ISSN: 1550-7521
ISSN: 1550-7521
John A. Hatcher
University of Minnesota, Duluth USA
Visit for more related articles at Global Media Journal
A study of 528 news items from 11 countries explores how anonymous and unnamed sources were used by journalists during the buildup to the Iraq War. A quarter of all sources appearing in news items were not identified by name. The use of unnamed sources corresponded with a decrease in ideas opposing the war and a tone that presented the war as being more positive and unavoidable. The findings raise questions about whether anonymous and unnamed sources serve the perceived whistleblower function in political discourse.
Iraq war, anonymous sources, content analysis, whistleblower
The image conjured up by the term “anonymous source” may be that of Deep Throat, the unnamed and, up until his death in 2008, unknown government source who guided journalists at The Washington Post in their reporting of the Watergate scandal that eventually led to the resignation of President Richard M. Nixon. Deep Throat is portrayed by journalism practitioners as a whistleblower: a source who helps reporters get information that cannot be obtained in any other way, who is in obvious opposition to political leaders, and who assists the media in their role as political watchdogs (Brown, 2005).
However, a rich body of work exploring the role of media in a political system shows that, driven by ideals that place high value on credible, verifiable information, journalists favor official sources who tend to frame an issue in a way that is congruent with political leaders; these sources speak in a tone that tends to reinforce the position of government leaders rather than challenge it (Bennett, 1990; Schudson, 2003). The goal of this study is to explore the long-held assumption that journalists break from their routines when they use unnamed and anonymous sources in political reporting. It asks: Does the presence of anonymous sources in political reporting lead to news content that challenges the ideas of political leaders? Or, is there evidence that anonymous sources are being used by political actors to reinforce and even strengthen their own positions? This study explores the use of anonymous sources using a long history of media sociology research that explains the journalist-source relationship as one intimately tied to the political institution and the dominant messages of the sources that Schudson refers to as “parajournalists” (2003, p. 3).
While there has long been a debate over the use of unnamed sources, little is known about exactly how the presence of anonymous sources in news reporting affects political discourse. This study uses news coverage of the buildup to the Iraq War to explore these questions; it analyzes 528 news items published in 22 newspapers from 11 countries representing five languages. It explores whether the use of unnamed sources was more prevalent in some countries than others, whether these sources encouraged the use of certain ideas to interpret the news event and whether these sources swayed the tone of media messages.
On Sept. 4, 2002, then-U.S. President George W. Bush announced that Iraqi leader Saddam Hussein was a threat to world security because he possessed weapons of mass destruction (“Moves towards U.S. military action,” 2002). He urged the United States and other countries to join together in a coalition that would remove Hussein from power. Bush’s announcement triggered a global discourse over the question of whether to invade Iraq. During the seven months prior to the invasion, the world’s media were fixated firmly on this global debate, offering the ideal opportunity to explore just how unnamed and anonymous sources shaped political discourse.
Hachten (1999) says that, like never before, global news events are instantaneously observed by media who, with the aid of technology, make information immediate and transnational. Hatchen believes there now exists one global style of journalism that is especially apparent in the coverage of wars and major disasters. Technology also makes global messages about such events homogenous as news channels no longer respect boundaries of culture or state (Hallin & Mancini 2004). Aiding technology, they assert, is a global Americanized media, concentrated in the hands of a few (Humphreys, 1996).
This study defines the unnamed or anonymous source as any source appearing in a news item who is not identified by name. To date, the body of research looking at the use of anonymous sources can be put into one of two categories: (1) how often unnamed sources are used in reporting; (2) whether readers question the credibility of news stories that include unnamed sources. This study examines the use of such sources from the perspective of media sociology research and particularly Bennett’s concept of media indexing (1990) and Schudson’s notion of the parajournalist (2003). The goal is to explore whether a whistleblower role can be detected in source use and to study this relationship in a comparative analysis of a global news event.
Research regarding the use of sources in the coverage of major news events finds that the unnamed sources used by journalists would likely be expert and official sources (Tuchman, 1978; Lawrence, 2000; Salomone, Greenberg, Sandman & Sachsman, 1990; Durham, 1998). In general, journalists depend on sources to help them make decisions about what’s news, how to think about (frame) that news and, indirectly, to determine the tone of the message. Lawrence (2000) notes that the official sources reporters rely on for information often have great ability to manage the news. Many journalists see it as their professional obligation “to gather information provided by legitimate sources and to convey that information as ‘neutrally’ as possible" (Lawrence, 2000, p. 52). The goal is objectivity in newsgathering, but the result, Lawrence asserts, is often news that could be characterized by Iyengar’s (1991) conception of news reports that merely give an accounting of what happened without making connections outside of the news event.
Tuchman (1978) finds the desire to obtain facts viewed as verifiable pushes journalists toward sources that have official capacity. The facts presented by unofficial sources may be just as accurate, but Tuchman notes that they are harder for a reporter to substantiate. Salomone, Greenberg, Sandman and Sachsman (1990) find many of the standard elements required for a quality news article – such as accuracy and assessment of a threat – are more likely to come from a traditional news source or expert. Journalists also rely on official sources to frame a story – to put it into context and give an event meaning (Durham, 1998).
Idealized, journalism defines events in their broadest terms: Diverse voices interpret events and show how isolated incidents connect to a broader world (Entman, 1993; Iyengar, 1991; Neuman, Just & Crigler, 1992). However, journalists work in a dynamic environment in which factors both inside and outside the newsroom weigh daily on the decisions they make regarding what stories to write and, ultimately, how to write them (Shoemaker & Reese, 1996. Entman (1989) says a journalist may yearn for work that attempts to “illuminate the powerful,” (p. 125) but falls short due to pressures from a desire to be objective, to be accountable and to produce a profitable news product. Entman believes these news practices are highly resistant to change, primarily because of the limits of audience interest and the journalists’ close relationship with political elite.
The term framing has been used in so many ways that it has almost lost its value. The concepts of framing, metaphor, idea elements, theme, focus and topic could be used interchangeably to describe the aspect of news content that has to do with the choices a journalist makes in the creation of a news story. This study develops its definition of framing from Entman (1993): “To frame is to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described” (p. 52).
Using this definition, coupled with what Tuchman (1973) observes, framing can be seen as a dynamic and evolving process, best understood by analyzing the choices a journalist makes in deciding which sources a journalist includes and what information those sources use to define a news event. Gitlin (1980, p. 6) says that frames are “principles of selection, emphasis and presentation composed of little tacit theories about what exists, what happens, and what matters.” Gamson and Modigliani (1989) say that journalists frame news events using five devices: metaphors, exemplars, catchphrases, depictions, and visual images.
Schudson (2003) describes a journalist-source relationship in which news organizations rely on expert sources to decide what events are newsworthy and what frames or ideas should be ascribed to those events. Journalists, he notes, are besieged by a vast array of “parajournalists” (p. 3) from myriad political and corporate institutions who have a powerful hand in shaping news. Journalists rely on official government sources to make sense of the world. “News is produced by people who operate, often unwittingly, within a cultural system, a reservoir of stored cultural meanings. It follows conventions of sourcing – who is a legitimate source, speaker or conveyor of information to a journalist” (p. 190).
Bennett (1990) uses the concept of indexing to describe how the boundaries of discourse are established by the political elite: “Mass media news professionals, from the boardroom to the beat, tend to ‘index’ the range of voices and viewpoints in both news and editorials according to the range of views expressed in mainstream government debate about a given topic” (p. 106). In other words, official sources decide what the news is and what ideas should be employed to interpret the news event. Indexing is most prevalent, Bennett asserts, in news events that are of great importance to the interests of the state – including military actions and foreign affairs. This routine would make it unlikely for media to include ideas that would fall outside the acceptable range of discourse.
In retrospect, media organizations have been forced to scrutinize their coverage of the buildup to the Iraq War and concede that they may have been misled by unnamed and anonymous sources. In May of 2004, the editors of The New York Times told readers they had not done an adequate job of reporting critically on the question of going to war in Iraq: “… we have found a number of instances of coverage that was not as rigorous as it should have been” (“The Times and Iraq,” 2004, May 26, para. 3). In part, the Times concluded, it erred by relying on sources and informants without verifying or questioning their claims with greater scrutiny. Indeed, the media’s questionable coverage of the buildup to the war, coupled with several high-profile scandals in the United States, have prompted news organizations to reexamine how anonymous and unnamed sources are used in news reporting. Many news organizations have responded by instituting more stringent policies, and recent analyses have documented a decline in the use of anonymous sources in reporting (Martin-Kratzer & Thorson, 2007). The State of the Media Report (2005) found that U.S. newspapers decreased their use of anonymous sources from 29 percent to just 7 percent in one year (more recent State of the Media reports have not revisited the use of anonymous sourcing).
Esposito (1999) notes that the use of anonymous sources is at least a century-old practice in journalism – one on which there has never been agreement. Longtime media critic Ben Bagdikian (2005) writes that one of the most famous anonymous “leaks” of information came on the eve of the attack on Pearl Harbor, when President Roosevelt released information to The New York Times that peace talks with Japanese negotiators were not progressing well and that an attack on U.S. soil was a possibility. Bagdikian (2005) and others say that anonymous sources provide a crucial role in the reporting process for whistleblowers who risk their livelihoods and even their lives in the release of information. Opponents, such as USA Today founder Al Neuharth, argue that only in rare cases in history has anonymity been warranted, suggesting that media are too quick to grant anonymity, leading to a loss of both credibility and accountability (Shepard, 1994).
Most research has found that, as a practice, journalists use anonymous sources frequently. Wulfemeyer (1985) examined the news magazines Time and Newsweek, finding that 80 percent of all stories analyzed used some kind of anonymous or unnamed source and concluding that journalists were not being any less cautious or transparent in their reporting methods in spite of recent scandals. Wulfemeyer and McFadden (1986) found the use of unnamed sources varied by medium: newsmagazines used the most, television network news broadcasts used them in about half of their news stories and newspapers used them the least. Rosentiel (2005) speculated that part of the difference across media may have something to do with the way in which television news’ compressed format forces it to look for ways to condense information, eliminating specific identifiers such as the names of sources.
Wulfmeyer and McFadden (1986) conclude that if sources don’t appear in reporting then, by default, the news organization becomes the one accountable for the credibility of the information. Esposito (1999) believes the internet and a 24-hour news cycle are pressuring journalists to rely more on these kinds of sources. Anonymous and unnamed sources, he concluded, don’t allow readers to assess the credibility of a source or the information they provide.
Even within the broad definition of an unnamed source, there are distinctions that journalists use that may help readers determine the credibility of the source and the information in a news report (Adams, 1962). A source identified as being a “White House source,” for example, may have greater credibility to a reader than an ambiguous attribution. However, when the credibility of an institution is tarnished – as the White House’s was through events such as Watergate and Vietnam — Riffe (1980) found, the credibility of unnamed sources from these institutions also suffered. However, the impact of an unnamed source is not uniform. Esposito (1999) speculated that these kinds of sources may be more prevalent in high-profile stories. Further, the level of controversy connected to a news event also may influence how closely a reader scrutinizes the credibility of a source (Fedler & Counts, 1981).
While the use of unnamed sources may be an important ideological discussion for journalists, some research suggests readers don’t object to or even notice the use of unnamed sources (Policinski, 2005). News organizations report that they rarely receive complaints from readers about the use of anonymous sources (Wilson, Babcock & Pribek, 1997). Rains (2007) finds that readers looking at information on health information Web sites gave anonymous sources the same level of credibility as named sources, and concluded that readers do not use “critical information seeking” (p. 209) skills when accessing information.
The Iraq War, one of the world’s most-covered media events (Tumber & Palmer, 2004), drew more than 3,000 journalists to produce news that was instantaneous and global. Coverage of a conflict of this magnitude encouraged journalists to seek out official sources who presented the war as a foregone conclusion, ignoring metaphors that discussed attempts to find alternatives to the war, the United Nations refusal to support an invasion as well as peaceful protests of the war (Lule, 2004). The result, Lule concluded, was that media failed to analyze the debate or offer any historical context.
Analysis of Iraq War media coverage shows journalists took their cues from their sources. Pfau, Haigh, Gettle, Donnelly, Scott, Warr and Wittenberg (2004) looked at media coverage from reporters who were embedded with troops and found that, as might be expected, this type of coverage is more favorable in tone toward individuals and the military. However, this kind of coverage was not more positive overall. Aday, Livingston and Hebert (2005) compared U.S. media coverage with the Qatar-based Al-Jazeera television station and concluded that cultural and ideological differences shape how objectively a news event is covered by media.
In general, they find, media cover the “whiz-bang” (p. 16) aspects of the war, but miss many other important story lines.
The theoretical model for this study, shown in Figure 1, is built on Schudson’s (2003) and Bennett’s (1990) descriptions of how political actors dictate the ideas used and establish the tone of the message in public discourse. One research question and four hypotheses are tested.
A source is any person or document to which information is attributed. An unnamed or anonymous source is any person or document to which information is attributed but who is not specifically identified by name.
A news frame describes the ideas used to interpret an event. They are thought of in this study as belonging to one of two broader categories based on whether they support or oppose the dominant message of the political elite – in this case, the White House view of the threat of Iraq.
The tone of a news item is defined as the series of attributes used to describe how strongly a news item presents information in a way that encourages a particular attitude about a news event.
RQ1: Did some countries use more unnamed sources than others in their coverage of the buildup to the Iraq War?
This question, the only one that looks at source use at the country level, explores whether some countries may have had a greater stake in the news event and therefore relied more on unnamed and anonymous sources for analysis.
H1: The more unnamed sources in a news item, the more likely the item will be given greater prominence.
Prominence is how much importance a news item is given based on its placement in a newspaper. Since journalists profess to use unnamed sources only when accessing important information that cannot be obtained in any other way, it is expected that news items including these kinds of sources would be deemed to have greater newsworthiness.
H2: The more unnamed sources in a news item,
a. the more organizational sources.
b. the fewer unaffiliated sources.
It’s expected that news items containing information attributed to unnamed and anonymous sources would require further substantiation. As such, organizational sources, such as government sources and military sources, would be used to verify information supplied by unnamed sources.
H3: The more unnamed sources in a news item,
a. the greater the use of pro-war ideas.
b. the less the use of ideas that oppose war.
It’s expected that many unnamed and anonymous sources are presenting information that is being leaked by government leaders – in this news event, the agenda of U.S. officials. As such, these sources are expected to champion the ideas that leaders are espousing as the correct choice. This prediction is contrary to the notion of the unnamed source serving as a whistleblower, suggesting these sources are reinforcing the ideas of political actors.
H4: The greater the use of unnamed sources in a news item,
a. the less neutral the tone of the news item.
b. the more the overall tone of a news item presents going to war as a good idea.
c. the more the idea of going to war is seen as being unavoidable.
Unnamed sources are predicted to champion the ideas of government officials making an argument for going to war. As such, the ideas will be presented in a way that will favor the option of taking military action against Iraq.
The event timeframe from which news items were collected started on Sept. 4, 2002, when U.S. President George Bush pledged to launch an intensive round of diplomacy with world leaders and the United Nations “to deal with the threat” of Iraqi leader Saddam Hussein (“Moves towards U.S. military action,” 2002). It concluded on March 19, 2003, when Bush declared war on Iraq ("Iraq Crisis, 2002-2003," n.d.). During these seven months, countries around the world debated whether to join or even support the United States in an invasion of Iraq. The entire world was embroiled in a passionate policy debate that forced governments to act, journalists to report and the public to respond.
A number of factors were considered when selecting the countries to include in this analysis: the Netherlands, India, Canada, Israel, Australia, New Zealand, the United States, the United Kingdom, Japan, France, and Spain. Each of these countries was faced with the same decision of whether to join the coalition that would invade Iraq. However, these countries also offer variation in culture, geography and, most importantly, their public opinion about the war and, ultimately, in whether they joined the U.S. invasion.
Two newspapers were chosen from each country. All of the newspapers, shown in Table 1, identified themselves as being general-interest newspapers. The newspapers selected for this study were the largest circulation newspapers from which data could be found for the identified news event using the Lexis-Nexis database. While it would be desirable to sample from even more newspapers from each country, the market penetration of these newspapers combined in each country can be said to be representative of the news content in each of these countries. All of these newspapers identified themselves as being general-interest newspapers, rather than specialty, entertainment or sports-oriented magazines. The French, Dutch and Spanish language newspapers were published in their native languages, which led to some challenges that were addressed in the area of intercoder reliability.
Table 1: Newspapers used in content analysis.
News items were collected from each newspaper using a keyword search (“Iraq” or the equivalent in the language in which the newspaper was published) in the Lexis-Nexis database. The search term is rather broad, but it was deemed more important to cull stories not related to the conflict, than it was to risk missing some stories related to the conflict that might not have been found if the search criteria were more narrow. Since the goal was to understand how the issue of Iraq was portrayed in public discourse, only news items that were clearly not about Iraq in any way were excluded.
Editorials, letters to the editor, columns and other opinion pieces were included. While journalists do not necessarily produce these items, they are still part of the content included by the news organization and are, therefore, subject to the same gatekeeping routines as other elements in a news product.
A systematic, random sample of news items from each newspaper was conducted, ensuring variation across the time period. A total of 25 news items from each of the 22 newspapers was sampled.
A source is any person or document to which information is attributed. Each source was classified into one of 11 different categories based on its organizational affiliation. Each source also was coded as being either a named or unnamed source based on whether it was identified by name in the news item. A number of previous studies were used as a foundation for the source coding scheme. Broussard, Shanahan and McComas (2004) classified sources into eight categories based on their affiliation with a particular group or agency. Kim and Weaver (2003) used a similar approach, identifying sources in the reporting of the Asian economic crisis based on their position and their country of origin.
A frame is an idea element in a news item. The phrase was the recording unit for frames, with information aggregated to the item level. This strategy has been used in numerous framing studies with success (Broussard et al., 2004; Kim, 2004; Voakes, Kapfer, Kurpius, & Chern, 1996; Luther & Miller, 2005; Lee & Maslog, 2005). Using previous studies of coverage of the buildup to the Iraq War, a list of idea elements was generated and categorized into story frame categories. Luther and Miller’s (2005) computer-generated analysis of news frames and subframes in news coverage during the buildup to the Iraq War was particularly important in the creation of these framing lists. This list was then tested on news items in several pilot studies that looked at news items that appeared during the time period identified for this study to ensure that the list was exhaustive and each category was mutually exclusive. These idea elements were eventually collapsed into pro-war frames (terrorism, liberation, weapons of mass destruction and impending war) and oppose-war frames (proof WMDs, Iraqi perspective, public opposition, alternatives to war and global tension).
To measure tone, each news item was rated using five different five-point scales, which were treated as interval-level variables. Four of these measures were then converted to a summative index (Cronbach’s alpha = .96) based on the score for all measures of tone. The fifth index was treated as a separate measure because the reliability of the index was not acceptable when it was included. The design for these indexes was based on Moy and Pfau (2000) who use a global attitude measure of six, seven-interval scales to create an index that measures tone. That same measure is also used by Pfau, Haigh, Gettle, Donnelly, Scott, Warr and Wittenberg (2004) in a study of coverage of the Iraq war.
It took nearly six months to design a coding scheme, modify that scheme and train coders to reach acceptable levels of reliability.[2] The procedure followed Neuendorf’s (2002) guidelines for achieving a reliable, replicable coding design. The initial coding scheme was developed drawing on previous studies. Practice coding was then conducted, with adjustments made to the codebook, instructions to coders and the coder training based on the discussion of this work. A pilot study was conducted to test reliability. Final coder training was conducted with further modifications to the coding scheme. Coding of a final sample of 16 percent of the final data set was conducted with reliability tests reported. PRAM, an automated reliability program, allowed coding to be inspected routinely.
Reliability estimates for all general news item characteristics between the two English coders was found to be acceptable or higher. For sources, reliability results using Pearson’s correlation coefficient averaged .88. Reliability for unnamed sources was .79, and .95 for total sources in a news item. The measures of story frames averaged .83 using Cohen’s kappa, though the framing category of global impacts of war was .68. Measures of tone are adequate, with the lowest of these correlations being “unavoidable” and “foolish” tones having Pearson’s correlation coefficients of .71.
The coding of non-English news items occurred after most of the English-language items had been analyzed. Coders were selected and coder training was conducted using practice news items that were not included in the analysis. Coders were not permitted to proceed to coding of non-English items until reliability for each variable was .70 or higher using Cohen’s kappa for nominal variables and Pearson’s correlation coefficient for ratio and interval-level measures.
Table 2 shows the prominence and type of news item and also whether the item was published in a country that joined the Coalition of the Willing. Nearly 15 percent of all news items were front-page articles; 64.6 percent were placed on inside pages. For nearly 20 percent of all news items, information on prominence was not provided. The news items included a mixture of hard news stories (74.2 percent), editorials (6.3 percent), commentary (7.2 percent), columns (5.7 percent) and letters to the editor (3.8 percent). The news items represent a nearly equal number of news items from countries that were on opposing sides in the debate on whether to go to war in Iraq, as the coalition member variable shows.
Table 2: Percentages for general item characteristics.
Table 3 shows news items averaged 633.61 words with a standard deviation of 446.38. An average of 3.75 sources was found in each news item, but this also had large variation with a standard deviation of 3.30. Journalists attributed information to 1,980 sources. Of these sources, the most common type were organizational sources in comparison to the unaffiliated voices of citizens as sources. Of the 3.75 sources in each news item, 28 percent were sources who were either anonymous or whose names were not given to readers (e.g. “a government official”).
Table 3: Means and standard deviations for news item variables.
Each news item was analyzed based on whether idea elements fit into one of eight news frame categories. Coders decided whether these frames were present or not present. These frames were then collapsed into pro-war and oppose-war categories. News items included almost equal use of the pro-war and oppose-war frames, with each story averaging 1.62 ideas that fit into the pro-war category and 1.58 ideas designated as oppose-war ideas.
Coders analyzed news items using five different measures of tone, all of which had means ranging from 3.00 to 3.5, showing balanced or neutral coverage when seen as a whole. Using four of these five measures, an overall tone index was created[6] that finds the overall tone of coverage was 12.43, with 12 being what would be classified as balanced or neutral on a scale of 4 to 20. However, the standard deviation of all of these measures shows that there was variation among the news items. The fifth measure of tone – whether the war was seen as avoidable or unavoidable is the furthest from neutral, with a mean of 3.50 showing that news items tended to present the likelihood of war as unavoidable. Finally, to measure how far from balanced or neutral news items were – regardless of the direction of that tone – a measure of neutrality finds that on a scale of 0 to 6, news items averaged 2.93.
Comparing News Items at the Country Level
Research Question 1 finds that some countries used more anonymous and unnamed sources in reporting than others. The analysis of variance comparisons shown in Table 4 find that the United States, Great Britain and Canada, all countries that became members of the U.S.-led coalition, used unnamed sources more in their reporting than other countries. The United States used the most unnamed sources, averaging two per news item (p<.001). France and Spain used the fewest unnamed sources in reporting. France was the most vocal opponent to the war, while Spain, which did join the coalition, did so in spite of great public opposition.
Table 4: One-way analysis of variance for key source variables by country.
Country-level comparisons (N=11) do not allow the use of more rigorous, multivariate statistics to control for the length of news items or the number of total sources. When these country-level differences were tested using proportional comparisons (unnamed sources to total sources; unnamed sources to total words in a news items), the differences in the use of unnamed sources all but vanish. A Sheffe post-hoc test of the analysis of variance finds that the only country-level differences significant at the 95 percent confidence level exist between France and Australia and France and the United Kingdom.
Influence on Item Prominence
Hypothesis 1 is supported. Table 5 shows that front-page news items used more unnamed sources than news items published on pages of less prominence. Front-page news items used 1.53 unnamed sources on average (p<.01), while news items published on inside pages used an average of .95 unnamed sources. This finding is tempered by comparing the use of total sources in a news item, which shows a statistically significant difference between the number of sources used in front-page and inside pages overall. Front-page news items averaged almost five sources per news item, while inside pages averaged 3.58 sources per item.
Table 5: Independent t-test for news item characteristics by prominence. [21]
Influence on Use of Other Sources
Hypothesis 2 looks at whether unnamed sources were used in conjunction with other kinds of sources and finds no support. Table 6 shows that unnamed sources have a positive correlation with the use of organizational sources (r=.66, p<.001) and with the use of total sources (r=.70, p<.001) overall and the length of a news item. However, H2a and H2b are not supported when compared at the multivariate level.
Table 6: Pearson’s correlation coefficients for source and news item characteristics
Influence on Use of Pro-War and Oppose-War Ideas
Hypothesis 3a is not supported. Unnamed sources did not explain how much news items included ideas that supported the idea of going to war in bivariate analysis shown in Table 7. However, unnamed sources were found to predict whether ideas opposing the war were used in news items, supporting H3b. Unnamed sources does not have a significant correlation with oppose-war ideas, however, the variable is significant (p<.001) in a regression equation that controls for total sources, item length and coalition membership. The b of -.15 shows in Table 8 that as the use of unnamed sources increases, the use of ideas opposing the war decreases in an equation that explains 8 percent of the variance. Total sources (b =.10) also predicts oppose-war ideas, showing an increase in sources leads to an increase in ideas opposing the war.
Table 7: Pearson’s correlation coefficients source and news item characteristics and news frames.
Table 8: Hierarchical regression analysis of news item and source characteristics on oppose-war news frames, N=528.
Influence on Tone of Message
Hypothesis 4 tests the influence of unnamed sources on the tone of the message. It finds that unnamed sources predict how strongly a news item will present the war as a positive idea and as unavoidable outcome. These relationships are shown in Tables 9, 10, and 11.
Table 9: Pearson’s correlation coefficients for news item, source, frame and tone variables.
Table 10: Hierarchical regression analysis of news item, source and frame variables on tone index[46], N=528.
Table 11: Hierarchical regression analysis of news item, source and frame variables on avoidable tone, [52] N=528.
H4a is supported in bivariate analysis: Unnamed sources has a negative correlation (r=-.12, p<.01) with how neutral a news item is. As the use of unnamed sources increases, the news item becomes more neutral. However, this relationship is not present in multivariate analysis.
H4b is supported when the direct path between the source and tone is tested, but not when the indirect path is tested. The use of unnamed sources has a negative correlation with overall tone (r=-.11, p<.01); as the use of unnamed sources increases the tone of the message becomes more favorable toward the idea of going to war. In bivariate analysis, the tone of the news item also has significant correlations (p<.001) with both the pro-war and oppose-war frames. As the use of pro-war frames increases (r=-.35), the tone of the news item becomes more favorable toward the war. In contrast, as oppose-war frames increase (r=.34), the tone of the message becomes less favorable. In multivariate analysis, Table 10 shows that the use of unnamed sources is significant when the direct path of a source’s influence on the tone of a message is tested. However, when the use of pro-war and opposewar frames are held constant, it is the use of these frames that remain predictors of the tone of the message in relationships that parallel those found in the bivariate comparisons. In a regression that explains 27 percent of the variance, as pro-war frames increase, the tone of the message becomes more favorable (b=-1.47, p<.001). As oppose-war frames increase, the tone of the message becomes less favorable (b=1.57, p<.001).
H4c is supported. The use of unnamed sources is found to predict whether news items presented the war as being an unavoidable outcome. Unnamed sources has a statistically significant correlation of .17 with the variable measuring the view of the war as being unavoidable. As the use of unnamed sources increases, the news item presents the war as being more unavoidable. In a regression equation that explains 13 percent of the variance, Table 11 shows that this relationship remains (b=.08).significant at the 95 percent confidence level. The use of pro-war and oppose-war frames are also found to have an influence on the view of the war as being unavoidable. As the use of pro-war frames increases (b=.29, p<.001), the idea of going to war becomes more unavoidable. As the use of oppose-war frames increases (b=-23, p<.001), the view of the war becomes seen as more avoidable.
Anonymous and unnamed sources have helped journalists expose news events that could not have been uncovered by any other means. They are news events in which the journalist is serving as the public watchdog, and the source is taking great risk to disclose some type of wrongdoing (Pincus, 2005; Bagdikian, 2005). If this description accurately reflects how and why these types of sources are used then one could expect that when news items include the voices of unnamed and anonymous sources they would also contain information that would challenge the arguments being made by political leaders. What’s more, it would suggest that anonymous and unnamed sources would be an exception in news reporting. But this study found no support for such a model. To the contrary, these findings are better explained by those who believe unnamed sources are an effective tool for political actors rather than for journalists.
To summarize, this study of news coverage of the buildup of the Iraq War found that news items that used more unidentified sources were stories that were less likely to including perspectives that opposed the idea of an invasion. Stories with more unidentified sources were stories in which the tone of the message was one that made the idea of going to war seem unavoidable; the view of an invasion was seen in a more positive light.
These relationships did not vary from country to country, suggesting that newsgathering routines regarding unidentified source use are uniform for journalists in the countries used in this analysis. These relationships were also not affected by whether the news item was published in a country that ultimately decided not U.S.-led invasion.
The results also suggest that use of these kinds of sources was fairly common during this time period. Rather than appearing only frequently and as a last resort, unidentified sources were a standard part of the newsgathering process. During the seven months prior to the U.S.-led invasion of Iraq, one of every four sources in the 528 news items published in 22 different newspapers from 11 countries was either unnamed or anonymous.
While these findings are far from definitive, they challenge the whistle-blowing role of the anonymous and unnamed source. The findings offer an opportunity to place unnamed and anonymous source use into previous theoretical discussions regarding political reporting and agenda setting (Entman, 1993; 1989). Journalists require sources they can trust to provide them with information they deem credible and accurate (Tuchman, 1973). Journalists covering politically volatile issues look to official sources as they seek to interpret a news event (Schudson, 2003: Bennett, 1990).
Agenda-setting theory shows how the journalist-source relationship is often dominated by political actors who hold great power in deciding not only what issues are discussed but also what ideas are used to interpret those issues (Shoemaker & Reese, 1996; Entman, 1993). Political reporting is characterized as being event-driven and lacking frames that add context to a news event (Iyengar, 1991), which in this case meant the absence of discussions of alternatives to going to war and doubts about the existence of weapons of mass destruction.
What’s most intriguing about how anonymous and unnamed sources were used in these news items is the possibility that they might have allowed political actors to have even more sway in the agenda-setting process. Anonymity may allow political leaders to “leak” ideas that they may not want their names attached to and that, as some have suggested, may give them even more power to direct the discussion of a news event from behind the scenes (Bagdikian, 2005; Shepard, 1994).
It is important to put the time period of this analysis into the context of the overall coverage of the war in Iraq because journalists do not always stick to the routines suggested by this agenda-setting process. As famous cases in journalism history, such as the Pentagon Papers and Watergate show, there are times when the public officials’ interpretation of news events comes into question. This skepticism empowers journalists to deviate from standard newsgathering routines. This is what Lawrence (2000) describes as a triggering event: Moments in the life of a news event when the explanations of political actors simply don’t add up. When this happens, Lawrence writes, journalists begin to challenge official interpretations of news events and look for outside sources to offer other explanations. For example, the videotaped footage of Rodney King being beaten by police led to riots in Los Angeles and prompted media to begin questioning police officials who wanted to categorize such incidents as isolated events. Journalists began making connections between these events nationwide and to question policing techniques. This contextual view of news is what Iyengar (1991) classifies as thematic framing of news rather than episodic framing, which might be seen as more reactionary.
It would make sense to see such triggering events as the time when journalists would look to anonymous sources as they scrutinized ideas being put forth by political leaders. Such sources would empower the journalist to scrutinize ideas dominating political discourse and offer new perspectives based on information gleaned from sources who had vital inside information.
It’s likely that during the period of time that was the focus of this study, the press was more accepting of the explanations being offered by political officials. If there was a triggering event in the coverage of the Iraq War, it came later, when the lack of discovery of weapons of mass destruction prompted journalists to begin to question the ideas owned by political leaders. Ironically, that scrutiny led, in part, to concerns that journalists had been manipulated by anonymous sources.
The strength of this study is that it looks at news coverage with great variation in the country of publication, newspapers and time frame. In doing so, it attempts to go beyond previous research of the use of anonymous sources that has been concerned either with how often journalists use unnamed sources or how readers perceive the credibility of these sources. However, there are some obvious limitations as to what can be said from these results. While the relationships outlined here allow us to control for various alternative explanations, this analysis does not make one important connection between the specific ideas in the news article and the owner of those ideas. It would be useful to further explore the use of anonymous sources by connecting each source to the specific ideas being attributed to them in news articles. It would also be valuable to see if the relationship between anonymous sources and issue ownership exists in more routine political reporting. This project also underscores the need to do more comparative work to parse out country-level differences in how unnamed and anonymous sources are used.
The temptation may be to use these findings to blame the media for the limited discourse that occurred prior to the Iraq War. This would ascribe far more power to media than many believe they have in such affairs – and, in fact, when seen as a whole, news stories included almost equal pro- and anti-war frames presented in a tone that was, on average, neutral in its coverage. While it’s convenient to “blame” news media when they fail to critically evaluate or challenge the dominant messages of political discourse, Schudson suggests that: “To hold news organizations accountable for news is something like holding parents accountable for the actions of their children – it is convenient to locate responsibility somewhere, and it reminds news organizations (or parents) that they have a serious job to do for which they will be judged. Still, they sometimes have to work with unyielding materials” (2003, p. 14). It might make more sense to consider journalism within the context of a cultural setting, reflecting and transmitting ideas and information but not capable of stepping outside of its situation. As Bennett notes, the existing media routines are unlikely to change without major shakeup and the resurgence of a courageous and independent press (1990).
These findings should help journalism practitioners see the powerful influence unnamed sources have on the public discussion of an issue, though not, perhaps, in the way they envision. Would questions regarding proof of weapons of mass destruction or the alternatives to war have been discussed in greater depth and in a tone that was more critical if journalists had been less willing to let government sources hide behind anonymity when making claims about the reasons to go to war? Perhaps greater scrutiny regarding the use of these kinds of sources would have prevented incidents such as the one involving USA Today’s foreign correspondent, Jack Kelley, who admitted to fabricating many of the unnamed and anonymous sources in his reporting (Martin-Kratzer & Thorson, 2007).
In the aftermath of the Iraq War, journalists said they had learned their lesson. Scholars began to monitor the use of anonymous sources and many media companies re-examined when they would permit sources to speak on the condition of anonymity. Media ethicists used the Iraq War to remind journalists to ask why a source is requesting anonymity and consider the impact of this choice (Rosentiel, 2005). But the debate continues. Clark Hoyt, public editor for The New York Times, said his newspaper continues to use anonymous sources even when the same information could have been obtained in other ways, “It allows unnamed people to provide quotes of marginal news value and to remain hidden with little real explanation of their motives, their reliability, or the reasons why they must be anonymous” (2010, p. WK10). Bagdikian (2005) believes that it’s naïve to think anonymous sources can be abandoned entirely, but that news organizations should become less reliant on them and do more in-depth reporting on their own.
[1] Dotted lines represent the dynamic nature of these relationships. These paths are not tested in this study.
[2] Neuendorf (2002) notes there is no agreed measure of acceptable reliability. Agreement was deemed acceptable when reliability was .70 or higher using Cohen’s kappa for nominal variables and Pearson’s correlation coefficient for ratio and interval variables.
[3] Coded based on where the news item ran in the newspaper.
[4] Coded 1-7 based on type of news item..
[5] Dummy variable coded 1 if news item was published in a country that agreed to be included as a member of the “coalition of the willing” (“Coalition Members,” 2003).
[6] Cronbach’s alpha of .96 using four measures of tone. Scale of avoidable/unavoidable treated as separate measure of tone to keep reliability high.
[7] Number of words in news item.
[8] Each source was coded as being one of the 11 source groups that were then collapsed into organizational and unaffiliated sources.
[9] The total number of sources was tallied from the list of all sources appearing in a news story.
[10] Anonymous or unnamed source to which information is attributed (e.g. a White House source, a government official, an anonymous government source).
[11] Items were coded based on whether they contained idea elements fitting into eight broader categories that were then collapsed into two categories: pro-war and oppose war.
[12] Each item was coded as: 5 (negative), 4 (somewhat negative), 3 (balanced or neutral), 2 (somewhat positive), 1 (positive).
[13] Each item was coded as: 5 (unavoidable), 4 (probably unavoidable), 3 (balanced or neutral), 2 (probably avoidable), 1 (avoidable).
[14] 5 (foolish), 4 (probably foolish), 3 (balanced or neutral), 2 (probably wise), 1 (wise)
[15] 5 (bad), 4 (probably bad), 3 (balanced or neutral), 2 (probably good), 1 (good)
[16] 5 (justified), 4 (probably justified), 3 (balanced or neutral), 2 (probably unjustified), 1 (unjustified)
[17] An index of four measures of tone to create an additive index of 4 to 20 (Avoidable/unavoidable excluded to increase reliability). A low measure is a positive view of the war. Cronbach’s alpha of .96.
[18] Recoding any measure as being opinionated in either direction away from neutral. Median of scale is 0 and one point in either direction away from that is the same from scale of 0 to 6.
[19] Anonymous or unnamed source to which information is attributed (e.g. a White House source, a government official, an anonymous government source).
[20] The total number of sources was tallied from the list of all sources appearing in a news story.
[21] Items coded 1-4 based on where the story ran in the newspaper.
[22] Anonymous or unnamed source to which information is attributed (e.g. a White House source, a government official, an anonymous government source).
[23] The total number of sources was tallied from the list of all sources appearing in a news story.
[24] Number of words in news item.
[25] Anonymous or unnamed source to which information is attributed (e.g. a White House source, a government official, an anonymous government source).
[26] Each source was coded as being one of the 11 source groups that were then collapsed into organizational and unaffiliated sources.
[27] Number of words in news item.
[28] Items were coded based on whether they contained idea elements fitting into eight broader categories that were then collapsed into two categories: pro-war and oppose war.
[29] Anonymous or unnamed source to which information is attributed (e.g. a White House source, a government official, an anonymous government source).
[30] Each source was coded as being one of the 11 source groups that were then collapsed into organizational and unaffiliated sources.
[31] The total number of sources was tallied from the list of all sources appearing in a news story.
[32] Number of words in news item.
[33] Items were coded based on whether they contained idea elements fitting into eight broader categories that were then collapsed into two categories: pro-war and oppose war.
[34] Anonymous or unnamed source to which information is attributed (e.g. a White House source, a government official, an anonymous government source).
[35] The total number of sources was tallied from the list of all sources appearing in a news story.
[36] Number of words in news item.
[37] Dummy variable coded 1 if news item was published in a country that agreed to be included as a member of the “coalition of the willing” (“Coalition Members,” 2003).
[38] Recoding any measure as being opinionated in either direction away from neutral. Median of scale is 0 and one point in either direction away from that is the same from scale of 0 to 6.
[39] An index of four measures of tone to create an additive index of 4 to 20 (Avoidable/unavoidable excluded to increase reliability). Cronbach’s alpha of .96.
[40] Each item was coded as: 5 (unavoidable), 4 (probably unavoidable), 3 (balanced or neutral), 2 (probably avoidable), 1 (avoidable).
[41] Anonymous or unnamed source to which information is attributed (e.g. a White House source, a government official, an anonymous government source).
[42] Each source was coded as being one of the 11 source groups that were then collapsed into organizational and unaffiliated sources.
[43] The total number of sources was tallied from the list of all sources appearing in a news story.
[44] Number of words in news item.
[45] Items were coded based on whether they contained idea elements fitting into eight broader categories that were then collapsed into two categories: pro-war and oppose war.
[46] An index of four measures of tone to create an additive index of 4 to 20 (Avoidable/unavoidable excluded to increase reliability). A low measure is a positive view of the war. Cronbach’s alpha of .96.
[47] Anonymous or unnamed source to which information is attributed (e.g. a White House source, a government official, an anonymous government source).
[48] Each source was coded as being one of the 11 source groups that were then collapsed into organizational and unaffiliated sources.
[49] Number of words in news item.
[50] Dummy variable coded 1 if news item was published in a country that agreed to be included as a member of the “coalition of the willing” (“Coalition Members,” 2003).
[51] Items were coded based on whether they contained idea elements fitting into eight broader categories that were then collapsed into two categories: pro-war and oppose war.
[52] Each item was coded as: 5 (unavoidable), 4 (probably unavoidable), 3 (balanced or neutral), 2 (probably avoidable), 1 (avoidable).
[53] Anonymous or unnamed source to which information is attributed (e.g. a White House source, a government official, an anonymous government source).
[54] Each source was coded as being one of the 11 source groups that were then collapsed into organizational and unaffiliated sources.
[55] Number of words in news item.
[56] Dummy variable coded 1 if news item was published in a country that agreed to be included as a member of the “coalition of the willing” (“Coalition Members,” 2003).
[57] Items were coded based on whether they contained idea elements fitting into eight broader categories that were then collapsed into two categories: pro-war and oppose war.
John Hatcher, Ph.D., is an assistant professor of journalism at the University of Minnesota Duluth. Previously, he studied mass communication and political communication at Syracuse University’s S.I. Newhouse School of Public Communications and The Maxwell School of Citizenship and Public Affairs. He has written about the sociology of news for academic and professional publications including the Journal of Global Mass Communication, the Columbia Journalism Review, and American Journalism. He has 15 years of experience as a newspaper reporter, editor and columnist. For a list of recent and pending publications, visit: http://www.d.umn.edu/~jhatcher.
Copyright © 2024 Global Media Journal, All Rights Reserved
+44-330-822-4832