Assessing Ideological,Professional,and Structural Biases in Journalists’ Coverage of the 2010 BP Oil Spill

Abstract

Previous studies based on aggregate data have not found consistent evidence that journalists’ personal beliefs and attitudes bias their coverage. This study, however, uses individual-level survey data on Gulf Coast journalists’ beliefs and attitudes toward the BP oil spill, matched with a content analysis of respondents’ stories about the disaster, and community structure data. The study examines the effect that journalists’ perceptions of professional norms and the social and economic contexts of the communities in which they work had on their attitudes toward and coverage of the crisis.

Keywords

bias oil spill community structure professional roles

Charges persist, particularly in partisan media,¹ that reporters’ political ideology biases their coverage of important issues, including the 2010 BP Gulf oil spill, that year’s most-followed domestic news story.² Some academic studies have found support for the “liberal bias hypothesis,”³ though others have challenged those findings.⁴ Previous studies, however, have not matched individual journalists’ political beliefs with their coverage. They have relied on aggregate data, which may not accurately reflect how individual journalists’ beliefs shape their coverage. Thus, this study examines whether Gulf Coast journalists’ political beliefs biased their coverage of the BP oil spill by matching individual-level survey responses with a content analysis of the stories individual journalists wrote.

Previous studies of journalistic bias suggest even these individual-level data are unlikely to produce strong support for the “liberal media hypothesis.”⁵ This study, however, does not propose that one can predict differences in coverage based on political ideology alone. Journalists, after all, are not completely autonomous of the professional and community contexts in which they are situated.⁶ The study combines individual-, professional-, and community-level perspectives to suggest that any bias in journalists’ work likely comes from multiple, subtler influences. Specifically, it examines whether contextual factors, including perceived professional norms and the social and economic structure of journalists’ communities, influenced their coverage of the BP oil spill.

Literature Review

Journalists’ Beliefs, Attitudes, and Reporter Bias

A story can have a slant—that is, favor a particular aspect of a story over another—without being biased.⁷ For example, political coverage may simply give more positive coverage to the stronger campaign. But coverage is said to be biased when some factor external to the issue or subject being covered—journalists’ political ideology, for example—produces that slanted coverage. Claims of bias are most common when news coverage touches on a controversial issue, such as offshore oil drilling, that exhibits a strong partisan divide.⁸ Scholars, though, have found little empirical support for the claim that journalists are biased, either toward liberal or conservative causes. D’Alessio and Allen’s meta-analysis, for example, found what they called a “probably insubstantial” pro-liberal bias in network news coverage.⁹ Lowry and Shidler did find that Bill Clinton received more favorable coverage when he ran against George H. W. Bush in 1992 and Bob Dole in 1996.¹⁰ Because, as a whole, journalists lean slightly left of center politically and vote Democrat,¹¹ Lowry and Shidler interpreted this coverage as reflecting journalists’ liberal media bias.¹² However, without a direct link between an individual reporter’s political beliefs and the slant of that reporter’s coverage, one cannot determine whether the coverage was slanted due to journalists’ political beliefs, or simply because Clinton was the superior candidate and more deserving of positive coverage.¹³ This is perhaps why there is more evidence of bias in years Democrats won the presidency (i.e., when Democrats were stronger candidates), but not in years Republicans won.¹⁴

A quasi-experiment that matched individual journalists’ political ideologies and their news judgments did find that journalists’ news decisions correlate strongly with their individual political ideologies.¹⁵ More conservative journalists were more likely to frame a hypothetical chemical plant controversy in a manner favorable to the chemical industry and against government regulation, whereas liberals favored government regulation. That experiment, however, did not examine actual news coverage,¹⁶ nor did a previous survey-based study that found Gulf Coast journalists’ attitudes toward the BP oil spill were moderately correlated with their personal political and environmental beliefs.¹⁷ Thus, the current research matches individual journalists’ responses to a survey with a content analysis of the stories they wrote to examine whether coverage of the oil spill was affected both by their individual-level political beliefs and attitudes toward the environment.

Professional Roles

Different professions, including journalism, have unique cultures, including shared values and norms of how one is to carry out one’s work.¹⁸ Based on surveys of journalists’ values and norms, Weaver and colleagues identified four types of journalists: the Interpreters believe journalists’ primary function is to provide analysis of complex problems, Adversarialists believe journalists should be skeptical of public officials and business interests, the Disseminators seek to serve the largest possible audience, and the Populist Mobilizers want to give their readers a voice in the media and encourage them to be actively involved in community affairs.¹⁹

Previous studies of journalists’ professional roles have been primarily descriptive in nature.²⁰ Watson, however, found that journalists’ affiliation with the Populist Mobilizer role positively predicted Gulf Coast journalists’ support for additional regulation of the oil industry following the BP oil spill (r = .24).²¹ That study, however, did not examine how journalists’ attitudes shaped their coverage.

Community Structure

To build and maintain information sources and audience for their coverage, journalists must also be aware of and responsive to the larger social structure in which they work. In smaller, more homogeneous (i.e., less structurally pluralistic) communities, power is more likely to be concentrated among a small group of elite actors, and local media are more reticent to cover conflicts, including environmental problems.²² But in more structurally pluralistic communities, society is more reliant on the media to communicate among larger, more dispersed, and more diverse social groups. In these pluralistic communities, media cover conflict more openly.

Studies have found that structural pluralism not only affects the amount of coverage environmental problems receive, but also how those issues are framed.²³ Griffin, Dunwoody, and Gehrmann, for example, found that newspapers in more structurally pluralistic communities were more likely to use thematic frames in their coverage of local industrial contamination.²⁴

Griffin and Dunwoody, however, found that economic reliance on manufacturing—measured as the percentage of the local workforce employed in manufacturing—was a stronger predictor than structural pluralism of how newspapers framed contamination; newspapers in more economically reliant communities were less likely to link industrial contamination to threats to human health.²⁵ Watson also found structural pluralism negatively predicted Gulf Coast journalists’ attitudes toward the oil industry following the BP oil spill, but that local communities’ economic reliance on the oil industry was a stronger predictor of journalists’ attitudes.²⁶

These studies have not tested how community structure variables enter or influence the news-gathering process. Watson suggested that journalists are socialized into the local social structure, which, along with their personal beliefs, shapes their attitudes toward issues in the news, which in turn affect their coverage of those issues.²⁷ Watson’s study, however, examined only journalists’ attitudes, not their coverage.

Journalists’ Coverage of the Oil Spill

Previous studies of journalistic bias have focused primarily on whether the tone of a story was positive, negative, or neutral.²⁸ This study focuses more broadly on whether coverage was “critical,” examining the extent to which coverage focused on BP’s responsibility for the oil spill, journalists’ use of thematic frames, and their use of unofficial sources.

Kensicki’s content analysis found that while newspapers focused on industry as the cause of pollution, government was overwhelmingly portrayed as being responsible for addressing pollution’s consequences.²⁹ She concluded that this pattern supports business as usual, shifting attention away from the changes needed to address the root causes of pollution (i.e., forcing industry to clean up its act).

How a story is framed can also shift how the audience assigns responsibility for a given problem. According to Iyengar, episodic stories—stories framed as isolated incidents—result in responsibility being assigned to individual actors.³⁰ On the other hand, thematic frames, which explore larger causes and consequences underlying a particular issue—for example, putting the BP oil spill in context of the government’s record of inspecting offshore oil drilling operations—are more likely to cause audiences to attribute responsibility for social problems to systemic failures. Exposure to thematic framing of climate change’s effects has also been found to increase support for additional regulations on greenhouse gas-emitting industries.³¹

Entman and Rojecki also found that stories that use unofficial sources are more likely to raise critical questions.³² Their content analysis of the nuclear freeze movement found that two-thirds of those stories that raised concerns over nuclear weapons attributed those concerns to unofficial sources. Unofficial sources, however, were quoted in only 12% of stories about the nuclear freeze movement. In a content analysis of newspaper coverage of the 1989 Exxon Valdez oil spill, Smith also found that official sources made up 60% of journalists’ sources, and that these government and oil industry sources were most likely to say the crisis had been overblown and to rate Exxon’s response to the crisis favorably.³³

While these variables help describe the degree of criticalness present in journalists’ coverage, previous studies have also shown that these variables are affected by the social structure in which journalists work. Journalists in less structurally pluralistic communities are less likely to use unofficial sources³⁴ or thematic frames³⁵ in their coverage. Journalists who work in less pluralistic and more economically dependent communities are also less likely to hold a polluting industry responsible for an environmental problem.³⁶

Hypotheses

Previous studies have not found strong, consistent evidence of ideological bias in journalists’ coverage.³⁷ The aggregate-level data these studies relied upon, however, fail to link adequately individual journalists’ beliefs and attitudes with their coverage. This study links individual-level survey data on journalists’ attitudes toward the oil industry with a content analysis of these individual journalists’ stories. This study retests the bias hypothesis using a path model, a simplified version of which is shown in Figure 1, which links individual-level survey data with a content analysis of those journalists’ coverage of the BP oil spill.

Figure 1.

Path model: Individual and structural biases in Gulf Coast newspaper journalists’ coverage of the BP oil spill.

H1: Journalists’ positive attitudes toward the oil industry will negatively predict their critical coverage of the oil industry; journalists with more positive attitudes will be less likely to write negative stories (H1a), focus on BP’s role in the oil spill (H1b), use thematic frames (H1c), or use unofficial sources (H1d).

Watson also found that journalists’ preferred professional roles—specifically their support for the Populist Mobilizer role—also affected journalists’ attitudes toward the oil industry.³⁸ While this study also tests the indirect effect of these beliefs, journalists’ preferred professional roles might also have direct effects on journalists’ coverage of the BP oil spill.

RQ1: Do journalists’ preferred professional roles predict how they covered the 2010 BP oil spill?

This study also examines the effect of community structure on journalists’ coverage of the BP oil spill. Previous studies have found that newspapers’ critical coverage of polluting industries was positively associated with structural pluralism and negatively associated with communities’ economic reliance on the oil industry.³⁹

H2: Communities’ economic reliance on the oil industry will negatively predict the criticalness of journalists’ coverage of the BP oil spill; journalists in communities that are more reliant on the oil industry will be less likely to write negative stories (H2a), focus on BP’s role in the oil spill (H2b), use thematic frames (H2c), or use unofficial sources (H2d).

H3: Communities’ degree of structural pluralism will positively predict the criticalness of journalists’ coverage of the BP oil spill; journalists in larger, more heterogeneous communities in which social power is likely to be more dispersed will be more likely to write negative stories (H3a), focus on BP’s role in the oil spill (H3b), use thematic frames (H3c), and use unofficial sources (H3d).

Previous studies have not examined how community structure’s influence enters the news-gathering process. This study tests Watson’s suggestion that these community-level variables might enter the news-gathering process by affecting journalists’ attitudes toward the oil industry, which in turn affect journalists’ coverage of the BP oil spill.⁴⁰ If this hypothesis is correct, the community structure variables’ indirect effect on journalists’ coverage will be stronger than the direct effects.⁴¹

H4: The indirect effects of communities’ economic reliance on the oil industry and structural pluralism on journalists’ coverage of the BP oil spill, via journalists’ attitudes toward the oil industry, will be larger than the direct effects.

Method

Survey

The America’s News database was searched for Gulf Coast newspaper stories published between April 20, 2010, the day of the Deep Water Horizon explosion, and September 20, 2010, the day after BP sealed the leaking oil well. Articles that used the key phrases “BP,” “oil spill,” or “Deep Water Horizon,” and whose headlines and lead paragraphs suggested the story was about the BP oil spill, were downloaded. Authors’ bylines were copied into an Excel spreadsheet. Reporters’ mailing and e-mail addresses were copied from their stories and newspapers’ websites. A total of 683 unique bylines with valid e-mail addresses were recorded.

Starting November 5, 2010, journalists received a pre-notification letter introducing the study by mail. The survey was e-mailed to the journalists one week later. Two hundred twenty (32.3%) journalists completed the survey.⁴²

Survey Measures

Political ideology

Political ideology was measured using a single-item, seven-point scale: “How would you characterize your political ideology, from left (1) to right (7)?”⁴³ This question was used because it was believed that due to the professional norm of objectivity, journalists might react negatively to more direct, multi-item measures of their personal political beliefs (M = 3.41, SD = 1.09).

Environmental ideology

Environmental ideology was measured using the four items from the New Environmental Paradigm that Dunlap and colleagues found had the highest correlations with the full twelve-item scale⁴⁴: “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). Respondents indicated their level of agreement with each statement on a five-point scale anchored by “strongly disagree” (1) and “strongly agree” (5) (Cronbach’s α = .68, M = 3.78, SD = 0.57).

Professional roles

Journalists’ preferences for the different professional roles were measured using seven questions selected from Weaver and colleagues’ survey of American journalists based on the questions’ face validity.⁴⁵ Respondents indicated their level of agreement with each statement on a five-point scale anchored by “strongly disagree” (1) and “strongly agree” (5). These questions measured four potential journalistic roles: the Adversarialist (two items, r = .98), the Populist Mobilizer (three items, α = .78), the Interpreter, and the Disseminator. Roles were not mutually exclusive; that is, journalists could embrace multiple professional roles (see items and their means in Table 1).

Table 1.

Journalists’ Preferred Professional Roles.

	M	SD
Adversarialist
Journalists should always be critical of corporations	2.29	1.10
Journalists should always be critical of government	2.34	1.12
Total	2.31	1.11
Populist Mobilizer
Journalists should give people an opportunity to express their views	4.10	0.81
Journalists should motivate people to get involved in the community	3.41	0.98
Journalists should point to possible solutions to community problems	3.83	0.87
Total	3.78	0.75
Interpreter
Journalists should provide analysis of complex problems	4.69	0.66
Disseminator
Journalists should concentrate on serving the largest possible audience	3.64	0.90
N	164

Note. Items measured on a five-point Likert-type scale, strongly disagree (1) to strongly agree (5).

Attitudes toward oil drilling

This portion of the survey adapted thirteen questions from public opinion surveys about oil drilling,⁴⁶ energy policy and government regulation,⁴⁷ and industry responsibility⁴⁸ (see Table 2 for question wording and descriptive statistics). Respondents indicated how strongly they agreed with each statement on a five-point Likert-type scale anchored by “strongly disagree” (1) and “strongly agree” (5) (α = .79, M = 2.68, SD = 0.56).

Table 2.

Journalists’ Attitudes toward the Oil Industry, Descriptive Statistics.

	M	SD
The collapse of the Deep Water Horizon oil platform and the oil spill in the Gulf of Mexico was a rare accident.	3.43	0.90
U.S. energy policy should continue to encourage production of more domestic oil supplies.	3.27	0.90
The United States should encourage exploration for new offshore oil fields.	3.09	0.93
U.S. regulators should allow continued offshore drilling for oil at current levels.	3.07	0.86
Tragic as the Deep Water Horizon accident was, we cannot let it get in the way of developing domestic oil supplies.	2.96	0.94
Oil drilling companies are generally concerned with limiting their environmental impact.	2.84	0.90
The oil drilling industry generally complies with government regulation.	2.82	0.86
Oil drilling companies are committed to protecting the public.	2.60	0.86
The government does not do enough to regulate the oil drilling industry (reverse-coded).	2.40	0.86
Government regulation of the oil industry is adequate.	2.38	0.89
Strict environmental laws and regulations cost too many jobs and hurt the economy.	2.26	0.78
Government regulation of business is necessary to protect the environment (reverse-coded).	1.99	0.77
U.S. energy should shift attention away from fossil fuels to sources of renewable energy (reverse-coded).	1.96	0.80
Total attitudes toward oil industry	2.68	0.56
N	164

Note. Items measured on a five-point Likert-type scale, strongly disagree (1) to strongly agree (5).

Demographics

In addition to age, gender, and income, journalists were also 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.

Content Analysis

Respondents to the survey had written 1,829 stories, 1,000 of which were randomly sampled and coded. The large range (1-82) in the number of stories individual journalists had written made it difficult to devise a meaningful summary of each journalist’s coverage. Thus, each story was treated as a unique case. Individual- and community-level data were matched with the stories based on bylines and the name of the community where a newspaper was located. A variable that grouped articles by the same author controlled for the violation of the statistical assumption that each case represents an independent observation.

Because stories were randomly sampled, not all journalists who completed the survey were represented in the final data set. The final data set included 164 unique authors. The individual characteristics of the reporters in the final data set did not differ significantly from those characteristics of all journalists who responded to the survey.

The first four paragraphs substantially about the BP oil spill in each story were coded, based on the assumption that these first paragraphs typically include a story’s lead and nutgraph, which state the story’s purpose and summarize its main points.⁴⁹ Thus, this study did not include all elements of the stories, but rather stories’ primary focus and stories’ most prominent elements.

The author conducted the content analysis, but a second trained coder analyzed a randomly selected sample of 10% (N = 100) of the stories in order to establish intercoder reliability. Reliability was measured based on simple agreement and Krippendorff’s alpha. Intercoder reliability exceeded the minimum .80 level of agreement for all variables.⁵⁰

Evaluative tone

Positive stories are those stories that emphasize desirable outcomes: for example, that the environmental impacts of the oil spill were not as severe as originally feared they would be. Negative frames are those that emphasize negative outcomes: for example, setbacks in BP’s efforts to cap the well. Neutral frames either did not indicate whether outcomes were either positive or negative, or mentioned both simultaneously (simple agreement = 89%; α = .80).

Episodic/thematic frames

Using Iyengar’s typology,⁵¹ episodic frames are those that focus on a single event: for example, the latest report of where tar balls had washed ashore. Thematic frames examine broader trends and implications beyond an isolated incident: for example, a story that explores the potential long-term negative economic impacts of additional government regulation of the oil industry (95%; α = .84).

Story subject

Each story was coded based on whether it focused primarily on BP’s role in the spill or the spill’s effect on BP (such as the effects of the spill on the company’s share prices); the spill’s impact on the rest of the oil industry; the government’s role in the oil spill, including government regulation and oversight of the oil industry, government hearings related to the oil spill, and so on; the environmental or health impacts of the spill; the economic impacts of the spill; and environmental or energy policy issues related to the spill. Only the first subject mentioned (i.e., a story’s primary subject) in each story was coded (86%; α = .82).

Sources

Lastly, the dichotomous presence/absence of “official” and “unofficial” sources was coded. Official sources included BP and other oil industry representatives, elected officials, and “other” government officials, such as employees of the Coast Guard. Unofficial sources included members of non-profits, environmental activists, and volunteers; independent scientists and engineers; and the employees of local businesses, including fishermen and oil industry employees not commenting in an official capacity (91%; α = .82).

Community Structure

Structural pluralism

Structural pluralism was measured based on the distribution of the population across different social categories using Blau’s 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.⁵² The index measures the probability, on a 0-to-1 scale, that two individuals drawn at random with replacement from the population will be from two different social groups. Census data were drawn from the 2009 American Community survey (ACS) five-year estimates based on the county where a newspaper was based.⁵³ Race was divided into two groups, white and non-white; education was divided into five groups, less than high school, high school graduate, some college (including associate’s degree), college graduate, and advanced degree (master’s, JD, PhD, etc.); income was also divided into five groups, residents with a household median income less than $19,999, $20,000-$34,999, $35,000-$49,999, $50,000-$74,999, and more than $75,000; and industries were divided into three groups based on the percentage of the working population employed, the goods-producing (construction and manufacturing), services-producing (information services, transportation, retail trade, wholesale trade, etc.), and agricultural sectors.

Population was also used as an indicator of a community’s degree of structural pluralism. To preserve the 0-to-1 scale of Blau’s index so that the final structural pluralism index is more easily interpretable than it would be if the measures were all standardized, each community’s population was measured as a ratio of the population of Houston, Texas, the largest community in the sample.

While population is perhaps the best indicator of structurally pluralistic communities in the United States,⁵⁴ it is also highly correlated with circulation, which might also affect the dependent measures. Thus, to make sure that the most pluralistic communities are not just the largest, but those that are also the most diverse racially, economically, and so on, the five individual structural pluralism indicators were added together and divided by five to give each pluralism measure equal weight in the final index (M = 0.38, SD = 0.06).

Economic reliance on the oil industry

Following Griffin and Dunwoody, 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,⁵⁵ according to the ACS (M = 0.01, SD = 0.01).This measure covers a broader swath of employment outside the oil industry. Due to the oil industry’s dominance in the Gulf Coast region, however, it is assumed that most individuals in this category were employed in the oil drilling industry. For the sake of brevity, this measure is referred to as the percentage of the workforce employed in the oil industry throughout this study.

Results

The average journalist in the final sample was a white (83.5%, n = 137) male (54.3%, n = 89) reporter (76.8%, n = 126), 43 years old (M = 43.12, SD = 11.18), who leaned slightly left politically (M = 3.41, SD = 1.09). He held a bachelor’s degree in journalism (69.5%, n = 114), had been a reporter for 20 years, and had worked at his current newspaper for 11 years. He made between $40,000 and $50,000 a year. This profile is very similar to Weaver and colleagues’ national survey of journalists.⁵⁶

These reporters had some concern for the state of the environment (M = 3.77, SD = 0.57) and little affinity for the oil industry following the BP oil spill (M = 2.68, SD = 0.56). The Adversarialist role was least popular among journalists (M = 2.31, SD = 1.12); the Interpreter role received the most support (M = 4.74, SD = 0.59).

As shown in Table 3, journalists’ political ideologies were positively associated with their attitudes toward the oil industry following the BP oil spill; more conservative journalists had more positive attitudes toward the oil industry (β = .335, p < .001). Journalists’ pro-environmental beliefs were negatively associated with their attitudes toward the oil industry (β = −.336, p < .001). Journalists’ preferred professional roles were not significantly associated with their attitudes toward the oil industry. Journalists’ attitudes toward the oil industry were also positively associated with the percentage of the local workforce employed in the oil industry. Journalists who worked in communities that relied more heavily on the oil industry for employment had more positive attitudes toward the oil industry (β = .247, p < .001).

Table 3.

Standardized Coefficients of Path Analysis Predicting Gulf Coast Newspaper Journalists’ Attitudes toward the Oil Industry and Their Coverage of the BP Oil Spill.

	Journalists’ attitudes toward the oil industry	Content analysis variables
	Journalists’ attitudes toward the oil industry	Story tone	Focus on BP’s responsibility	Thematic frames	Reliance on unofficial sources
Attitudes toward the oil industry		.126*	.002	.114	.040
Political ideology	.335***
Environmental beliefs	−.336***
Professional roles
Interpreter	−.038	−.072	.098*	.234*	−.140**
Disseminator	−.054	−.076	−.057	−.071	.054
Adversary	−.058	−.063	−.055	.069	. 097
Populist Mobilizer	.186	.126	−.184***	.081	.080
Community structure
% of workforce employed in oil industry^a	.247***	.098	.091	−.112	.082
Structural pluralism^a	.131	−.124*	.087	.065	.081

Note. χ²(20, N = 987) = 22.556, p = .311; RMSEA = .011; CFI = .977; TLI = .922. RMSEA = root mean square error of approximation; CFI = comparative fit index; TLI = Tucker–Lewis index.

% of the workforce employed in the oil industry and structural pluralism were log-transformed to correct for non-normality.

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

In regard to journalists’ coverage, overall they were significantly more likely to focus on the government’s (n = 318) over BP’s (n = 115) responsibility for the oil spill, χ²(1, N = 1000) = 202.86, p < .001. The least frequent subject in journalists’ coverage was environmental or energy policy (N = 13). Journalists were also significantly more likely to use episodic (n = 941) rather than thematic frames (n = 59), χ²(1, N = 1000) = 774.40, p < .001, and to rely on official (n = 774) rather than unofficial sources (n = 386), χ²(1, N = 1000) = 635.20, p < .001. Approximately half the journalists’ stories were negative (n = 499), and approximately 37% (n = 368) were neutral; just 13% (n =133) were positive.

Hypotheses and Research Questions

Data analysis was conducted using MPlus (version 6), a software that can handle categorical dependent variables (i.e., story tone, BP linkage, thematic frames, and unofficial characters) and can group non-independent observations—in this instance, more than one story written by the same author—into clusters. Creating these clusters controls for violation of the statistical assumption of independent observations.⁵⁷

Before conducting the multivariate analysis, thirteen cases that were missing data for the survey measures were deleted, the most conservative approach to dealing with missing data that affects fewer than 5% of the cases.⁵⁸ Thus, the multivariate analysis included 987 newspaper stories and 156 unique authors (i.e., clusters).

While the model was a good fit,⁵⁹ it explained a relatively small proportion of the variance in journalists’ focus on BP (R² = .095), thematic frames (R² = .092), the tone of their stories (R² = .062), or their use of unofficial characters (R² = .045). The model did explain a much larger proportion of the variance in journalists’ attitudes toward the oil industry (R² = .451). The results of the path model are shown in Table 3.

H1 predicted that journalists’ positive attitudes toward the oil industry would negatively predict whether they wrote negative stories about the oil spill (H1a) and whether their stories focused on BP’s role in the oil spill (H1b), used thematic frames (H1c), or used unofficial sources (H1d). As shown in Table 3, only H1a was supported: journalists with more positive attitudes toward the industry wrote more positive stories about the oil spill (β = .126, p < .05).

RQ1 asked whether journalists’ preferences for the different professional roles predicted how they covered the 2010 BP oil spill. As shown in Table 3, journalists who embraced the Interpreter role were less likely to use unofficial sources (β = −.140, p < .01), were more likely to use thematic frames (β = .234, p < .05), and more likely to focus on BP’s role in the oil spill (β = .098, p < .05). Journalists who embraced the Populist Mobilizer role were significantly less likely to focus on BP’s role in the oil spill (β = −.184, p < .001). The other roles were not significantly associated with journalists’ coverage.

H2 predicted that communities’ economic reliance on the oil industry will negatively predict whether journalists wrote negative stories about the oil spill (H2a) and whether their stories focused on BP’s role in the oil spill (H2b), used thematic frames (H2c), or used unofficial sources (H2d). As shown in Table 3, the percentage of the local workforce employed in the oil industry was not significantly associated with journalists’ coverage of the BP oil spill. Thus, H2 is not supported.

H3 predicted that communities’ degree of structural pluralism would positively predict whether journalists wrote negative stories about the oil spill (H3a) and whether their stories focused on BP’s role in the oil spill (H3b), used thematic frames (H3c), or used unofficial sources (H3d). As shown in Table 3, only H3a was supported. Journalists in larger, more pluralistic communities did write significantly more negative stories about the oil spill (β = −124, p < .05).

H4 predicted that the indirect effects of communities’ economic reliance on the oil industry and structural pluralism on journalists’ coverage of the BP oil spill, via journalists’ attitudes toward the oil industry, would exceed the direct effects. Only structural pluralism had a direct effect on the tone of journalists’ stories (β = −.124, p < .05). The indirect effect of structural pluralism on the tone of journalists’ stories, via journalists’ attitudes toward the oil industry, can be calculated by taking the product of the path from structural pluralism to journalists’ attitudes (β = .131) and the path from journalists’ attitudes to story tone (β = −.124), which equals −.016. The direct effect is larger. Thus, H4 is not supported.

Conclusion

According to Niven,

There are hundreds of articles published every year in U.S. newspapers on media bias, most alleging that newspapers favor liberals and Democrats over conservatives and Republicans . . . [but] despite the importance of this coverage of bias and this belief in bias, scholarly research on this topic has been hamstrung by limitations of method.⁶⁰

This study responded to the need for methodological advancement in how bias is measured, matching individual-level survey data with a content analysis of the stories individual journalists wrote. As compared to aggregate-level data, this individual-level survey and content analysis data establish a more methodologically sound link between journalists’ ideology and their coverage than previous studies of bias.

This study’s results, though, should be interpreted cautiously. It examined one non-randomly selected subset of newspaper journalists, in one region, reporting on a specific topic. Nonetheless, this study corroborates what others have found: There is minimal partisan bias in journalists’ coverage. More conservative, less environmentally concerned journalists had more positive attitudes toward the oil industry and in turn wrote slightly more positive stories about the spill. Journalists’ attitudes toward the oil industry, however, did not affect whether journalists focused on BP’s role in the disaster, their use of thematic frames, or their use of unofficial sources.

The fact that ideology was a strong predictor of journalists’ attitudes toward the oil industry suggests that, personally, these journalists certainly are not ideologically neutral. But whether it is due to professional norms that guide their work or the watchful eye of an editor, journalists’ personal attitudes by and large were not reflected in their coverage of the oil spill.

Despite these research findings, the debate over bias in mainstream news coverage is unlikely to subside. On one side of the debate, there are too many politicians, commentators, and average Americans who perceive that mainstream journalists’ coverage is biased. On the other hand, American journalists claim their work can be objective.⁶¹ The public is likely to continue to be skeptical of such claims, because as suggested above, journalists are not ideologically neutral. That does not relate directly to journalists’ claims that their coverage is objective—in theory, one can have strong attitudes that do not affect one’s work. But this study did find, however small, trace evidence that journalists’ personal ideology did have an effect on the tone of their coverage. Perhaps American journalists’ insistence that their work is objective simply adds fuel to the fire of the “liberal media” debate, particularly when there are bits of evidence that journalists are not drones who can completely separate their professional coverage from their personal beliefs. There is, though, consistent evidence that individual-level ideological bias does not play a significant role in shaping journalists’ reporting.

There are, however, other factors that shape the production of media content, which are worthy of researchers’ attention. Beyond ideology, this study found that journalists’ coverage is also shaped by the professional and social contexts in which they work. Journalists who more strongly endorsed the Interpreter professional role used more thematic frames, delving more deeply into the underlying causes and consequences of the BP oil spill. However, these Interpreters were also less likely to use unofficial sources who are most likely to raise critical perspectives, reinforcing officials’ explanations over independent, critical perspectives.

There were also some contradictions in stories of those journalists who embraced the Populist Mobilizer role; that is, they saw it as their role to give readers a voice in the media and to encourage them to be actively involved in community affairs. These journalists were less likely to focus on BP’s role in the crisis (perhaps because they viewed the government as more credible, and as a representative of the public’s voice), but they were not more likely to use unofficial sources (i.e., give the public a voice in their coverage). While most previous work on journalists’ professional roles has been descriptive in nature,⁶² this study suggests journalists’ beliefs about their role in society can also have predictive value. Future studies might also examine how well journalists’ work matches up with their self-conceived roles. (One might expect, for example, that Populist Mobilizers would give greater voice to unofficial sources in their coverage.) The method used here of matching individual-level survey and content analysis data would be useful to such endeavors.

In regard to the community context in which journalists work, journalists in larger, more heterogeneous (i.e., structurally pluralistic) communities, in which social power is more likely to be dispersed, appeared to be freer to write negative stories involving a powerful corporate interest, such as BP. Previous studies of structural pluralism’s effect on news coverage have not offered an explanation of how such a structural bias enters the news-gathering process.⁶³ This study did not find support for the suggestion that structural pluralism shapes journalists’ attitudes toward issues in the news, which in turn affect journalists’ coverage (i.e., that journalists’ attitudes mediate the relationship between structural pluralism and coverage). It is possible the lack of this individual-level path suggests that what biases exist at the community level are more institutional—whether at the level of media outlets, or the profession as a whole—than individual in nature. Future research should further examine how the influence of community structure enters the news-gathering process.

This study did not address more institutionalized forms of bias, which would not manifest in comparisons of individual journalists’ coverage. However, despite a methodologically better match between individual journalists’ personal beliefs and their coverage, this study found results very similar to other studies that have concluded there is little evidence of individual-level biases in mainstream journalists’ coverage.⁶⁴ Thus, perhaps there is little value in continuing to look for small traces of individual-level bias in journalists’ coverage (though audiences’ perceptions of bias deserve further study).⁶⁵ One might make similar conclusions about researching the role of community structure or journalists’ preferred professional roles based on the size of the effects observed in this study. However, if journalists are part of a very complex professional, social, economic, and political system, as previously suggested, one would expect that any one factor would have a small impact on journalists’ coverage. Further studies of community structure are worthwhile because there are still important theoretical advances to be made in understanding through what means community structure influences the news-gathering process. And due to the descriptive nature of previous studies of journalists’ preferred professional roles, there are likewise advancements to be made in that arena.

Footnotes

Acknowledgements

The author thanks associate editor Dr. Julie Andsager and the anonymous reviewers for their assistance in improving this manuscript.

Declaration of Conflicting Interests

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

Funding

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

Assessing Ideological,Professional,and Structural Biases in Journalists’ Coverage of the 2010 BP Oil Spill

Abstract

Keywords

Literature Review

Journalists’ Beliefs, Attitudes, and Reporter Bias

Professional Roles

Community Structure

Journalists’ Coverage of the Oil Spill

Hypotheses

Method

Survey

Survey Measures

Political ideology

Environmental ideology

Professional roles

Attitudes toward oil drilling

Demographics

Content Analysis

Evaluative tone

Episodic/thematic frames

Story subject

Sources

Community Structure

Structural pluralism

Economic reliance on the oil industry

Results

Hypotheses and Research Questions

Conclusion

Footnotes

Acknowledgements

Declaration of Conflicting Interests

Funding

Notes