Reading from a journal article and answering questions

Corporate Ownership and News Bias Revisited: Newspaper Coverage of the Supreme Court’s Citizens

United Ruling

CATIE SNOW BAILARD

The clear financial benefits accrued to owners of television stations as a result of the Citizens United v. Federal Elections Commission (FEC) decision opens the door to an important question: Did the degree to which media corporations benefited from the changes in campaign finance law influence their news outlets’ coverage of the Citizens United decision? In other words, is it possible to identify variation in how media outlets covered the Supreme Court decision that correlates with the degree to which those outlets’ parent companies profited from the resulting increase in campaign spending? Answering this question will provide an important and much-too-uncommon opportu- nity to systematically test for bias in news coverage. Replicating the method used by Gilens and Hertzman (2000) in their own test of coverage of the 1996 Telecommunications Act, this analysis reveals that newspapers belonging to media corporations that own more television stations covered the Citizens United ruling systematically differently—and more favorably—than those with few or no television stations. This has important implications for the degree to which the news produced by increasingly conglomerated and corporatized media companies may eschew neutral or balanced coverage in favor of news frames that promote their own financial interests.

Keywords campaigns, Citizens United v. FEC, elections, media bias, news, political advertising

“I think Thomas Jefferson would have said, ‘The more speech, the better.’ That’s what the First Amendment is all about, so long as the people know where the speech is coming from.… You can’t separate the speech from the money that facilitates the speech.”

“What the Supreme Court did in Citizens United is to say to these same billionaires: ‘You own and control the economy, you own Wall Street, you own the coal companies, you own the oil companies. Now, for a very small percentage of your wealth, we’re going to give you the opportunity to own the United States government.’”

In Citizens United v. Federal Elections Commission (2010), the Supreme Court ruled by a 5–4 decision that independent political expenditures by corporations and

Catie Snow Bailard is Assistant Professor, School of Media and Public Affairs, George Washington University.

Address correspondence to Catie Snow Bailard, Assistant Professor, George Washington University, School of Media and Public Affairs, 805 21st Street NW, Suite #400, Washington, DC 20052. E-mail: [email protected]

Political Communication, 33:583–604, 2016 Copyright © Taylor & Francis Group, LLC ISSN: 1058-4609 print / 1091-7675 online DOI: 10.1080/10584609.2016.1142489

583

unions qualified as protected speech under the First Amendment, effectively opening the door to unlimited campaign spending by these groups. Under the new law, any indivi- dual or group could now donate any sum of money to a super political action committee (i.e., Super PAC), who could spend that money as they saw fit to attack or support any candidate or proposition up until Election Day—so long as this spending was indepen- dent of the national parties and the candidates’ own campaigns. This overturned previous restrictions on electioneering put in place by McCain and Feingold’s Bipartisan Campaign Reform Act of 2002, which had limited what could be said and when in advertisements sponsored by these groups. In addition, while current law still requires public disclosure of the identities of contributors to Super PACs, nonprofit corporations and local employee associations that qualify as “social welfare groups,” according to the Internal Revenue Code section 501(c)(4), are exempt from disclosure requirements in the wake of the Citizens United ruling.

In the years following this ruling, political spending in elections has mushroomed. The total cost of the 2012 elections topped out at $6.3 billion—an increase of more than 50% over the $4.15 billion price tag of the 2004 elections (Center for Responsive Politics, 2012). A substantial portion of this increase is a result of independent political spending by the outside groups that were empowered by the Citizens United ruling, many of whom did not disclose their donors. Specifically, outside spending in 2012 surpassed $1.3 billion dollars—whereas more than $600 million of this came from Super PACs who are required to disclose the source of their funds, more than $300 million was also spent by the non-disclosing social welfare groups (Center for Responsive Politics, 2012).

As the quotes presented earlier illustrate, Citizens United is a flashpoint for heated debate. While it has drawn scorn from citizens and politicians alike—including a memorable public remonstration of the Supreme Court members by President Barack Obama during a State of the Union address—the ruling also has many strong propo- nents. Supporters of the decision include the National Rifle Association, the United States Chamber of Commerce, analysts with the Cato Institute and Heritage Foundation, and the American Civil Liberties Union. In response to critics’ protests that the ruling would increase government corruption, for example, Cato Institute analysts John Samples and Ilya Shapiro (2010) argued that, although the decision may lead to costlier campaigns, it is important to remember “that none of this money will go directly to candidates for office. It will go instead to broadcasting or otherwise communicating speech about candidates and issues. Such increases in spending should be welcome because studies have shown that more spending—more political communication—leads to better-informed voters.”

Another perspective in the debate regarding the consequences of Citizens United suggests that, whether or not more money in politics is a good thing in theory, the deluge of spending that followed this ruling may not make all that much difference in practice. In this vein, the Sunlight Foundation calculated the “returns on investment” that outside groups received for their 2012 general election spending. They found that, “After outside groups spent more than $1.3 billion in independent expenditures to influence the outcome of the election, we now get to see just what all that money bought them—or didn’t. Turns out some of the smart money wasn’t so smart after all when it came to making political bets” (Young, 2012). For example, of the more than $100 million spent by American Crossroads, only 1.29% of the races that received a portion of their funds ended in the desired result, with the U.S. Chamber of Commerce only faring marginally better at 6.9% (Young, 2012).

584 Catie Snow Bailard

Whatever the consequences for the candidates, policy groups, citizens, and the American democratic process prove to be, there is one group that has indisputably benefited from the Citizens United ruling—anyone working in a field associated with the campaign industry. According to the Center for Media and Democracy, nearly $500 million of the $1.3 billion dollars spent by outside groups in the 2012 election was funneled through just six media companies, who typically used the money to produce ads or to purchase the slots where the ads ultimately aired (Dooling, 2012). Other undeniable winners in the deluge of political spending post-Citizens United are the television stations that air these political advertisements. Specifically, in 2012 political groups spent nearly $3.4 billion on television advertising—with $2.8 billion going to local television stations, $104 million spent on national networks, and $467 million spent at cable networks (Lieberman, 2012). According to McChesney and Nichols (2012), “Back in the 1960s and ’70s TV candidate advertising constituted an almost imperceptible part of total TV advertising revenues”; however, in 2012 “political advertising will account for over 20% of TV station ad revenues.” In the words of industry insider Eric Greenburg, “Political advertising and elections are to TV what Christmas is to retail” (quoted in McChesney & Nichols, 2012). Thus, it is not surprising that in the wake of the Citizens United ruling, “Every media channel and media outlet, from local stations to the networks, is licking their lips over this feast of spending.” (Kip Cassino, quoted in Dexheimer, 2012)

The clear financial benefits accrued to owners of television stations as a result of the Citizens United decision opens the door to an important question: Did the degree to which media companies benefited from the changes in campaign finance law influence their news outlets’ coverage of the Citizens United decision? In other words, is it possible to identify variation in how news outlets covered this Supreme Court decision that correlates with the degree to which those outlets’ parent companies profited from the resulting increase in campaign spending?

Answering this question will provide an important and far-too-uncommon opportunity to systematically test for bias in news coverage. In an era when cries of media bias abound, the opportunities to systematically and empirically test for biased coverage remain elusive. Nevertheless, as media consolidation continues apace, whether the interests belonging to the handful of corporations that own the majority of America’s media outlets seep into the news coverage that their outlets produce has clear import in a political system founded on the ideals of robust and informed political deliberation, a marketplace of ideas, and a watchdog press that is vigilant on behalf of citizens.

The problem in testing for bias, however, remains formidable. There simply is no objective and absolute standard of what unbiased coverage would look like with which to compare the news coverage that does exist. This is, in part, because what constitutes biased coverage very much depends upon whom you ask. It is for this reason, according to Gilens and Hertzman (2000), that “the more general arguments often heard about whether the press has a conservative or a liberal bias are almost by their nature irresolvable…. Being unable to agree on what ‘impartial’ or ‘unbiased’ coverage would look like, it is hardly surprising that we disagree about whether actual coverage is biased, and if so, in what direction” (p. 371).

However, rather than abandon the scientific pursuit entirely, Gilens and Hertzman (2000) offer a compelling alternative: “In the absence of an ‘objective standard’ against which media bias could be identified, the most promising approach is to focus on issues for which different media owners have different interests, asking whether news content differs in accord with those differing interests” (p. 371). In this vein, Gilens and Hertzman tested whether the degree to which various media corporations stood to benefit from the

Corporate Ownership and News Bias Revisited 585

loosening of TV station ownership restrictions, a key component of the 1996 Telecommunications Act, predicted variation in how that act was covered by the news- paper outlets that belonged to those corporations. Their analysis revealed that companies that stood to gain more financially from the loosening of ownership restrictions covered the 1996 Telecommunications Act significantly differently from those that did not stand to benefit. “In short, very different pictures of the likely effects of this legislation were being painted by the different newspapers examined, pictures that served to further the interests of the newspapers’ corporate owners rather than the interests of their readers in fair and complete coverage of an important public policy issue” (2012, p. 383).

This analysis provided compelling insight into the capacity for corporate owners’ interests to slant the content of news produced by their respective outlets. Unfortunately, there have been relatively few opportunities since to continue and expand this line of analysis. This is largely due to the fact that the majority of outlets are owned by a shrinking number of corporations (Bagdikian, 2004). As a result, it is rare when the interests belonging to this handful of corporations vary from one another in clear, systematic, and measurable ways. Quite simply, in most cases, regulatory policies, judicial decisions, and tax laws tend to advantage or disadvantage these large corporations to relatively commensurate degrees.

It is for this reason that the Citizens United ruling provides a compelling opportunity to empirically investigate the degree to which the financial interests of media corporations may shape the news that Americans receive about pertinent policies and issues. Since the majority of the financial windfall precipitated by the deluge of political spending in the wake of the Citizens United ruling was spent on political advertising on television, this provides a clear case in which media corporations’ interests vary from one another in systematic and measurable ways. Media corporations with a greater number of television stations have benefited directly from this ruling and they have benefited to a greater degree than corporations with fewer or no television stations. And, the reason for this is self- evident: the more television stations a media corporation owns, the more airtime they have available to sell to political advertisers. Accordingly, this variation offers a rare and important opportunity to systematically test for the influence of corporate owners’ interests on the content produced by their news outlets.

In the following sections, I briefly review the current state of research on media bias. I then replicate and expand the method employed by Gilens and Hertzman to analyze the content of newspaper coverage of the Citizens United ruling. The findings of this analysis reveal measurable differences in the content of the coverage, which vary according to the degree to which the corporate owners of the newspapers benefited from the ruling. Mirroring Gilens and Hertzman’s findings in their own analysis of coverage of the 1996 Telecommunications Act, newspapers belonging to media corporations that own more television stations covered the Citizens United ruling systematically differently from—and more favorably than—those with few or no television stations.

As an additional test of the robustness of these findings, I conduct a second analysis of the content of coverage published by this same set of newspapers regarding another recent, high-profile, and controversial Supreme Court decision: Burwell v. Hobby Lobby Stores, Inc. This additional analysis addresses important omitted-variable bias concerns that, rather than financial interests, there may be alternative, confounding variables that are actually driving the slant of the Citizens United coverage, such as the ideological compo- sition of the newsrooms or their targeted audiences or other constraints related to adver- tising or circulation demands.

586 Catie Snow Bailard

Hobby Lobby is an ideal case for a comparative content analysis, since both cases deal with corporate personhood and abut with prominent civil liberties issues (freedom of speech and freedom of religion, respectively). However, unlike Citizens United, the Hobby Lobby decision does not have clear financial implications for the corporate owners. Therefore, if there is another factor driving the content of coverage, we should expect to see similar slants in the coverage of these two different decisions. Instead, the findings reveal that, whereas the coverage of Citizens United was clearly correlated with the financial interests of the parent corporations, the Hobby Lobby coverage did not follow a similar trajectory. The results of this additional test assuage omitted-variable bias concerns and support the conclusion that the parent corporations’ financial interests did seep into their newspaper outlets’ coverage of the Citizens United decision.

Literature

Gilens and Hertzman outlined three categories that comprised the body of extant media bias literature at the time they published their study in 2000. The first includes case studies in which media owners’ interests clearly influenced the news content produced by their outlets (Bagdikian, 1997). While illustrative, this anecdotal approach lacks a systematic mechanism to uncover and test the full universe of bias that is potentially driven by the interests of media owners. The second set of research examines bias that the news media as a whole exhibits in covering specific issues, often in terms of whether media generally favors a liberal or conservative tilt (Bennett, 1988; Bozell & Baker, 1990; Cohen & Solomon, 1993; Herman & Chomsky, 2002; Lichter, Rothman, & Lichter, 1986; Parenti, 1986; Soderlund & Schmitt, 1986). While instructive, this approach is somewhat handi- capped by the absence of an objective standard of unbiased news with which to make a credible empirical comparison. The final category represents the handful of studies that follow the approach recommended by Gilens and Hertzman, which identify a specific area where media corporation owners’ interests diverge from one another in clear and measur- able directions and then compare the coverage produced by the news outlets belonging to these corporations accordingly (Burriss & Williams, 1979; Pratt & Whiting, 1986; Snider & Page, 1997)

One additional category of research on media bias overlooked in this summary is the consideration of how journalistic norms and specific economic features of media compa- nies and their markets may shape which stories are told and how. Within this vein, Fico and Cote (1999) demonstrate how the news-gathering practices of journalists and news- room norms may contribute to a surfeit of one-sided or structurally imbalanced news stories about campaigns. An earlier analysis of local newspaper coverage of controversies (Fico, Lacy, & Simon, 1989) found that the size of journalists’ workload increased imbalance in these news stories. In addition, this study found that group ownership decreased fairness in how these stories were reported, but intercity competition had the reverse effect of increasing fairness in these news stories.

Since the publication of Gilens and Hertzman’s study, partisan media has experienced a resurgence in the United States—primarily thanks to the growth of cable television and the Internet blogosphere (Baum & Groeling, 2008). Accordingly, the literature has wit- nessed a sizable increase in the number of studies that fall into the second category of literature characterized by Gilens and Hertzman. These studies include analyses that operationalize and measure partisan or ideological bias in the media generally (Entman, 2007; Gentzkow & Shapiro, 2010; Groseclose & Milyo, 2005), in regard to presidential elections (D’Alessio & Allen, 2000), as well as in regard to specific policy areas, such as

Corporate Ownership and News Bias Revisited 587

the economy (Larcinese, Puglisi, & Snyder, 2011), immigration (Branton & Dunaway, 2009), or the environment (Dispensa & Brulle, 2003).

A related set of literature tests the effect of profits and corporate ownership on dimensions of news coverage other than bias. Since these papers do not test for bias specifically, they do not fall squarely into Gilens and Hertzman’s third category. However, these studies do offer important insight into the potential for corporate ownership and market pressures to shape news content. For example, Dunaway (2008) finds that corpo- rate ownership (as opposed to private ownership) and higher levels of market competition are associated with a decrease in issue coverage in campaign news produced both by newspapers as well as television stations. Another set of studies consider the effect of profit motivation on the substantive content of the news produced by outlets (Hamilton, 2011; Zaller, 1999), as well as the effect of profit motivation on the quality of journalistic output (Klinenberg, 2005; McManus, 1995).

More broadly, classic studies employing sociological and organizational theory argue that news production is not simply an act of journalists mirroring reality, but rather a product of the interaction between the organizational structures, market and financial pressures, and professional norms that characterize the media industry. For example, seminal work by Epstein (1974) found that “the pictures of society which are shown on television as national news are largely—though not entirely—formed and shaped by organizational considerations,” of which economic considerations figure heavily (p. 258). In another foundational study, Gamson, Croteau, Hoynes, and Sasson (1992) argue, “Media empires are not simply a result of the market system; they also serve as cheerleaders for it. Bottom-line pressure to turn a profit plus the need to protect the image of corporations as good citizens will continue to put pressure on journalists to create media content that is politically safe” (p. 379). Finally, McManus (1995) employs market theory to highlight the intersections where market norms are likely to trump journalistic norms, “for mass-mediated news supported by advertising, achieving the greatest return requires a subordination of most journalism norms to market norms…. High-minded owners-inves- tors could direct that sometimes market norms give way to journalism norms. But under most market conditions, they would have to be willing to accept less return” (pp. 327–328).

In All the News That’s Fit to Sell: How the Market Transforms Information Into News, Hamilton (2011) delves further into the financial motives and market structures that shape the news that audiences receive. Although much of the discussion focuses on cost-benefit calculus made by journalists in crafting news that will attract the largest possible audience, Hamilton also considers the economic motives of ownership. First, he explores and substantiates the assumption that the primary motive of owners is profit maximization, particularly in light of the growth of publicly held media corporations with boards accountable to shareholders in recent decades. The implication of this for the present analysis is that profit-maximizing owners are likely to be aware of and favorable to legislation and judicial decisions that promise to increase the profitability of their corpora- tion, such as the Citizens United decision.

However, the question remains, through what mechanisms might owners’ preferences trickle down and shape the coverage produced by their news outlets? Hamilton offers a simple answer: Journalists have incentives to produce coverage that does not harm the interests of their corporate owners. “News workers may be reluctant to provide unfavor- able news coverage of the parent company. Allegations of this nature have been made about ABC’s treatment of stories about Disney World and NBC’s handling of information about the nuclear power industry, a sector important to its parent company General

588 Catie Snow Bailard

Electric” (p. 25). Moreover, Hamilton stipulates that journalists do not have to consciously weigh the variety of profit-maximizing considerations each time they write a story; instead, “A journalist will not explicitly consider each of these economic questions in crafting a story. The stories, reporters, firms, and media that survive in the marketplace, however, will depend on the answers to these questions…” (p. 7).

In Media, Markets, and Democracy, Baker (2001) suggests another avenue through which the preferences of profit-maximizing media owners might shape the news produced by their outlets: the need to appease their advertisers. “As the media’s dominant paymaster, advertisers influence media enterprises to give audiences editorial content that advertisers want them to receive…. Advertisers also influence media… to avoid content that dis- parages the advertisers’ products or political agenda” (p. 25). Accordingly, Baker asserts that there are instances when the desires of advertisers will supersede the interests of the audience in the production of news. This provides another avenue through which the preferences of profit-maximizing owners might shape the content of the news produced by their outlets, since media corporations that own more TV stations are likely to have more developed relationships with political advertisers than those that do not own TV stations. Thus, media companies that have more established and reciprocally dependent relation- ships with political advertisers (who have long paid hefty sums to these corporations to advertise on their TV channels) are likely to have incentives to produce favorable coverage of Citizens United (which serves the interests of those political advertisers), shaping their news coverage of the decision accordingly.

In summary, the incentives to create coverage that is favorable to the interests of the media owners is likely a result of journalists’ incentives to weigh the preferences of their profit-maximizing owners (both explicitly and implicitly) as well as the preferences of their advertisers.

This body of research provides a solid foundation for understanding and testing the effect of corporate ownership on news content. The present study contributes to this field by testing whether the different degree to which media corporations financially benefited from the Citizens United ruling influenced the content of news coverage about the ruling produced by those corporations’ newspaper outlets. The findings of this analysis provide suggestive evidence of the degree to which media owners’ interests may seep into and slant news coverage more generally and regularly.

Analysis

In this section, I replicate and expand upon the methodology employed by Gilens and Hertzman in their analysis of coverage of the 1996 Telecommunications Act in order to test whether media corporations that benefited differently from the large increase in political spending precipitated by the Citizens United ruling produced newspaper coverage that varied accordingly. This begins by identifying the 100 largest media companies in terms of media-derived revenue (Johnson, 2010). Next, following Gilens and Hertzman’s model, I categorized the companies included in this list into three sets—those with no television stations, those with limited television stations (i.e., between 1 and 18 stations), and those with a substantial number of television stations (i.e., 19 or more stations) (Columbia Journalism Review, 2013).1 Next, I identify which of these companies owned at least one daily newspaper (The Pew Research Center’s Project for Excellence in Journalism, 2013) with full text archived in either the LexisNexis or ProQuest databases. This rendered a list of 12 media companies with 33 newspapers. (Please see Table 8 in the supplemental Appendix for a full list.)

Corporate Ownership and News Bias Revisited 589

I then compiled all articles referencing the Citizens United ruling from these news- papers via LexisNexis and ProQuest searches of the phrase “Citizens United” between the dates of March 1, 2009, and February 28, 2011. This two-year time frame is necessary due to the multiple stages of this case and its implications for political spending, which was first argued March 24, 2009, then reargued September 9, 2009, then decided on January 1, 2010, with the tangible consequences of which first becoming manifest in the fall 2010 election.

Editorials and opinion pieces were excluded from the analysis in order to ensure a more difficult and appropriate test for slanted coverage. Whereas, as dictated by their nature, editorials advocate particular perspectives or positions, the expectation of straight news is that of balance and objectivity. Therefore, if an analysis of non-editorial content reveals skewed coverage, this will provide a much more compelling empirical demonstra- tion of the potential for the financial interests of media corporations to distort the range of perspectives made available to the public through their news outlets. This is not to suggest that an analysis of editorial content would not provide another valuable opportunity to test for systematic variance in coverage that reflects corporate owners’ self-interest; however, this is not the focus of the present analysis. Accordingly, this search yielded 439 non- editorial and non-opinion news pieces for analysis that mentioned the Supreme Court case directly.

Next, two graduate student coders read a sample of newspaper articles referencing the Citizens United ruling from newspapers that were not included in this analysis in order to identify positive and negative consequences and implications of the ruling that were mentioned in news coverage. Based on this analysis, we constructed a comprehensive list of potential positive and negative ramifications associated with the decision, which are listed in Tables 1 and 2.2 After establishing inter-coder reliability, the coders (who were blind to the parent corporation of the newspapers) independently coded half of total articles constituting this analysis.3

Findings

This analysis begins with two tests of the overall tone of the content of these news stories. The first test employs an ordinary least squares (OLS) regression to test the relationship between the number of TV stations owned by the newspaper’s parent corporation and the overall balance of positive to negative consequences mentioned in the articles—operatio- nalized as the aggregate number of positive consequences mentioned minus the aggregate number of negative consequences mentioned in each article (so that positive integers represent many more positive mentions than negative mentions in an article, zero signifies that there were an equal number of positive as negative mentions, and negative values indicate more negative consequences mentioned compared to positive). The findings of this regression follow the trajectory predicted by the financial interests of the parent corporations (p ≤ .01). (Please see Table 3.) For example, whereas 17% of the articles in papers owned by companies with no TV stations mentioned at least one more positive consequence than negative consequence, 28% of the articles published in newspapers belonging to parent companies with 19 or more TV stations mentioned at least one more positive consequence than negative.

In the second test, the dependent variable is operationalized as the percentage of positive consequences mentioned out of the total number of potential consequences (negative and positive) mentioned in a given article. The results of this test also correspond with the degree to which the parent companies benefited from the ruling—with the

590 Catie Snow Bailard

percentage of positive mentions significantly higher in articles published by newspapers belonging to parent companies that owned more TV stations (p ≤ .01). (Please see Table 3.)

Turning to more granular tests of the content of these news articles, Kendall tau-b tests—which compare associations of ordinal variables across two or more groups—reveal significant differences in the number of positive consequences mentioned in the articles published by newspapers owned by corporations with no television stations, those with a limited number of stations, and those with a substantial number of stations. Moreover, this differential correlates with the divergent financial interests belonging to each set of companies.

Table 1 Positive consequences of Citizens United ruling mentioned in newspaper articles

Citizens United No TV Stations

Limited TV

Stations

Substantial TV

Stations

Tau-b (Asymptotic–

SE)a

Logistic Coefficient

(SE)b

Ends censorship and/or restores free speech

34% 28% 48% .13 1.02 (.05) (.001)

p ≤ .01*** p ≤ .01*** Levels the playing field for

corporations to compete politically with other groups

9% 9% 18% 1.03 .10 (.008) (.05) p ≤ .01***

p ≤ .04** Gives corporations and

individuals freedom to do with their money as they please

20% 21% 25% .04 1.004 (.05) (.008)

– p ≤ .57

Rightfully recognizes corporations as people

2% 3% 5% .06 1.02 (.05) (.01)

– p ≤ .29 Allows for more voices in

the policy dialogue 9% 7% 18% 1.03

.10 (.008) (.05) p ≤ .01***

p ≤ .04** Gives voices to job creators

and/or those who promote economic growth

0.5% 2% 4% .10 1.06 (.04) (.02)

p ≤ .01*** p ≤ .01*** Articles that mention any

positive consequences 45% 47% 64% .15 1.46

(.04) (.17) p ≤ .01*** p ≤ .01***

Articles that mention 2 or more positive consequences

20% 18% 34% .12 .005 (.05) (.001)

p ≤ .01*** p ≤ .01***

Note. Inter-coder reliability scores for each code indicated in parentheses. aAll tau-b test p-values are reported as two-tailed tests. bAll logistic results in this table are derived

from bivariate regressions with the number of TV stations owned by parent corporations as the discrete independent variable (with all results reported in odds ratio).

Corporate Ownership and News Bias Revisited 591

Please note that Kendall tau-b tests are used as the primary models in this portion of the analysis in order to remain consistent with the methodology employed by Gilens and Hertzman. However, to further test the robustness of these findings, each of these tests is re-run employing a logistic bivariate regression (reported in odds ratio) with the discrete number of TV stations owned by the parent corporation serving as the explanatory

Table 2 Negative consequences of Citizens United mentioned in newspaper articles

Citizens United No TV Stations

Limited TV

Stations

Substantial TV

Stations Tau-b (ASE)a

Logistic Coef. (SE)b

Creates unlimited corporate influence and/or drowns out voices of average citizens

50% 50% 45% –.03 .99 (.05) (.006) – p ≤ .04**

Restricts the democratic process 14% 13% 15% .01 1.00 (.05) (.009) — p ≤ .99

Makes possible future detrimental changes to election processes

6% 5% 8% .02 1.003 (.05) (.01) – p ≤ .81

Drives up costs of Campaigns 17% 11% 14% –.05 .99 (.05) (.009) – p ≤ .16

Eliminates financial disclosures that are positive for the political process

23% 31% 17% –.04 .98 (.04)– (.008)

p ≤ .04** Treats corporations as people, which

is not valid 10% 4% 11% .01 1.01

(.05) (.01) – p ≤ .24

Allows foreign money into our campaigns

7% 15% 9% .04 1.002 (.04) (.01) – p ≤ .82

Increases attack advertising 13% 11% 18% .06 1.0005 (.05) (.009) – p ≤ .96

Increases risk of corruption 12% 9% 15% .03 1.01 (.05) (.009) – p ≤ .16

Articles that mention any negative consequences

65% 69% 62% –.01 .99 (.05) (.006) – p ≤ .06*

Articles that mention 2 or more negative consequences

45% 42% 42% –.03 .99 (.05) (.006) – p ≤ .1*

aAll tau-b test p-values are reported as two-tailed tests. bAll logistic results in this table are derived from bivariate regressions with the number of TV stations owned by parent corporations as the discrete independent variable (with all results reported in odds ratio).

592 Catie Snow Bailard

variable. In each case, the results retained or increased in significance in the logistic tests. (Results of both the Kendall tau-b and logistic regressions are included in Tables 1 and 2.)

The tests reveal that, across all six of the positive consequences tested, the number of positive mentions in newspaper articles belonging to media companies with substantial television holdings is substantially larger than the number of positive mentions in papers belonging to parent companies with no stations. And, this differential reaches statistical significance for four of these positive consequences: that Citizens United will end censor- ship and/or restore free speech, level the playing field for corporations to compete politically with other groups, allow for more voices in the policy dialogue, and give voices to job creators and/or those that promote economic growth.

Turning back to more summary measures, a test of the number of articles that mentioned any positive consequences at all is also reflective of the owners’ financial interests (p ≤ .01). Whereas 45% of the articles published in newspapers with owners with no TV stations mentioned at least one positive consequence, 62% of the articles in papers belonging to owners with substantial TV holdings mentioned at least one positive consequence. This same pattern is also evident in the aggregate number of positive consequences mentioned in the three sets of articles (p ≤ .01). For example, more than 33% of the articles from newspapers belonging to owners with substantial TV stations mention two or more positive consequences of the decision; however, only 20% of the articles published by newspapers with parent companies with no TV stations mention at least this many positive conse- quences. Thus, across a range of measures, newspapers owned by parent companies that benefited from the increase in political spending precipitated by the ruling were significantly and substantially more likely to reference positive consequences in their coverage of the Citizens United decision than companies that benefited less from the ruling.

Turning now to mentions of negative consequences, the Kendall tau-b tests of these differentials do not reach statistical significance for any of the nine potential negative consequences tested. However, the logistic regressions do reveal significant differences for mentions of two potential negative consequences of the decision (i.e., Citizens United will

Table 3 OLS regressions of summary measures of tone of Citizens United articles

Coefficient SE p-value 95% Confidence Interval

First Model Number of TV Stations .02 .006 .01*** .01–.03 Intercept –.98 .11 .01 –1.2–.78 N: 439 Adj. R-squared: .03 Second Model Number of TV stations .005 .001 .01*** .003–.007 Intercept .32 .02 .01 .27–.36 N: 344 Adj. R-squared: .05

Notes. Dependent variable in first model is the aggregate number of positive consequences mentioned in an article minus aggregate number of negative consequences mentioned in that article (mean = –.73, SD = 1.94). Dependent variable in second model is the percentage of all consequences mentioned in an article that are positive (mean = .38, SD = .36). This test only includes articles that mention any consequences at all (i.e., it excludes articles with a zero in the denominator as a result of not mentioning any consequences, positive or negative, at all). For all tables, *** indicates p ≤.01; ** indicates p ≤.05; * indicates p ≤.10.

Corporate Ownership and News Bias Revisited 593

create unlimited corporate influence and/or drown out voices of average citizens, and it will eliminate financial disclosures that are positive for the political process). And, these differentials also travel in the direction that corporate financial interests would predict: fewer negative consequences were mentioned in newspapers whose parent corporations own more TV stations.

In addition, tests of summary measures of the presence of negativity in these articles reveal that the number of articles that mention any negative consequences at all is also negatively correlated with the number of TV stations owned by the parent corporation of the newspaper (p ≤ .06), as is the number of articles that mention at least two or more negative consequences (p ≤ .1). This lends tentative support to the tendency for news- papers belonging to corporations who benefited more from the decision to feature less negative consequences in their coverage compared to newspapers whose parent company did not financially benefit from the decision.

In summary, the finding that there are moderate differences across the three sets of newspapers in terms of negative consequences of the Citizens United decision mentioned in their articles and several large and significant differences in the number of positive consequences mentioned confirms that the coverage produced by these three sets of newspapers painted distinctly different pictures of the likely consequences of this ruling. In addition, these different pictures vary according to the degree to which the parent companies of these newspapers profited from the ruling. Corporations with more television stations, which equates to more airtime to sell to political advertisers, were significantly more likely to mention positive consequences associated with the Citizens United ruling compared to companies with fewer or no television stations, and they were also somewhat less likely to mention negative consequences relative to their counterparts who stood to gain less financially from the ruling. Thus, as Gilens and Hertzman (2000) stipulate in their own analysis, although there remains no objective standard of unbiased coverage with which to compare these findings, what is “clear is that papers’ coverage of this issue was strongly related to the ownership interests of their corporate parents” (p. 380).

Multivariate Analyses

Following the example of Gilens and Hertzman, this section employs OLS regressions to determine whether there may be certain confounding factors that better explain the variation in news coverage that was uncovered in the previous set of tests. In other words, it may be the case that there are other factors correlated with TV station ownership, which actually drive the apparent relationship between owners’ financial interests and the content of the news coverage of Citizens United. In their analysis, Gilens and Hertzman shift the unit of analysis to the level of the newspaper and employ multivariate analyses to test two potential alternative explanations: weekly circulation (Alliance for Audited Media, 2010) and the political slant of the newspaper. I replicate this here, and I also include an additional control: the amount of independent campaign expenditures in the 2010 election in the state where the newspaper is located. Last, I further test the robustness of the findings by using an alternative explanatory variable: the proportion of the parent companies’ total revenue that is generated by television (Johnson, 2010).

Weekly circulation, as Gilens and Hertzman explain, is a proxy for the size of the newspapers’ “news holes.” Higher circulation equates to higher revenues from subscrip- tions and advertising, which increases the amount of space (i.e., number of inches)

594 Catie Snow Bailard

available for news content. Thus, more space to fill could provide more opportunity for news coverage to mention negative or positive consequences of the Citizens United ruling.

In their analysis, Gilens and Hertzman also control for the political slant of the newspaper by means of presidential endorsements, which they argue is a simple proxy for whether a news outlet favors a more liberal or conservative tilt, and which is likely correlated with their stance on the Citizens United ruling. While I do run a regression in which I control for 2008 presidential endorsements (and find commensurate results), I include here instead a more robust measure of the political slant of a newspaper. This measure was generated by Gentzkow and Shapiro (2010), and “measures the frequency with which newspapers use language that would tend to sway readers to the right or to the left on political issues” (p. 36).

I also control for the amount of independent expenditures spent in the state where each newspaper is located in the 2010 election. This is a necessary addition, since it is also plausible that the coverage produced by newsrooms may have been influenced by the amount of political spending that occurred in their state.4 After all, the visibility of the ruling’s impact and its potential consequences likely hinged on the degree to which the local airwaves and electoral experience was more tangibly affected by the increase in spending brought by the Citizens United decision.

The dependent variable in this multivariate analysis is the differential between the average number of positive consequences minus the average number of negative con- sequences mentioned in the articles published by each newspaper. This means that a positive value indicates that, on average, there was that number more positive than negative consequences mentioned in that paper’s coverage of the decision. A zero would indicate the same number of positive as negative consequences mentioned on average. And, a negative value denotes more negative than positive consequences men- tioned on average in the stories produced by each paper.

This regression reveals a significant and sizable effect of substantial TVownership on the average difference between the positive and negative consequences mentioned in each newspaper’s stories about the Citizens United decision. (Please see Table 4.) Relative to newspapers owned by companies with no TV stations, newspapers belonging to corpora- tions with a large number of TV stations were much more likely to publish stories that, on average, mentioned more positive than negative consequences of the decision (p ≤ .04). It is worth noting, however, that the differential for corporations with limited TV stations fails to achieve significance and does not follow the trajectory that skewed coverage would predict.

The next model utilizes the percentage of the parent corporation’s revenue derived from television as an alternative explanatory variable, since the degree to which media companies benefited from the Citizens United ruling also hinges on how dependent that station is on television revenue. Moreover, considering revenue as an alternative explanatory variable also enables consideration of the effect of revenue from cable service provision on news outlets’ output. This is an important addition, since cable service providers also directly benefited from the increase in political spending. “Cable companies such as Comcast are courting campaign dollars by promising even more precisely aimed ad placement than is possible on local TV. As a result cable operators have quadrupled political advertising revenues since 2004, to $467 m, and doubled their share of the pot” (The Economist, 2012). Accordingly, this model tests the effect of percentage of revenue derived from television—both through TV station ownership as well as from cable service provision—on the average differential in the positive to negative consequences mentioned in the articles published by each paper. This analysis reveals that having a higher percentage of the parent companies’

Corporate Ownership and News Bias Revisited 595

revenue derived from television significantly increases the average number of positive consequences relative to negative consequences mentioned in the stories published by a particular newspaper regarding the Citizens United ruling (p ≤ .06). (Please see Table 4.)

Hobby Lobby: An Additional Analysis

Despite the results of the multivariate regressions—which control for the effects of size of circulation, ideological slant, and the amount of independent expenditures in a news- paper’s home state—there remain reasonable omitted-variable bias concerns. Essentially, rather than financial interests, is there another factor that is actually driving this slant in coverage, for which this analysis fails to account?

The most effective way to address this concern is to conduct an additional analysis of the news coverage of a comparable court case across this same set of newspapers. Ideally, this case should share meaningful similarities with the Citizens United case; however, it

Table 4 Results of multivariate OLS regressions

Coefficient SE p-

value 95% Confidence

Interval

First Model Limited TV Stations –.21 .68 .76 –1.62–1.2 Substantial TV Stations 1.45 .68 .04** .05–2.84 Political Slant of Newspaper 3.82 7.32 .61 –11.32–18.97 Average Weekday Circulation (ten

thousands) –.0008 .005 .87 –.01–.008

2010 Election Independent Expenditures (millions)

–.02 .01 .19 –.05–.01

Intercept –2.14 3.54 .55 –9.47–5.19 N: 29 Adj. R-squared: .06 Second Model % of Revenue Derived From Television 2.92 1.5 .06* –.18–6.04 Political Slant of Newspaper 4.45 6.87 .52 –9.7–18.6 Average Weekday Circulation –.00006 .005 .99 –.01–.01 2010 Election Independent Expenditures

(State) –.01 .01 .29 –.04 to .01a

Intercept –2.57 3.26 .44 –9.31–4.16 N: 29 Adj. R-squared: .05

Notes. Dependent variable is the differential between the average number of positive consequences minus the average number of negative consequences mentioned in that newspaper’s articles (mean = –.74, SD = 1.23). Four newspapers were excluded from these regressions because three of the newspapers did not have political slant scores calculated by Gentzkow and Shapiro (2010), and one did not have any publicly available information for average weekday circulation in 2010. For all tables, *** indicates p ≤.01; ** indicates p ≤.05; * indicates p ≤.10.

aThe graduate student coders independently coded 75 of the total articles in order to assess inter- coder reliability (ICR). The coders achieved an average agreement of 88% for these measures. For a detailed list of the ICR scores of each code, please refer to the Supplemental Material online.

596 Catie Snow Bailard

should differ from the Citizens United case in that the decision has no bearing on the financial interests of the corporate owners. Burwell v. Hobby Lobby Stores, Inc. (in which the Supreme Court ruled that closely held private companies are exempt from a law if its owners religiously object to it and if there is a less restrictive means of furthering the law’s interest) meets these parameters. Both cases are high-profile and controversial, both occurred within a similar historic time frame, and both deal with corporate personhood and prominent civil liberties issues (i.e., freedom of speech and freedom of religion, respectively).

However, unlike Citizens United, the Hobby Lobby decision does not have clear financial implications for the corporate owners. Thus, if there is another factor driving the content of coverage, which this analysis has failed to account for, we should expect to see similar slants in the news coverage of these two decisions produced by the newspapers considered in this analysis. If this additional analysis fails to reveal a similar slant, however, this assuages omitted-variable bias concerns and lends convincing support to the conclusion that financial interests did, indeed, slant the coverage of the Citizens United case.

To conduct this analysis, I replicated the same steps as those of the Citizens United content analysis. First, I compiled all articles referencing the Hobby Lobby ruling from these newspapers via LexisNexis and ProQuest searches of the phrase “Hobby Lobby” between the dates of March 1, 2013, and February 28, 2015.5 This search produced 489 articles that mentioned the Supreme Court case directly. (Editorials and opinion pieces were again excluded from the analysis.) Next, two graduate student coders read a sample of articles referencing the Hobby Lobby ruling from newspapers that were not included in this analysis in order to identify the positive and negative implications of the ruling mentioned in the articles, which are listed in Tables 5 and 6.6 After establishing inter- coder reliability, the coders (who were blind to the parent corporation of the newspapers) independently coded roughly half of total articles constituting this analysis.7 The results of the analysis reveal markedly different content of coverage relative to that of the Citizens United ruling. It is true that newspapers belonging to corporations with more television stations do mention three of the six positive consequences significantly more often than newspapers belonging to corporations that do not own any stations (which accords with the findings of the Citizens United analysis). Unlike the Citizens United analysis, articles published about Hobby Lobby by corporations that own more stations were also signifi- cantly more likely to report three of the six negative consequences. In other words, articles published by newspapers with parent corporations that owned more TV stations were more likely to report both positive and negative consequences than the other newspapers included in this analysis. This is supported by an additional test confirming that these newspapers were also more likely to publish articles that mention any consequences at all —for both positive and negative consequences. (Please see Tables 5 and 6.)

An additional analysis of the balance of positive to negative consequences mentioned in the articles further highlights differences in the content of news coverage of Hobby Lobby versus Citizens United. Whereas tests of the Citizens United coverage revealed a strong positive correlation between the number of TV stations a corporation owns and the number of positive consequences relative to negative consequences mentioned in the articles published by the newspapers belonging to those corporations, this trend is not evident in these same newspapers’ coverage of Hobby Lobby. Moreover, a replication of the multivariate analyses, controlling for ideological slant, independent expenditures, and circulation—as well as a second test with the alternative independent variable—also fail to reveal the same pattern in Hobby Lobby coverage that was evident in the analysis of the Citizens United decision. (Please see Table 7.)

Corporate Ownership and News Bias Revisited 597

In summary, whereas there is a clear trend in coverage across these newspapers when reporting on the positive and negative consequences of Citizens United, which correlates with the financial interests of the newspapers’ corporate owners, a similar trend is not evident in these same newspapers’ coverage of the Hobby Lobby decision. This assuages omitted-variable bias concerns, and provides compelling support for the capacity for the financial interests of media corporations to shape the content of the news published by their outlets.

Table 5 Positive consequences of Hobby Lobby ruling mentioned in newspaper articles.

Hobby Lobby No TV Stations

Limited TV

Stations

Substantial TV

Stations

Tau-b (Asymptotic-

SE) Significancea

Logistic Coefficient

(SE) Significanceb

Protects religious freedom of business owners

49% 57% 59% .09 1.01 (.04) (.006)

p ≤ .04** p ≤ .06* Rightfully recognizes

corporations as people 8% 15% 11% .05 1.003

(.04) (.01) – p ≤ .76

Will not hurt employees’ well-being because Hobby Lobby treats it employees well

3% 6% 3% –.003 1.004 (.04) (.02) – p ≤ .8

Protects owners from having have to choose between their business and religion

46% 63% 54% .08 1.004 (.04) (.006)

p ≤ .07* p ≤ .47

Is a narrow ruling, so it will not lead to a slippery slope

22% 24% 29% .07 1.01 (.04) (.007)

p ≤ .09* p ≤ .05** Is not a war on women 4% 3% 6% .05 1.02

(.05) (.01) – p ≤ .18

Articles that mention any positive consequences

62% 80% 76% .14 1.02 (.04) (.007)

p ≤ .01*** p ≤ .01*** Articles that mention 2 or

more positive consequences

42% 54% 50% .08 1.009 (.04) (.006)

p ≤ .08* p ≤ .14

aThe codebook also included an “other” category in order to identify any other positive or negative consequences that were mentioned in the news coverage. However, this “other” category was only selected for either positive or negative mentions in less than 1% of the cases. Thus, it can be reasonably concluded that the list of positive and negative consequences included in the codebook is reasonably comprehensive of the consequences mentioned in news coverage. All tau-b test p-values are reported as two-tailed tests. bAll logistic results in this table are derived from bivariate regres- sions with the number of TV stations owned by parent corporations as the discrete independent variable (with all results reported in odds ratio).

598 Catie Snow Bailard

Conclusion

The unprecedented deluge in political spending precipitated by the Citizens United ruling provides a unique and important opportunity to test for the degree to which media corpora- tions’ financial interests may seep into the news produced by their outlets. Since the majority of the increase in political spending went to political advertising on TV, media corporations that owned more TV stations with which to sell airtime to advertisers benefited more

Table 6 Negative consequences of Hobby Lobby ruling mentioned in newspaper coverage

Hobby Lobby No TV Stations

Limited TV

Stations

Substantial TV

Stations

Tau-b (Asymptotic-

SE) Significancea

Logistic Coefficient

(SE) Significanceb

Is part of a larger war on women

39% 36% 43% .03 .005 (.04) (.001)

– p ≤ .01*** Enables owners to impose

their religious beliefs on their employees

31% 37% 36% .05 .004 (.04) (.001)

– p ≤ .01*** Is a slippery slope that will

open the door to other troubling lawsuits and rulings

21% 35% 36% .15 .0007 (.04) (.001)

p ≤ .01*** p ≤ .57

Is flawed because corporations do not have religious rights

9% 19% 20% .14 .0006 (.04) (.0005)

p ≤ .01*** p ≤ .29 Treats corporations as

people, which is not valid

4% 8% 8% .07 .003 (.04) (.001)

p ≤ .09* p ≤ .01*** Will be used

disingenuously by owners to cut their costs

2% 2% 2% .007 .002 (.04) (.0004)

– p ≤ .01*** Articles that mention any

negative consequences 53% 62% 68% .14 .005

(.04) (.001) p ≤ .01*** p ≤ .01***

Articles that mention 2 or more negative consequences

33% 38% 39% .05 1.007 (.04) (.006) – p ≤ .26

aThe codebook also included an “other” category in order to identify any other positive or negative consequences that were mentioned in the news coverage. However, this “other” category was only selected for either positive or negative mentions in less than 1% of the cases. Thus, it can be reasonably concluded that the list of positive and negative consequences included in the codebook is reasonably comprehensive of the consequences mentioned in news coverage. All tau-b test p-values are reported as two-tailed tests. bAll logistic results in this table are derived from bivariate regres- sions with the number of TV stations owned by parent corporations as the discrete independent variable (with all results reported in odds ratio).

Corporate Ownership and News Bias Revisited 599

immediately and substantially from the ruling compared to media companies with limited or no TV stations. Being that it is a rare occurrence when the small number of media corpora- tions’ interests vary from one another in a clear, measurable, and systematic manner, this case lends itself nicely to the method utilized by Gilens and Hertzman (2000) in their analysis of newspaper coverage of the 1996 Telecommunications Act.

Before concluding, however, it is worth revisiting that this approach does not make any sort of claim to test for “bias” by comparing the coverage as it exists to some sort of impossible and unknown standard of what “unbiased” coverage would look like. As stated at the outset, this sort of test is not feasible. Instead, “bias” in this case simply means that newspapers covered the decision differently, and these differences reflected the degree to which their parent corporations benefitted financially from the ruling. Thus, there is no determination that one group of newspapers were necessarily more or less objective than another in any sort of abstract sense; but, rather that the coverage produced by these

Table 7 Results of multivariate OLS regressions for Hobby Lobby

Coefficient SE p-

value 95% Confidence

Interval

First Model Limited TV Stations .14 .2 .49 –.28 to–.57 Substantial TV Stations –.04 .2 .84 –.45–.37 Political Slant of Newspaper 3.47 2.12 .12 –.94–7.88 Average Weekday Circulation

(thousands) .0002 .0002 .31 –.0002–.0005

2010 Election Independent Expenditures (millions)

.001 .004 .77 –.007–.009

Intercept –1.51 1.007 .15 –3.6–.57 N: 27Adj. R-squared: –.03 Second Model

% of Revenue Derived From Television

.37 .53 .48 –.71–1.47

Political Slant of Newspaper 3.07 1.95 .13 –.98–7.11 Average Weekday Circulation

(thousands) .0001 .0001 .56 –9.55x10a to 9.43x10a

2010 Election Independent Expenditures (millions)

–1.42x10b 1.3x10b .29 –.009–.008

Intercept –1.27 .91 .18 –3.16–.62 N: 27Adj. R-squared: .008

Note. Dependent variable is the differential between the average number of positive consequences minus the average number of negative consequences mentioned in that newspaper’s articles (mean = .13, SD = .36).

aThe graduate student coders independently coded 100 of the total articles in order to assess inter- coder reliability (ICR). The coders achieved an average agreement of 90% for these measures. For a detailed list of the ICR scores of each code, please refer to the Supplemental Material online. bAll tau-b test p-values are reported as two-tailed tests.

600 Catie Snow Bailard

newspapers varied systematically and did so in line with the financial interests of their parent corporations.

The tests revealed that newspapers belonging to companies with substantial TV holdings covered the ruling significantly more favorably on balance than newspapers belonging to companies with limited or no TV holdings. And, these findings maintain even after controlling for a number of potentially confounding factors and also substituting the percentage of total revenue derived from TV sources (including both owning TV stations and cable service provision) as an alternative explanatory variable. An additional analysis of these newspapers’ coverage of the Hobby Lobby decision does not reveal a similar slant, assuaging omitted-variable bias concerns.

Another point worth revisiting before concluding is the finding that, whereas the coverage of the Citizens United decision meaningfully varied in terms of the number of positive arguments mentioned (across a broad range of individual and aggregate mea- sures), the number of negative mentions varied only marginally. So, what might potentially explain why we see a larger difference in positive mentions than in negative mentions? One answer could be that the presence or absence of positive arguments is less conspic- uous than negative arguments, which is supported by research on the effects of political advertising that found that negative advertisements tend to be more likely to be recalled than positive advertisements (Shapiro & Rieger, 1992). This suggests that varying the number of positive versus negative arguments might represent a more subtle approach to shaping the tone of content, which would be preferable to news outlets that want to protect their image of objectivity.

Another potential answer to this question is that newspapers belonging to corporations with more television stations have better access to resources that enable them to tell more complex, nuanced stories, which results in more mentions of positive consequences. However, if this were the case, we should also expect to see a similar distribution of negative and positive mentions in the Hobby Lobby stories, which the analysis did not. A more plausible explanation in line with economic theories of news production might be that the negative consequences of the Citizens United decision are somewhat more widely known and broadly accepted; however, media corporations with many TV stations, that have long enjoyed profitable relationships with political advertisers, may have more awareness of and more incentive to report the potentially positive ramifications of the decision as a direct result of those relationships, shaping their news coverage accordingly.

Ultimately, a conclusive answer to this question will require further research aimed at uncovering the processes, incentives, and motives that explain this finding. Nevertheless, the finding of this study are clear: the composite picture painted by these newspapers of the implications of Citizens United for the American political system—which is a product of the relative balance of both negative and positive arguments that make up the whole— varies in relation to the degree that the corporate owner benefited from the decision.

In closing, although there are a limited number of cases that enable this sort of test for slanted news coverage, the potential for the financial interests of media corporations to influence how their news outlets cover issues is troublingly broad. Consider, for example, the slew of legislation and policies, both domestic and foreign, which influence market structures and trade policies with clear implications for large corporations. Moreover, the fact that the majority of American media companies belong to even larger conglomerations with holdings that extend far beyond the media sphere gives additional reason for pause.

Whatever the full range of consequences of Citizens United proves to be for the future of the American democracy, this ruling provides important insight into the possibility for the financial interests of media corporations to slant the news content produced by their

Corporate Ownership and News Bias Revisited 601

outlets. This has important implications for the degree to which the news produced by increasingly conglomerated and corporatized media companies may eschew neutral or balanced coverage in favor of news frames that promote their own financial interests—a troubling prospect for those who believe the news has a duty to provide citizens with a balanced and objective perspective on the issues and policies facing the nation. This also has troubling implications for the quality and diversity of perspectives the American citizens receive from their news media more generally. Thus, echoing Gilens and Hertzman’s (2000) concluding words, if newspaper coverage of the Citizens United ruling examined here “is indicative of a general tendency within the news media, we have much to be concerned about” (p. 384).

Acknowledgments

I would like to thank the editors and reviewers for their insightful suggestions, which improved this manuscript markedly. I would also like to thank my terrific research assistants for their diligent work on this project: Garrett Estrada, Todd Kominiak, Molly Pfaffenroth, and Nathan Wolfson.

Supplemental Material

Supplemental data for this article can be accessed on the publisher’s web site: http://www. dx.doi.org/ 10.1080/10584609.2016.1142489

Notes

1. For the purpose of this analysis, I increased the thresholds for the latter two sets of companies relative to those used by Gilens and Hertzman. This is because, due to the very loosening of television ownership restrictions that were the focus of Gilens and Hertzman’s analysis, the number of TV stations that can and do belong to any single media corporation has increased markedly since 1996 (Bagdikian, 2004). Thus, in the present analysis limited ownership demarcates companies that own between 1 and 18 television stations, and substantial television ownership is comprised of corporations that own 19 or more stations (Columbia Journalism Review, 2013).

2. The codebook also included an “other” category in order to identify any other positive or negative consequences that were mentioned in the news coverage. However, this “other” category was only selected for either positive or negative mentions in less than 1% of the cases. Thus, it can be reasonably concluded that the list of positive and negative consequences included in the codebook is reasonably comprehensive of the consequences mentioned in news coverage.

3. The graduate student coders independently coded 75 of the total articles in order to assess inter-coder reliability (ICR). The coders achieved an average agreement of 88% for these measures. For a detailed list of the ICR scores of each code, please refer to the Supplemental Material online.

4. The statistics for the amount of independent expenditures in the 2010 election for each state are provided by the Sunlight Foundation and can be accessed at http://reporting.sunlightfoundation. com/outside-spending-2012/states/.

5. Since there were several instances of news stories that were not at all about the case but instead simply about a Hobby Lobby store opening in a particular neighborhood, it was necessary to employ additional search terms to narrow the results. Thus, the search terms used to identify relative articles were “Hobby Lobby” AND “case” OR “decision” OR “verdict” OR “court” OR “ruling.”

6. The codebook also included an “other” category in order to identify any other positive or negative consequences that were mentioned in the news coverage. However, this “other” category was only selected for either positive or negative mentions in less than 1% of the cases.

602 Catie Snow Bailard

7. The graduate student coders independently coded 100 of the total articles in order to assess inter-coder reliability. The coders achieved an average agreement of 90% for these measures. For a detailed list of the ICR scores of each code, please refer to the Supplemental Material online.

References

Alliance for Audited Media: The New Audit Bureau of Circulation. (2010). Total circ. Retrieved from http://auditedmedia.com/news/research-and-data/

Bagdikian, B. (1997). The media monopoly. Boston, MA: Beacon Press. Bagdikian, B. H. (2004). The new media monopoly: A completely revised and updated edition with

seven new chapters. Boston, MA: Beacon Press. Baker, C. E. (2001). Media, markets, and democracy. Cambridge, UK: Cambridge University Press. Baum, M. A., & Groeling, T. (2008). New media and the polarization of American political

discourse. Political Communication, 25(4), 345–365. doi:10.1080/10584600802426965 Bennett, W. L. (1988). News: The politics of illusion. New York, NY: Longman. Bozell, L. B., & Baker, B. H. (1990). And that’s the way it isn’t. Alexandria, VA: Media Research

Center. Branton, R. P., & Dunaway, J. (2009). Slanted newspaper coverage of immigration: The importance

of economics and geography. Policy Studies Journal, 37(2), 257–273. doi:10.1111/psj.2009.37. issue-2

Burriss, L. L., & Williams, J. P. (1979). Use of network news material by cross-owned newspapers. Journalism Quarterly, 56(Autumn), 567–571.

Center for Responsive Politics. (2012). Outside spending. Retrieved from http://www.opensecrets. org/outsidespending/fes_summ.php

Cohen, J., & Solomon, N. (1993). Adventures in medialand: Behind the news, beyond the pundits. Monroe, ME: Common Courage Press.

Columbia Journalism Review. (2013). Who owns what. Retrieved from http://www.cjr.org/resources/ D’Alessio, D., & Allen, M. (2000). Media bias in presidential elections: A meta‐analysis. Journal of

Communication, 50(4), 133–156. doi:10.1111/j.1460-2466.2000.tb02866.x Dexheimer, E. (2012, June 5). Political advertising: Influencing business, not just politics. Medill

Reports. Retrieved from http://newsarchive.medill.northwestern.edu/chicago/news-206534.html Dispensa, J. M., & Brulle, R. J. (2003). Media’s social construction of environmental issues: Focus

on global warming–A comparative study. International Journal of Sociology and Social Policy, 23(10), 74–105. doi:10.1108/01443330310790327

Dooling, W. (2012, December 4). Where did all those super PAC dollars go? 1/3 of all outside money moved through handful of media firms. The Center for Media and Democracy’s PR Watch. Retrieved from http://www.prwatch.org/news/2012/12/11868/where-did-all-those-super- pac-dollars-go-13-all-outside-money-moved-through-handf

Dunaway, J. (2008). Markets, ownership, and the quality of campaign news coverage. Journal of Politics, 70(4), 1193–1202. doi:10.1017/S0022381608081140

The Economist. (2012, Sep 8). Of mud and money: The only sure winners of November’s elections are media firms. Retrieved from http://www.economist.com/node/21562211

Entman, R. M. (2007). Framing bias: Media in the distribution of power. Journal of Communication, 57(1), 163–173. doi:10.1111/jcom.2007.57.issue-1

Epstein, E. J. (1974). News from nowhere: Television and the news. New York, NY: Vintage Books. Fico, F., & Cote, W. (1999). Fairness and balance in the structural characteristics of newspaper

stories on the 1996 presidential election. Journalism & Mass Communication Quarterly, 76(1), 124–137. doi:10.1177/107769909907600109

Fico, F., Lacy, S., & Simon, T. (1989). Relationships among economic, newsroom, and content variables: A path model. Journal of Media Economics, 2(2), 51–66. doi:10.1080/ 08997768909358185

Corporate Ownership and News Bias Revisited 603

Gamson,W.A., Croteau, D., Hoynes,W., & Sasson, T. (1992).Media images and the social construction of reality. Annual Review of Sociology, 18, 373–393. doi:10.1146/annurev.so.18.080192.002105

Gentzkow, M., & Shapiro, J. M. (2010). What drives media slant? Evidence from U.S. daily newspapers. Econometrica, 78(1), 35–71. doi:10.3982/ECTA7195

Gilens, M., & Hertzman, C. (2000). Corporate ownership and news bias: Newspaper coverage of the 1996 Telecommunications Act. The Journal of Politics, 62(2), 369–386. doi:10.1111/ 0022-3816.00017

Groseclose, T., & Milyo, J. (2005). A measure of media bias. The Quarterly Journal of Economics, 120(4), 1191–1237. doi:10.1162/003355305775097542

Hamilton, J. T. (2011, October 23). All the news that’s fit to sell: How the market transforms information into news (Kindle Locations 630-631). Princeton, NJ: Princeton University Press. Kindle Edition.

Herman, E., & Chomsky, N. (2002). Manufacturing consent: The political economy of the mass media. New York, NY: Random House.

Johnson, B. (2010, September 27). 100 leading media companies. AdAge: Media News. Retrieved from http://adage.com/article/media/100-leading-media-companies-2010/146004/

Klinenberg, E. (2005). Convergence: News production in a digital age. The Annals of the American Academy of Political and Social Science, 597(1), 48–64. doi:10.1177/0002716204270346

Larcinese, V., Puglisi, R., & Snyder, J. M. (2011). Partisan bias in economic news: Evidence on the agenda-setting behavior of U.S. newspapers. Journal of Public Economics, 95(9–10), 1178– 1189. doi:10.1016/j.jpubeco.2011.04.006

Lichter, S. R., Rothman, S., & Lichter, L. S. (1986). The media elite. Bethesda, MD: Adler & Adler. Lieberman, D. (2012, November 30). Political groups spent record $3.37B on TV for 2012

campaign: Analyst. Deadline Hollywood. Retrieved from http://deadline.com/2012/11/politi- cal-advertising-tv-2012-campaign-379988/

McChesney, R. W. & Nichols, J. (2012). The bull market: Advertising. Monthly Review: An Independent Socialist Magazine, 63(11). Retrieved from http://monthlyreview.org/2012/04/01/ the-bull-market/

McManus, J. (1995). A market‐based model of news production. Communication Theory, 5(4), 301– 338. doi:10.1111/comt.1995.5.issue-4

Parenti, M. (1986). Inventing reality: The politics of the mass media. New York, NY: St. Martin’s Press.

The Pew Research Center’s Project for Excellence in Journalism. (2013). The state of the news media 2013. Retrieved from http://www.stateofthemedia.org/2013/

Pratte, A., & Whiting, G. (1986). What newspaper editorials have said about deregulation of broadcasting. Journalism Quarterly, 61(1), 56–65.

Samples, J., & Shapiro, I. (2010, January 21). Free speech for all. DC Examiner. Retrieved from http://www.cato.org/publications/commentary/free-speech-all

Shapiro, M. A., & Rieger, R. H. (1992). Comparing positive and negative political advertising on radio. Journalism & Mass Communication Quarterly, 69(1), 135–145. doi:10.1177/ 107769909206900111

Snider, J. H., & Page, B. I. (1997). Does media ownership affect media stands? The case of the Telecommunications Act of 1996. (No. 97–12). Institute for Policy Research at Northwestern University, Evanston, IL.

Soderlund, W. C., & Schmitt, C. (1986). El Salvador’s civil war as seen in North and South American press. Journalism & Mass Communication Quarterly, 63(2), 268–274. doi:10.1177/ 107769908606300203

Young, L. (2012, December 17). Outside spenders’ return on investment. Sunlight Foundation Reporting Group. Retrieved from http://reporting.sunlightfoundation.com/2012/ returnon_investment/

Zaller, J. (1999). Market competition and news quality. In Annual Meetings of the American Political Science Association, September, Atlanta, GA.

604 Catie Snow Bailard

Copyright of Political Communication is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.