or (2) foreign countries; mitigation that is accomplished, being executed, or being developed but cannot be identified as domestic or foreign was coded as (3) international.

Audience-based frames.

As mentioned earlier, a line of research has developed frame typologies that mirror how the audience understands news (Cappella & Jamieson, 1997; Iyengar, 1991; Neuman, Just, & Crigler, 1992; Norris, 1995; Schnell & Callaghan, 2005). These audience-based frames demand special attention to their measurement because of their somewhat elusive and interpretive nature. In general, the more latent a variable is, the more multifaceted it is, and consequently, the more difficult its measurement becomes (Riffe, Lacy, & Fico, 2005; Tankard, 2001). This study adapted the multi-item coding instrument developed by Semetko and Valkenburg (2000) to measure six audience-based frames: conflict, human interest, non-human interest, morality, economic consequences, and responsibility. Based on a factor analysis, only items with a factor loading higher than .6 were adapted. When coding the variable morality, for example, coders answered “yes” or “no” to three questions: (1) Does the story contain any moral or ethic message regarding climate change? (2) Does the story make reference to humanistic or religious morality regarding climate change? (3) Does the story offer specific social prescription about how to behave to deal with climate change? A composite score was later calculated to measure the frame’s prevalence.

The multi-item scales were designed to measure the prevalence of the six frames for two reasons. First, climate change coverage is often multilayered and therefore may have multiple frames in one story. Second, using multiple items to measure a variable helps to capture its conceptual complexity while showing the prevalence of coexisting frames in a story. For example, three yes/no questions measured the prevalence of the conflict frame. The prevalence score for the conflict frame was calculated by averaging the unweighted scores on the three individual items. Therefore, the prevalence scores of all six frame variables ranged from 0 (no frame prevalence) to 1 (strong frame prevalence). A binary coding strategy was used because it tended to yield higher intercoder reliability.

Attribution of responsibility variables.

Similarly, this study borrowed Semetko and Valkenburg’s (2000) multi-item design to measure the prevalence of attribution of responsibility. Six types of attribution are used—individual responsibility, government responsibility, industry responsibility, organization responsibility, and responsibility of humans in general—each has three items to measure compositely its visibility. To take government responsibility as an example, the coders answered “yes or no” to the following questions. (1) Does the story suggest that government(s) (domestically or internationally) has the responsibility to mitigate climate change? (2) Does the story describe government’s contribution to climate change mitigation, such as policy change and funding research programs? (3) Does the story describe what government(s) can do to help alleviate the negative impacts of climate change? Only stories that mentioned all three aspects of government responsibility would receive the highest composite prevalence score of 1.

The variable government responsibility boundary distinguished whether mitigation responsibility was attributed to domestic, foreign, or international governments. For instance, the United Nations was coded as an international government; international cooperation initiated by governments of different countries was coded as international governments.

Validity and Reliability

A two-stage process was conducted to examine the validity and reliability of the coding. A pilot study was first done to test the multi-item frame identification instruments and intercoder reliability. A pilot sample of 70 stories including newspaper articles from all five sources was drawn in a different timeframe, in order not to interfere with the coding of the final sample. A bilingual coder, who reads English and Chinese proficiently, coded the pilot sample. Then a varimax-rotated Principal Component Analysis showed solid factorial validity—a form of construct validity—of both audience-based and responsibility frames with all the factor loadings higher than .70. For the instrument’s internal consistency, the Cronbach’s alphas ranged from .80 to .95 (six audience-based frames and five responsibility frames), indicating satisfactory consistency of the instruments.

Then two other bilingual coders coded the same set of stories. All the variables except for year and country were processed to test intercoder reliability. A satisfactory omnibus Cohen’s Kappa of .85 was reached with item-level Ks ranging from .70 to .97. The three coders discussed the items with lower Ks and made minor changes to minimize wording ambiguity.

The fact that both the audience-based frames and responsibility frames derived from framing effect research (Iyengar, 1991; Neuman et al., 1992; Price, Tewksbury, & Powers, 1997) establishes what Krippendorff (2004) called “semantic validity.” This kind of validity accesses “the degree to which the analytical categories of text correspond to the meanings these texts have for particular readers or the role the play within a chosen context” (p. 323). It is high when the “users of the texts … serve as sources of validating evidence of the categories that a content analysis employs” (p. 323).

The second stage involved another round of reliability tests, which were done during the final coding after the entire sample had been collected. One coder coded the sample (N = 444), and the other two coders coded a set of randomly selected articles (n = 89). Again, a satisfactory omnibus K of .83 was reached with item-level Ks ranging from .74 to .92 (see Appendix).

Results

RQ1 asked how skepticism toward climate change varied by country from 2005 to 2008. A 2 (US vs. China) × 4 (2005–2008) factorial ANOVA showed a significant main effect of country, F(1, 436) = 48.12, p < .001,  = .10. On a scale of 1 (non-skeptical) to 3 (very skeptical), the US newspapers featured more skepticism (M = 1.28, SD = .52) than their almost non-skeptical Chinese counterpart (M = 1.05, SD = .23). Even though year also had a significant main effect, F(3, 436) = 6.18, p < .001,  = .04, a strong interaction between country and year, F(3, 436) = 5.15, p < .01,  = .03, offered a more detailed picture. As Figure 2 visualizes, while the Chinese newspaper remained non-skeptical over the four years, the US newspapers became less skeptical toward climate change. A trend test confirmed this longitudinal decline, t (238) = –4.66, p < .001.

Figure 2 Longitudinal changes in skepticism toward climate change in the US and Chinese newspapers.

Note: The prevalence of skepticism ranges from 1 (non-skeptical) to 3 (strongly skeptical).

RQ2 looked at how the uses of micro-issue salience (definition, cause, morality, and remedy) varied by country. The two countries differed only in definition, χ²(1, N = 444) = 8.72, p < .01, φ_c = .14; the Chinese newspapers were more likely to define climate change (64.4% of all Chinese stories) than were the US newspapers (50.4% of all US stories). As seen in Figure 3, in both countries, more than half of the stories touched on definition and remedy, whereas less than 20% of them mentioned its causal and moral dimensions.

Figure 3 Micro-issue salience by country.

Note: Sample size N = 444. Percentages were calculated within each country and do not add to 100% because multiple micro-issues may be present in one story.

RQ3 distinguished among domestic, foreign, and international mitigation achievement and looked at how their coverage varied. Of the US newspaper stories that mentioned mitigation achievements (n = 144), 66.7% reported mitigation achievements domestically in the US, whereas only 12.5% described foreign achievements and 20.8% international achievements (see Table 1). Between-country differences were detected, χ² (2, n = 260) = 8.62, p < .05, φ_c = .18. Although one in two (49.1%) Chinese news stories mentioned domestic achievement, two in three US stories (66.7%) did so, indicating even closer attention to domestic accomplishment.

Table

Variables	Variable-level Cohen’s Kappa	Item-level Cohen’s Kappa
Skepticism	.74
Micro-issue salience
Definition	.83
Cause	.76
Morality	.80
Remedy	.89
Mitigation boundary	.91
Audience-based frames
Conflict	.81	.81, .84, .79
Human interest	.90	.90, .88, .91
Non-human interest	.87	.88, .87, .85
Morality	.76	.75, .74, .78
Economic	.85	.83, .89, .84
Responsibility	.77	.77, .75, .80
Responsibility attribution
Individual	.76	.74, .75, .79
Government	.88	.86, .88, .91
Industry	.84	.84, .89, .79
Organization	.80	.80, .77, .84
Scientist	.90	.89, .92, .88
All humans	.76	.75, .75, .79
Government responsibility boundary	.87

Table 1 Newspaper coverage of mitigation boundary by country.

	Domestic	Foreign	International	Total
US Newspapers	96 (66.7%)^a	18 (12.5%)	30 (20.8%)	144 (100.0%)
Chinese Newspapers	57 (49.2%)	26 (22.4%)	33 (28.4%)	116 (100.0%)
Total	153 (58.9%)	44 (16.9%)	63 (24.2%)	260 (100.0%)

Note: χ²(2, N = 260) = 8.62, p (two-tailed) <.05; Cramer’s V = .18.

^a N (Row%).

RQ4 probed into how much variances of audience-based frames (conflict, human interest, non-human interest; economic, morality, and responsibility) could be explained by country and year. A 2 (country) × 4 (year) MANOVA shows that country had a significant multivariate main effect, Wilk’s Lambda = .85, F(6, 431) = 12.62, p < .001,  = .15. As follow-up tests revealed (see Table 2 for F, p, and values), the conflict frame was twice more prevalent in the US newspapers than it was in the Chinese press (M = .39 vs. M = .18). On the other hand, the Chinese newspapers focused more on human interest (M = .32 vs. M = .23), non-human interest (M = .42 vs. M = .21), and responsibility (M = .58 vs. M = .52).

Table 2 Factorial analyses of variance of audience-based frames and attribution of responsibility by country and year.

Variable		SS	df	MS	F	p
Country	Audience-based frames
	Conflict	5.70	1	5.70	32.62	.000***	.07
	Human interest	1.24	1	1.24	7.33	.007**	.02
	Non-human interest	3.89	1	3.89	22.02	.000***	.05
	Morality	.17	1	.17	1.58	.21	.00
	Economic	.12	1	.12	1.80	.18	.00
	Responsibility	1.58	1	1.58	7.54	.006**	.02
	Attribution of responsibility
	Individual	.04	1	.04	.47	.494	.00
	Government	.19	1	.19	.86	.354	.00
	Industry	1.13	1	1.13	14.15	.000***	.03
	Organization	.41	1	.41	7.81	.005**	.02
	Scientist	.25	1	.25	2.71	.101	.01
	All humans	.00	1	.00	.01	.909	.00
Year	Audience-based frames
	Conflict	1.29	3	.43	2.46	.063	.02
	Human interest	.53	3	.18	1.05	.371	.01
	Non-human interest	2.66	3	.89	5.01	.002**	.03
	Morality	.47	3	.16	1.44	.231	.01
	Economic	.52	3	.17	2.55	.055	.02
	Responsibility	7.65	3	.26	12.18	.000***	.07
	Attribution of responsibility
	Individual	1.08	3	.36	4.60	.004**	.03
	Government	1.72	3	.58	2.56	.055	.02
	Industry	.31	3	.10	1.30	.276	.01
	Organization	.14	3	.05	.89	.447	.01
	Scientist	.05	3	.02	.19	.901	.00
	All humans	.13	3	.04	.76	.520	.01
Country × Year	Audience-based frames
	Conflict	.35	3	.12	.66	.575	.01
	Human interest	.59	3	.20	1.16	.326	.01
	Non-human interest	.53	3	.18	1.00	.391	.01
	Morality	.94	3	.31	2.85	.037*	.02
	Economic	.63	3	.02	.31	.818	.00
	Responsibility	2.59	3	.86	4.13	.007**	.03
	Attribution of responsibility
	Individual	.03	3	.01	.02	.950	.00
	Government	.47	3	.16	.70	.555	.01
	Industry	.27	3	.09	1.12	.339	.01
	Organization	.10	3	.03	.37	.776	.00
	Scientist	.10	3	.03	.37	.776	.00
	All humans	.12	3	.04	.72	.539	.00

Note: Two-tailed tests of significance:

*** p < .001

** p < .01

* p < .05.

Significant country–year interactions in morality and responsibility provided two contrasting pictures, one with divergence and the other with convergence. As seen in Figure 4, in the Chinese newspapers, climate change was increasingly framed as a moral issue since 2006, whereas the US newspapers showed a sharp decline in this regard since 2005. The two trends continued to diverge till 2008, when the morality frame was three times more prevalent in China than in the US (M = .27 vs. M = .08). The responsibility frame, however, was nearly invisible in the US newspapers in 2005 (M = .12) but sharply increased over the following years. On the other hand, responsibility was rather prevalent in the Chinese newspaper in 2005 (M = .57), dropped in 2006, but rebounded in the following two years. In 2008, the frame’s prevalence on both countries converged to a new high (M = .68 and M = .67 in the US and China, respectively).

Figure 4 Interactions of morality and responsibility frames by country and year.

Note: Frame prevalence for morality and responsibility ranges from 0 to 1.

RQ5 examined how attribution of responsibility (individual, government, organization, industry, scientist, and all humans) differed by country and year. As Figure 5 shows, the newspapers in both countries attributed mitigation responsibility mostly to government, followed by scientist as the second largest responsible group (in the US, organization tied with scientist at the second place). The multivariate main effect for country, F(6, 431) = 4.17, p < .001,  = .06, was accompanied by significant univariate effects on industry, F(1, 436) = 14.15, p < .001,  = .03, and organization, F(1, 436) = 7.81, p < .01,  = .02. Specifically, industry was attributed to fight climate change more heavily in the US than it was in China (M = .14 vs. M = .03); similarly, organization was attributed more often in the US as well (M = .08 vs. M = .02). Year had a univariate effect for on individual responsibility, F(3, 436) = 4.60, p < .01,  = .03, showing two parallel increases in both countries over the four years. No country and year interaction was found (see Table 2 for F, p, and values).

Figure 5 Attribution of responsibility by country.

H1 distinguished among domestic, foreign, and international governments, and hypothesized that the Chinese news media were more likely than their US counterpart to attribute mitigation responsibility to non-domestic governments. A crosstab analysis was done on newspaper articles of both countries that attributed responsibility to governments (n = 225). Of the Chinese newspaper articles (n = 112), a majority (84.8%) of them attributed the responsibility of mitigating climate change either to specific foreign governments (for example, the US government) or to international governments, whereas a majority (77.0%) of the US newspaper articles (n = 113) attributed the responsibility to domestic governments (the US Government). Significant differences were detected, χ² (2, N = 225) = 8.62, p < .001, φ_c = .62, indicating a substantial contrast between the Chinese and US newspapers’ coverage (See Figure 6). Therefore, H1 was supported.

Figure 6 Attribution of responsibility to types of government.

Discussion

This study confirms and extends the existing literature on climate change communication. Its longitudinal design and the use of theory-driven analytical tools distinguish itself from earlier studies on climate change representation in China, such as “snapshot” analyses of Chinese newspapers (e.g. Jia, 2007) and descriptive studies largely based on qualitative observation and interviews (e.g. Tolan, 2007). The study’s comparative approach shows culturally situated contrasts between the two most important stakeholders in climate change rhetoric.

The dwindling skepticism toward climate change from 2005 to 2008 is in line with the “evolutionary” shift (Boykoff, 2007) to a less skeptical tone in the US news media, while such skepticism was barely visible in the Chinese media (Jia, 2007; Painter & Ashe, 2012; Tolan, 2007). One would easily attribute the cross-nation discrepancy to the “informational bias” criticism toward the US media (Antilla, 2005; Boykoff, 2006) and the mouth piece identity of the Chinese media. However, researchers should be cautious about making this simplistic explanation without looking into each country’s state-media dynamics.

The underlying rationale of journalistic objectivity is for a politically independent press to serve as a watchdog. For a highly politicized and polarized issue like climate change (Gallup, 2014; McCright & Dunlap, 2011), leaning toward either side would readily suggest political inclination, regardless of how science navigates the ontological dimension of issue. Therefore, American journalists were implicitly encouraged to “under-report an issue than be perceived as having a political or environmental agenda” (Friel, 2005, p. 51). In fact, the decreased representation of skeptical voices from 2005 to 2008 already caused growing public concerns. According to a 2009 Gallup poll, significantly more Americans thought the news media exaggerated the seriousness of global warming than said it underestimated the threat: 41% vs. 28%, reversing the gap in 2006, 30% vs. 38%. Heavy coverage of the political dimension of climate change leads to divorcing science from the public discourse, which then warrants the necessity to cover “all sides” so as to defend the political detachment of the press, which eventually further perpetuates the politicalization of the issue. In the case of climate change representation, rather than a promise to the marketplace of ideas where “truth would prevail” (Milton, 1886), the practice of balanced reporting looks more like a delusive self-fulfilled prophecy that reinforces what political powers, not the science community, think is true.

Conversely, the Chinese press’ unequivocal support for climate change reflects in part the lack of counterbalancing powers in the one-party state. But the question of why the central government took such a stance on the issue needs to be answered in the context of the political reality in China. Note that protecting the Communist Party’s sovereignty has been one of the guiding principles of environmental policy-making (Yu, 2008). Despite the absence of skepticism toward climate change in the Chinese media, climate change is now a politicized issue, as Chinese media scholar Xiguang Li contended, because the government has embraced the US Democratic Party line on the issue to gain “public support by being seen to care for matters” that affect the ordinary citizen (de Burgh & Rong, 2012, p. 32).

The politicalization of climate change in China also finds support in this study. As reviewed earlier, the government has been consistently calling for leading commitment from developed countries as well as their technical and financial assistance to help developing countries combat climate change. The stance can be seen in (1) the newspapers’ focus on domestic mitigation achievement and (2) the heavy responsibility attributed to foreign and international governments. This theme of “domestic accomplishment, international responsibility” also echoes Chinese environmental journalists’ concern that holding the government accountable for their mitigation efforts was the most contentious aspect of environmental reporting (de Burgh & Rong, 2012).

The US Government, instead of foreign and domestic governments, was attributed as the primary source of mitigation in the US newspapers. The practice may have stemmed partly from the Americans media’s adoption of a political focus on elites when reporting global issues and their lack of interest in articulating international relations in those stories (Hafez, 2007). Industry and NGOs were held responsible more often in the US but received little attention in the Chinese media. It seems that climate change mitigation is seen as an issue that involves a multitude of social institutions, whereas in China, its efficacy is largely at the government’s discretion.

Similar cultural sensitivity should also be used when interpreting the convergence in the morality frame (see Figure 4). The rise of the morality frame in the US newspapers echoes other findings that both secular and religious morality often appeared in recent climate change coverage in the US media (Hart & Feldman, 2014), in hopes of reducing polarization in public perception (Feinberg & Willer, 2013). In China, the consistent and one-dimensional presence of the morality frame focused instead on developed countries’ moral obligation to “fix” the problem they had created. The position seems logical in the context of state-led nationalism (Zheng, 1999) and the Communist Party’s “self-proclaimed status as the monopoly holder of truth” and moral standards (Zhao, 2004, p. 181). It can also be seen as a continuation of China’s historical introspection as a victim of Western development during its colonial past in the nineteenth and early twentieth century. Interestingly, similar colonial experiences may be strongly associated with a nationalistic tone in climate change communication across borders, regardless of differences in sociopolitical and media systems. For example, India, an enormous economy with growing carbon emission, tends to believe that “there is a neo-colonial desire to suppress India’s growth through unrestrained climate change impacts and restrictive policy” (Billett, 2010, p. 15).

Of particular concern is the finding that the Chinese media’s active role in defining climate change and disseminating scientific knowledge in relation to their lack of scientific training and direct access to scientific literature (de Burgh & Rong, 2012; Jia, 2007). Under increasingly sophisticated Internet censorship and restricted access to academic search engines like Google Scholar (Makinen, 2014), original scientific research, along with ideologically deviant materials, may fall victim to state control, making Chinese journalists more susceptible to conforming to the official rhetoric and misinforming the public, even if the opposite is what they intend.

The hybrid analytical framework provided two advantages for this study. First, it helped the research questions and the hypothesis progress coherently. Second, it integrated the universality of the generic frames and the depth of the issue-specific frames, revealing both the commonality and the sociopolitically driven differences in the two countries’ coverage. On a larger scale, the theoretical contribution of the framework resides in its modular design. The framework answered Borah’s (2011, p. 256) call for consolidating frames by making conceptual connections among existing frame families and turning more specific frames into “modules” of more generic frames. The modular structure not only offers flexibility and expandability for an analytical framework but also encourages theorists to consider the interoperability of the frames they are developing with other frames. For example, under the modular structure, theorists may ask “how do my frames connect with existing frames and perhaps serve as one of its modules?” or “how accommodating or expendable are my frames for future researchers to add modules to them?” This practice, it is hoped, would discourage capricious frame development and frame fragmentation, resulting in better cross-study comparability and more robust theory-building.

Several limitations of this study suggest directions for further research. First, even though the four-year time frame has allowed us to see interesting longitudinal trends, a longer time frame encompassing multiple attention cycles would provide insight of how the media interacted differently with external factors, such as political events, extreme weather, major climate science achievement, through intercycle comparisons.

Second, the language proficiency of the three bilingual coders eliminated the need for translation and therefore benefited the intercoder reliability, but their similar education background as US-trained Chinese graduate students may have led systematic bias in the coding process. An attack against the US Government published in the China Daily, for example, would probably suggest different interpretive cues and elicit different emotional responses from American and Chinese citizens. Moreover, the coders’ study in the US might also cause them to react differently from native Chinese who do not have the experience. For future studies, we suggest a more thorough consideration of the target audiences of analyzed publications and a more diversified mix of cultural backgrounds of the coders.

Third, only two major Chinese newspapers were included. Even though they more or less represent the Chinese Government’s voice, recent studies have found more editorial freedom on environmental issue in local and regional media outlets (de Burgh & Rong, 2012; Jiang, 2010). Chances are that the heightened public concerns about environmental issues, which are often suppressed in national outlets (Jiang, 2010), created a centrifugal force that allows those smaller outlets to deviate from the official narrative. Moreover, the lack of coverage of NGOs in China made us wonder if they were under-represented, as scholars documented rising influence of environmental NGOs (Yang, 2005, 2008) though online activism (Stalley & Yang, 2006; Yang, 2003). Attention to their online presence would provide not only a wider view of environmental agents in China but also an opportunity to examine how they shape, through collaboration or competition with established voices, the public discourse of climate change in the country.

Fourth, this study examined only the text of newspaper stories. Since audiences, especially the younger generations, are learning about climate change increasingly through visual media (Schäfer & Schlichting, 2014), a focus on such media across platforms would bring researchers to a new frontier of how visuals translate to the public perception of climate change and how they are produced from a news production standpoint.

Acknowledgements

The author thanks the anonymous reviewers for their insightful comments and the following individuals for their guidance throughout the research project: Dr Dennis Lowry, Prof. William Freivogel, Dr Thomas Johnson, Dr Xigen Li, and Dr Uche Onyebadi. Special thanks also go to Dr Oliver Witte for proofreading the manuscript.

Notes

1. Two terms in Chinese – “global warming” (“Quan Qiu Bian Nuan” 全球变暖) and “climate change” (“Qi Hou Bian Hua” 气候变化) —were used to search for relevant articles. Search results rank ordered by year and relevance allowed similar screening and sampling methods to perform.

Appendix: Reliability for coded items

Note: Each variable under audience-based frames and responsibility attribution was measured by three items compositely and therefore has three item-level Kappa scores.