Few observers of the news business would seriously dispute the claim that news content is more often the product of cooperation than conflict between journalists and the elites in the field that they cover (Bennett 1996:119). The assumption that relationships between journalists and elites are essentially adversarial is clearly outdated. For the bulk of the information reported as political news, reporters depend on official government sources or a small number of experts with their own access to official government sources (Lee and Solomon 1990: 17). Government-produced information transmitted via policy briefings, news conferences, and increasingly routinized "leaks" provides reporters with the bulk of their information. "Most political events are so predictably scripted that reporters can condense them easily into formulaic plot outlines, (which official) did what (official action), where (in what official setting), for what (officially stated) purpose, and with what (officially proclaimed) result" (Bennett 1996: 153-154). Not too surprisingly, then, this continuous and mutually beneficial interaction very often results in reporters and editors adopting the language and opinions of their elite sources. Words and phrases employed by elites to describe events are likely to find their way into the news stories about those events.
- Thus, the potential bias in the content of news coverage attributable to the use of official government or expert sources is very real. Whether any particular word or phrase appears in any particular news article probably has little importance. News coverage constitutes a massive, continuous stream of information. However, when reporters and editors consistently select one set of one words or phrases over alternative sets of words and phrases in their news coverage, they generate a "socio-linguistic map that can be read (Hart 2000: 24)." Whether print journalists are fully cognizant of their crucial roles as interpreters of events and opinion makers, some interpretation is present in every column inch of newsprint (ibid., 184).
- This bias matters for two reasons. First, it violates the fundamental journalistic norms of objectivity, balance, ‘fairness,’ and independence from government control which are all central to the liberal model of news coverage. Adherence to these liberal norms of journalism are intended to separate the private sector news organizations from government organs or public relations agencies.[1]
Second, when the terminology of elites becomes the terminology of reporters and editors, the result is a public discourse in which elite perspectives of events exclude alternative, critical perspectives of those same events. Such alternative perspectives are crucial if the public is to attain a holistic understanding of issues and conflicts.
- When popular political commentators charge that political news coverage is biased, they typically do so without presenting systematic, empirical evidence of bias. If evidence is offered, it invariably comes in the form of anecdote, and this phenomenon will be discussed at length later. The audience for such commentary is usually highly partisan, and thus unlikely to demand more than the claim supported by anecdote. Even in the absence of convincing systematic evidence, charging bias is simply too attractive a polemical device to forego. Although they do so much less frequently, scholars may also charge that political news coverage is biased, and they too typically fail to offer systematic evidence of bias. Where charges of bias made by popular political commentators may differ from those made by scholars is that the former are heavily discounted as attempts to mold public opinion, while the latter are normally accorded greater credibility as factual. In any case, and with whatever purpose, such claims of bias are generally not supported by systematic evidence, thus weakening any charge that is presented.
- The empirical research described in this paper was prompted by a single instance in which bias in political news coverage was charged by probably the most renowned and prolific scholar of politics in Sri Lanka, K.M. de Silva. Writing about news coverage of the Indian intervention in Sri Lanka, de Silva asserts that:
The regional power equation was also skewed because of a factor that has seldom, if ever, been considered in evaluations of India’s role as a regional power, namely Delhi’s position as a regional information center. A great many of the world’s major newspapers have their South Asia correspondents based in Delhi, and when regional issues affecting India emerge the reporting can often be influenced by information obtained from, if not supplied by, the South Block through official or unofficial channels. . . . [T]he reporting on the Indian intervention in Sri Lanka uncritically reflected Indian sources. The smaller countries of South Asia, Bangladesh, Nepal, and Sri Lanka, have all had to live with the reality that the principal news center in South Asia is Delhi, and that reporting often carries a Delhi flavor, if not a Delhi bias. (1995: 330)
Whether the news coverage of South Asia is subject to a pro-Indian ‘New Delhi’ bias is an interesting question; a pro-Indian bias in international news coverage would represent another arrow in India’s quiver of foreign policy power resources. Influence over the content of international news coverage means influence over both elite and mass opinion in advanced industrial countries which may become important players in regional politics. - While de Silva offers no systematic evidence of bias, the proposition is plausible for two reasons. First, distortion in international news coverage across geographic regions has been demonstrated. Ovsiovitch (1993: 682-686) established that human rights coverage in the New York Times, Time, and the CBS Evening News was overwhelmingly focused on Eastern Europe and Latin America, rather than Africa, Asia, and the Middle East. If distortion is evident across regions, then it is entirely reasonable to expect distortion in the coverage across countries within regions that already receive scant attention.
- Second, major daily newspapers and wire services have reduced their operating costs by cutting the numbers of their foreign correspondents over the past three decades.[2]
As Grier (1996) ruefully observes, "U.S. media now employ about 1,500 foreign correspondents -- about the number of people that work in one wing of a large metro shopping mall." The commercial logic is that posting a smaller number of reporters in the capital of the largest country in a region is less expensive than posting a larger number of reporters in a larger number of countries. Inevitably, this small corps of reporters is more dependent on information from official government sources because it is easily gathered.
- If the bulk of the international news coverage of South Asia comes from reporters posted in New Delhi, then that would explain any consequent pro-Indian bias. However plausible the proposition, it still begs the question. Is there a pro-Indian ‘New Delhi’ bias in news coverage of South Asia? Our paper presents empirical findings which answer that question.
- Investigating the empirical question is also important because it may reveal something of the underlying sources of bias in other international news coverage. Critics of the news media typically identify two sources of bias. First, they attribute the bias they detect in news coverage to the political ideological orientations of reporters, editors, and news media executives (Demertzis 1999). In this increasingly dated view of American news media, reporters are thought to be ideologically liberal while editors and news media executives are thought to be ideologically conservative. Thus the content of domestic U.S. news coverage is understood as the product of interaction between liberal reporters on the one hand and conservative editors and executives on the other. Another, more current, critique is that international news coverage in American news media is flawed because reporters, editors and news media executives share a common parochial perspective in which they tend to project "American concerns, neuroses, and assumptions onto other parts of the world," with the result that "world events are shoehorned into concepts derived from American domestic politics" (Fallows 1996: 197-198). Shackled by their ethnocentrism, American journalists are rendered incapable of recognizing, understanding, and reporting many important events beyond the borders of the United States.
- Second, critics attribute the bias they detect in news content to a more proximate cause; the close cooperation that exists because of the continuous interaction between reporters and editors on the one hand and government and expert elites on the other. Reporters covering international news coverage are thought to be particularly susceptible to influence by national policymakers (Seib 2000). In effect, the reporters and editors of a national media source slant international coverage to support their national foreign policy. This effect would be most powerful when reporters are in Washington rather than posted abroad and when both U.S. national interest and foreign policy are clearly articulated (Kurspahic 1995: 96-97).
- When we conceived of and executed our project in 2000, we selected news coverage of the long-running conflicts in Indian Kashmir and Sri Lanka because, at that time, neither of these potential sources of bias should have been operative. Neither liberals nor conservatives in the U.S. had any strong interests in the outcomes of either of these conflicts. With the possible exceptions of nuclear non-proliferation and human rights, these conflicts were remote from American liberal or conservative ideological or policy interests. Unlike conflicts in the Balkans, Middle East, Latin America, East Asia, or Africa, they had little reason to "root for" either side in these South Asian conflicts. Although the U.S. government has chosen sides in both conflicts, favoring governments over insurgents, the depth of U.S. diplomatic and military involvement in both South Asian conflicts was comparatively slight. Reporters, editors, and news media executives thus lacked two potential reasons for biasing news coverage. What remained was a single possible systematic source of bias -- that most reporters are posted in New Delhi and thus subject to influence by official sources in India.
- As mentioned earlier, typically the only evidence offered to support the claims of bias in news coverage that are made in popular political commentary or scholarly works is anecdotal. For example, Sremec (1999:18-25) assails United States news coverage of the conflicts in the former Yugoslavia for anti-Serbian bias citing a handful of articles which she describes as inaccurate or sensationalist. Similarly, criticizing Indian news coverage of the war in Sri Lanka for pro-Tamil separatist bias, Pisayena and Sendhera (1987:95-101) offer what they assert are ‘randomly selected’ sensationalist headlines. The basic methodological problem with anecdotal evidence is that it is subject to a different sort of bias -- selection bias. When researchers search for evidence with which to support or to contradict a proposition, they are very likely to find what they are seeking. Unfortunately, they are likely to ignore contrary evidence in the process.
- To avoid selection bias, we tested our hypothesis that news coverage of South Asia exhibits a pro-Indian ‘New Delhi’ bias by using an empirical content analysis research design. Empirical content analysis consists of systematically collecting and then analyzing quantitative data drawn from text.
- To test our hypothesis we chose to compare news coverage of the conflicts in Sri Lanka and Kashmir because the conflicts are similar in several ways. They both involve efforts by armed movements with extensive external support from communities in neighboring states to draw new international boundaries. In addition, both conflicts have taken approximately the same number of lives in approximately the same time period.
- For our study, we collected all of the Associated Press wire service articles dealing with Sri Lanka and Kashmir which appeared from January 1999 to January 2000, a period of intense violence in both conflicts. The Associated Press was selected as a news source because it provided a sufficient number of articles on both of the conflicts for comparative analysis and because it is an important non-governmental international news source for many national, regional, and local newspapers in the U.S. as well as for many newspapers outside the U.S.; "Even the smallest papers typically subscribe to the Associated Press, which produces a flood of interesting foreign stories every day" (Grier 1996). What the wire services present to their subscribers influences the content of news that appears in newspapers (Whitney and Becker 1991: 233-236).[3]
- A total of 161 news articles was collected, including 96 about the conflict in Sri Lanka and 65 about the conflict in Kashmir. Press releases were excluded because they do not purport to present objective news. Table 1 presents descriptive statistics for the data used in our study, including the total numbers of news articles by conflict, mean word lengths, and mean number of descriptors.
Table 1. Descriptive Statistics
Sri Lanka
Kashmir
Total Articles
96 65 Total Number of Words
33176 32458 Minimum Number of Words
107 136 Maximum Number of Words
889 950 Mean Number of Words
345 499 Standard Deviation
192 155 - Descriptors, or individual words used to designate actors in the conflicts, were selected as the unit of content because they provide a means of empirically identifying bias for comparison. A unit of content "can be a word, sentence, paragraph, image, article, television program, or any other description of content based on a definable physical or temporal boundary, or symbolic meaning" (Riffe, Lacy and Fico 1998: 58). Note that despite the discrepancy in the numbers of articles on the two conflicts, the total numbers of words about them were surprisingly similar. The total numbers of words used in the articles about the Kashmir and Sri Lanka conflicts were 32,458 and 33, 176, respectively.
- We coded each article for length as measured in numbers of words, for the numbers of individual descriptors used to describe participants in the conflicts, and for whether descriptors designated government actors or anti-government actors. Acronyms were counted as individual words. The following descriptors appeared more than 5 times in the articles on Kashmir: Guerrillas, Rebels, Separatists, Insurgents, Infiltrators, Intruders, Religious Fighters, Freedom Fighters, Forces, Troops, Soldiers and Police. The following descriptors appeared more than 5 times in the articles on Sri Lanka: Terrorists, Guerillas, Suicide Bombers, Militants, Rebels, Separatists, Insurgents, Combatants, Fighters, Forces, Troops, Soldiers, and Commandos. Descriptors that occurred fewer than 5 times were coded as Other. For the Kashmir articles the Other category included Recruits/Trainees, Attackers, Holy Warriors, Terrorists, Officer and POW. For the Sri Lanka articles the Other category included Attackers, Gunmen, Cadres, Assassin, Bomber, and Sniper.
- Although different descriptors might be used to identify actors performing the same kinds of action, some descriptors carry connotations that are more positive or more negative than other descriptors. Differences in connotation attached to the various anti-government descriptors are greater than those for government descriptors. For example, the anti-government descriptors ‘terrorist,’ ‘guerrilla,’ and ‘rebel’ may be used to describe the same actor. However ‘terrorist’ typically connotes violence against non-combatants, ‘guerrilla’ typically connotes more selective violence which may be directed against both combatants and non-combatants, and ‘rebel’ typically connotes opposition to government which may include violence directed at combatants.
- Government descriptors such as ‘soldier’ and ‘troops’ connote legitimate authority and perhaps honorable service. Arguably, because the connotations attached to anti-government descriptors differ more dramatically, their systematic, selective use may contravene the norms of journalistic neutrality and objectivity and thus signal bias in news coverage.
- To check the validity of the differences in connotation of anti-government descriptors, we surveyed 27 faculty in the social and natural sciences, humanities, and business as well as 67 undergraduate students to ascertain their subjective impression of the threat implicit in the different anti-government descriptors that appeared more than 5 times in the articles. Respondents were asked to rank order the threat implicit in the 11 anti-government descriptors from ‘1' (one) to ‘11' (eleven), with ‘1' being the most threatening and ‘11' being the least threatening. The means for their responses, which are reported in Table 2, confirm that some descriptors carry a greater subjective connotation of threat than others.
Table 2. Threat Connotation
Faculty
Students
Suicide Bomber
1.7
2.1
Terrorist
2
2.1
Guerrillas
5.4
3.6
Intruders
6
7.3
Insurgents
6.1
7.3
Infiltrators
6.2
6.4
Rebels
6.9
6.6
Fighters
7.6
7
Combatants
7.8
6.9
Freedom Fighters
8.1
8.2
Separatists
8.3
8.3
n=27
n=67
- For both sets of respondents, ‘suicide bomber’ and ‘terrorist’ were, respectively, ranked highest and next highest in subjective threat. But what is most important is that ‘terrorist’ connoted more threat than ‘guerrilla,’ and that ‘guerrilla’ connoted more threat than ‘rebel.’ Interestingly, students ranked ‘intruders’ and ‘insurgents’ less high in subjective threat than did faculty.
Tables 3a and 3b present figures for individual descriptors used in articles on the conflicts in Sri Lanka and Kashmir respectively, including the numbers and percentages of descriptors for both government and anti-government actors. Examination of the frequencies of descriptors in Table 3a reveals a clearly defined conflict in Sri Lanka. Distinct sets of descriptors were used to designate government and anti-government actors. Contrast that with the frequencies of descriptors on Table 3b, which suggest a much less clearly defined conflict in Kashmir, where many of the same descriptors frequently used to describe government actors were frequently used to describe anti-government actors. Note also that the spread across anti-government descriptors is markedly greater in the Kashmir data than in the Sri Lanka data. These differences may be explained by the role of the government of Pakistan in supporting the anti-government actors in Kashmir. Had we selected a year in the period between 1987 and 1990 when the Indian Army intervened in Sri Lanka, then the data might have presented a similarly muddy image of that conflict.
Table 3a. Individual Descriptors: Sri Lanka Anti Government Government Total
Percent
Total
Percent
Terrorists
12
1.8
0
0
Guerrillas
90
13.2
0
0
Suicide Bombers
14
2
2
,7
Militants
8
1.2
0
0
Rebels
478
70.2
0
0
Separatists
35
5.2
0
0
Insurgents
8
1.2
0
0
Combatants
2
.3
0
0
Fighters
11
1.6
0
0
Forces
3
.45
19
6.7
Troops
1
.15
92
32.3
Soldiers
2
.3
161
56.5
Commandos
0
0
9
3.2
Others*
17
2.5
0
0
*Attackers, Gunmen, Cadres, Assassins, Bombers, Snipers
Table 3b. Individual Descriptors: Kashmir Anti Government Government Total
Percent
Total
Percent
Guerrillas
111
24.4
0
0
Rebels
41
9
0
0
Separatists
12
3
0
0
Insurgents
24
5.3
0
0
Infiltrators
50
11
0
0
Intruders
20
4
0
0
Religious Fighters
80
17.6
0
0
Freedom Fighters
5
1.1
0
0
Forces
21
4.6
22
13.5
Troops
30
6.6
54
31.1
Soldiers
46
10
72
44.2
Police
0
0
5
3
Others**
14
3.1
10
6.1
**Recruits, Trainees, Attackers, Holy Warriors, Terrorists, Officers, POWs
- The most telling comparison between the articles on the two conflicts involves the frequencies with which the different anti-government descriptors were used. ‘Rebel’ was the most frequently used descriptor for anti-government actors in Sri Lanka. ‘Rebel’ appeared 478 times and made up approximately 70% of the total anti-government descriptors for that conflict. ‘Guerrilla’ was the most frequently used descriptor for anti-government actors in Kashmir. ‘Guerrilla’ appeared 111 times and made up approximately 25% of the total anti-government descriptors for that conflict. ‘Guerrilla’ was only the second most frequently used descriptor in the Sri Lanka articles, appearing 90 times and making up about 13% of anti-government descriptors in that conflict. ‘Religious fighter’ was the second most frequently used anti-government descriptor in the Kashmir articles, appearing 80 times and making up about 18% of anti-government descriptors in that conflict. ‘Infiltrators,’ ‘rebels,’ and ‘insurgents,’ were the third, fourth, and fifth most frequently used anti-government descriptors in the Kashmir articles. Although ‘terrorist’ and ‘suicide bomber’ do appear in the articles on Sri Lanka and not in the articles on Kashmir, they appear infrequently.
- The overwhelming majority of articles used more than one descriptor for anti-government actors but individual descriptors dominated in most articles. In one article from December 18th, 1999, the descriptor ‘rebel’ was used 14 times to describe the anti-government actors in Sri Lanka (Ansar). The following text is typical of many articles: "The bombing came a day after Sri Lankan troops fought back an assault by rebels in boats on a strategic military base in the north, leaving 128 rebels and soldiers dead, the Defense Ministry said today." In a comparable article from June 21st, 1999, the descriptor ‘guerrilla’ was used 6 times to describe anti-government actors in Kashmir (Guruswamy). The text from this article, also typical, reads, "In the Mushko valley farther east...Indian troops have pushed back the guerrillas to within two miles of the cease-fire line. The guerrillas had infiltrated up to four miles inside Indian territory."
- Based on differences in connotation, the interpretation of the different frequencies with which the anti-government descriptors were used is straightforward. Anti-government actors are presented as less threatening in the Sri Lanka conflict than in the Kashmir conflict. Where most of the anti-government actors in Sri Lanka are designated as less threatening ‘rebels,’ the largest number in Kashmir are presented as more threatening ‘guerrillas.’ This points to a pro-Indian bias in news coverage because the government of India is portrayed as facing a more threatening challenge than is the government of Sri Lanka.
- Empirical content analyses of news coverage tend not to result in straightforward findings because they are the products of an investigation of the language employed by reporters and editors whose primary task in the commercial news business is to make news entertaining. Achieving objectivity and neutrality in news coverage are secondary tasks of commercial news journalism. And achieving consistency and precision in the use of descriptive language are clearly not tasks that command the attention of reporters and editors. That is why finding systematic and selective use of different descriptors, as we have done here, reveals so much about the presence of bias in news coverage.
- As our findings appear to confirm, the content of international news coverage is strongly biased, presumably by the interaction between journalists working for a U.S. news source and official elites from another country, thereby revealing the extraordinarily porous and subordinate nature of the commercial news media. Rather than simply adopt norms which attempt to make journalists working for U.S. news sources less receptive to information provided by the official elites of countries in which they are posted, there are indeed constructive responses to this problem. Those responses would include posting larger numbers of international correspondents as well as posting them in more countries in a region and adopting norms which stress the value of information drawn from multiple elites, including the leaderships of anti-government movements.
Notes
Some critics conclude that there is no effective difference. According to John Pilger, "Ninety percent of all world news and current affairs now comes to us from fewer and richer and more powerful sources. Three agences, Associated Press, Reuters and Agence France Presse, supply most of the world's 'wore services' news. One is American, one is British, one is French. Reuters and AP make huge profits selling financial and corporate information; their newsrooms have become centres of the 'free market' crusade" (1998: 285). Back
McChesney (1999: 54-55) directs the same criticism at domestic news media. Back
One reviewer points out that our method might be less reliable in analyzing news coverage from Asian countries where English is less commonly used and wire service stringers might use the same synonym repeatedly in their translations from official sources. Back
Works Cited
- Ansar, Aasif. "Sri Lankan President Injured in Bomb Blast," Associated Press. December 18th, 1999.
- Bennett, W. Lance., 1996. News: The Politics of Illusion, 4th Ed., New York: Longman.
- de Silva, K.M., 1995. Regional Powers and Small State Security: India and Sri Lanka, 1977-90. Baltimore: The Johns Hopkins University Press.
- Fallows, James. 1996. Breaking the News: How the News Media Undermine American Democracy. New York: Vintage Books.
- Grier, Peter. "The Motto on Foreign News Coverage: Through a Lens Darkly and Infrequently." Christian Science Monitor. 88, 44: 1. (Jan. 1, 1996.)
- Guruswamy, Krishnan. "India Fights to Hold on to Crucial Himalayan Peak," Associated Press. June 21st, 1999.
- Hart, Roderick P., 2000. Campaign Talk: Why Elections Are Good For Us. Princeton: Princeton University Press.
- Kurspahic, Kemal. 1995. "Neutrality vs. Objectivity in Bosnia." Media Studies Journal. 9(4): 91-98. (1995).
Lee, Martin A., and Norman Solomon. 1990. Unreliable Sources: A Guide to Detecting Bias in News Coverage. Carol Publishing.
- McChesney, Robert A., 1999. Rich Media, Poor Democracy: Communication Politics in Dubious Times. Urbana: University of Illinois Press.
- Ovsiovitch, Jay S., 1993. "News Coverage of Human Rights," Political Research Quarterly,46(3): 671-689.
- Pilger, John. 1998. Hidden Agendas. New York: The New Press.
- Piyasena, S. and R.Y. Sendheera. 1987. India ‘We Tamils’ and Sri Lanka. Delhi: Sri Satguru Publications.
- Riffe, Daniel, Stephen Lacy and Frederick G. Fico. 1998. Analyzing Media Messages; Using Quantitative Content Analysis in Research. Mahwah, NJ: Lawrence Erlbaum Associates, Publishers.
- Seib, Philip. 2000. "Politics of the Fourth Estate," Harvard International Review, 22, 3: 60-64.
- Sremac, Danielle S., 1999. War of Words: Washington Tackles the Yugoslav Conflict. Westport: Praeger.
- Whitney, D. Charles, and Lee B. Becker. 1991. ""Keeping the gates" for Gatekeepers: The Effects of Wire Services," in David L. Protess and Maxwell McCombs, eds., Agenda Setting: Readings on Media, Public Opinion, and Policymaking. Hillsdale: Lawrence Erlbaum Associates.
- Bennett, W. Lance., 1996. News: The Politics of Illusion, 4th Ed., New York: Longman.
Back to Table of Contents, Vol. 6 Issue 3
Back to Jouvert Main Page