Abstract
Numerous international studies have considered the impact of media reporting of suicide, specifically examining whether news reports of suicide can lead to imitation. Pirkis and Blood systematically reviewed 42 such studies, and concluded that the body of evidence suggested a causal association. Few of these studies were conducted in Australia [1, 2].
Less attention has been given to the questions of the extent and nature of media reporting of suicide. Pirkis and Blood examined the literature on these questions [1, 2], and found the extent of reporting to be highly variable, and the nature of such reporting to be sensationalist, over-emphasizing celebrity suicides, suicides by unusual or violent methods, and suicides by young people [3–11]. Few of these studies considered quality of media reporting in any systematic way, and, once again, few were conducted in Australia.
The media monitoring project was commissioned by the Mental Health and Special Programs Branch (MHSPB) of the Commonwealth Department of Health and Aged Care to address gaps in knowledge with respect to the extent, nature and quality of reporting of suicide by the Australian media. The primary aim of the media monitoring project was to establish a baseline picture of how the Australian media reports suicide, with a view to informing future strategies intended to optimize reporting of suicide.
The media monitoring project was closely aligned to a second media initiative of the MHSPB. Prior to the start of the media monitoring project, a kit known as Achieving the Balance [12] had been developed and launched, with the aim of promoting awareness of suicide-related issues among media professionals.
Method
The media monitoring project collected media items over 12 months, beginning on 1 March 2000.
Item retrieval
A list of suicide-related search terms was provided to Media Monitors, a media retrieval service that identified relevant newspaper, radio and television items for the project. Trained readers, viewers and listeners scanned selected media, retrieving items on a daily basis.
Media Monitors scanned all national metropolitan daily newspapers and all Victorian suburban and regionals, and worked through the entire paper. There was coverage of all national television and radio networks, but the retrieval was restricted to news and current affairs programmes.
Newspaper items were provided to the project team as complete press clippings. Radio and television items were provided as broadcast summaries. The broadcast summaries provided information about the source of the item, and provided a précis of its content. For a sample of radio and television items, full transcripts or audio/video tapes were requested in order that quality ratings could be made (see below).
Data extraction
Three trained coders extracted identifying and descriptive information for each item, and made quality ratings, entering these data into an Access database.
Identifying information
Identifying information was extracted for all items from press clippings and broadcast summaries, and included details such as the media type (that is, newspaper, television or radio).
Descriptive information
Descriptive information was extracted from press clippings and broadcast summaries for all items, and included, for example, details on the content of the story.
Quality ratings
Quality ratings were made for a randomly selected 10% of items (stratified by newspaper, television and radio). Specifically, a purposedesigned computer program was written that flagged randomly selected items for quality rating at the point at which the initial identifying information was entered. The press clippings provided sufficient information for these quality ratings to be made for newspaper items, but the broadcast summaries alone did not allow ratings of quality to be made with confidence for television and radio items. For this reason, full transcripts or audio/video tapes were sought for those broadcast items for which quality ratings were to be made.
The quality of suicide items was rated on a set of nine dimensions that operationalized criteria in Achieving the Balance.
Data analysis
Simple descriptive statistics were calculated to provide a baseline picture of how the media reports suicide.
Results
Extent and nature of media reporting of suicide
In total, 4813 items on suicide were retrieved: 1162 (24.1%) were newspaper items; 608 (12.6%) were television items; and 3043 (63.2%) were radio items.
Media characteristics
Table 1 provides a profile of the items by media characteristics. Newspaper items typically appeared in the body of metropolitan papers, were prepared by the paper's own reporters, and took the form of news items. The majority of television items were short segments appearing on non-government funded stations during the evening news. Radio items also tended to be short and be broadcast on nongovernment funded stations, but more commonly occurred in the morning. As with the television items, most were broadcast on news programmes, but a greater proportion occurred on current affairs programmes, particularly shows with a talk-back format. No data were available on the reach of these media in terms of readership or audience numbers.
Table 1. Profile of items
Focus
Table 2 describes the focus of the items, reflecting the fact that suicidal behaviours are frequently considered in a hierarchy from suicidal ideation (‘thoughts of suicide, which can vary from transient notions about life being meaningless to intense preoccupation with taking one's own life’ [13]) through attempted suicide (‘a selfinflicted, nonhabitual act that has the potential to be fatal, but in actuality is not’ [14, 15]) to completed suicide (‘death that is the result of an act perpetrated by the victim, with the intention of achieving this outcome’ [16]). For all media types, the majority of items were about completed suicide only (60.8% of newspaper items, 68.7% of television items and 66.2% of radio items).
Table 2. Item characteristics
Content
Table 2 also provides a breakdown of the content of the items on suicide, allowing for multiple responses. Irrespective of media type, items most commonly described a particular instance in which an individual had completed or attempted suicide, or had experienced suicidal ideation (41.7% of newspaper items, 38.5% of television items and 31.4% of radio items). Items describing murder-suicides or mass suicide could be regarded as subsets of the stories describing an individual's experience of suicidality, since they are distinguished only by the fact that the suicidal individual took the lives (or tried to take the lives) of others (in the case of murder-suicide) or shared the suicidal act with companions (in the case of mass suicide). Television items of this nature were relatively common (11.5% and 15.6%, respectively), radio items less common (8.1% for both), and newspaper items still less common (1.6% and 1.7%, respectively).
Common across all media were items describing policy or programme initiatives. These were defined broadly, and ranged from national level policy to local events aimed at suicide prevention. Onequarter to one-third of all newspaper, radio and television items included content about policy or programme initiatives.
Many newspaper items (23.4%) provided a statistical overview of completed or attempted suicide in the population. In some cases, this was the sole focus of the item. In other cases, reference was made to suicide statistics in passing, as background to a story on, say, why a particular policy or programme had been initiated. Items providing a statistical overview of suicide in the population were less common on radio (17.3%), and still less common on television (9.0%).
Geographical reference point
Table 2 also presents information on the geographical reference point of the item. The majority of items (92.5% of newspaper items, 72.9% of television items and 85.4% of radio items) referred to a story occurring in the Australian context.
At-risk groups
Consideration was given to whether particular demographic groups known to be at heightened risk of suicide featured prominently. It should be noted that an individual item might make reference to more than one demographic group, and that many items did not specifically make reference to any group. As Table 2 shows, several demographic groups stood out in terms of their frequency of presentation. Males and young people commonly featured in suicide stories, as did people living in rural areas. So too did psychiatric patients and people in custody. These patterns held across all media types, although television items were less likely than items presented via the other two media to make specific reference to at-risk groups.
Method
Table 2 also shows the different methods referred to in the suicide items. It should be noted that some items referred to more than one method, and others did not refer to any method. Reference to a specific method was more common in newspaper items than items presented on television or radio. Hanging as a method featured in 11.2% of newspaper items. Also of note is the relatively high proportion of items that fell into the ‘other’ category, not only for newspaper items (6.2%), but also for television items (9.2%) and radio items (3.6%). These were varied, but included cutting/stabbing/slashing, drowning and selfimmolation.
Quality of media reporting of suicide
In total, 504 suicide items (10.5% of all items) were randomly selected to be rated on the nine dimensions of quality. It was possible to rate 415 (82.3%) of these (75.4% of television items, 78.0% of radio items and 96.8% of newspaper items). The rated items were not split by media type in the following analyses, on the grounds that the observations would have been too few for meaningful analyses.
Quality of reporting on individual dimensions
Table 3 shows the quality of reporting on the nine dimensions. It should be noted that ratings of quality were more readily made on some dimensions than others, so the n does not always equal 415.
Table 3. Quality of reporting on individual dimensions
• phrases that suggested that completed suicide was a desirable outcome (for example, ‘unsuccessful suicide attempt’ ‘failed suicide attempt’, ‘botched suicide pact’);
• sensationalist terminology describing suicide rates (for example, ‘suicide epidemic’, ‘suicide rates are out of control’, ‘distressing new trend’, ‘suicide is at an all-time high’); and
•inappropriate terminology used to describe the mental health status of those who completed or attempted suicide (for example, ‘a crazed Australian tried to kill himself’, ‘basket case’, ‘a suicidal loner … is disturbed’, ‘he went mad and shot himself’).
All 415 items were rated on this dimension. The majority (83.1%) were not inappropriately located.
This dimension was only relevant to newspaper items. The majority of the 122 items rated (70.5%) did not use the word ‘suicide’ in the headline.
The number of items rated on this dimension was limited by its relevance to items about an individual's experiences (including items on murder-suicide and mass suicide), as well as by the fact that it did not make sense to ask the question of radio items. The majority of the 96 items rated on this dimension adhered to the above suggestion, with only 13.5% including a photograph, diagram or footage related to the suicide.
This dimension was generally only considered relevant in the case of items about an individual's experiences (including items on murdersuicide and mass suicide), although additional items were sometimes included (for example, items presenting statistical overviews of method-specific suicide rates). Items performed less well against this dimension of quality than against some other dimensions. In 50% of cases, the method of self-harm was described in detail.
There were relatively few stories about celebrity suicides, limiting the utility of this dimension to 34 items. However, in the majority of these (91.2%), reference was actively made to the fact that the person who died by suicide or attempted suicide was a celebrity.
Arguably, this dimension required the most subjective judgements on the part of the coders. In total, they felt this dimension was applicable to 302 items. In around half of these items the opportunity was taken to increase mental health literacy, but in the other half, this opportunity was missed.
All 415 items were rated on this dimension. Only in a minority of items (6.5%) was information on help services provided. Not infrequently this was only a brief mention.
Assessment on this dimension required the item to be about an individual's experiences, including murder-suicide and mass suicide. In the majority of rated items (82.0%), the bereaved were not interviewed.
The TQS was calculated by summing all of the actual scores on those dimensions for which a ‘yes’ or ‘no’ response was available, dividing this by the sum of all of the potential scores on those dimensions for which a ‘yes’ or ‘no’ response was available, and converting the result to a percentage. A high score was indicative of good quality.
Calculating TQSs in this way yielded a picture of variability across items. The scores ranged from 0% (scored by five items) to 100% (scored by eight items). The median was 57.1%.
Discussion
Interpretation of findings
The media monitoring project found that reporting of suicide was extensive, and that the nature of reporting was highly variable in terms of focus and content. Suicide items tended to be about completed suicide (as opposed to attempted suicide or suicidal ideation), and most commonly involved content related to an individual's experiences, policy/programme initiatives and/or statistical overviews of suicidality in the population. There were some differences across media types with regard to item content. For example, newspaper articles were more likely to present a statistical overview than were television or radio items, possibly because articles of this type are best supported by visual information that can be taken in at the chosen pace of the audience.
It is of interest to consider this ‘mass mediated reality’ in the light of the ‘official reality’. Fishman and Weimann coined these terms to describe the situation portrayed by the media and the actual situation, respectively [11]. In some cases, it is fair to say that the mass mediated reality parallels the official reality. For example, young males are known to be at heightened risk of suicide, compared with other demographic groups. The steady increase in annual suicide rates in Australia over time has largely been accounted for by an increase in the rate for males aged 20–24 [17]. Similarly, psychiatric patients are known to be at increased risk of suicide, with rates up to 25 times higher than those for the general population [18]. There are also instances in which the mass mediated reality diverges from the official reality, however. It is known, for example, that the annual rate of completed suicide in Australia is 13 per 100 000 [17]. The annual rates of attempted suicide and suicidal ideation are much higher, at 300 per 100 000 and 2900 per 100 000, respectively [19]. Yet the vast majority of items on suicide concern completed suicide. Similarly, the methods of suicide commonly reported did not always reflect official reality. Hanging was the most commonly reported and is the most common method used, but reference was frequently made to quite rare methods, such as self-immolation. Issues such as newsworthiness and public interest would appear to have a bearing on the extent to which suicide items reflect the official reality.
Where suicide items could be rated for quality on a given dimension, they showed considerable variability across dimensions. The majority of suicide items did not have examples of inappropriate language, were not inappropriately located, did not use the word ‘suicide’ in the headline, and did not use explicit photographs/diagrams or footage. However, around half of the suicide items provided a detailed discussion of the method of selfharm and portrayed suicide as merely a social phenomenon. Where items concerned the suicide of a celebrity, reference was commonly made to that person's celebrity status. Most failed to provide information on help services. There is scope for media professionals to concentrate on these facets of reporting of suicide.
When the quality ratings for each item were combined into a TQS, the results were also variable, with some very high quality items, some very poor quality items and a broad range in between. These findings run contrary to those of other studies that tend to suggest that media reporting of suicide is overwhelmingly sensationalist [3–11]. These differences might be accounted for by differences in location, since the majority of the above studies were conducted overseas. Alternatively, they might be explained by differences in sampling or methodology. The current study was greater in scope and more comprehensive than previous studies.
Formal examination of predictors of quality was beyond the scope of this paper, but the impressions of the coders are worthy of note. Their consensus opinion was that content was the greatest predictor of quality, irrespective of factors such as media type and funding source. In particular, items that provided a statistical overview of completed suicide in the population, described murder-suicides or mass suicides, or presented legal issues associated with suicide tended to sensationalize the issue. By contrast, and perhaps contrary to prior expectation, items about an individual's suicide tended to treat the issue with sensititivity.
Study limitations
The media monitoring project was the largest of its kind in Australia, and possibly internationally. Considerable effort was made to ensure methodological rigour, but some limitations must be acknowledged.
It was necessary to restrict the study to three relatively traditional media. Ideally, it would have been desirable to include media like magazines and the Internet. Even within the three media types under study, some inclusion/ exclusion criteria had to be used for reasons of economy, raising issues of generalisability.
The study relied on Media Monitors to retrieve all items. Media Monitors is an established media retrieval service, whose experienced readers, viewers and listeners were carefully briefed as to the scope of the task, were provided with a consistent list of terms relating to suicide, and retrieved material in a timely fashion. Media Monitors performed well in ad hoc checks designed to ensure comprehensiveness. Despite this, it is possible that some items were missed.
In assessing the extent of media reporting of suicide, it is acknowledged that the figures are absolute and not relative. It was beyond the scope of the study to consider suicide items as a proportion of all items.
The identifying and descriptive data, and the quality ratings, all relied on coders' interpretation of media items. Efforts were made to ensure consistency between coders, in that coders were trained in a uniform manner and regular meetings were held to discuss issues of interpretation. Despite this, it is possible that different coders may have interpreted certain information differently.
Ten per cent of all items were randomly selected to be rated for quality, but it was only possible to rate about 80% of these. Although this figure is reasonably high, it should be noted that the items that were unavailable were likely to be of a particular kind. In the main, they were television and radio items broadcast on smaller stations. They were unable to be rated because it was not possible to retrieve full transcripts or audio/video tapes of the items. This differential effect may have impacted upon the findings.
In some instances, it was difficult to make judgements regarding quality because information was only available regarding the end product – the item itself – and not the processes that led to it. For example, in responding to the dimension used to assess suicide items that asked ‘Are the bereaved interviewed?’, the coders could not determine whether their privacy was respected. This may have resulted in overestimates of poor quality on some dimensions.
Future research
It would be useful to take the current data on the extent, nature and quality of media reporting of suicide, and consider its impact. This would involve considering whether poorer quality suicide items were associated with short-term increases in completed and attempted suicide rates in the area in which the item was printed or broadcast, and, conversely, whether better quality suicide items were associated with short-term decreases in rates. Although studies of this kind have been conducted (see review by Pirkis and Blood [1, 2]), few have occurred in Australia. None have been conducted on this scale, and none have taken into account the quality of the item.
Conclusions
The media monitoring project has shown that reporting of suicide is extensive across all media types. There is considerable variability in terms of the nature of suicide items. There is also considerable variability in terms of the quality of suicide items. In general good items outnumber poorer items. However, there are still opportunities for improving media reporting of suicide.
Footnotes
Acknowledgements
The media monitoring project was made possible by funding from the Mental Health and Special Programs Branch of the Commonwealth Department of Health and Aged Care. The authors wish to acknowledge the input of members of the Media Reference Group throughout the project.