Is Twitter an Alternative Medium? Comparing Gulf Coast Twitter and Newspaper Coverage of the 2010 BP Oil Spill

Abstract

This study compares Gulf Coast journalists and Twitter users’ coverage of the BP oil spill. In addition to examining authors’ attitudes toward and coverage of the BP oil spill, the study examines community-level variables that shaped attitudes and coverage. The community structure literature has suggested that news media in smaller, more homogeneous communities, which are economically dependent on a polluting industry (as are many communities along the Gulf Coast), are more reticent to be critical in their coverage of pollution. Scholars have suggested, though, that the Internet transcends local geography and that the Internet is more open to alternative perspectives. This study suggests, though, that while the distribution of online content may make local geography less relevant, its production is still rooted in local communities. As a result, Tweets about the oil spill were shaped by many of the same social and economic forces that shaped journalists’ coverage.

Keywords

oil spill social media Twitter community structure alternative media

Even if the news media fail to foresee a crisis, fulfilling their surveillance function (Lasswell, 1948) would mean that they should at least raise questions about the causes and consequences of the crisis. In the case of the 2010 BP oil spill, those causes included cost- and corner-cutting by BP, lax regulatory oversight, and energy policies that encourage the pursuit of oil in increasingly remote reserves without adequate regard for the risks involved. (The Deep Water Horizon oil rig had been described as “fail safe.”)¹ Consequences included both environmental and economic damage to an already fragile Gulf Coast (Bureau of Ocean Energy Management, Regulation, and Enforcement, 2011; Freudenburg & Gramling, 2011).

However, consistent with previous research findings, it is expected that local newspapers’ coverage of the crisis would be tempered by the community structure in which those newspapers are embedded. Specifically, newspapers in small, homogeneous communities economically dependent on polluting industries would be reticent to raise critical questions about the causes and consequences of environmental contamination (Donohue, Olien, & Tichenor, 1985; Griffin & Dunwoody, 1995, 1997; Tichenor, Donohue, & Olien, 1980).

In addition to being among the year’s most followed domestic news stories, the 2010 BP oil spill was also the most discussed topic on social media for more than a month following the spill (Pew Research Center’s Project for Excellence in Journalism, 2010). It has been suggested that the Internet, which would include social media, differs from traditional news media in several important ways. First, it “overrides geography” (Reese, Rutigliano, Hyun, & Jeong, 2007, p. 235), freeing online media from the social constraints of local communities, as well as lowering the distribution constraints that traditionally have exiled alternative media to a “radical ghetto” (Downey & Fenton, 2003, p. 199). Furthermore, online networks, such as those forged by social media, make new forms of radical organizing and action possible (Dahlberg, 2007). Thus, Twitter has the potential to serve as an alternative medium, challenging those social, economic, and political influences that restrict critical coverage in the mainstream press, especially in smaller communities that are more economically reliant on the oil industry (Atkinson, 2010; Atton, 2002).

An alternative medium is best defined in contrast to a mainstream medium (Harcup, 2003). Thus, this study compares the extent to which Gulf Coast newspaper reporters and social media users raised questions about the causes and consequences of the disaster. This study focuses on Gulf Coast communities because the BP oil spill was a particularly salient and controversial issue in many of these most affected communities.

Due to the lack of previous literature about how community structure might affect social media coverage of the spill, this study first proposes a set of hypotheses about Gulf Coast newspapers’ coverage of the disaster. It then compares that coverage, as well as the social, economic, and political factors that shaped the newspaper coverage, to Gulf Coast Twitter users’ tweets about the spill. If Twitter acted as an alternative medium, there should be significant differences in both the content of, and factors that shaped, newspaper stories and tweets about the BP oil spill.

Literature Review

Journalists’ Political and Environmental Attitudes

Previous studies have suggested that journalists are not able to completely separate their personal beliefs from the stories they produce (D’Alessio & Allen, 2000; Patterson & Donsbach, 1996). Thus, it is expected that journalists’ political and environmental attitudes will to some degree influence their attitudes toward the oil industry, which in turn will affect their coverage of the BP oil spill.

Political and environmental attitudes are different constructs. Political attitudes represent beliefs about how political affairs should be structured to bring about desired social and economic ends (Jost, Federico, & Napier, 2009). Environmental attitudes describe what individuals believe about the natural world and humans’ proper relationship with nature (Dunlap, Van Liere, Mertig, & Jones, 2000). Environmental and political attitudes, however, are correlated. Research has found that political conservatives are generally less concerned than more liberal individuals about the environment, engage in fewer pro-environmental behaviors, and oppose policies meant to improve environmental conditions (Dunlap et al., 2000).

Feygina, Jost, and Goldsmith (2010) suggested that conservatives’ environmental attitudes may be due to the fact that a conservative ideology favors more structure and order and that ideological conservatives are more likely to engage in system justification. Feygina et al. defined system justification as a “motivation to perceive . . . systems as fair, legitimate, beneficial, stable, as well as desire to maintain and project the status quo” (p. 327). As the authors pointed out, acknowledging environmental problems, such as Americans’ heavy reliance on fossil fuels and lax regulatory oversight of oil drilling, requires acknowledging a problem with the status quo. Action to address environmental problems, such as using more alternative energy sources, requires changing the status quo, which is the antithesis of the stability conservatives prefer. It is not surprising then that attitudes toward the BP oil spill sharply divided the American public down partisan lines, with conservatives expressing much less anger toward BP following the oil spill than political liberals (Pew Research Center for the People & the Press, 2010). As a result, this study posits that more conservative journalists will be less concerned about the environment than more liberal journalists, and will have more positive attitudes toward the offshore oil drilling industry, even after the BP oil spill. In turn, these journalists with more positive attitudes toward the oil industry will also write more positive stories about the BP oil spill.

Community Structure and Coverage of the Environment

Journalists, though, are not completely independent individuals. They are also influenced by the social structure of the communities in which they work. According to Tichenor et al. (1980; Donohue et al., 1985), in smaller, more homogeneous communities, conflicts are tightly controlled by a small group of community leaders. As a result, conflict, including environmental contamination, is downplayed in the news. In larger, more heterogeneous communities, different racial groups, religious groups, industries, and so forth vie for and share social power. Furthermore, larger, heterogeneous communities are more dependent on the media to communicate across these diverse groups and serve a feedback function. Thus, media in larger, heterogeneous communities have greater latitude to openly and critically cover community conflicts (Donohue et al., 1985; Tichenor et al., 1980).

Structural pluralism—typically measured based on a community’s population and the distribution of its residents across different census categories, for example, income ranges, industries of employment, race, and so forth (Nah & Armstrong, 2011)—has previously been shown to affect coverage of environmental contamination (Griffin & Dunwoody, 1995, 1997; Tichenor et al., 1980). Griffin and Dunwoody (1997) found that newspapers in more pluralistic communities are more likely to focus on scientific evidence of human health risks associated with environmental contamination than were newspapers in more homogeneous communities. In a separate study, however, Griffin and Dunwoody (1995) found that a community’s economic reliance on polluting manufacturing, measured as the percentage of local residents employed in manufacturing, is a stronger predictor than is structural pluralism of how a newspaper will cover an environmental controversy. In their analysis of U.S. newspapers in communities of various sizes, they found that even in the most structurally pluralistic communities, newspapers in communities that relied heavily on manufacturing for local employment were less likely in their coverage to link a report about toxic releases to risks to human health.

In addition to examining how journalists’ definitions of their professional roles shaped their attitudes toward, and coverage of, the BP oil spill, Watson (2014a) also examined how journalists’ coverage of the BP oil spill was shaped not only by their political and environmental attitudes but also by the community in which they worked. Watson (2012) previously found that more politically conservative, less environmentally concerned Gulf Coast newspaper reporters who lived in communities that were more reliant on the oil industry for local employment had more positive attitudes toward the oil industry following the BP oil spill. Building on that earlier survey, Watson (2014a) found that these journalists wrote more positively valenced stories about the disaster. This current study builds on these previous studies by comparing newspaper journalists’ and Twitter users’ coverage of the BP oil spill, examining both whether Twitter served as an alternative medium by raising issues downplayed by the mainstream press, and whether tweets about the oil spill were shaped by different social, economic, and political factors.

Twitter as an Alternative Medium

Critics of mainstream media (MSM) have argued that as a function of their goals of reaching the largest general-interest audience possible and selling commercial advertising, mainstream commercial newspapers omit voices who are critical of the status quo (consistent with the structural pluralism hypothesis; Hamilton, 2000; Harcup, 2005). However, there is a lack of consensus as to how to define alternative media, in part because what some might label “alternative” media (e.g., metro alternative weeklies) oftentimes share significant characteristics with MSM.

There is general agreement that the fundamental characteristic of alternative media is that they create an alternative discourse by not only adhering to different production processes but also creating content that significantly differs from MSM’s coverage (Atkinson, 2010; Atton, 2002). In terms of production, alternative media are less hierarchical, more participatory, and often less commercially driven. Alternative media also produce an alternative discourse by using sources and serving audiences who are omitted, or at least marginalized, in mainstream news coverage (Hamilton, 2000; Harcup, 2003, 2005). That is, alternative media can be primarily defined in contrast to MSM (Harcup, 2003). Thus, this study compares Twitter to mainstream newspaper coverage.

Precisely because they are less commercially oriented, alternative media traditionally have not been able to achieve mass distribution. Thus, they have had a marginal existence within the larger political discourse. Widespread, inexpensive distribution via the Internet, however, has been touted as having the potential to revolutionize alternative media (Downey & Fenton, 2003). It has also been stated that the Internet’s virtual network “overrides geography” (Reese et al., 2007, p. 235). This suggests that the Internet should be unaffected by those local social and economic factors that constrain traditional local news media’s coverage of controversial issues, especially in smaller communities more economically reliant on particular industries.

Others contend that the realities of the Internet fall short of its radical potential. Writing about the “corporate colonization” of the Internet, Dahlberg (2005) pointed out that despite the diversity of content from noncommercial sites, American Internet users disproportionately visit mainstream commercial websites. Dahlberg concluded that “(t)his situation goes against the vision of the Internet operating as an alternative medium to the mass media, as a space where positions and critique excluded offline are foregrounded” (p. 172). Wu, Hofman, Mason, and Watts (2011) came to similar conclusions about the fact that despite having 100 million-plus active users on Twitter, just 20,000 “elite” users—celebrities, corporations, major media—create more than half of tweets. The authors concluded that “while attention that was formerly restricted to mass media channels is now shared amongst other ‘elites,’ information flows have not become egalitarian by any means” (p. 6).

As Sparks (2001) pointed out, the Internet is large and diverse enough that if one seeks them out, one can certainly find examples of the Internet serving as an alternative medium. However, this study is not interested in activities at the fringes, but rather those perspectives that dominate on the social networking website. Thus, this study focuses on the most followed users on Twitter in a given community, who are most likely to act as opinion leaders of the public discourse.

Analyzing Coverage

This study examines four variables in the media’s coverage of the 2010 BP oil spill: what larger issues were raised in the coverage, the coverage’s tone (positive, negative, or neutral), whether authors used episodic or thematic frames, and authors’ reliance on official or unofficial sources. These variables are not only associated with how the audience may perceive an issue but have also been shown by previous studies to be affected by the degree of structural pluralism in a community where a newspaper is based (Donohue et al., 1985; Griffin & Dunwoody, 1995, 1997; Hindman, Littlefield, Preston, & Neumann, 1999; Tichenor et al., 1980).

The first two variables are based on the second-level agenda-setting literature, which suggests that the media’s coverage not only focuses the public’s attention on specific topics (e.g., the oil spill) but also affects how the public evaluates those subjects (Kim, Scheufele, & Shanahan, 2002; Sheafer, 2007). For example, if the media’s coverage of the oil spill were to primarily focus on BP’s responsibility for the oil spill, the audience may be more likely to focus on BP when evaluating the spill. Furthermore, if the media primarily focus on negative aspects of the oil company’s response, the audience may be more likely to have negative attitudes toward BP’s response to the crisis.

This study also examines whether newspaper stories and tweets were framed episodically or thematically. Entman (1993) provided perhaps the most succinct definition of framing:

To frame is to select some aspects of 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/or treatment recommendation for the item described. (p. 52)

According to Iyengar (1991, 1996), an episodic frame treats a news event as an isolated story; it focuses on, for example, the immediate decisions and events that led to the explosion aboard the Deep Water Horizon oil platform. A thematic frame explores larger causes and consequences underlying a particular issue. A thematic story about the BP oil spill would, for example, put the disaster in the context of BP’s overall safety record, the government’s inspection record, and broader questions about the proper role of government regulation and oversight.

The media’s use of particular frames is significant because those frames can be associated with audiences’ moral evaluations of social conflicts and policy recommendations intended to resolve those conflicts (Hart, 2010; Kim et al., 2002). Iyengar (1991, 1996), for example, found that thematic frames are more likely to be associated with the attribution of societal responsibility for a given issue. Hart (2010) also found that thematic framing of environmental issues was positively associated with public support for environmental regulations, specifically additional taxation of greenhouse gas-emitting industries.

Lastly, this study examines the sources mentioned in authors’ newspaper stories or tweets, as these sources help define the tone and framing of the coverage. In Smith’s (1993) study of coverage of the 1989 Valdez Spill, he found that government and oil industry officials made up 60% of journalists’ sources. (Scientists comprised only 8.3% of sources, and environmentalists only 7.8%.) In addition, government and industry sources were most likely to say that the Valdez crisis had been overblown, and rate Exxon’s response to the crisis favorably. Entman and Rojecki (1993) found very similar results in their analysis of news coverage of the antinuclear power movement. Newspapers in smaller, more homogeneous communities are least likely to use unofficial sources—those sources most likely to raise critical questions—in their coverage (Hindman et al., 1999).

Hypotheses and Research Questions

Due to the lack of literature exploring the effects of these community-level variables on social media, this study first sets up a series of hypotheses and research questions about Gulf Coast newspapers’ coverage of the 2010 BP oil spill. These results are then compared with tweets about the oil spill, answering the overarching research question: In the context of the 2010 BP oil spill, did Twitter act as an alternative medium?

Griffin and Dunwoody (1997) found that coverage of environmental contamination is more likely to be linked to industry and be more critical of industry in more structurally pluralistic communities. Thus, it is hypothesized that coverage of the BP oil spill will also differ based on a community’s degree of structural pluralism. Journalists in more pluralistic communities will be more likely than journalists in less pluralistic communities to write negative stories about the BP oil spill; that is,

Hypothesis 1a (H1a): Journalists in more pluralistic communities will be more likely to focus on BP’s role in the disaster.

Hypothesis 1b (H1b): They will be more likely to use thematic frames.

Hypothesis 1c (H1c): They will be more likely to write negative stories.

Hypothesis 1d (H1d): They will be more likely to use unofficial sources.

Griffin and Dunwoody (1995) found that even in more pluralistic communities, newspapers in communities that relied more heavily on manufacturing for local employment were less likely to link local environmental contamination to local industry and frame that coverage critically. Watson (2014a) also found that journalists who work in communities that are reliant on the oil industry for employment wrote more positively valenced stories about the BP oil spill. Thus, it is hypothesized that newspaper reporters in communities that rely on the oil industry more heavily for local employment will be less likely to focus on BP’s role in the crisis and to frame its coverage critically.

Journalists in communities with greater percentages of the local workforce employed in the oil industry will write more positive stories about the BP oil spill than journalists in communities in which a lower percentage of the workforce is employed in the oil industry; that is,

Hypothesis 2a (H2a): Journalists in communities that are more reliant on the oil industry for local employment will be less likely to focus on BP’s role in the disaster.

Hypothesis 2b (H2b): They will be less likely to use thematic frames.

Hypothesis 2c (H2c): They will be less likely to write negative stories.

Hypothesis 2d (H2d): They will be less likely to use unofficial sources.

In addition, this study tests the hypothesis that community-level variables indirectly affect, along with journalists’ personal political and environmental attitudes, journalists’ BP disaster coverage by directly shaping their attitudes toward offshore oil drilling. Those attitudes then in turn shape the stories these journalists write. Researchers consistently find that attitudes about environmental issues, including the BP oil spill, are divided along partisan lines: Both in the American public generally, and among journalists specifically, more politically conservative individuals expressed greater anger toward BP following the oil spill than did more liberal individuals (Pew Research Center for the People & the Press, 2010; Watson, 2014a). According to Feygina et al. (2010), this is in part because political conservatives value structure and stability. As a result, they engage in system justification, downplaying environmental concerns, such as the environmental consequences of the BP oil spill, that require acknowledging a problem with the status quo and a need to change it. Because previous studies of potential bias in journalists’ work have found that journalists are not capable of completely filtering their personal beliefs out of their news coverage (D’Alessio & Allen, 2000; Patterson & Donsbach, 1996), it is hypothesized that journalists’ political and environmental beliefs will affect their attitudes toward BP following the spill and this will in turn shape their coverage of the disaster:

Hypothesis 3a (H3a): Journalists in more structurally pluralistic communities will have more negative attitudes toward the oil industry than journalists in less pluralistic communities.

Hypothesis 3b (H3b): Journalists in communities with greater percentages of the local workforce employed in the oil industry will have more positive attitudes toward the oil industry than journalists in communities that rely less on the industry for local employment.

Hypothesis 4 (H4): More politically conservative journalists will have more positive attitudes toward the oil industry following the BP disaster than less conservative journalists.

Hypothesis 5 (H5): Journalists more concerned about the environment will have more negative attitudes toward the oil industry following the BP disaster than journalists who are less concerned about the environment.

In turn, journalists with more positive attitudes toward the oil industry following the BP oil spill will be less likely than journalists with negative attitudes toward the oil industry to write negative stories about the disaster; that is,

Hypothesis 6a (H6a): Journalists with more positive attitudes toward the oil industry will be less likely to focus on BP’s responsibility for the disaster.

Hypothesis 6b (H6b): They will be less likely to use thematic frames.

Hypothesis 6c (H6c): They will be less likely to write negative stories about the spill.

Hypothesis 6d (H6d): They will be less likely to use unofficial sources.

Lastly, this study poses a series of research questions to determine if Twitter can be considered an “alternative medium” in the context of the BP oil spill, or if Twitter reflects similar viewpoints and is shaped by similar social, economic, and political forces, as mainstream journalism coverage:

Research Question 1 (RQ1): Is there a significant difference between Twitter users’ and journalists’ attitudes toward the oil spill?

Research Question 2 (RQ2): Are there significant differences between Twitter users’ and journalists’ coverage of the oil spill?

Research Question 3 (RQ3): Are there significant differences in the social, economic, and political factors that shape Twitter users’ and journalists’ attitudes toward, and coverage of, the BP oil spill?

Method

This study is part of a larger research project and builds on the methods used in Watson’s previous studies (2012, 2014a, 2014b, 2014c) studies of coverage of the BP oil spill. It combines data from a survey of journalists and Twitter users who wrote about the BP oil spill, a content analysis of those respondents’ stories and tweets, and census data describing those authors’ communities. From a list of the survey respondents’ newspaper stories (n = 1,829) and tweets (n = 6,437), 1,000 stories and 1,000 tweets were randomly sampled. These 2,000 newspaper stories and tweets were used as the cases in the final data set; survey and community data were assigned to these cases by matching the story/tweet author’s name and the name of the community with the survey responses and census data. The range in the number of the stories/tweets written by the authors in the study made it difficult to devise a meaningful summary score of authors’ coverage, which is why individual stories were used as the cases. Having different cases authored by the same person violates the statistical assumption of independent observations, but the violation of this assumption was controlled for in the final analysis by grouping newspaper stories and tweets written by the same authors into clusters in MPlus (Asparouhov & Muthen, 2006). The final data set included 404 clusters: 164 unique journalists and 240 unique Twitter users.

Surveys

Survey of journalists

By searching the America’s News database from April 20, 2010, the day of the Deep Water Horizon explosion, until September 20, 2010, the day after BP sealed the leaking oil well, 682 Gulf Coast journalists who covered the BP oil spill were identified. The following keywords were used to search Florida, Alabama, Mississippi, Louisiana, and Texas newspapers: BP, oil spill, and Deep Water Horizon. Headlines and stories’ lead paragraphs were read to determine if the story was about the Gulf oil spill. Both news and opinion articles were included in order to capture an overall picture of the impression a given newspaper’s coverage is likely to have on a reader. Email and postal addresses were gathered from the bottom of stories and newspapers’ websites.

A pre-notification letter was mailed to journalists on November 5, 2010, followed by an emailed link to the survey on November 10, 2010, and then followed by six reminder emails. The survey was completed by 220 journalists, for a response rate of 32.3%. While the response rate is modest, the demographic profile of journalists who responded to the survey was very similar to Weaver, Beam, Brownlee, Voakes, and Wilhoit’s (2007) national survey of journalists. Thus, it is likely that the sample is relatively representative of journalists as a whole.

Survey of Twitter users

First, the 250 most followed users in Gulf Coast communities that also had a newspaper in the America’s News database were identified from Twitiholic.com. Then Twitter’s API was used to download all of those users’ tweets (more than 14.2 million). The downloaded tweets were searched by the same keywords used to search for newspaper stories about the spill. There were 25,501 tweets about the BP oil spill, written by 4,396 unique authors in 110 cities.

Starting April 22, 2011, authors were invited to participate in the survey by sending identical Twitter messages with a link to a web-based survey to each user. The last message, including two reminder tweets, was sent June 10, 2011.

The survey was completed by 731 users (a 16.6% response rate), though 28 users who did not give user names or answered the survey but did not live in cities included in the study—perhaps because the survey link was retweeted—were eliminated. The final survey data included 703 Twitter users. It is possible that this sample of Twitter users is somewhat skewed toward either political extreme. It is hard to persuade Twitter users to respond to a survey with a 140-character tweet. Because the oil spill was a controversial issue with strong ideological viewpoints on both ends of the political spectrum, it is possible that those Twitter users who were more passionate about the issue were more likely to respond to the survey. This is reflected in larger standard deviations on most measures for the Twitter users than the journalists, which were the result of a greater number of responses at either extreme.

Survey measures

Other than the demographic questions and political ideology, all survey measures used a 5-point Likert-type scale where 1 equals strongly disagree and 5 equals strongly agree. Political ideology was measured on a 7-point scale where 1 equals very liberal and 7 equals very conservative. Items were reverse coded where noted.

Political ideology

Political ideology was measured using a single item, adapted from Patterson and Donsbach (1996): “How would you characterize your political ideology, from left (1) to right (7)?” This question was chosen because it was believed that the journalists in the study might have reacted more negatively to a more direct question that asked them to identify a preferred political party (M = 3.61, SD = 1.362).

Environmental ideology

The 12-item New Ecological Paradigm (NEP) scale is designed to measure “‘primitive beliefs’ about the nature of the earth and humanity’s relationship with it” (Dunlap et al., 2000, p. 427). To keep the overall length of the survey manageable, this study selected the four individual items that Dunlap et al. (2000) found had the highest correlation with the scale consisting of all items. These items were as follows: “Humans are severely abusing the environment”; “The balance of nature is strong enough to cope with the impacts of modern industrial nations” (reverse coded); “The so-called ‘environmental crisis’ has been greatly exaggerated” (reverse coded); and “Humans will eventually learn enough about how nature works to be able to control it” (reverse coded) (α = .749, M = 3.795, SD = 0.755).

Attitudes toward oil drilling

This portion of the survey adapted 13 questions from public opinion surveys about oil drilling (“The Oil Spill in the Gulf,” 2010), energy policy and government regulation (Bolsen & Cook, 2008), and industry responsibility (Miller & Sinclair, 2009; see Table 1 for question wording and descriptive statistics; α = .912, M = 2.675, SD = 0.746).

Table 1.

Journalists’ and Twitter Users’ Attitudes Toward the Oil Industry Following the 2010 BP Oil Spill (Descriptive Statistics).

	Journalists		Twitter users		Total
	M	SD	M	SD	M	SD
Oil drilling companies are generally concerned with limiting their environmental impact	2.84	0.899	2.45	1.142	2.61	1.066
Oil drilling companies are committed to protecting the public	2.60	0.856	2.35	1.091	2.45	1.009
The government does not do enough to regulate the oil drilling industry (reverse coded)	2.40	0.863	2.46	1.192	2.44	1.070
The oil drilling industry generally complies with government regulation	2.82	0.860	2.80	1.106	2.81	1.012
Government regulation of the oil industry is adequate	2.38	0.889	2.46	1.112	2.43	1.027
U.S. regulators should allow continued offshore drilling for oil at current levels	3.07	0.855	3.23	1.222	3.17	1.090
The U.S. should encourage exploration for new offshore oil fields	3.09	0.929	3.18	1.296	3.14	1.161
Strict environmental laws and regulations cost too many jobs and hurt the economy	2.26	0.782	2.44	1.229	2.37	1.073
Government regulation of business is necessary to protect the environment (reverse coded)	1.99	0.771	2.00	0.937	2.00	.873
U.S. energy policy should continue to encourage production of more domestic oil supplies	3.27	0.902	3.48	1.124	3.40	1.044
U.S. energy should shift attention away from fossil fuels to sources of renewable energy (reverse coded)	1.96	0.798	1.68	0.947	1.80	0.899
The collapse of the Deep Water Horizon oil platform and the oil spill in the Gulf of Mexico was a rare accident	3.43	0.900	3.15	1.232	3.26	1.116
Tragic as the Deep Water Horizon accident was, we cannot let it get in the way of developing domestic oil supplies	2.96	0.942	3.04	1.327	3.00	1.185
Total attitude toward oil industry	2.68	0.563	2.67	0.849	2.68	0.746
n	164		240

Demographics

Respondents were also asked their age, race, and income. Journalists were asked whether they held a journalism degree, how many years they had been a journalist, their tenure at their current newspaper, their primary job function (reporting, commentary, or editing), what beat they were assigned to, and whether they had any special training in covering environmental or energy issues.

Twitter users were asked how long they had used Twitter and what industry they worked in, specifically if they worked in the oil, tourism, or fishing industries, which were particularly impacted by the BP oil spill.

The 164 journalists in the sample represented 45 newspapers in 44 different Gulf Coast communities. The majority of journalists in the sample were White (83.5%, n = 137), male (54.3%, n = 89) reporters (76.8%, n = 126). The mean age was 43 years old, and the mean political ideology was slightly left of center (M = 3.410/7, SD = 1.090). A total of 65% (n = 114) of the journalists held a bachelor’s degree in journalism. The journalists had worked in the industry for an average of 20 years, and served at their current papers for an average of 11. The mean salary was between US$40,000 and US$50,000 a year.

Among those reporters who covered the BP oil spill, 26% had specialized training in covering either energy (n = 17) or environmental (n = 25) issues; less than 13% were energy (n = 6), environmental (n = 12), or outdoor reporters (n = 3). These reporters had a mild degree of concern for the state of the environment (M = 3.767, SD = 0.570). They had little affinity for the oil industry following the BP oil spill (M = 2.681, SD = 0.563).

The 240 Twitter users in the sample represented 51 unique Gulf Coast communities. (Both samples together represented 65 unique communities.) They had an average of 3,589 followers (range = 34 to 119,420; SD = 11,920.168) on the social networking website. The majority of Twitter users in the sample were also male (57.7%, n = 138) and White (82.4%, n = 196). The Twitter users, however, were significantly younger than the journalists (journalists, M = 43.42, SD = 13.275; Twitter users, M = 39.46, SD = 10.423; t(283.995) = 3.145, p = .002, equal variances not assumed). They also earned significantly more money than the journalists, with an average annual income in the US$50,000 to US$60,000 range. The most frequent occupations among the Twitter users were advertising/marketing/public relations (26.1%, n = 62), journalism/media/publishing (11.7%, n = 28), and IT/technology/software development (10.8%, n = 26). Only 16.7% (n = 40) of Twitter users worked in nonprofessional occupations, including creative occupations (music/art/design, etc.; 5%, n = 11), tourism/hospitality (2%, n = 5), and “other” nonprofessional occupations (10%, n = 24). Only three of the Twitter users worked in the oil industry.

Like journalists, Twitter users also tended to lean slightly left, albeit significantly less so than the journalists (M = 3.75/7, SD = 1.508, t(400.697) = −2.561, p = .011). There were no significant differences in the two groups’ attitudes toward the environment, or their support for the oil industry following the BP oil spill. However, it is worth noting that even when journalists and Twitter users did not differ significantly, the statistics for the Twitter users had significantly larger standard deviations, which suggests that the Twitter users represent a more diverse group. As previously noted, however, that diversity primarily reflects more cases at either extreme of a measure. That is, Twitter users were more polarized.

Content Analysis

The 220 reporters who responded to the survey had written 1,829 stories during the period of the study and the 703 Twitter users who responded to the survey authored 6,437 tweets. From these two groups, 1,000 of each were randomly selected and coded by the author. A second coder coded a randomly selected sample of 100 (10%) from each of the newspaper stories and tweets to establish intercoder reliability. At the end of the study, the author recoded a randomly selected sample of 100 (10%) from each to establish intracoder reliability. As shown in Table 2, reliability exceeded the recommended .80 cutoff for all variables used to answer research questions and test hypotheses (Riffe, Lacy, & Fico, 2005).

Table 2.

Frequency of Present/Absent Content Analysis Variables in Oil Spill Coverage.

	Stories			Tweets
Variable	n	%	Reliability^a	n	%	Reliability^a
Linkage			.821/.860			.865/.907
BP	155	15.5		323	32.3
Government oversight/regulation	318	31.8		203	20.3
Other oil industry	27	2.7		29	2.9
Environmental/health risk	189	18.9		165	16.5
Environmental/energy policy	13	1.3		14	1.4
Economic	131	13.1		21	2.1
Other	167	16.7		245	24.5
Thematic frame	59	5.9	.835/.841	126	12.6	.825/.897
Tone			.800/.875			.819/.859
Positive	133	13.3		99	9.9
Negative	499	49.9		570	57.0
Neutral	368	36.8		331	33.1
Official sources^b	774	77.4	.793/.944	559	55.9	.960/.900
BP official	355	35.5		394	39.4
Other oil industry official	33	3.3		11	1.1
Elected official	243	24.3		134	13.4
Other government official	439	43.9		107	10.7
Unofficial sources^b	386	38.6	.819/.941	176	17.6	.914/1
Non-Governmental Organization/activist/volunteer	70	7.0		19	1.9
Independent scientist/engineer	81	8.1		16	1.6
Local business/fisherman/oil industry employee	140	14.0		28	2.8

Note. NGO = nongovernmental organization.

Coder 2 coded a 10% random sample of the content; Coder 1 recoded a random sample of 10% of the content at the end of the study. The first number reported is the Krippendorf’s alpha for the intercoder reliability check; the second is the alpha for the intracoder reliability check.

Cateogires of sources were coded as being either present or absent in a given story/tweet, rather than tallying the total number of sources. Thus, sub-categories of sources (i.e., BP official, elected official, etc.) do not add up to equal the overall category (i.e., official or unofficial sources).

Content analysis protocol

The first four paragraphs, which contain the story’s “lead,” and “nut graph,” which summarizes the main points of a story, were coded.

Tweets were coded in their entirety.

Episodic/thematic frames

Using Iyengar’s (1991 definition, episodic stories and tweets were those that focused on a single event. Thematic stories and tweets were those that examined broader trends and implications beyond an isolated incident. Stories and tweets were coded as being either episodic or thematic.

Linkages

This portion of the study focuses both on the assigning of responsibility for the oil spill—whether stories/tweets assigned primary responsibility to BP or the government—and broader health, environmental, economic, and policy issues raised in the coverage (see Table 2 for a list of “linkages” used in this study). Only the first linkage in each story/tweet was coded.

Evaluative tone

Positive frames are those that emphasize desirable outcomes—BP making progress on capping the well, the fact that the ecological effects of the oil spill were not as bad as they could have been, and so forth. Negative frames are those that emphasize negative outcomes—setbacks for BP’s efforts to cap the well, ecological impacts of the oil spill on Louisiana’s fragile wetlands, and so forth. Neutral frames either did not mention positive or negative outcomes, or mentioned both simultaneously.

Sources

Lastly, coders recorded whether different official and unofficial sources were mentioned in newspaper and Twitter coverage of the BP oil spill. Sources are any individual or organization mentioned in a tweet or news story. A list of official and unofficial sources can be found in Table 2. Each category of source was coded as being either present or absent.

Structural Pluralism

Although the presence of multiple social groups does not guarantee the distribution of social power (Gandy, 1999), Blau’s (1977) diversity index,

D i v e r s i t y = 1 - \sum_{i = 1}^{R} P_{i}^{2},

where P_i is the proportion of the population in a given category/group, was used to measure the potential distribution of social power in a community. The index measures the probability, on a 0 to 1 scale, that two individuals in the population drawn at random (with replacement) belong to different groups.

The indicators of structural pluralism, drawn from the 2009 American Community Survey (ACS) 5-year estimates, included population, race, educational attainment, income, and major industry sectors (Nah & Armstrong, 2011). (See Table 3 for an explanation of how these measures were divided into different social groupings.) Data describe the counties where journalists work and Twitter users live, because the county is a better geographic representation of a media market than an individual city.

Table 3.

Coefficients of the Path Model Predicting Journalists’ and Twitter Users’ Oil Spill Coverage.

Path		Stories^a		Tweets^a
From	To	(StdYX Est.^b)	p	(StdYX Est.^b)	p
Structural pluralism^c,d	Government link	−1.132 (−.066)	.258	−.149 (−.008)	.895
	BP link	2.652 (.150)	.011	.644 (.035)	.552
	Thematic frames	−0.986 (−.057)	.313	−.862 (−.047)	.552
	Story tone	−2.228 (−.128)	.031	−2.165 (−.118)	.030
	Unofficial sources	.609 (.035)	.608	.264 (.014)	.782
% workforce in the oil industry^d	Government Link	.024 (.014)	.843	.001 (.000)	.996
	BP link	.217 (.121)	.064	−.071 (−.044)	.481
	Thematic frames	.243 (.138)	.016	−.105 (−.065)	.393
	Story tone	.197 (.111)	.103	.010 (.006)	.915
	Unofficial sources	−.158 (−.090)	.088	−.064 (−.040)	.459
Structural pluralism^c,d	Attitudes toward the oil industry	.648 (.083)	.381	1.690 (.097)	.300
% workforce in the oil industry^d	Attitudes toward the oil industry	.205 (.257)	.001	.530 (.345)	.016
Journalists’ political conservatism	Attitudes toward the oil industry	.132 (.308)	.000	.164 (.291)	.137
Journalists’ pro-environmental attitudes	Attitudes toward the oil industry	−.288 (−.368)	.000	−.625 (−.643)	.000
Attitudes toward the oil industry	Government link	.042 (.019)	.771	.318 (.303)	.000
	BP link	−.077 (−.034)	.703	−.152 (−.044)	.020
	Thematic frames	−.277 (−.125)	.027	.140 (.135)	.040
	Story tone	.260 (.117)	.073	.370 (.352)	.000
	Unofficial sources	.104 (.047)	.501	−.069 (−.066)	.157

The model fit for the journalists’ data was χ²(10) = 11.173, p = .672, Root Mean Square Error of Approximation (RMSEA) = .000, Comparative Fit Index (CFI) = 1.000, Tucker Lewis Index (TLI) = 1.033; the model fit for the Twitter users’ data was χ²(14) = 22.580, p = .068, RMSEA = .018, CFI = .855, TLI = .544; the model fit for the baseline model containing both groups was χ²(28) = 33.190, p = .229, RMSEA = .014, CFI = .969, TLI = .904. RMSEA = root mean square error of approximation; CFI = comparative fit index; TLI = Tucker-Lewis index.

Coefficients are standardized using the variance of both the X and Y variables.

Structural pluralism was measured using Blau’s (1977) index. Race was divided into White and non-White; educational attainment into less than high school, high school graduate, some college (including associate’s degree), college graduate, and advanced degree (master’s, JD, PhD, etc.); income into the quintile distribution of the median U.S. household income; percentage of the population employed in different industry sectors into the industry sectors used by the federal Bureau of Labor Statistics (2011): goods-producing (construction and manufacturing), services-producing (information services, transportation, retail trade, wholesale trade, etc.), and agriculture. To preserve the 0 to 1 scale of Blau’s index, communities’ population was measured as the percentage of the Houston, Texas, population, the largest city in the study. To ensure that communities that ranked highest in structural pluralism are not simply the largest, but ranked highly on other indicators, the structural pluralism measures were added together and then divided by 5 so that each pluralism measure had equal weight in the final index.

These variables were nonnormal; the error associated with census measures is also correlated with the measure itself. To correct both of these problems, these variables were log transformed (see Tabachnick and Fidell, 2007, for a discussion of commonly used data transformations); significant paths are boldfaced.

Economic Reliance on the Oil Industry

Economic reliance on the oil industry was measured as the percentage of the local workforce employed in “mining, quarrying, and oil and gas extraction” industries, according to the ACS (M = 0.006, SD = 0.014; U.S. Census Bureau, 2010). More direct measures of a community’s economic reliance on the oil industry are not available because the Bureau of Economic Analysis restricts community-level data, lest they reveal proprietary data about individual companies (Bureau of Economic Analysis, 2011).

Missing Cases

Fifteen cases—3 newspaper stories and 12 tweets (less than 5% of the cases)—with missing data for the survey measures were deleted (Tabachnick & Fidell, 2007). Thus, the multivariate analyses included 997 newspaper stories and 988 tweets.

Results

Just 1.3% of the newspaper coverage examined the environmental and energy policy issues associated with the BP oil spill, and just 18.9% of the newspaper coverage of the BP oil spill examined the environmental risks associated with the disaster. Very few of these stories explored the long-term effects and policy implications of the oil spill—94.1% of the newspaper stories used episodic frames. Newspaper stories were more likely to focus on the government’s (31.8%) rather than BP’s (15.5%) role in the disaster. While the news media are often criticized for overemphasizing negative news, only 49.9% of stories were framed negatively (36.8% were framed neutrally, and 13.3% positively). Official sources were quoted in more than three quarters of the coverage (77.4%); unofficial sources were mentioned in 38.6% of the coverage.

More tweets than newspaper stories focused on BP’s role in the disaster (χ²(1, N = 2000) = 77.590, p < .001), used thematic frames (χ²(1, N = 2000) = 26.738, p < .001), and had negative tones (χ²(1, N = 2000) = 10.130, p = .001). Tweets were less likely to mention unofficial sources (χ²(1, N = 2000) = 109.137, p < .001). Other patterns of the media, however, were identical: They used more episodic than thematic frames, and they were more likely to rely on official than unofficial sources (see Table 2).

Hypotheses and Research Questions

The hypothesized paths are shown in Figure 1. To answer the hypotheses and research questions, this model was first fitted to the journalists’ data. Then the model was fitted to the Twitter users’ data to test if the model fit both groups. Lastly, it was fitted to both groups simultaneously in order to test for invariance of means/thresholds and path coefficients. This last step answers whether there are statistically significant differences between journalists’ and Twitter users’ attitudes toward the oil industry following the BP oil spill, their coverage of the spill, and those individual- and community-level variables that shaped their attitudes and coverage.

Figure 1.

Path analysis model: How community-level variables and authors’ personal attitudes bias their coverage of the BP oil spill.

As shown in Table 3, the proposed model was a good fit for the journalists’ data (Browne & Cudeck, 1993; Hu & Bentler, 1999). That said, the model did not explain a large proportion of the variance in journalists’ use of government links (R² = .004), BP links (R² = .046), thematic frames (R² = .026), tone in their stories (R² = .036), or unofficial sources (R² = .008). The model did explain a much larger proportion of the variance in journalists’ attitudes toward the oil industry (R² = .416).

H1a to H1d predicted that the degree of structural pluralism in a community where a journalist is based would positively predict journalists’ critical coverage of the BP oil spill. As shown in Table 3, only H1a and H1c were supported. Journalists who worked in more structurally pluralistic communities were more likely to focus on BP’s role in the disaster (β = .150, p = .011). They were also more likely to write negative stories about the oil spill (β = −.128, p = .031). Structural pluralism did not affect journalists’ use of thematic frames (H1b) or their use of unofficial sources (H1d).

H2a to H2d predicted that local communities’ reliance on the oil industry for employment would negatively predict journalists’ critical coverage of the BP oil spill. As shown in Table 3, these hypotheses were not supported. Journalists in communities that were more economically dependent on the oil industry were actually more likely to use thematic frames (β = .138, p = .016; see H2b), perhaps because the oil spill was more salient in these oil industry-dependent communities, which led journalists in these communities to focus more on big picture issues surrounding the spill. The other paths (H2a, H2c, and H2d) were not significant.

H3a predicted that the degree of structural pluralism in journalists’ communities would negatively predict journalists’ positive attitudes toward the oil industry. As shown in Table 3, the hypothesis was not supported (β = .083, p = .381).

H3b predicted that the degree to which journalists’ communities rely on the oil industry for their economic bases would positively predict journalists’ positive attitudes toward the oil industry. As shown in Table 3, this hypothesis was supported; journalists in more economically dependent communities had more positive attitudes toward the oil industry (β = .257, p = .001).

H4 predicted that journalists’ political conservatism would positively predict their attitudes toward the oil industry. As shown in Table 3, this hypothesis was supported; more conservative reporters held more favorable attitudes toward the oil industry (β = .308, p < .001).

H5 predicted that journalists’ environmental ideologies would negatively predict their positive attitudes toward the oil industry. As shown in Table 3, this hypothesis was supported; more pro-environmental reporters held more critical attitudes toward the oil industry (β = −.368, p < .001).

H6a to H6d predicted that journalists’ positive attitudes toward the oil industry would negatively predict their critical coverage of the BP oil spill. As shown in Table 3, only H6b was supported. Journalists with more positive attitudes toward the oil industry were less likely to use thematic frames (β = −.125, p = .027). The other paths (H6a, H6c, and H6d) were nonsignificant.

Group differences

After establishing that the path model was also an acceptable fit for the Twitter users (see notes in Table 3), the model was fit simultaneously to both groups, allowing the parameters to vary between the groups (Model 1). As shown in Table 3, this baseline model was a good fit.

To answer the final research questions, parameters were constrained between the two groups to examine whether they were similar or different from one another. RQ1 and RQ2 were answered by constraining the groups’ means and thresholds² to be equal (Model 2); RQ3 was answered by constraining the path coefficients to be equal between the groups (Model 3). The chi-square of the more constrained model was compared with a less constrained model using a chi-square difference test to examine if constraining the parameters resulted in a significant loss of goodness of fit. A significant change in the goodness of fit between the models would indicate a significant difference in the constrained parameters between the two groups.

In a multigroup comparison path analysis, if constraining a set of parameters, for example, means/thresholds, significantly changes the model’s fit, individual constraints are relaxed in order of the absolute value of the largest derivatives³ of the individual parameters, until there is a nonsignificant difference between the more constrained and the less constrained models. The groups are different in regard to those parameters that have been relaxed in the final model.

When the means of journalists’ and Twitter users’ attitudes toward the oil industry and the tone of their stories/tweets were constrained to be equal in Model 2, there was not a significant loss of model fit (Δχ²(8) = 10.100, p = .258). There were no significant differences in journalists’ or Twitter users’ attitudes toward the oil industry (RQ1); there were also no statistically significant differences in the likelihood that individual journalists and Twitter users would use BP and government linkages, negative frames, thematic frames, or unofficial sources (RQ2).

RQ3 asked if there were significant differences in the social, economic, and political factors that shaped Twitter users’ and journalists’ attitudes toward the oil industry and the groups’ coverage of the BP oil spill. To answer this question, Model 2 was further constrained to hold the path coefficients equal between the two groups (Model 3), which resulted in a significant loss of goodness of fit (Δχ²(19) = 42.993, p = .001). Thus, there were significant differences between the two groups.

The largest derivative for the constrained parameters was for the path from attitudes toward the oil industry to thematic frames (.023). Thus, the constraint that this coefficient be equal between the groups was relaxed (Model 4), though when compared with Model 2, there was still a significant loss of goodness of fit (Δχ²(18) = 31.504, p = .025). Therefore, the constraint from the journalists’ attitudes toward the oil industry to government linkages (derivative = .012) was also relaxed (Model 5). Again, there was a marginally significant loss of goodness of fit when compared with the less constrained model (Model 2; Δχ²(17) = 27.277, p = 0.054). The constraint with the next largest derivative (.011) for the path from attitudes toward the oil industry to BP linkages was relaxed (Model 6). When compared with Model 2, doing so resulted in no significant loss of goodness of fit (Δχ²(16) = 22.134, p =.139).

The data demonstrate that the effect of the groups’ attitudes toward the oil industry on their use of thematic frames did have significantly different effects on journalists’ and Twitter users’ use of thematic frames. While journalists’ positive attitudes toward the oil industry made them significantly less likely to use thematic frames (β = −.125, p = .027), Twitter users with more positive attitudes toward the oil industry were more likely to use thematic frames (β = .135, p = .040). The groups also differed significantly in terms of how their attitudes toward the oil industry affected their focus on either BP’s or the government’s role in the oil spill crisis. Twitter users who had more positive attitudes toward the oil industry were less likely to focus on the role of BP (β = −.044, p = .020) and more likely to focus on the role of the government (β = .303, p < .001). The use of these two linkages among the Twitter users was significantly and negatively correlated (tetrachoric correlation = −.833, p < .001). Thus, it appears to an extent that the two linkages are substitutes for one another. When writing about the failure to cap the leaking well, for example, Twitter users focused primarily on either BP or the government. Those Twitter users with more favorable attitudes toward the oil industry focused on the government’s response to the crisis to shift attention away from BP. Journalists’ attitudes toward the oil industry did not significantly affect their use of either BP (β = −.034, p = .703) or government linkages (β = .019, p = .771), although journalists’ use of BP and government linkages was also significantly and negatively correlated (tetrachoric correlation = −.797, p < .001).

Discussion

Freudenburg and Gramling (2011) wrote that the BP oil disaster is “a challenge to take a closer, more clear-eyed look at our [energy] policies . . . to respond to the oil-darkened waters with clearer thinking” (p. 7). To the extent there were “lessons to be learned” from the BP oil spill, they were not likely to be found in the Gulf Coast journalists’ coverage that was part of this study. Nearly all (94%) of journalists’ stories were framed episodically, covering the BP oil spill as an isolated, episodic event, rather than as part of larger energy and regulation policies. There was almost zero coverage of the energy or environmental policy issues related to the oil spill (n = 13).

Even if the journalists in this study were inclined to do so, one might wonder if these journalists would be equipped to cover these critical issues. Only a quarter of the journalists had specialized training in covering either environmental or energy-related issues; less than 13% of the journalists covered energy or environmental beats.

As hypothesized, journalists in more pluralistic communities were more likely to focus on BP’s role in the crisis and frame their coverage critically. Put differently, journalists in larger, more heterogeneous communities, where social power is potentially more dispersed, appear freer to be critical in their coverage of the oil spill, including toward the oil giant, than journalists in less pluralistic communities.

However, what is most striking about this current study is the similarity between journalists’ and Twitter users’ attitudes toward, and coverage of, the BP oil spill. One would expect some differences, if only due to the difficulty of applying equivalent content analysis measures across very different media. Individual journalists and Twitter users, though, did not differ in their attitudes toward the oil industry, or their coverage of the BP oil spill. That is, Twitter did not serve as an alternative medium in these Gulf Coast communities.

Journalists and Twitter users did differ in how their attitudes toward the oil industry affected their use of thematic frames: Journalists with more positive attitudes were less likely to use thematic frames; Twitter users were more likely to use thematic frames. The other significant difference between the two groups is that Twitter users with more positive attitudes toward the oil industry were significantly more likely to focus on the government’s role in the oil spill and less likely to focus on BP’s. Journalists’ attitudes toward the oil industry did not significantly influence whose role in the crisis they focused on more frequently. That being said, the forces that shaped journalists’ and Twitter users’ coverage of the crisis were more similar than different, undermining the notion, at least in the context of Gulf Coast coverage of the BP oil spill, that Twitter represented an alternative medium. These findings also underscore that while the Internet may largely override local geography in terms of the distribution of online content, the producers of that online content still live in traditional, off-line communities that can have an influence on the production of online content.

Future studies should explore at greater depth why and how journalists’ and Twitter users’ attitudes converge to be so similar. The current study’s data suggest one possible reason these journalists and the most followed Twitter users are so similar. The social strata that command the most follows in social media are very similar to those groups that dominate traditional media. The average journalist and Twitter user were both White male professionals, who are slightly left of center politically, and significantly wealthier than the average per capita income of other residents of the Gulf Coast states (U.S. Census Bureau, 2010). These demographic similarities support Wu et al.’s (2011) contention that “while attention that was formerly restricted to mass media channels is now shared amongst other ‘elites,’ information flows have not become egalitarian by any means” (p. 6). It should be further noted that the lack of egalitarianism is not an inherent characteristic of the Internet, or the doing of some elite, wealthy force. Rather, in the case of the most followed Twitter users in this study, it reflects who, given seemingly endless choices of alternative voices, individual Twitter users choose to follow. That is, simply because alternative media and individual voices have greater potential distribution online does not mean that they will achieve greater readership or attention from online users.

The findings of this study, however, should be interpreted with caution due to the study’s limitations, which in many ways are inevitable when studying a crisis that unfolds in real time. Neither the journalists nor Twitter users in this study represent a scientific random sample, in part because it was not certain if a survey of a relatively small group of professional journalists and previously untested methods for surveying Twitter users would yield enough responses to model multivariate relationships between variables in the study.⁴ Fortunately, both surveys yielded sufficient samples to produce robust models of the relationships between variables within the samples. Findings, however, should not be generalized beyond the Gulf Coast newspaper journalists and most followed Twitter users in the sample.

It should also be clear that the variables used to explore similarities and differences between the journalists and Twitter users in this study represent a fraction of those variables that could be used to compare the groups. The complexity of the content variables in this study was restricted by the need to have measures that could be applied equivalently across two very different media, and by the challenges of inferring meaning from a 140-character tweet. Yet, future studies might try to come up with different content analysis strategies and other methodological approaches for exploring other similarities and differences between “new” and more established media. These comparisons are valuable because they can serve as a basis for understanding how changes in communication technologies are, and are not, fundamentally changing communication about public affairs issues.

Footnotes

Declaration of Conflicting Interests

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

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Brendan R. Watson (PhD, University of North Carolina-Chapel Hill) is an assistant professor in the School of Journalism and Mass Communication at the University of Minnesota-Twin Cities. His research interests include community information needs, how social media are/are not changing communication about local public affairs issues and quantitative research methods.

References

Asparouhov

Muthen

(2006). Multilevel modeling of complex survey data. In Proceedings of the Survey Research Methods Section, American Statistical Association (pp. 2718-2726). Retrieved from http://www.amstat.org/sections/SRMS/Proceedings/y2006/Files/JSM2006-000803.pdf

Atkinson

J. D.

(2010). Alternative media and politics of resistance: A communication perspective. New York, NY: Peter Lang.

Atton

(2002). Alternative media. London, England: Sage.

Blau

(1977). Inequality and heterogeneity: A primitive theory of social structure. New York, NY: Free Press.

Bolsen

Cook

F. L.

(2008). Public opinion on energy policy: 1974–2006. Public Opinion Quarterly, 72, 364-388. doi:10.1093/poq/nfn019

Browne

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 136-162). Beverly Hills, CA: Sage.

Bureau of Economic Analysis. (2011). Regional data: GDP & personal income [Data set]. Retrieved from http://www.bea.gov/iTable/iTable.cfm?reqid=70 ;step=1 isuri=1 acrdn=2

Bureau of Labor Statistics. (2011). Employment by major industry sector [Data set]. Retrieved from http://www.bls.gov/emp/ep_table_201.htm

Bureau of Ocean Energy Management, Regulation, and Enforcement. (2011). Report regarding the causes of the April 20, 2010 Macondo well blowout. Retrieved from http://www.boemre.gov/pdfs/maps/DWHFINAL.pdf

10.

Dahlberg

(2005). The corporate colonization of online attention and the marginalization of critical communication? Journal of Communication Inquiry, 29, 160-180. doi:10.1177/0196859904272745

11.

Dahlberg

(2007). The Internet, deliberative democracy, and power: Radicalizing the public sphere. International Journal of Media & Cultural Politics, 3, 47-64. doi:10.1386/macp.3.1.47/1

12.

D’Alessio

Allen

(2000). Media bias in presidential elections: A meta-analysis. Journal of Communication, 50, 133-156. doi:10.1111/j.1460-2466.2000.tb02866.x

13.

Donohue

G. A.

Olien

C. N.

Tichenor

P. J.

(1985). Reporting conflict by pluralism, newspaper type, and ownership. Journalism Quarterly, 62, 489-499. doi:10.1177/107769908506200305

14.

Downey

Fenton

(2003). New media, counter publicity and the public sphere. New Media & Society, 5, 185-202. doi:10.1177/1461444803005002003

15.

Dunlap

R. E.

Van Liere

K. D.

Mertig

A. G.

Jones

R. E.

(2000). Measuring endorsement of the new ecological paradigm: A revised NEP scale. Journal of Social Issues, 56, 425-442. doi:10.1111/0022-4537.00176

16.

Entman

R. M.

(1993). Framing: Toward clarification of a fractured paradigm. Journal of Communication, 43, 51-58. doi:10.1111/j.1460-2466.1993.tb01304.x

17.

Entman

R. M.

Rojecki

(1993). Freezing out the public: Elite and media framing of the U.S. anti-nuclear movement. Political Communication, 10, 155-173. doi:10.1080/10584609.1993.9962973

18.

Feygina

Jost

J. T.

Goldsmith

R. E.

(2010). System justification, the denial of global arming, and the possibility of “system-sanctioned” change. Personality and Social Psychology Bulletin, 36, 326-328. doi:10.1177/0146167209351435

19.

Freudenburg

W. R.

Gramling

(2011). Blowout in the Gulf: The BP oil spill disaster and the future of energy in America. Cambridge, MA: The MIT Press.

20.

Gandy

O. H.

Jr. (1999). Community pluralism and the “tipping point”: Editorial responses to race and related structural change. In Demers

D. P.

Viswanath

(Eds.), Mass media, social control, and social change: A macrosocial perspective (pp. 139-158). Ames: Iowa State University Press.

21.

Griffin

R. J.

Dunwoody

(1995). Impacts of information subsidies and community structure on local press coverage of environmental contamination. Journalism & Mass Communication Quarterly, 72, 271-284. doi:10.1177/107769909507200202

22.

Griffin

R. J.

Dunwoody

(1997). Community structure and science framing of news about local environmental risks. Science Communication, 18, 362-384. doi:10.1177/1075547097018004005

23.

Hamilton

(2000). Alternative media: Conceptual difficulties, critical possibilities. Journal of Communication Inquiry, 24, 357-378. doi:10.1177/0196859900024004002

24.

Harcup

(2003). “The unspoken—said”: The journalism of alternative media. Journalism, 4, 356-376. doi:10.1177/14648849030043006

25.

Harcup

(2005). “I’m doing this to change the world”: Journalism in alternative and mainstream media. Journalism Studies, 6, 361-374. doi:10.1080/14616700500132016

26.

Hart

P. S.

(2010). One or many? The influence of episodic and thematic climate change frames on policy preferences and individual behavior change. Science Communication, 33, 28-51. doi:10.1177/1075547010366400

27.

Hindman

D. B.

Littlefield

Preston

Neumann

(1999). Structural pluralism, ethnic pluralism, and community newspapers. Journalism & Mass Communication Quarterly, 76, 250-263. doi:10.1177/107769909907600205

28.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55. doi:10.1080/10705519909540118

29.

Iyengar

(1991). Is anyone responsible? How television frames political issues. Chicago, IL: University of Chicago Press.

30.

Iyengar

(1996). Framing responsibility for political issues. Annals of the American Academy of Political and Social Science, 546, 59-70.

31.

Jost

J. T.

Federico

C. M.

Napier

J. L.

(2009). Political ideology: Its structure, functions, and elective affinities. Annual Review of Psychology, 60, 307-337. doi:10.1146/annurev.psych.60.110707.163600

32.

Kim

Scheufele

D. A.

Shanahan

(2002). Think about it this way: Attribute agenda-setting function of the press and the public’s evaluation of a local issue. Journalism & Mass Communication Quarterly, 79, 7-25. doi:0.1177/107769900207900102

33.

Lasswell

H. D.

(1948). The structure and function of communication in society. In Bryson

(Ed.), The communication of ideas (pp. 37-51). New York, NY: Institute for Religious and Social Studies.

34.

Miller

B. M.

Sinclair

(2009). A model of public response to marketplace advocacy. Journalism & Mass Communication Quarterly, 86, 613-629. doi:10.1177/107769900908600310

35.

Nah

Armstrong

C. L.

(2011). Structural pluralism in journalism and media studies: A concept explication and theory construction. Mass Communication and Society, 14, 857-878. doi:10.1080/15205436.2011.615446

36.

The oil spill in the Gulf. (2010, May 11). CBS News. Retrieved from http://www.cbsnews.com/8301-503544_162-20004732-503544.html

37.

Patterson

T. E.

Donsbach

(1996). News decisions: Journalists as partisan actors. Political Communication, 13, 455-468. doi:10.1080/10584609.1996.9963131

38.

Pew Research Center for the People & the Press. (2010). Little change in opinions about global warming. Retrieved from http://people-press.org/report/669/

39.

Pew Research Center’s Project for Excellence in Journalism. (2010). 100 days of gushing oil. Retrieved from http://www.journalism.org/analysis_report/100_days_gushing_oil

40.

Reese

S. D.

Rutigliano

Hyun

Jeong

(2007). Mapping the blogosphere: Professional and citizen-based media in the global news arena. Journalism, 8, 235-261. doi:10.1177/1464884907076459

41.

Riffe

Lacy

Fico

F. G.

(2005). Analyzing media messages: Using quantitative content analysis in research (2nd ed.). Mahwah, NJ: Lawrence Erlbaum.

42.

Sheafer

(2007). How to evaluate it: The role of story-evaluative tone in agenda setting and priming. Journal of Communication, 57, 21-39. doi:10.1111/j.0021-9916.2007.00327.x

43.

Smith

(1993). News sources and power elites in news coverage of the Exxon Valdez oil spill. Journalism Quarterly, 70, 393-403. doi:10.1177/107769909307000214

44.

Sparks

(2001). The Internet and the global public sphere. In Bennett

W. L.

Entman

R. M.

(Eds.), Mediated politics: Communication in the future of democracy (pp. 75-95). Cambridge, UK: Cambridge University Press.

45.

Tabachnick

B. G.

Fidell

L. S.

(2007). Using multivariate statistics (5th ed.). Boston, MA: Allyn & Bacon.

46.

Tichenor

P. J.

Donohue

G. A.

Olien

C. N.

(1980). Community conflict and the press. Beverly Hills, CA: Sage.

47.

U.S. Census Bureau. (2010). 2005-2009 American Community Survey 5-year estimates [Data set]. Retrieved from http://factfinder.census.gov/faces/nav/jsf/pages/download_center.xhtml

48.

Watson

B. R.

(2012). Ideologies drive journalists’ attitudes toward oil industry. Newspaper Research Journal, 33, 6-22.

49.

Watson

B. R.

(2014a). Assessing ideological, professional, and structural biases in journalists’ coverage of the 2010 BP oil spill. Journalism & Mass Communication Quarterly, 91, 792-810. doi: 10.1177/1077699014550091

50.

Watson

B.R.

(2014b). Normallizing community structure’s restraint on critical Tweets about a polluting industry. Journal of Broadcasting and Electronic Media, 54, 581-600. doi: 10.1080/08838151.2014.966359

51.

Watson

B.R.

(2014c). When critical voices should speak up. Journalism Practice, 8, 842-854. doi: 10.1080/17512786.2014.892723

52.

Weaver

D. H.

Beam

R. A.

Brownlee

B. J.

Voakes

P. S.

Wilhoit

G. C.

(2007). The American journalist in the 21st century. Mahwah, NJ: Lawrence Erlbaum.

53.

Hofman

J. M.

Mason

W. A.

Watts

D. J.

(2011). Who says what to whom on Twitter. Retrieved from http://labs.yahoo.com/publication/who-says-what-to-whom-on-twitter/