Entrepreneur: Start & Grow Your Business

Stages of a crisis and media frames and functions: U.S. television coverage of the 9/11 incident during the first 24 hours.


by Li, Xigen

Much research is devoted to determining how news media frame information so that it affects audiences' understanding and interpretation of issues. A number of studies also look at media functions under various situations. On September 11, 2001, continuous television coverage of the most aggressive terrorist attack on America to date began within seconds of the initial plane crash into the World Trade Center. This provided a unique opportunity to understand how television media cover a crisis of unprecedented magnitude. This study looks at how television outlets framed 9/11 during the first 24 hours, the functions they performed in the national crisis, and how the stages of the crisis affected coverage frames and media functions as unfolding events brought attention to new issues.

Literature Review

When the social order is seriously disrupted, people usually want more information than the media can provide (Neal, 1998). During crises, the public becomes almost totally dependent on the media for news that may be vital for survival and for important messages from public and private authorities. They look to the media for information, explanations, and interpretations (Graber, 1980, p. 228). For example, after President Kennedy's assassination, public uncertainty about the future of the U.S. government resulted in greater need for interpretation, explanation, and consolation (Schramm, 1965).

Media Functions in a Crisis

The National Research Council Committee on Disasters and the Mass Media (1980, p. 10) postulated that the press performed six functions during a crisis: 1) warning of predicted or impending disasters; 2) conveying information to officials, relief agencies, and the public; 3) charting the progress of relief and recovery; 4) dramatizing lessons learned for the purpose of future preparedness; 5) taking part in long-term public education programs; and 6) defining slow-onset problems as crises or disasters.

Researchers say the media have many functions depending on the audience's needs. For example, in addition to transmitting information, the media perform a "social utility function" (Dominick, 1996, p. 47) by providing companionship and emotional support in the absence of other human beings. Others (Entman, 1991 ; Hertog, 2000; Iyengar & Simon, 1993; Ungar, 1998) found the media performed different functions within different crisis situations. The selection of issues and the emphasis they receive tend to differ among media, but all forms of media include information on the primary issues (Lowery & DeFleur, 1995, p. 341). When dealing with breaking news, such as a crisis, the change in reporting routines affects the type of information that journalists disseminate. Journalists who covered the breaking news of the 9/11 terrorist attacks were found to assume multiple roles in delivering information. The content of breaking news reported live was fundamentally different from the content of news stories that were produced with more time to check for violations of journalistic standards (Reynolds & Barnett, 2003).

Graber's notion that there are three stages of media coverage of a crisis seems to reaffirm the media functions listed (1980, p. 229). During the first stage, media are the primary information source not only for the general public, but also for public officials involved with the crisis. Media's key roles are to describe what has happened and help coordinate the relief work. Their top priority is to get accurate information, which relieves uncertainty and calms people (pp. 233-234). In the second stage, media coverage focuses on making sense out of the situation. Plans are formulated and implemented to address the needs of the victims and repair the damage. Graber says the third stage overlaps with the first two. In an effort to provide context, the role of media is to place the crisis in a larger, longer-term perspective.

Studies found between-media difference in U.S. media coverage of various issues including presidential campaigns (G. H. Stempel & Windhauser, 1989) and Canadian elections (Husselbee & Stempel, 1997). However, studies also showed similarity in selection of stories among newspapers (Riffe, Ellis, Rogers, Van Ommeren, & Woodman, 1986). Competing media will conform under certain circumstances (Bigman, 1948). These findings suggest between-media difference in coverage on issues of social significance. But to what degree television media differ in functions they perform in a crisis situation remains unanswered.

Frame Analysis and Sources' Role in Framing

Frame analysis is the most common approach to examining media content. It is based on the assumption that journalists filter information in ways that affect an audience's understanding or interpretation of issues, stories, or events (Lowery & DeFleur, 1995, p. 327). By selecting certain facts from a continuous flow of information, emphasizing specific issues or events over others, and presenting issues or events in specific orders, journalists have the ability to influence attitudes, beliefs, and behavior in a number of ways.

Researchers have looked at media frames from various perspectives. "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." (Entman, 1993, p. 52) Framing, then, can be described as a "story angle or hook"; it is "the central organizing idea or story line that provides meaning to an unfolding strip of events and weaves a connection among them" (Gamson & Modigliani, 1987, p. 143). Media use "certain perspective and frames" in news coverage to help people organize and understand news information (King, 1997, p. 29). News is often presented from a point of view that changes the viewer's understanding or interpretation of events and evokes emotions (Nimmo & Combs, 1985, pp. 17-18; Norris, 1995, p. 359).

A number of studies have focused on news content and how it is framed (Entman, 1993; Fico & Freedman, 2001; Iyengar, 1991; Iyengar & Simon, 1993; Larson, 1984; Nacos, 1994; Norris, 1995; Pan & Kosicki, 1993; Semetko & Valkenburg, 2000; Tewksbury, Jones, Peske, Raymond, & Vig, 2000; Ungar, 1998). Framing research shows that four frames are more common than others: conflict, human interest, responsibility, and economic consequences (Valkenburg & Semetko, 1999, p. 551). The responsibility frame is the most frequently used, followed by the conflict frame. Economic and human interest frames were significantly lower in use (Semetko & Valkenburg, 2000). However, in a crisis involving a large number of casualties, human interest could emerge as a dominant frame. Media functions suggested by The National Research Council Committee on Disasters (1980, p. 10) and the social utility function (Dominick, 1996, p. 47) concurred in pointing to the attention that media paid to human interest in the coverage of disasters, which could lead to human interest frames overshadowing others.

Sources play an important role in media framing. Studies found source use is related to media frames (Liebler & Bendix, 1996). Other studies found unusual source selection in a crisis and its effect on framing (Andsager & Powers, 1999; Colby & Cook, 1991; Entman, 1991; Lasorsa & Reese, 1990; Nacos, 1994). Media used a wider variety of sources when covering an anti-American terrorist act than when covering other foreign policy issues (Nacos, 1994). In the coverage of AIDS, the typical AIDS story tended less to sensationalize than to reassure because government officials and high-ranking doctors were major sources (Colby & Cook, 1991). Under the crisis situation of 9/11, a wider range of sources was used and such source reliance is likely to affect media frames.

Frame Dynamics and Stage of Crisis

The studies reviewed here indicate that various factors affect media frames. Whereas the findings of these studies offer insight into the coverage pattern of important issues in a crisis, there is still much to be understood. For example, in a rapidly developing national crisis, do news frames emerge with patterns similar to those in other crisis situations? What sources played the most important roles in framing the news? To what degree do new events bring about changes in media frames? One of the major flaws in frame analysis of news coverage is that most of the studies examined media frames from a static perspective; that is, the media frames were considered constant throughout the process. In fact, when covering a rapidly changing crisis, media are likely to follow the changes and present varying frames as events unfold. This study seeks to specify the changing construction of media frames during a rapidly evolving crisis.

Several studies have touched on frame dynamics in media coverage of various issues including welfare (Gilens, 1999), the Clinton health care package (Jacobs & Shapiro, 2000) and police brutality (Lawrence, 2000). Media could shift framing strategies from presenting frightening information to a containment or calming approach when "dread-inspiring events are developing in unpredictable and potentially threatening ways" (Ungar, 1998, p. 36). None of these studies specifically focused on frame dynamics in media coverage of a national crisis or offered an elaborate view on frame dynamics. Chyi and McCombs (2004) proposed "frame-changing" in their study of the coverage of the Columbine school shooting, arguing that during any news event's life span, media often reframe the event by emphasizing different attributes. Muschert and Carr (2006) extended the study of frame-changing across similar events and between more and less salient events. Although both studies examined frame dynamics and offered useful ideas of sequence-related frame changes associated with time, neither of them examined frame-changing in news coverage as events evolved through stages.

Unlike previous studies of media frames in breaking news, this study introduces stage of crisis as a key factor affecting the frame of coverage, and attempts to reveal the frame dynamics through analysis of television coverage of 9/11. This study proposes that media frames are dynamic rather than static. Previous studies basically considered frames as unitary properties of news content and few studies examined media frames according to their nature and sophistication. Frames are not identical in their modes of formation and in their properties, especially in situations where events change rapidly. Two dimensions of frame properties were identified: frame nature and frame sophistication, which change over the stages of crisis.

Frame nature refers to the degree to which a frame probes into issues concerned when certain facts and issues are presented to form the frame. According to Reynolds and Barnett's (2003) study of the first 5 hours of television coverage of 9/11, the journalists worked in a situation drastically different from traditional reporting routines. As the pace of coverage slowed in the later stages, journalists gathered more information about the events and had more time to think through the facts. They gradually resumed reporting routines, which would result in substantial media frame changes. There are three types of frame nature associated with the stages in a crisis. The descriptive frame corresponds to the news media's function of informing and relieving uncertainty during the first stage of a crisis, mainly through explaining what has happened. The attributive frame corresponds to the news media's function of making sense out of the situation during the second stage. The affective frame relates to the news media's function of understanding the event from a long-term view and the impact of the incident during the third stage of crisis. Media frames change from descriptive to attributive to affective during different stages of crisis.

Frame sophistication refers to the level of complexity of a story angle or salience. For example, the disaster frame in the coverage of a crisis is easily identifiable, whereas frames of human interest and economic impact are relatively more difficult to identify because they are on higher levels of frame sophistication. Informed by Reynolds and Barnett's (2003) findings on news content produced under stressful situations, journalists are likely to report what they observe directly and will produce stories with lower frame sophistication during the earlier stages of a crisis. As the coverage of a crisis extends in scope and depth, media frames will go from relatively simple to more sophisticated. Examination of frame sophistication will reveal frame dynamics in addition to substantial frame changes through the stages of a crisis.

The literature suggests that news media are used mainly as a source of information in a crisis situation. According to the social utility function, media also fulfill needs other than the need for information. In the early stages of a crisis, aside from providing information, media could also serve as a source of consolation and guidance. As the events unfold, media may perform their social utility function differently during different stages of a crisis.

According to the literature on the influence of government sources on news content, government sources play a major role in the coverage of a national crisis. However, with the magnitude of a disaster like 9/11, which involved more parties and issues than typical political events, government officials were not able to get as much information as they would under less stressful situations and were also searching for information in order to take appropriate actions. Television outlets had to rely on a wider range of sources to cover the rapidly changing events. This could translate to a diminished role of government sources in the early stage of a crisis situation like 9/11.

Previous studies showed between-media differences as well as similarities in news coverage. During a national crisis, media coverage will vary due to organizational and other societal factors. Media also try to get accurate information, relieve uncertainty and calm people (Graber, 1980) rather than sensationalize events and scoop each other. The key issues involved in the crisis were politics, public safety, and antiterrorism. No television outlet could afford to deviate considerably from others at such critical times in informing the public. The crisis situation is likely to put rivals into conformity with each other (Bigman, 1948). It is proposed television media will vary slightly in coverage frames and media functions performed, but not drastically.

Rapid changes of news events during the first 24 hours and observable stages of the coverage provided a unique opportunity to examine how television media framed the news events, what functions they performed in this unprecedented situation, and to what degree stages of the crisis affected coverage frames and media functions in a national crisis situation.

Based on the literature and previous research, the following hypotheses are proposed:

[H.sub.1]: Television media frames change from descriptive to attributive to affective during different stages of a crisis.

[H.sub.2]. Television media frame sophistication rises as coverage moves through the stages of a crisis.

[H.sub.3]: Television coverage emphasizes human interest more than political and economic factors in a crisis situation involving tragedy.

[H.sub.4]: Television media's use of a wider range of sources in a crisis leads to a diminished role of government sources under rapidly changing situations.

[H.sub.5]: Television media in a crisis situation serve as a source of consolation and guidance in addition to an information source.

[H.sub.6]: Television media vary, but not extensively, in coverage frames and media functions performed in a national crisis.

Method

This study used content analysis to examine the first 24 hours of television coverage of the attacks on the World Trade Center and the Pentagon on September 11, 2001. Television was chosen because it had been the dominant medium for informing Americans about the terrorist attacks and the war on terrorism that followed (G. Stempel & Hargrove, 2002). The news coverage of five television outlets, ABC, CBS, NBC, CNN, and FOX News, was selected for content analysis. These outlets were selected because of their dominant status in television news coverage in the United States. They also include three different types of television outlets: the established broadcast networks, a relatively established cable channel (CNN), and a relatively new cable channel (FOX). The first 24 hours were chosen because the time period contained the most critical stages and the most intensive coverage of the incident, and reflected changes in media coverage due to rapid development.

The news content of the five television outlets during the first 24 hours was acquired through Vanderbilt University's video library. A total of 2,647 stories were identified from the first 24 hours of coverage, including 745 stories from ABC, 612 from CBS, 427 from NSC, 657 from CNN, and 206 from FOX News.

The study unit was a news story, defined as a group of studio and field shots that specifically address one topic or issue and run consecutively. The stories were identified according to the basic building blocks: anchor reader, voice-overs, voice-overs with sound bites and packages. Due to the unusual situation, news coverage sometimes ran without clear building blocks. The following cues were then used to identify a story: when the anchor or reporter changed the topic and started reporting on a different aspect of the event instead of merely mentioning something briefly, and the coverage of the topic ran for a significant amount of time (at least 30 seconds). The stories identified ran from 30 seconds to 12 minutes.

The following key variables were coded:

Stage of Coverage. The first 24 hours of television coverage were divided into three stages: the first stage, 8:48 a.m. to 11 a.m., when the disaster struck; the second stage, 11 a.m. to 3 p.m., when the media tried to make sense out of the situation; and the third stage, 3 p.m. to 7 p.m., when the media started to observe the event from a long-term perspective. The three stages were consistent with Graber's (1980) definition of the stages of crisis coverage. The later three periods, 7 p.m. to 12 a.m., 12 a.m. to 6 a.m., and 6 a.m. to 9 a.m. were an extension of the third stage. The unequal lengths of the later three periods conform to the intervals of normal TV programming. According to Graber (1980), fewer changes in coverage will occur during the last three periods. The inclusion of the later three periods extends understanding of the frame dynamics and media functions during the later stage of this crisis of unprecedented magnitude.

Coverage Frame. Following Entman's (1993) definition, the coverage frame is defined as the aspects of a perceived reality identified through a story that makes these aspects more salient in the news coverage. One frame was selected from a story through the story angle or focus identified in the lead. If the story angle or focus could not be identified from the lead, the rest of the story was consulted until the salience of the story was identified. Coverage frame contained a total of nine categories. The coding decisions were made based on whether the story angle or focus identified fell into the defining realm of one of the nine categories, which included 1) Political: national security, government policy, or international relations; 2) Economic: economic impact brought by 9/11; 3) Criminal: criminal acts and investigation; 4) Environment: environmental hazard, threat to public health; 5) Safety: public and personal safety, and the ways to survive; 6) Human interest: human well-being, human feeling, family, and love; 7) Religion: religious faith and act, Islam, Muslim, and other religions; 8) Disaster: damage and casualty; and 9) Other: all other framing aspects that do not fit into above categories.

Frame Nature. The frames were categorized into three types according to the degree to which a frame probed into issues concerned when certain facts and issues were presented to form the frame: 1) descriptive; 2) attributive; and 3) affective. The descriptive frame corresponds to news media's function of informing and relieving uncertainty during the first stage of a crisis, mainly explaining what has happened. Disaster is a descriptive frame. The attributive frame corresponds to news media's function of making sense out of the situation during the second stage. Political, criminal, and religious are attributive frames. The affective frame relates to news media's function of understanding the event from a long-term view and the impact of the incident during the third stage of crisis. Economic, environment, safety, and human interest are affective frames.

Frame Sophistication. The frames were categorized according to their level of complexity when certain facts and issues were selected and presented to form a frame. Three levels were identified. Disaster as a low-level frame does not require much work in perceiving the facts involved and selecting certain facts and issues. Political and criminal as medium-level frames involve more investigations and fact-filtering processes. Human interest, economic and environment frames, all high-level frames, go beyond the observable facts and issues and entail more thought-provoking processes.

Content Orientation. Content orientation is used to identify main functions that news media perform when covering a crisis. It contains four categories, 1) fact; 2) analysis; 3) consolation; and 4) guidance. Content orientation was identified through the lead of a story. If it could not be identified from the lead, the whole story was consulted. The coding decisions were made by identifying whether a story primarily focused on 1) describing facts about what was happening; 2) analyzing information, facts, or events; 3) consoling or comforting the audience to make them feel safe and secure; or 4) providing guidance to the audience on what to do. A story could perform more than one function. But only the main focus was considered as content orientation.

Source. A source was defined as a name of a person or an organization associated with direct or indirect quotes in a story. The types of sources coded include government officials, the President, experts, witnesses, business, airline officials, international, relatives of victims, and other sources such as tourists.

Seven coders were trained using a coding protocol and by following the procedures for coding news content prescribed by Rifle, Lacy, and Fico (2005). Ten percent of the coding content of ABC and CNN, which contained about 1 hour of the news coverage from the two outlets, was used for an intercoder reliability check. Although a random sample from each of the five outlets for the intercoder reliability test could generate more reliable results, ABC and CNN, one broadcast network and one cable channel, were found to be sufficient for the following reasons: 1) A preliminary check showed the nature of 9/11 produced homogeneity of coverage across all outlets; 2) Coding categories were constructed by extensive review of the news content of all five outlets; 3) Coder training used news stories from all five outlets. Working with the sample from the two outlets also made this project more manageable as it produced a huge number of television stories. Scott's Pi was used to test the intercoder reliability for nominal variables. The intercoder reliability test results were: 1) Stage of coverage .96; 2) Coverage frame .80; 3) Frame nature .80; 4) Frame sophistication .80; 5) Content orientation .94; and 6) Source .84.

Results

[H.sub.1], that television media frames change from descriptive to attributive to affective during different stages of a crisis, was supported. During the first stage, from 8:45 a.m. to 11 a.m., the coverage was framed primarily with the descriptive frame disaster (56.8%). Other types of frame, political (14.8%), criminal (12.5%), and safety concerns (8.95%) were less noticeable. During the second stage of the coverage, from 11 a.m. to 3 p.m., the descriptive frame disaster (37.5%) declined dramatically whereas the attributive frames rose significantly. The political frame increased from 14.8% at 11 a.m. to 28.5% at 3 p.m., although the criminal frame remained constant (11.6%). The affective frame emerged in the later hours of the coverage. The economic frame rose from 0.9% at 11 a.m. to 4.1% after midnight. The safety frame increased significantly from 8.9% in the earlier stages to 19% the next morning. The human interest frame rose from 1.3% during the first stage to 8.8% by midnight. (Figure 1).

Fewer striking events occurred at night although the coverage of 9/11 continued throughout the night, which resulted in less systematic differences in coverage frames. Coverage after 7 p.m. varied less but presented its own unique pattern. The attributive political frame decreased continuously (from 17% to 12.2%). The affective economic frame became evident (from 2.9% to 4.1%). The other two affective frames, safety and human interest, rose respectively from 11.3% to 19% and from 8.8% to 12.9% the next morning.

[H.sub.2], that television media frame sophistication rises as coverage moves through stages of a crisis, was supported. The frames identified during the first 24 hours were found to be different in their levels of sophistication. Frames associated with stages of a crisis in the coverage of 9/11 went from relatively simple to more sophisticated. During the first stage, disaster stories dominated, which involved more unified work in selecting facts and framing the stories, whereas stories with political and criminal frames were still evolving. As the coverage proceeded, the disaster frame declined and political and criminal frames increased, and more sophisticated work was required in framing the stories. During the third stage, stories framed as human interest increased significantly, and issues concerning the environment and economy also began to surface. These high-level frames involved more thought-provoking processes from journalists.

[FIGURE 1 OMITTED]

[H.sub.3], that television coverage emphasizes human interest more than political and economic factors in crisis situations involving tragedy, was not supported. About 6.6% of the stories were framed as human interest. The three major frames of coverage were political (17.9%), criminal (11.9%), and safety (11.8%). Approximately 40% of the stories were framed as stories of a disaster. Although these stories might be associated with the well-being of people, human interest was not found as a major frame of stories during the first 24-hour coverage. Overall, the political frame was more evident than human interest in the coverage of the first 24 hours. However, as the event unfolded and the coverage moved to later stages, human interest frames increased dramatically from the earlier stages to become as prevalent as political frames. Although the economic frame also rose in the later stages of the coverage, human interest was emphasized more than the economics frame through the coverage of the first 24 hours. (Figure 1)

[H.sub.4], that television media's use of a wider range of sources in a crisis leads to a diminished role of government sources under rapidly changing situations, was supported. The four most frequently used sources were government officials (19.3%), witnesses (10.8%), the President (6.7%), and experts (5.8%). Government officials were the dominant sources among all sources used. However, their relatively small percentage (19.3%) showed that government officials were not as powerful as in the coverage of other crises or political events. When examining the relationship between specific sources used in the coverage of 9/11 and media frames, the diminished role of government sources was even clearer. Government officials (33.7%) were associated most with the political frame. The stories framed as disaster used witnesses (22.4%) as the major source, with government officials (16.8%) the second. Criminal frames used more experts (25.7%) than government officials (21.3%) as the major sources. The stories framed as economic cited more business sources (11.9%) than government officials (4.5%). The stories framed as human interest were dominated by witnesses (13.2%) and victims' relatives (18.4%) compared to government sources (5.2%). (Table 1)

[H.sub.5], that television media in a crisis situation serve as a source of consolation and guidance in addition to an information source, was not supported. The television outlets served primarily as an information source rather than a consoling and guiding source during their coverage of the first 24 hours. More than 80% of the stories were identified as primarily presenting facts, whereas only 14.8% of the stories offered analysis. In terms of coverage time, 68% was devoted to presentation of facts, whereas 17% was devoted to analysis. The coverage devoted to consoling or easing stress and anxiety of the audience (2.0%) and to guiding the audience in a crisis situation (1.2%) was negligible. When looking at the changes of media functions by stages of crisis, no changes were found in the stories primarily presenting facts. An average of 81% of the stories was devoted to presenting facts across the three stages of the crisis. Analysis news items decreased gradually. Differences were found in the coverage devoted to consolation; there were almost no stories devoted to consolation during the first stage of the crisis. As the coverage continued, stories of consolation increased from 2.8% in the second stage to 3.5% in the third stage. Consolation decreased after 12 a.m. the next day (Table 2).

[H.sub.6], that television media vary, but not extensively, in coverage frames and media functions performed in a national crisis, was supported. There was a difference in coverage frames among the television outlets ([chi square] = 72.04, df = 32, p < .01). But the relationship between coverage frame and television outlets is relatively weak (Cramer's V = .08, p < .01), which means coverage frames did not vary substantially across the outlets. As for the four major coverage frames identified, disaster, political, criminal, and safety, some outlets deviated from other outlets in one or two of the frames. CNN and FOX had more stories framed as political than ABC, CBS, and NBC. Whereas three networks (ABC, CNN, and FOX) had a similar number of stories framed as criminal, CBS had the most stories (13.6%) with the criminal frame, and NBC had the fewest (9.6%). ABC, CBS, and NBC devoted more attention to the safety frame than CNN and FOX did. Whereas stories with the human interest frame did not receive much space from all the outlets, ABC (7.5%) and CBS (7.8%) had more stories with the human interest frame than the other three outlets (Table 3).

There was a slight difference in media functions performed among the television outlets in the crisis situation. The television outlets were similar in performing the primary function, presenting facts in their news coverage (around 82% on average), whereas FOX deviated from other outlets in presenting fewer facts in its coverage (65.4%). FOX provided much more analysis (25.7%) in its coverage than other outlets (around 14%). FOX also led the television outlets in providing consolation (3.9%) and guidance (2.9%) whereas other outlets provided much less consolation and negligible guidance in their coverage (Table 4).

Discussion and Conclusions

The findings from the five television outlets' coverage of 9/11 offer fresh insights about how U.S. media covered a rapidly changing national crisis. The findings confirm the functions of television media during a crisis (Graber, 1980; National Research Council Committee on Disasters and the Mass Media, 1980), and suggest that media frames of crisis in television coverage are dynamic rather than static, especially under situations where events change rapidly. The results also suggest the stage of crisis has an effect on media function. The television coverage shifted focus as events brought attention to issues that had not been at stake during the earlier stages.

Support of [H.sub.1] and [H.sub.2] is the most important finding of this study, which reveals how frame nature and frame sophistication changed during the stages of a crisis. The stage of the crisis was found to be an important factor associated with frame nature and frame sophistication. The findings confirm Graber's (1980) theory of three stages of crisis covered by news media and go beyond Graber's observation by revealing the dynamic nature of media frames.

One noticeable finding was that the less systematic difference in coverage frame occurred at midnight compared to earlier stages of the crisis. Structural factors in newsrooms could be responsible for the differences. There were more journalists working during the day than in the middle of the night, and events were thus followed closely and promptly. Media shifted their focus of coverage due to less breaking news and less access to sources at midnight, which could lead to coverage topics being contained in a relatively narrow range with less variation.

The findings regarding frame dynamics in the coverage of a crisis could advance understanding of framing in the following aspects. First, they offer empirical evidence of the effects of the stage of a national crisis on media frames in covering the crisis. Coverage of rapidly developing events is likely to be framed differently as the events unfold. Lawrence (2000, p. 9) discussed the difference between the coverage of "routine news" and "event-driven news." Framing dynamics are more likely to be associated with event-driven news. The coverage of 9/11 went through a series of events with new issues brought up continuously. Journalists had to make constant decisions about what to select and how to present the facts (Mogensen, Lindsay, Li, Perkins, & Beardsley, 2002). When an event goes through rapid development, media coverage frames are likely to follow with a series of frames taking turns in dominance.

Second, the findings shed light on important dimensions of media framing. The coverage of a crisis with identifiable stages calls for variable coverage frames corresponding to specific media functions at each stage. Frame nature changes revealed that media frames correspond to media functions during the different stages of a crisis. The frame changes from descriptive to attributive to affective through different stages of the crisis during the first 24 hours were consistent with media function at each stage of the crisis (Graber, 1980).

Third, frame changes associated with stages of a crisis in the coverage of 9/11 underwent a pattern from relatively simple to more complicated frames. The findings suggest that, in the coverage of a national crisis, coverage frames not only change substantially in their nature over the stages of the crisis, but also proceed toward a higher level of sophistication as journalists learn more about the events and select facts with more deliberation.

The findings on effects of stages of crisis have important implications for frame analysis of media content. Although frame analysis deals mostly with issues that are not changing rapidly, a dynamic view of frame changes over time allows frame analysis to reflect the coverage more thoroughly. These frame dynamics apply not only to television media, as in the case of 9/11 with rapidly changing events, but also to any other news media that cover ongoing events or issues. Stages of crisis could turn into stages of development as ongoing events and issues are covered by any news media. It is important to take into account the effect of time periods on media frames. Even for "routine news," when coverage of an issue lasts over a period of time, coverage frames are likely to vary. A dynamic view in frame analysis may reveal more insight on how media frame news events and provide a better understanding of how such frames may eventually influence the audience.

When a tragedy involves human casualties, human interest is likely to be a central issue. The failure to find support for [H.sub.3] indicates another priority in the coverage of a national crisis. Human interest may give way to issues bearing more weight. In the coverage of 9/11, the human interest frame became evident during the later stages, when the focus of the coverage shifted and events brought media attention to issues that had been less noticeable during the earlier stages. The changes in the framing of human interest during different stages also suggest a shift in level of human interest in a tragedy of national magnitude. Human interest normally refers to human feelings, well-being, and family, or love. As the events of national magnitude unfolded, the media were not able to focus their attention on individuals during the early stages of the crisis. When human interest surfaced as a noticeable frame, human well-being was not attached to only a small number of people or seen through stories about individuals. Instead, human interest was presented from a national perspective and attached to the well-being of all people being affected by the crisis. The magnitude of the crisis defined the human interest frame on a national rather than individual level as seen in the coverage of other human tragedies.

Support of [H.sub.4] revealed the role of government sources in a crisis situation involving national interest. Government officials were one of the two major sources used in the coverage of 9/11. However, the fact that 19.3% of the stories used government officials suggests that during an unprecedented crisis the capacity of government officials as sources is limited in certain aspects. Government officials also need to sort information from all possible sources and may be unable to offer accurate information at certain points. Media need to rely on a variety of sources to provide accurate and useful information. The findings suggest a diminished role of government sources during various stages of a national crisis. The findings are also consistent with Nacos' (1994) argument that media use different methods when covering an anti-American terrorist act than when covering other foreign policy issues. The variations in source use during different stages of the crisis indicate that whether a wider range of source will be used in the coverage depends not only on the nature of the incident but also the functions media perform during different stages of the coverage.

Graber (1980) noted that media provided guidelines for what to do in a crisis situation. The findings for [H.sub.5] suggest that in a crisis of unprecedented magnitude, media's social utility function is contingent upon the primary need of the audience. The coverage of 9/11 showed a clear order in priorities of media functions, and that the need for consolation and guidance may be overridden by the need for accurate and informative facts. Thus, providing facts may be the fundamental task of media in a crisis of national magnitude, especially during the early stages. The finding regarding increasing number of stories of consolation during the later stages confirms media's social utility function and the effect of stages of crisis on media functions. Whereas media primarily served as a source of information during the early stages of the crisis, the media function of consoling the public became more evident as events unfolded. The finding suggests that media functions in a crisis also shift during different stages of a crisis. As the need for critical information is fulfilled during the initial stage of a crisis, the media social utility function would resume through the coverage during the later stages of the crisis.

The findings of [H.sub.6] confirmed the between-media difference in news coverage. They also shed light on rivals in conformity and to what degree media coverage might be similar in a crisis of national magnitude. During the early hours of their coverage of 9/11, the three television networks ABC, CBS, and NBC actually shared news footage. The issues at stake and the primary functions imperative for media to perform during the early stages helped reduce difference in coverage frames and functions media performed and homogenized them to some degree.

This study examined a total of 2,647 stories. Although all efforts were made to ensure that the measures of the variables were understood and the content was coded consistently, extra caution needs to be taken with the findings and conclusions because intercoder reliability only used content from two of the five television outlets. Further studies could look into a longer period of the coverage and the impact of other important variables on the coverage, such as how reporter-source relationships and different reporting modes affect coverage frames. Social utility is one of the major media functions in a crisis. Future studies could examine how media perform the social utility function under various crisis situations. Human interest is considered a major aspect of news value and a key frame of news coverage involving human activity. Further examination of the factors framing human interest and the relationship between human interest and other frames in the coverage may provide more insights on how media perform their functions in a national crisis.

References

Andsager, J. L., & Powers, A. (1999). Social or economic concerns: How news and women's magazines framed breast cancer in the 1990s. Journalism & Mass Communication Quarterly, 76, 531-550.

Bigman, S. K. (1948). Rivals in conformity: A study of two competing dailies. Journalism Quarterly, 25, 127-132.

Chyi, H. I., & McCombs, M. (2004). Media salience and the process of framing: Coverage of the Columbine school shootings. Journalism & Mass Communication Quarterly, 81, 22-35.

Colby, D. C., & Cook, T. E. (1991). Epidemics and agendas: The politics of nightly news coverage of AIDS. The Journal of Health Politics, Policy and Law, 16, 215-249.

Dominick, J. R. (1996). The dynamics of mass communication (5th ed.). New York: McGraw-Hill.

Entman, R. M. (1991). Framing U.S. coverage of international news: Contrasts in narratives of the KAL and Iran air incidents. Journal of Communication, 41(3), 6-27.

Entman, R. M. (1993). Framing: Toward clarification of a fractured paradigm. Journal of Communication, 43(4), 51-58.

Fico, F., & Freedman, E. (2001). Setting the news story agenda: Candidates and commentators in news coverage of a governor's race. Journalism and Mass Communication Quarterly, 78, 437-449.

Gamson, W. A., & Modigliani, A. (1987). The changing culture of affirmative action. In R. D. Braungart (Ed.), Research in Political Sociology (pp. 137-177). Greenwich, CT: JAI Press.

Gilens, M. (1999). Why Americans hate welfare: Race, media, and the politics of antipoverty policy. Chicago: University of Chicago Press.

Graber, D. A. (1980). Mass media and American politics. Washington, DC: Congressional Quarterly Press.

Hertog, J. K. (2000). Elite press coverage of the 1986 U.S.-Libya conflict: A case study of tactical and strategic critique. Journalism & Mass Communication Quarterly, 77, 612-627.

Husselbee, L. P., & Stempel, G. H. (1997). Contrast in U.S. media coverage of two major Canadian elections. Journalism & Mass Communication Quarterly, 74, 591-601.

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

Iyengar, S., & Simon, A. (1993). News coverage of the Gulf crisis and public opinion. Communication Research, 20, 365.

Jacobs, L. R., & Shapiro, R. Y. (2000). Politicians don't pander: Political manipulation and the loss of democratic responsiveness. Chicago: University of Chicago Press.

King, P. T. (1997). The press, candidate images, and voter perceptions. In M. E. McCombs, D. L. Shaw, & D. H. Weaver (Eds.), Communication and democracy: Exploring the intellectual frontiers in agenda-setting theory (pp. 29-40). Mahwah, NJ: Lawrence Erlbaum Associates, Inc.

Larson, J. F. (1984). Television's window on the world: International affairs coverage on the U.S. networks. Norwood, NJ: Ablex Pub. Corp.

Lasorsa, D. L., & Reese, S. D. (1990). News source use in the crash of 1987: A study of four national media. Journalism Quarterly, 67, 60-71.

Lawrence, R. G. (2000). The politics of force: Media and the construction of police brutality. Berkeley: University of California Press.

Liebler, C. M., & Bendix, J. (1996). Old-growth forests on network news: News sources and the framing of an environmental controversy. Journalism & Mass Communication Quarterly, 73, 53-65.

Lowery, S., & DeFleur, M. L. (1995). Milestones in mass communication research: Media effects (3rd ed.). White Plains, NY: Longman Publishers USA.

Mogensen, K., Lindsay, L., Li, X., Perkins, J., & Beardsley, M. (2002). How television news covered the crises: The content of CNN, CBS, ABC, NBC and Fox News on September 11. In B. S. Greenberg (Ed.), Communication and terrorism (pp. 101-120). New York: Hampton Press.

Muschert, G. W., & Carr, D. (2006). Media salience and frame changing across events: Coverage of nine school shootings, 1997-2001. Journalism & Mass Communication Quarterly, 83, 747-766.

Nacos, B. L. (1994). Terrorism and the media: From the Iran hostage crisis to the World Trade Center bombing. New York: Columbia University Press.

National Research Council Committee on Disasters and the Mass Media. (1980). Disasters and the mass media. Proceedings of the Committee on Disasters and the Mass Media Workshop. Washington, DC: National Academy of Sciences.

Neal, A. G. (1998). National trauma and collective memory: Major events in the American century. Armonk, NY: M. E. Sharpe.

Nimmo, D. D., & Combs, J. E. (1985). Nightly horrors: Crisis coverage by television network news. Knoxville: University of Tennessee Press.

Norris, P. (1995). The restless searchlight: Network news framing of the post-cold war world. Political Communication, 12, 357-370.

Pan, Z., & Kosicki, G. M. (1993). Framing analysis: An approach to news discourse. Political Communication, 10(1), 55-75.

Reynolds, A., & Barnett, B. (2003). This just in ... How national TV news handled the breaking "live" coverage of September 11. Journalism & Mass Communication Quarterly, 80(3), 689-703.

Riffe, D., Ellis, B., Rogers, M. K., Van Ommeren, R. L., & Woodman, K. A. (1986). Gatekeeping and the network news mix. Journalism Quarterly, 63, 315-321.

Riffe, D., Lacy, S., & Fico, F. (2005). Analyzing media messages: Using quantitative content analysis in research (2nd ed.). Mahwah, N J: Lawrence Erlbaum Associates, Inc.

Schramm, W. L. (1965). Communication in crisis. In B. S. Greenberg & E. B. Parker (Eds.), The Kennedy assassination and the American public: Social communication in crisis (pp. 1-25). Stanford, CA: Stanford University Press.

Semetko, H. A., & Valkenburg, P. M. (2000). Framing European politics: A content analysis of press and television news. Journal of Communication, 50(2), 93.

Stempel, G., & Hargrove, T. (2002). Media sources of information and attitudes about terrorism. In B. S. Greenberg (Ed.), Communication and terrorism (pp. 17-26). New York: Hampton Press.

Stempel, G. H. (1985). Gatekeeping: The mix of topics and the selection of stories. Journalism Quarterly, 62, 791-815.

Stempel, G. H., & Windhauser, J. W. (1989). Coverage by the prestige press of the 1988 presidential campaign. Journalism Quarterly, 66, 894-919.

Tewksbury, D., Jones, J., Peske, M. W., Raymond, A., & Vig, W. (2000). The interaction of news and advocate frames: Manipulating audience perceptions of a local public policy issue. Journalism & Mass Communication Quarterly, 77, 804-829.

Ungar, S. (1998). Hot crisis and media reassurance: A comparison of emerging diseases and Ebola Zaire. British Journal of Sociology, 49(1), 36.

Valkenburg, P. M., & Semetko, H. A. (1999). The effects of news frames on readers' thoughts and recall. Communication Research, 26, 550-569.

Xigen Li (Ph.D., Michigan State University) is an assistant professor in the School of Journalism at Southern Illinois University Carbondale. His research interests include societal and institutional influence on news content, and news media and the Internet. Table 1 New Source Associated With Coverage Frame (N = 2647) Source Political Economic Criminal Environ. Airline 0.4 0.0 1.6 0.0 Arab Group 0.2 0.0 0.3 0.0 Business 0.0 11.9 0.0 0.0 Expert 5.5 9.0 25.7 0.0 Gov. official 33.8 4.5 21.3 23.1 International 9.1 4.5 5.7 0.0 President 15.0 4.5 3.5 0.0 Relative 0.0 0.0 0.3 0.0 Witness 0.6 0.0 3.2 7.7 Other 8.0 11.9 17.5 7.7 Source Safety HumnInt Religion Disaster Airline 9.3 1.2 0.0 1.0 Arab Group 0.3 0.0 0.0 0.0 Business 0.3 0.6 0.0 0.2 Expert 3.5 2.3 5.0 1.4 Gov. official 19.5 5.2 5.0 16.8 International 0.6 1.7 30.0 0.6 President 2.9 1.2 0.0 7.4 Relative 1.3 18.4 0.0 0.8 Witness 1.6 13.2 0.0 22.4 Other 23.3 27.0 35.0 12.6 Source Other Total Airline 3.0 2.1 Arab Group 0.0 0.1 Business 0.9 0.5 Expert 4.4 5.8 Gov. official 13.5 19.3 International 3.0 3.3 President 2.2 6.7 Relative 1.7 1.9 Witness 5.2 10.8 Other 13.9 14.8 Note: * Percentages reflect how each source was used in the stories where specific frames were identified. Sources were not identified in some of the stories and thus total frequency does not add up to 100%. Table 2 Percentage of Story Primary Orientation by Stages of Crisis (N = 2647) Orientation 8-11 am 11-3 pm 3-7 pm 7-12 am Analysis 17.2 21.3 14.3 12.9 Consolation 0.9 2.8 3.5 2.9 Fact 79.2 73.6 81.3 79.2 Guide 2.2 1.6 0.9 1.0 Other 0.9 0.5 0.0 0.2 Total 100 100 100 100 Orientation 12-6 am 6-9 am Total Analysis 11.8 10.9 14.8 Consolation 0.8 1.4 2.0 Fact 85.7 85.8 81.4 Guide 0.6 0.7 1.2 Other 1.3