Media Reporting of Accidents

Abstract

It has often been argued that the media are too eager to speculate on the causes of major accidents. Rather than defer judgment until the outcome of official investigations, intense public interest persuades editors and journalists to identify and often promote speculative theories about the events leading to major failures. There have been few studies that investigate this media response in any detail. This paper provides one of the first detailed investigations of the media response to the failure of a safety-critical system. The following sections trace the development of causal hypotheses in two national newspapers in the days following the loss of the Air France Concorde, flight AFR4590. The study focuses on the articles and editorials presented by a mass-market tabloid and a lower circulation ‘broadsheet’ newspaper. Our analysis yields a number of insights into the relationship between safety professionals and the media. For instance, the origin of many causal hypotheses can be traced back to misunderstandings that stem from the experts’ inability to convey detailed technical information in a manner that is readily understood by journalists. In other situations it is possible to argue that experts deliberately create ambiguity in their statements to the press. This ambiguity may help to cover their uncertainty about the causes of an accident. When reported by the press, however, it can act to mislead the public who rely on media reports of adverse events.

Introduction

The role of the media in reporting the causes of major accidents has been widely criticised. For example, Leveson and Turner describe how “Stories about the Therac-25 have appeared in trade journals, newspapers, People Magazine, and on television's 20/20 and McNeil/Lehrer News Hour. Unfortunately, the previous accounts of the Therac-25 problems have been oversimplified, with misleading omissions” (ref.1). Such criticisms are also reflected in the ambivalent attitude of many accident investigation agencies towards the media. On the one hand, they are eager to publicize the results of investigations and improvements in safety statistics. On the other hand, many agencies are concerned to minimise the speculation that often emerges in the aftermath of major accidents. In consequence, most of these organisations publish detailed guidance on the material that their employees should release about the interim findings of any investigation. For instance, New Zealand’s Transport Accident Investigation Commission has summarised their policy as follows: “Only the barest factual information is released, because we need to confirm the facts, and to minimise speculation as to cause and contribution. Some witnesses become reluctant to volunteer information if the accident is subject to speculation in the news media. In general, the less publicity during an investigation, the better the information we get. We avoid publishing interim reports on the circumstances of accidents. Occasionally in the past we issued interim reports, containing a brief summary of the facts with the intention of reducing public uncertainty and speculation while the analysis was being carried out. Unfortunately the interim reports did not meet that objective. We will do our best to answer questions and provide updates to the news media and the public about the progress with an investigation, and the investigative process in general, but will avoid speculation on the causes and circumstances of an accident. If a news organisation or journal publishes incorrect information about an investigation or report we will seek corrective action” (ref. 2).

Editorial staff and journalists often feel that they have a duty to inform the public about the causes of major accidents. This concern is justified in terms of the ‘public interest’. For example, it can be important to reassure the public that it is safe to continue using particular modes of transport. Paradoxically, press coverage can also help to reduce speculation by informing the public about the possible causes of high-profile failures. The media also play an important role in ensuring that accidents are properly investigated. The significance of this varies from country to country. For example, the Chinese People’s Daily recently urged the media to take a more active role in publicising major accidents that might otherwise be hidden from the public; “Nowadays, we feel gratified that the Chinese media can face up to accidents. As a real representative of the people's fundamental interests, the Chinese Communist Party and government facing disasters and accidents, the media help to put the security and benefits of the masses in first place and draw lessons from them in order to prevent or mitigate any recurrence. As the mouthpiece of the Party and the people, the media has the duty, the responsibility and the rights to report facts about accidents to the people so as to help them gain positive insights from misfortunes that they may be unwilling to face” (ref. 3). The investigative role of the media is not restricted to the Chinese press. For example, both Le Parisien and the Times of London carried articles criticising the composition of the French Transport Ministry’s investigation team into the loss of Concorde Flight AFR4590. Members of this team had investigated the 1992 crash of an AirInter Airbus in Alsace. Their report focused on the inexperience of the pilots, however, a subsequent court case identified the failure of cockpit instruments as a primary cause in this accident.

Given the key role that the media play in informing the public about the causes of major accidents, it is surprising that so little research has been done into the coverage that they devote to the failure of safety-critical systems. Most studies focus on media bias. For instance, Singer and Endreny review the ways in which print and electronic media present many different hazards to the public (ref. 4). These range from toxic waste and famine to transportation accidents and experimental surgery. They identify a range of criteria that the media use to determine when hazards should be considered ‘risky’. Curtis provides a more focussed example of this work in his investigation of the New York Times’ coverage of major airline accidents between 1978 and 1994 (ref. 5). The New York Times publishes an annual index of stories. This provided the sole source for his research. All fatal events were listed under: Airlines - Accidents and Safety (1978-1982), Airlines (1983), or Airlines and Airplanes (1984-1994). He used this material to argue that the paper devoted disproportionate coverage to 25 fatal airline events involving hijacks sabotage or military action. These events averaged 53 references each. The remaining 160 other fatal events averaged 7.2 references. Curtis went on to argue that the New York Times focused more on events that occurred in the U.S. or that involved U.S. carriers. Accidents were also more likely to be reported as the number of fatalities increased.

In contrast to Curtis’ high-level review, we focus on the press coverage of a singe incident. The following analyse the press response to the loss of the Air France Concorde crash, flight AFR4590. The official enquiry into this accident found that the front right tyre of the left landing gear ran over a strip of metal shortly before rotation during takeoff from Charles de Gaulle Airport (ref. 6). The strip had fallen from another aircraft. Damage to the tyre created debris that was thrown against the wing. The debris ruptured a fuel tank and a major fire broke out under the left wing. Problems appeared on engine 2 and for a brief period on engine 1 but the aircraft took off. The crew shut down engine 2, following an engine fire alarm. They noticed that the landing gear would not retract. The aircraft flew for around a minute but was unable to gain altitude beyond 200 feet or speed beyond 200 knots. Engine 1 lost thrust, the aircraft’s angle of attack and bank increased sharply. The thrust on engines 3 and 4 fell suddenly and the aircraft crashed onto a hotel. Of course, this analysis and the evidence that supports it were not generally available to the journalists and editors that covered the accident as the story unfolded in the week after the accident. Instead, as we shall see, they had to piece together alternate versions of events as information emerged or was ‘leaked’ over this period.

We have chosen to focus on the loss of flight AFR4590 because it typifies the high-profile accidents that elicit considerable interest from the media. This case study also typifies a growing class of safety-critical, legacy systems that were once considered to embody ‘cutting edge’ technology. Rather than base our analysis of the media reaction on the contents of a single newspaper, we have chosen to compare two very different publications: The Times of London and The Sun. This decision is justified by the observation that editorial policy is largely driven by the need to pursue specific markets of readers. The Times is a well-known ‘broadsheet’; it has a large page size and presents an authoritative, ‘in-depth’ analysis of news and current affairs. It has a daily circulation of around 630,000 in August 2002. In contrast, The Sun is published in ‘tabloid’ format. It presents a more varied perspective, mixing news items with a greater proportion of celebrity coverage and current affairs. The wider appeal of the publication is reflected in daily sales of approximately 3,600,000 in August 2002. Both titles are published by the News International organisation. This, in theory, enables us to investigate the use of different techniques in different areas of the newspaper market. We are interested to compare the approach of The Sun and The Times because they serve different areas of the newspaper market. As Tiffen (ref. 7) observes “quality" newspapers emphasise international news, politics, business and major social institutions. Their readers may expect a more informed and sustained analysis than their counterparts in the “popular” press. In contrast, mass-market publications focus on crime, sport, sex and human interest. The presentation techniques used by newspapers also differ. The popular press relies more on large, dramatic headlines and pictures that are intended firstly to attract attention on the new shelves and then to draw the reader into particular stories (ref. 8).

Figure 1 – Page Coverage of AFR4590 by Date

Quantitative Comparisons: Media Interest in the Loss of AFR4590

Figure 1 provides an overview of the coverage in The Sun and The Times in the immediate aftermath of the loss of AFR 4509. It presents the total number of individual pages that contained references to the accident. This calculation is more complex than it might appear. The introduction of flexible production and composition techniques means that there are regional variations even though The Times and The Sun are national newspapers. The page totals in Figure 1 refer to editions that were sold in Scotland in July and August 2000. Although, these pages occur in sections that should be common across the distribution of the paper, it is possible that regional variations may affect these totals in other regions. We have also chosen to focus on the daily editions of these newspapers. Sunday editions have used different editorial teams, have additional pages for more extended coverage and often repeat material published in the dailies. Further problems complicate the quantitative analysis of media coverage following major airline accidents. For example, The Times includes a series of supplements. These are important for our study because one of these inserts focussed on media news whilst a second focussed on travel news. The former included an article on German and French media coverage of the accident. The later included coverage of the economic impact of the crash on British Airways and Air France. Figure 1 includes the coverage in supplements, adding 4 pages to The Times on July 27^th, 2 pages on July 28^th and a single page of coverage from the travel supplement on the 29^th July. It should also be noted that several pages of coverage towards the end of the week in Figure 1 refer not to articles written by the papers’ news staff but to letters that were sent by the public in response to previous coverage. These letters account for a single page of coverage in The Times on August 1^st and 2^nd.

Figure 1 records the total number of pages that included articles devoted to the loss of AFR4590. However, some pages contained very little information about the accident. For instance, page 13 was the only one to contain information about the accident in The Sun published on the 28^th July. The total area of text devoted on that page was approximately 157 cm^2.This is indistinguishable in Figure 1 from page 7 of The Times, which on the same day contained approximately 524 cm²of text at a smaller point size. Table 1, therefore, presents a more detailed breakdown of the area devoted to the accident in each of the papers over the week following the crash. The approximate total area in The Times’ broadsheet is 1,855 cm²and 945cm²for The Sun’s tabloid format. The Times devoted a smaller area to the accident than The Sun on the day after the incident. It might be argued that this reflects editorial policy to delay comment until further evidence is available. As we shall see, however, this argument is hard to sustain. Table 1 also illustrates a rapid ‘tail off’ in coverage in the aftermath of the accident. The readers’ letters explain the continuing interest in The Times on the 1^st and 2^nd August.

	26^th July		27th July		28th July		29th July		31st July		1^st August		2^nd August
	Sun	Times	Sun	Times	Sun	Times	Sun	Times	Sun	Times	Sun	Times	Sun	Times
Text	1718	1829	760	1934	157	1196	0	444	0	154	0	452	0	194
Images	5893	4000	1571	2146	144	1262	0	661	0	0	0	0	0	0
Headlines	2026	480	586	637	123	334	0	108	0	48	0	59	0	34

Table 1- Area Devoted to AFR4590 Excluding Supplements (cm²)

As mentioned in the introduction, Tiffen argues that ‘popular’ newspapers make greater use of images and headlines to sustain their readers’ attention than their ‘quality’ counterparts (ref. 7). This is most obvious in The Sun’s coverage for the 29^th July. A range of video and still footage was taken during the last moments of the flight. The Sun’s use of these images, arguably, reflects the papers’ format. However, it is important not to over simplify. Table 1 also shows that The Times made extensive use of this material in the immediate aftermath of the accident. In particular, Table 1 illustrates important changes in the proportion of images to text in the days immediately following the loss of AFR4590. The proportion of text devoted to the incident increases from 30% on the 26^th in the Times to approximately 40% on the 27^th. The proportion of images devoted to the incident falls from approximately 60% on the 26^th in the to 45% on the 27^th. However, the extensive use of images in both the ‘quality’ and ‘popular’ press can be explained in different ways. For instance, it might be argued that The Times’ use of images helps to ensure an adequate coverage of the incident without requiring excessive speculation about the causes. As more information becomes available, for instance on the 27^th and 28^th July, the need for images declines and the proportion of prose rises. Although these statistics illustrate significant changes in media interest during the week following the accident, they provide few insights into media arguments about the causes of the crash. The following section, therefore, presents a more qualitative analysis of the different hypotheses that were put forward in the days following the loss of AFR4590.

Qualitative Comparisons: Media Analysis of the Causes of the Loss of AFR4590

The official enquiry into the accident argued that a piece of metal fell from another aircraft onto the runway. This damaged Concorde’s tyres relatively late in the take-off sequence. Debris from the tyres, in turn, damaged the fuel tank and led to the fire (ref. 6). This information was not, however, available to the editors and journalists on the 25^th July. Instead, attention focussed on previous reports about cracks in the wings of the Concorde fleet. In the days following the crash, the papers revised their account as experts dismissed the cracks as a possible cause of the accident. Alternative causal hypotheses were presented to the public. These ranged from age-related issues, including the possibility of metal fatigue, through to fan-blade separation within the engine or problems involving the maintenance of a thrust reverser immediately prior to take off. Table 2 provides an overview of how these different hypotheses appeared in the two papers in the week following the accident.

	July 26th	July 27th	July 28th	July 29th	July 31st	August 1st	August 2nd
The Sun
Cracks in the wings	ü	ü
Age related issues	ü
Fan blade separation	ü		ü
Uncontrolled fuel release	ü		ü
Thrust reverser failure		ü

The Times
Cracks in the wings	ü
Engine fire	ü
Fan blade separation	ü	ü
Engine fire control	ü
Fractured fuel tank	ü	ü			ü	ü	ü
Hydraulic control failure	ü
Terrorism	ü
Human error	ü
Tyre blow-out	ü		ü		ü	ü	ü
Age-related issues	ü	ü	ü
Thrust reverser failure		ü	ü
Bird strike			ü
Fuel line failure			ü		ü
Maintenance staffing			ü
Runway surveillance					ü	ü
Reheater ignites fuel						ü

Table 2- Potential Causal Factors by Date Discussed

Table 2 illustrates a paradox in the coverage of the Concorde accident. The tabloid newspaper contains less speculation about the causes of the incident than its sister broadsheet. For example, The Times presented ten different causal hypotheses on the 26^th July while The Sun only published four. These differences are partly the result of differing editorial policies associated with the markets that these papers serve. A limited flow of information was provided by the official investigations in the immediate aftermath of the accident. The paper’s readership expects detailed coverage of such major events. It can, therefore, be argued that The Times had to rely on speculation in order to sustain its analysis. The coverage in The Times also reflects the uncertainty that existed in the aftermath of the accident. In contrast, the more limited coverage in The Sun offered less scope for speculation.

Table 2 illustrates further differences in the coverage of this accident. The Sun focuses on fan-blade separation as a potential cause of the engine damage and fuel leak that led to the loss of AF 4590. In contrast, the wider range of potential causes in The Times gradually begins to focus on the role played by a tire blowout, the fracturing of a fuel tank and ignition from the afterburners. The identification of these causal factors within a week of the incident either illustrates the exchange of information between the official investigations and the media or the fortuitous results of informed speculation. It is also important to emphasise that individuals with appropriate skills and experience can also make prescient statements even if they are not part of an official investigation team. The role of the afterburners was suggested in a letter to The Times from a fast-jet pilot in the RAF. Their comments long pre-dated the official report that could not determine whether fuel ignition had occurred from an arc triggered by a short-circuit in an electric harness close to the main landing gear or by fuel contact with sections of the reheating subsystem.

Figure 2 presents a Conclusion, Analysis, Evidence (CAE) diagram including direct quotations from The Sun about the role that cracks in the wings might have played in the course of the accident. CAE diagrams provide a means of representing and reasoning about the arguments that are made in the aftermath of accidents and incidents (ref. 9). The conclusion that cracks in the wing played a role in the accident is supported by a series of arguments that are grouped in the box at the top of Figure 2. Arguments that weaken or contradict this view are collected within the dotted box in the lower part of the diagram. Evidence used in these different causal arguments is presented on the right of Figure 2.

This more detailed analysis of the arguments made in The Sun show that, in contrast to many analysts’ expectations, the tabloid provides a relatively balanced view of this potential cause. Comments such as “the disaster came just a day after The Sun revealed cracks found in wings had forced British Airways to ground one of its seven Concordes” and balanced by comments from the Air France president who “denied the cracks exposed in The Sun were to blame for the horror – although investigators will not be ruling anything out”. Similarly, Joan Collins’ anxiety over news about the cracks are balanced by the BA’s chief pilot who is reported to observe that they are ‘non safety related’ and cause him ‘no concern’.

Figure 2 – Arguments Relating to the Presence of Cracks in the Wings, The Sun, July 26th

Figure 3 illustrates the arguments published in The Times on the 26^th July. As can be seen, the CAE diagram captures the more detailed analysis presented by the broadsheet. However, the volume of prose devoted to the analysis is support by similar evidence to that published in the tabloid Sun. This again illustrates the dilemma that faces journalists who cover accidents and incidents for ‘quality’ newspapers. Their readers expect a more sustained analysis even though the staff must rely on the same information that is available to their colleagues on ‘popular’ titles. The development of these CAE diagrams raised a number of problems in identifying different strands of argument about the potential causes of an accident. For example, Figure 3 contains arguments that directly reject the cracks in the aircrafts’ wings as a cause of the accident. On page seventeen it is stated “no evidence yet points to a connection with the recent admission that hair-cracks have developed on all seven of the British Airways Concorde fleet”. Other arguments do not directly contradict this version of events but instead offer competing hypotheses. For example, a former Concorde test pilot is reported as saying that a “turbine may have disintegrated or spun out of the engines”. It is possible to refine the CAE diagrams to capture some of these distinctions. For example, Figure 3 uses a heavier outline for arguments that directly contradict the role of the wing problems in the accident. However, the subtle differences in these different forms of argument also illustrate the ambiguity that can be created in the reader’s mind as they read the newspaper accounts of the Concorde crash. Although these accounts speculate about the causes of the incident, they typically express the speculation in terms of direct quotes from safety professionals. They also offer alternative accounts that illustrate the uncertainty over these expert opinions. It is important to remember these insights when we condemn media reports about the causes of accidents or incidents. It is also important to reiterate that this analysis must be formed in the immediate aftermath of the accident when information is scarce and rumour is in plentiful supply.

Figure 3 – Arguments Relating to the Presence of Cracks in the Wings, The Times, July 26^th

Conclusions

This paper has analysed part of the media response to the loss of Concorde flight AF 4590. We have focused on articles published in a tabloid, The Sun, and a broadsheet newspaper, The Times of London. There have been very few previous studies of this type. Our results confirm some of the criticisms but challenge other assumptions that safety professionals have made about the media reporting of incidents and accidents (ref. 9). In particular, we have noted the way in which an initial, high level of interest rapidly wanes as other new items attract the attention of editors and journalists. We have also seen the way in which analytical writing and photojournalism help to shape the media response to such major incidents. In the immediate aftermath of this accident, editors relied heavily on images to provide readers with information about the incident. In the following days, readers were already familiar with these images and more information became available about the potential causes. In consequence, the proportion of prose analysis of the incident increased.

A paradoxical finding in our work has been that the broadsheet account contains more speculation than the tabloid. We have argued that this is the result of a pressure to inform the readership about potential causes when little ‘hard’ information is available. Journalists seem to be aware of their dilemma and so speculation is, typically, presented in the form of direct quotes from experts and eyewitnesses. Finally, our work on this case study has revealed that the media quickly focused on the causes that were ultimately identified in the official report. This may reflect the unusual openness that characterised the early stages of the official investigation into this accident. However, this prescience forms a subject of continuing research.

A number of caveats must be made about this study. Firstly, we have only considered the media reaction in the week immediately following the accident. Further work is needed to analyse the subsequent reporting of the loss of AF 4590. Secondly, we focused on UK newspapers. The nature of aviation accidents often creates media interest in several different countries. In this accident, most of the victims were German. The aircraft was operated by a French company and crashed outside Paris. We are currently conducting a comparative study of the media reporting in these different countries. Thirdly, this paper has focused on the print media. More work is required to trace the causal analysis provided by broadcast services and Internet news agencies. Fortunately, the growth of publicly accessible digital archives has supported our work in this area. In particular, we would like to determine whether readers assign greater trust to the independence of broadcast media in the reporting of accidents as they do in the reporting of other news items (ref. 8). Having raised these caveats it is important to reiterate our central argument that unless we study the media reaction to major accidents then there is little chance we will ever understand their oversimplifications and ‘misleading omissions’ (ref. 1).

References

1. N. Leveson and C.S. Turner, An Investigation of the Therac-25 Accidents, IEEE Computer, Vol. 26, No. 7, July 1993, pp. 18-41.

2. New Zealand Transport Accident Investigation Commission, Publicity and media relations, Bulletin, May 2000. See http://www.taic.org.nz/bulletins/2000may.html

3. The Media have a Duty to Uncover Accidents: Opinion Piece, Chinese People's Daily, 08/27/2001. Available in translation http://www.china.com.cn/english/2001/Aug/18230.htm

4. E. Singer and P.M. Endreny, Reporting on Risk: How the Mass Media Portray Accidents, Diseases, Disasters, and Other Hazards, Russell Sage Foundation; May 1993.

5. T. Curtis, Airline Accidents and Media Bias: New York Times 1978-1994. Available from http://www.airsafe.com/nyt_bias.htm

6. Bureau d'Enquêtes et d'Analyses pour la Sécurité de l'Aviation Civile (BEA), Accident on 25 July 2000 at La Patte d.Oie in Gonesse (95) to the Concorde registered F-BTSC operated by Air France, Report f-sc000725a. http://www.bea-fr.org/anglaise/actualite/concorde-en.htm.

7. R. Tiffen, News and Power, Allen and Unwin, London, 1989.

8. Christian, H. (ed) The Sociology of Journalism and the Press, University of Keele Press, UK, 1980.

8. C.W. Johnson, The Failure of Safety-Critical Systems: A Handbook of Accident and Incident Reporting, Springer Verlag, London, in press and to appear 2003.

Biography

Prof. C.W. Johnson, MA, MSc, DPhil, CEng, Glasgow Accident Analysis Group, Dept. of Computing Science, University of Glasgow, Glasgow, G12 9QQ, Scotland, UK, telephone - +44 141 330 6053, facsimile - +44 141 330 4913, e-mail – johnson@dcs.gla.ac.uk, URL - http://www.dcs.gla.ac.uk/~johnson

Chris Johnson is Professor of Computing Science at the University of Glasgow. He heads one of the largest research teams specifically devoted to new generations of accident analysis techniques. He helped to author European guidelines for mishap reporting in Air Traffic Management. He has developed an incident analysis scheme for the UK Health and Safety Executive that is specifically designed to support the investigation of adverse events involving programmable systems across the process industries. In 2002, he held a NASA/ICASE fellowship analysing a series of mishaps, including the SOHO mission interruption.