As you would expect, in a time-constrained lecture it is impossible to address all the data and trends identified by this research. An in-depth review of the findings is due to be published in spring 2011 - for those of you who want to learn more.
In the course of this evening’s presentation I will first of all provide some context for mental illness and the media – and why research is necessary.
Next I will layout the specific objectives and methodology of my research before going through the key findings and what conclusions we might draw from these.
I will also speak briefly about the subsidiary research we did from 1900 and 1950 – an illuminating and fascinating exercise in its own right I have to say.
Reporting of mental health has been shown by research in a number of countries to perpetuate stigma and negative attitudes towards people with a mental illness. So too have depictions of mental illness in drama and entertainment.
It has been suggested strongly that stigma – defined in the dictionary as ‘a mark of shame, infamy, or degradation, with the term today signifying more of a psychological ‘mark’ than a physical ‘brand’ - can seriously affect many individuals’ quality of life and impede help-seeking and recovery.
In all there is great potential for what is published in the media to affect the lives of people with mental illness, often quite negatively.
It is hard enough to have a mental disorder, but there is convincing evidence that stigma and shame yield greater distress and limitations, beyond the symptoms and impairments themselves. One report from the US Surgeon General listed stigma as one of the foremost obstacles to improved mental health care. The report noted: “Stigma tragically deprives people of their dignity and interferes with the full participation in society”.
Bearing in mind this evidence and the fact that at least a quarter of people will experience some mental health difficulty in their lifetime, then the potential consequences of stigma are vast, damaging, and enduring.
The seriousness of the problem has been widely recognized not least with the advent of nationally coordinated anti-stigma campaigns such as Time to Change in Britain.
In terms of the news media, and its role in reporting and discussing mental illness and suicide, it is worth bearing a few things in mind. The past 15 years has seen news go truly global. Before the internet there was CNN and the BBC world service, but these did not have the ‘mass’ reach, instant publishing, or social networking capacity we are so familiar with now.
Now people regularly read newspapers and access all kinds of sources of reporting, blogging, podcasting and much more from multiple countries. This creates both opportunities and dangers. The potential is there for good coverage – and for the voices of people with a mental illness to reach more people. Indeed technology has also produced brand new platforms for people with mental illnesses to circumvent mainstream media and take their message directly to the audience such as podcasting.
However, for all the possibilities out there – when it comes to mainstream coverage - new technologies also mean poor coverage can spread at a rapid speed - and be accessed repeatedly.
No more dusty archives for questionable coverage such as this to be buried in. A recent example might be this Sun cover story earlier this year when Derrick Bird murdered a number of people in Cumbria.
Set aside for the moment that Bird did not have any prior diagnosis of a mental illness – but taken on its own this is a headline that caused many campaigners and people living with a diagnosis to complain. Here again, they said, was yet another example of an unfounded link being made by the press about violence and mental illness.
But here also was a headline that - thanks to the internet - would go global.
Technological developments make it all the more important that ongoing comparative research on coverage of mental illness is conducted cross-nationally.
This research project was carried out as a first step in that direction and to ascertain if indeed more research is warranted. Once you have heard the findings, I think you’ll agree there are grounds for further investigation.
But why does a journalist want to conduct a research project of this kind? The simple answer is that I was curious. The more detailed answer is that I was curious about some things in particular.
As many of you in this room know, there has been much excellent research conducted in Britain and elsewhere on how the media reports mental illness - and on the potential ramifications of such reporting. In the case of mental illness generally, stigma has been a primary focus. In the case of suicide, much of the concern has been around the impact reporting might have on, for example, copycat suicides.
The thing I really wanted to know however was not just how mental illness is written about – but how have things changed – if indeed at all. I also wanted to look beyond national boundaries to find out how we here in Britain compared to other countries. Are we better? Worse? Similar?
So what did we set out to do? What was the purpose of this research?
• What are the differences and similarities in the coverage of mental illness in the mainstream press in the US and the UK?
• Has that coverage altered over time and if so, in what ways?
• To see if there is any insight to be gained from examining much older coverage.
• To evaluate coverage of mental illness in a range of mainstream newspapers over a significant timeframe in the US and the UK.
• To establish if the way coverage is advertised - namely “headlines” - differs from the content of full articles. Are headlines, for example, more inflammatory or sensationalist?
• To chart trends in coverage over time within each country.
• To draw comparisons between the two countries.
• To identify if further research is warranted.
It should be noted that these objectives were tempered by a number of practical considerations – most notably the time available for design, implementation and analysis - and also the resources at my disposal to conduct the research. Important parameters were therefore set.
We wanted a significant time span to work with in order to collect data that would measure trends but that would not impede robust sampling. It would have been truly great to go back 50 years or more in great detail, but practicalities dictated a more recent timeframe for the bulk of the research.
Selecting two countries with a common language, widely understood cultural similarities, and comparable media environments, was important.
Again – as much as I would have liked a larger base of media outlets, limits had to be applied. Newspapers, rather than broadcast, radio or new-media were selected.
The mainstream newspaper sectors in the US and in the UK are undergoing major changes including transformation of coverage spawned by technological leaps forward, falling print circulations, and declining profits.
Still, print newspapers remain (and are likely to remain) widely read while online readerships have grown exponentially. Newspapers also continue to be enormously influential. The Pew Centre Project for Excellence in Journalism News Coverage Index, which measures coverage in the US across a swathe of media on and ongoing basis concludes the following:
“Specific newspapers, such as the New York Times and Washington Post, have an even greater influence on the national and international news agenda because they serve as sources of news that many other outlets look to in making their own programming and editorial decisions. So while the overall audience for newspapers has declined over recent years, newspapers still play a large and consequential role in setting the overall news agenda that cannot be strictly quantified or justified by circulation data alone.”
So – a quick run through our methodology. Again – there will be more detail in the spring report. And, clearly, there are huge needs out there for longer-term historical perspectives, for an examination of many more nations, and for scrutiny of multiple forms of media.
Sample: 9 newspapers were selected across two countries - 5 UK publications and 4 in the US.
Because measuring the “influence” of a given newspaper is inherently problematic, the publications included for the purposes of this research were selected based on circulation, but also on prominence with some account taken of geographic distribution and reach.
Although not strictly representative, the publications selected afforded a reasonable spread of major newspapers in each country.
The research did not seek to compare printed articles with online variations or online-only articles. Rather, the data captured were the final versions of articles and headlines as archived.
All 4 US titles are represented for each year sampled. In the UK sample, robust archive data was not available for some publications every year: namely, The Sun and The London Evening Standard and The Daily Mail.
Data for this study were collected using a combination of proprietary Lexus Nexus search software and newspaper website archives where data via Lexus Nexus were unavailable.
All publications were scanned for any mentions of mental health or related topics and for suicide. Search criteria comprised key words and terms relating to mental health and suicide for both headlines and ‘body’ copy. The text of articles was captured in their entirety – including headlines, sub headlines, body copy and cross-references. Therefore, headlines that did not explicitly include key words were still captured for the purpose of analysis.
To guard against flooding the sample with non-relevant material a number of broad filters were applied. For example, all articles that included the word ‘nuts’ but that were about food-related topics were discarded.
Two tiers of analysis were applied for this research. First a large sample of 429 headlines were captured and coded from 1985 to 2009. Headlines were the primary focus. As the ‘advertisement’ for articles, headlines play a significant role in influencing what readers choose to read. In this context we wanted to establish if headlines were more pejorative than articles and - if they were – whether this pattern has this changed over time.
Headlines were defined as the main headline at the top of an article but also, where they appeared, the ‘sub-heading’ or stand-firsts that are used by some publications to provide supplementary detail to the main headline.
A second tier of smaller, delimited ‘full-text’ samples– that is complete articles – was also coded. This allowed for a direct comparison with the tone of headlines.
The sample size was determined through the application of a two-stage approach: First, I selected which years to sample.
The time frame selected was the 24-year period of 1985 to 2009. I then selected three individual years within this frame, with intervals of no less than 10 years. The years were
Then, articles were selected within a three-month time frame for each publication for each year. The first quarter of each year was chosen.
I then used random sampling to select articles within each 3-month interval and these were used for the main headline analysis. Next, a smaller, delimited pilot of the full text of a sample of articles drawn from the headline sample was selected using parallel methods.
Three independent coders worked on this research. An odd number was chosen to test inter-coder reliability. All coders were trained extensively in the use of a detailed coding manual.
The coding manual (which was refined through multiple drafts) provided a detailed breakdown and careful explanation of all variables to be coded.
Thorough inter-coder testing was carried out to measure inter-coder reliability. A coding pilot was conducted using a limited selection of headlines and full text articles given to each coder after inter-coder training was complete. (There was no duplication with the main sample).
In addition to testing for inter-coder reliability this allowed for the identification of potential cultural barriers or misunderstandings the coders may have had during the pilot. This was particularly important as all three coders were from the US and may not have been regular readers of UK publications. However no marked problems with interpretation of text were identified.
Agreement levels for the pilot were extremely high and more than met accepted standards of reliability for textual analysis.
Next, a small, ‘blind’ pilot was included with the first wave of coding of the main sample as an additional test of inter-coder reliability. That is, coders were unaware that each had been given a small identical sample of headlines and articles to code.
This small blind sample was analysed prior to a second wave of coding.
This test also more than met accepted standards of reliability.
Two sets of variables were used in order to meet the objectives of this research. The first were Housekeeping Variables: ‘basic facts’ about the material being coded.
These housekeeping variables were defined as follows:
• The number of words in the text being analysed. That is - the length of an article. This was because the amount of space given - over to an article as well as its position within a publication - can be a key indicator of perceived ‘news worthiness’.
• We also documented the type of coverage. These were broken down to:
• Date of publication was also recorded as was the
• Type of publication: eg: tabloid vs broadsheet
The second set of variables was Critical Variables.
These related to the tone of an article or headline as well as the topics or mental health conditions mentioned.
•Main topics: This was an exhaustive list of issues related to mental illness including: suicide, murders or violent crimes committed by a person with a mental illness, the psychiatric profession, wellbeing, stigma, treatment issues, therapies & causes of mental illness.
•Secondary topics. These were topics that might be directly linked to the main topic in the headline or that constituted a related topic.
These included: gender, race, age, addiction, public policy, celebrity, the justice system, first-person stories, funding, wider healthcare provision, the pharmaceutical industry, disability & mental health professionals.
Secondary topics were coded because they offered a useful way to increase specificity and to broaden the scope of analysis - such as tracking the frequency of coverage of mental illness dealing with, say young people or celebrities.
•The next critical variable coded was Mental health ‘status’
That is: any condition or diagnosis specified or mental health term mentioned was recorded were broken down into categories. These were broken down as:
- Severe mental illnesses – including Bi-Polar Disorder and Schizophrenia
- Other common mental illnesses such as depression or anxiety disorders.
- Other rarer disorders – such as OCD and post-natal depression
- Generic references – such as mentally ill, psychiatric disorder, mental distress
- Wrongly labeled – where something was wrongly referred to as a mental illness. EG: learning disabilities.
And finally were it was unclear what, if any condition was being referred to.
By recording specific references to conditions this enabled us to track a number of trends including identifying conditions that featured most prominently and most often in stories.
The next thing recorded was the tone of both headlines and articles.
A four-point scale was used to guard against a key pitfall of textual analysis, namely, coders selecting the middle of a three-point or five-point spectrum as a default option. Every headline and article in the sample had to be given a code for tone. This ensured a 100% response to the sample. There was no opt-out option and no option to code for ‘other’ or ‘unsure’.
Where an article or message was deemed overall to be positive ie:
It contained some indications or expression of hope, progress, or of something good having happened.
There were 2 circumstances under which a neutral code could be applied:
A. When it consisted largely of statements of fact or provided information only.
An article or headline that was coded negative overall would have incorporated statements containing some indication or expression of:
A sensationalist article or headline was defined as one that met the same criteria as a for negative coding but was distinguished by hyperbole and/or the use of inflammatory or offensive language including:
For both headlines and articles, coders were required to document any word or phrase that prompted them to code as ‘sensationalist’.
The full text sample was coded for the ‘overall’ tone of the article.
Headlines however were coded according to 9 possible ‘messages’.
Where a clear link was made between mental illness and violence.
Where a clear link was made between mental illness and recovery.
Where a clear link was made between mental illness preventability.
Where a clear link was made between mental illness and factors that lead to it developing for example – use of illegal narcotics.
Where a clear link was made between mental illness prevalence.
Where a link was made between mental illness and an ability to function well.
Where a clear link was made between mental illness and treatments.
Where a clear reference was made to the attitudes of a person/group/population toward mental illness either or people with a mental illness.
9. Other – where none of the messages specified in the coding manual were clearly identifiable.
A smaller secondary analysis of coverage made up of a random sample of articles and headlines from two periods earlier in the 20th century was also conducted as part of this research project. This much smaller sample was intended to examine coverage from an earlier period- from each country - to see if any preliminary conclusions could be drawn regarding shifts in coverage over a substantial time frame.
Further, it was hoped the findings would indicate if more in-depth research was warranted.
The sample frame was based on a number of practical factors including time available for analysis. Titles were selected based on their prominence and cultural importance. However robustness of the newspapers’ archives was particularly important for accurate data capture. The number of publications is therefore limited.
The sample of headlines and articles was obtained using broad terms relating to mental health and suicide for two years; 1900 and 1950. All articles and headlines captured from the 12 month period of each year were included. This was because overall, the number of stories was – compared with contemporary publications – small. Due to the small sample size there was no need for random sampling with skip intervals. All articles and headlines captured were coded.
Fortunately, this research threw up a small mountain of rich, interesting data. Unfortunately that means a lot of statistics to trawl through and absorb – and a series of ever so exciting charts for you to look at. Make notes as we go along if anything is unclear or requires further explanation and we can talk about afterwards. As I said earlier – the great bulk of detailed analysis will be in the spring report.
Since stigma was very much at the core of this research, recording the tone of text - the impression it transmits to readers - was critical. As I mentioned, the sample was coded on a four-point scale. As a result, we were able to learn a lot about both overall tone and about changes over time.
First of all – how each country performed overall. On analysis of complete articles in the US over the 24 years, 42% of articles were found to be either negative or sensationalist – a substantial proportion.
When broken down over time the ratio of neutral articles to negative shifted in favour of netural between 1985 and 1995 suggesting some improvement. However, that ratio remained static
Between 1995 and 2009 indicating that progress may have stalled. At either end of the scale, there is little evidence of extreme reporting or of overtly positive reporting.
The UK articles sample overall reveals a more complex picture. Barely any articles were coded sensationalist while a substantial proportion – 21% were positive. When compared with the US however, a greater proportion of articles (45% as opposed to 35%) were coded negative.
Looking at how UK articles have changed over time the interesting change is between 1995 and 2009. The sample for 1985 was half that of the other two years due to archive limitations so makes direct comparisons difficult - but - where tabloid articles were included in 1995 and 2009 the number of negative articles jumps significantly.
BUT – if we extract tabloids from the UK sample and isolate data from broadsheets a dramatically different picture emerges.
More than 40% of all articles are positive if UK broadsheets are isolated. 67% of all articles are either positive or neutral. Compare this with the US – all a broadsheet sample – and
it is apparent that the UK broadsheets are much more likely to include overtly positive articles. A third of UK articles are still negative (33%) – almost the same as the US (35%) – but there are clearly distinct differences between the two countries.
Comparing overall UK articles with the broadsheet UK sample it is clear that tabloids do – as would probably be suspected – skew the results toward more negative content. It is also worth mentioning at this point that – bearing in mind the much greater print circulations of tabloids - these negative messages are potentially reaching a far larger proportion of the population (in print at least) than their broadsheet counterparts with ramifications for perpetuating stigma.
So on to the larger sample of text analysed for this research – headlines. As the ‘advertisement’ for articles the tone of headlines is arguably particularly significant when it comes to stigma as they send the first signal to a reader as to what they might expect.
To more fully understand what exactly headlines are communicating – as mentioned earlier - we sought to measure (not overall tone as in the case of articles) but the specific ‘messages’ contained within headlines.
Within the US sample of 166 headlines 237 individual messages were identified. It is the on the tone of these messages that the research focuses.
First to the US sample. Looking at the proportion of messages that are negative it is immediately apparent that headlines in the US press are more negative than the articles that accompany those headlines (58% for headlines compared with 35% for articles) suggesting a marked ‘mis-advertising’ of coverage. The higher proportion of positive messages compared to positive articles is worth noting but on the whole, further investigation as to why headlines are more negative than articles is warranted.
Indeed, if we look at how the advertising of articles has altered over time we can see an even clearer argument for further investigation as the number of negative message has increased steadily over time.
On to the UK then where 192 messages were identified across 151 headlines. What did we find?
The first thing to strike an observer about the results for headline messages across the UK sample is how the distribution of tone compares with that of articles. It is almost identical.
18% of headline messages are positive compared to 21% of articles for example. 47% of headline messages are negative compared to 45% of articles.
When headlines are compared with the US – again while the UK has more positive messages than the US - it also has a higher proportion of negative messages.
When tracking headlines over time the results for negative headlines are even more illuminating. (Again, setting aside the 1985 broadsheet only sample) there is a clear rise in the volume of negative messages between 1995 and 2009 – as was the case for articles.
When the broadsheet data is isolated from the headline sample something very significant is revealed when compared with what happened when the same was done for articles.
If you remember – when broadsheet articles were isolated there was a marked difference in the way tone was proportioned – most notably that broadsheet articles were demonstratively more positive than tabloids. This was not the case for headlines.
There were only marginal shifts on the Positive-neutral-negative-sensationalist spectrum for headlines when comparing the broadsheet sample with the sample as a whole.
However – a significant difference in the pattern of headline tone over time was discovered. The number of negative messages in broadsheets has declined steadily over the years while
it has increased when the whole sample including tabloids is taken into account.
When compared with the US sample – overall - UK Broadsheet headline messages were less negative (52% of UK messages were negative or sensationalist compared with 59% in the US). Further investigation into why this might be would be useful.
So - some brief conclusions regarding how articles and headlines compare in terms of tone. It is clear that in the UK headlines more closely resemble the article attached to them than do their counterparts in the US. In the US however, headlines tend to be more negative than articles suggesting an ‘over-selling’ of stories by employing negative messages in headlines.
Over time articles & headlines in the UK have become more negative with the exception of broadsheets which have done the opposite.
In the US, headlines have become steadily more negative while articles have become more neutral.
There are very few overtly sensationalist articles or headlines in any year in either country.
The UK has higher proportions of positive articles and headlines than does the US – largely, it has to be said, attributable to broadsheets.
I’ve talked so far about the overall results for messages in headlines but what about the specific messages? For example, which was the most common message communicated and how was it couched? Again, there will be more analysis of this in the spring but it is significant that
(excluding the option of ‘other’) Risk of violence is the most common message communicated by headlines in relation to mental in both countries – and – over time. There is a consistency to this link with violence that would suggest more work is needed on how the press arrives at such conclusions.
If, as campaigners continually attempt to communicate, there is not a discrete link between mental illness and violence, why does the media continue to make it?
Secondly there is the issue of the way in which messages of violence are communicated. Arguably it is only to be expected that reporting of violence is going to be negative – the question though is if the negativity is attributed to the act or the to the people perpetrating it and their membership of a particular group. Does the almost blanket negativity perpetuate stigma for the whole group?
Another facet of this research involved classifying ‘types’ of coverage in relation to mental illness. How, for example, do news stories stack up against features? Doe the immediacy and urgency of news mean that there is less time for reporters and editors to carefully consider the language they use or to include the voices of people with a mental illness a mechanism for balance?
As you can see from these charts, the US coverage is news dominated while in the UK there is a more even split between news and features.
When broken down by tone, what appears to be a somewhat even distribution of tone across the various types of coverage. However, once broken down further what emerges is a greater likelihood in both countries that news stories will be negative. In the US news accounts for 54% of overall coverage but 60% of negative coverage.
It is simply not possible to examine mental illness and newspapers without a specific reference to how language – frequently pejorative - is used by them. This slide has examples of such words – all of which were found within the sample we studied. Again, there is more research to be done on this, but the evidence to date from other research is that it carries enormous power to foster discrimination.
In his seminal book, ‘Media Madness, Public images of mental illness’, Professor Otto Wahl succinctly explains why depictions of mental illness in any media – and in particular the language employed to facilitate it – are worthy of our attention. He writes about how mass media has adapted to campaigns from different disadvantaged or devalued groups within society – for example women’s groups, gay rights groups and those advocating for people from ethnic minorities – and how the outcome was a reduction in negative stereotyping or those groups.
He notes that, largely thanks to vigorous advocacy from mental health campaigners, there are signs of progress for this group also. But he emphasizes that the depiction of mental illness continues to lag behind that of other groups who began campaigning earlier.
So what did we find when it came to language? The findings I have laid out for you on the high proportion of negative or sensationalist messages within headlines and articles give it away slightly. But it could be argued that statistics only partially illuminate.
MOST of the terminology over the years and in each country was generalist or benign – for example vague references to ‘mental illness’ or ‘mental breakdown’, ‘mental disorders’ or ‘mental health problems’ but others were more problematic. Words like ‘crazed’, ‘maniac’, ‘madman’, ‘nuts’ and even ‘monster’ were found in both headlines and in articles.
The UK sample was more likely to include sensationalist terms of this kind. It would be tempting to explain this by the inclusion of tabloid papers in the UK sample. And while tabloids were indeed more likely than broadsheets to include such terms (in 2009 the Sun gave us ‘Psycho Rapist’ and ‘Madman’ for example) such language was not exclusive to them. Broadsheets were also found to use Lunatic (1985) and ‘Crazed’ (2009).
In the US there were some problematic terms – for example ‘Violently Insane’ in a 1995 headline but on the whole the US sample was much less likely than the UK sample to use pejorative language.
All of this suggests, that particularly in the UK, much more work needs to be done if the use of language around mental health is to change in any significant way in newspapers.
There is always speculation around possible trends in the coverage of some mental health conditions. One possibility often discussed is if there have been more references to Bi-polar disorder since celebrities such as Stephen Fry and Alistair Campbell began speaking publicly in the media about their diagnoses.
While our research was not specifically tracking when certain conditions may have come to prominence in the press it did look at which conditions were mentioned in the years we examined and therefore allowed us to draw some conclusions about trends in coverage.
The first noticeable finding was that conditions were less likely to be overtly referred to in a headline than general references such as ‘mental illness’. This was the same in both countries and there was no pattern over time to indicate this was a changing.
Where a condition was mentioned in either a headline or an article it was more likely to be depression than anything else. Schizophrenia was the second most referred to condition but the number of mentions was very small compared to depression.
There was no evidence to suggest that depression is covered more now than in the past in either country. In both countries in fact depression was consistently the most mentioned condition. This suggests theories about celebrities “coming out” with a diagnosis of depression has not affected the overall level of attention the condition receives from newspapers.
[However – that’s not to say celebrities are not having an impact – more on that shortly].
It would seem that more investigation into why depression has persisted as the number one condition covered for such a long time would be instructive.
Finally on conditions – as might be expected – coverage of conditions linked to broader cultural changes or events such as the wars in Iraq and Afganistan (namely PTSD) appear in 2009 while not in the earlier years. In the case of PTSD this was more apparent in the US papers. Others that featured prominently in 2009 in both countries include ADHD, Stress and Anxiety disorders.
What were the topics most covered? As explained earlier, first we looked at what were defined as ‘main topics’. The aim was to see which were the most frequently covered and to see if any patterns emerged. Again, there will be much more detail on what we found in the spring – including changes over time - but I would like to take you through some of the key findings now.
First of all (excluding those articles or headlines coded as unclear or other) the most commonly reported topics in each country are strikingly similar for both headlines and articles.
The top three subjects in headlines in each country are identical with suicide the most prominent overall. Suicide for the purposes of this research was defined as any reference to a suicide or attempted suicide but also to suicide-related issues. That is to say that not every story coded was about an individual act of suicide. The sample also included stories relating to trends and statistics on say, rates of suicide or, for example, coverage of suicide prevention initiatives.
Mental health services – the second most common topic – was a broad category that included provision, delivery, type or quality of service. The scope of the topic may partially explain its prominence. Stories would have included subjects such as hospital closures, reports on the quality of services as well as coverage of individuals being sectioned.
Wellbeing – the third most covered area – was defined as where the focus was on the mental health or wellbeing of a whole group of people or the population at large.
Its prominence might be explained by a tendency to cover reports issued by organizations such as charities or by governments.
Murder by a person with a mental illness was a common topic in headlines in both countries – when taken with the results for messages which showed Risk of Violence as the number one message – that murder features so highly is perhaps not surprising. It may be the case that it is the number 4 topic in the UK due to the tabloid sample.
The prominence of this topic in headlines warrants further research as it goes to a central quandary for mental health campaigners – namely - is significant, undue prominence given to murders by someone with a mental illness compared to those by people without a diagnosis - and - does this skew public attitudes on the likelihood of someone with a mental illness to engage in violence.
So was there any difference in the topics focused on by headlines and the articles attached to them? There were some slight differences but perhaps the most interesting initial finding is that murder slips down the table for the UK sample but moves up the US table.
Further analysis of this is to be encouraged as it suggests that – in the UK at least – murder is being used to ‘sell’ an article but may not in fact be a primary aspect of the full story.
Psychiatry, the psychiatric profession and therapist or therapies (including alternative therapies) featured much more in the US than in UK coverage as did stories about overcoming adversity – perhaps a reflection of the American cultural preoccupation with the American dream and the little guy winning now and then.
And - when it came to directly addressing issues around stigma or discrimination – something I’m sure campaigners will be especially interested in – very few headlines or articles did so in either country. As a journalist I know I have written about the impact of stigma and discrimination. The question perhaps to be asked by any further research would be why is stigma not more extensively covered when it is such a big problem for people with mental illnesses and therefore a significant problem for our society?
Next then to secondary topics. Coders had a lot of options to choose from here – 30 in all – and there is not time to breakdown the results for all of them here.
That said – I’d like to highlight some of the significant findings on these topics.
There were almost twice as many Secondary topics as Main topics so it is perhaps not surprising that there was more variety in the way they were reported. As you can see – there were some similarities between the topics mentioned in headlines in the two samples with, gender, children and teens as well as inquests featuring frequently.
Interestingly, the UK press was much more likely to refer to the personal stories of an individual with a mental illness in headlines – what we defined as ‘first-person’. This is worth noting as previous qualitative research has highlighted a desire for more stories that include the views of people with a diagnosis of mental illness. For example, The Mind Over Matter Research by the DOH anti-stigma campaign, Shift based on broadcast and print coverage in 2005.
One noticeable difference in the 2009 sample was that the US press was more likely to mention the armed forces in headlines - in line with its greater tendency to focus on PTSD, mentioned earlier.
When it came to the content of articles, multiple secondary topics – and multiple mentions of secondary topics – were apparent. For the purposes of this research however, coders did not record the number of mentions - only if - a secondary topic was mentioned at all. Based on this coding, you can see there are similarities with headlines as would be expected: eg: inquests and children and teenagers features highly.
However, for both countries (and across years) there is considerable coverage of policy, legislation and either mental health in the context of the wider health system or mental health professionals.
Again – there are numerous patterns to be identified for comparative purposes from so much data and more of this will be explored in the spring. Here is a small sample though.
One of the most sensitive areas of mental health reporting is how it depicts people from ethnic minorities. Campaigners have been concerned for some time about first of all – whether the press covers enough how the mental health system engages with people from ethnic minorities – but also if the coverage of this group is more negative than for other groups.
The topic of race or ethnicity came up very rarely in text analysed for this research making it difficult to draw any robust conclusions on the subject. It may mean however indicate that future, discrete research – perhaps with a focus on analyzing specific stories – is warranted to garner substantive insight on this issue.
Other aspects of the secondary topic data identified by this research included how celebrities feature in coverage.
The results clearly show that recently – in the UK at least – celebrities with mental health problems are much more likely to be written about or interviewed than in previous years. This research did not track if this trend had resulted in a greater propensity for first-person stories more generally but this could be an area worth investigating further.
Another nugget that emerged – and that may be of interest to campaigners in both countries – is the dearth of coverage of so-called ‘Big Pharma’ or medication in general in relation to mental illness.
Certainly from my discussions with all kinds of people interested in the subject of mental illness, issues around for example prescribing of medication or the pharmaceutical industry generally come up frequently so it is interesting to note the lack of coverage in both countries.
And so to an important note on the coverage of suicide. There were many striking elements that emerged from the data we collected for this research and not least was that for suicide.
The sheer volume of suicide coverage would suggest a need for further, in-depth, analysis of this topic on its own. We were not evaluating suicide coverage in and of itself but rather tracking where it appeared in relation to mental illness but it is impossible not notice that it is a subject of enormous interest to editors. It is also – as many people in this room know - a very contentious issue when it comes to reporting – not least of course because of the risk of copycat suicides and issues around the implications of the reporting of excessive detail. Again, the objectives of this research did not include dissecting suicide coverage per se - but - as one of the more recurring topics identified it is worth a special mention.
As well as the overall volume of coverage it was also the case that the level of coverage was remarkably consistent across both countries. There was almost no difference in the amount of coverage of suicide overall. And, both broadsheets and tabloids had high volumes of coverage.
Much more will be available on the type and tone of reporting relating to suicide in my report in the spring but something worth mentioning initially is the use of cross-references:
Those small but – campaigners would argue – extremely important helpline numbers or website addresses that can be placed (usually) at the end of an article.
One the things I was keen to examine as part of this research was if newspapers had heeded recommendations from campaigners to include helpful information alongside articles about either mental health or suicide.
A telephone number for the Samaritans for example.
Inclusion of such references is widely regarded by mental health advocates as not just helpful but essential. Yet - we found that despite the issue being raised with editors and reporters – certainly in the UK – this is not happening.
Out of 166 US articles about mental health or suicide across the timeframe of this research just 7 cross references were included with the article. In the UK the number was 8 out of 151.
All cross-refs in UK publications were in 2009 suggesting that their inclusion is a relatively recent phenomenon. Those that did appear were spread across the sample publications indicating that there is no sector neglecting them entirely.
Despite this slight indicator of change however, the numbers were extremely small. This may be all the more surprising as the advent of the internet has meant fewer space restrictions so - in theory at least - there is less reason to not include helplines.
It may be the case that hyperlinks to organizations such as The Samaritans mentioned in articles perform the function that cross-refs once did in a print only platform in that they facilitate a click-through function - but this research was not tasked with measuring this.
It is worth asking the question though – should further research be done on this??
Does including a hyper-link carry the same value or perform the same function as a helpline when it may not be interpreted as a call to action in the same way??
So – we are almost finished. And I promise - only a few more charts to go. These are slightly different though.
I mentioned earlier that we traveled way back in time for a small element of this research. We took a glance at coverage in 1900 and 1950 in a few publications. Because of this limit – not all papers were around back then of course – it is difficult to make direct comparisons.
That said – there are interesting observations that can be made – and – questions that are thrown up.
Why for example were there no Sensationalist articles at all in 1900 yet a quarter of articles coded in 1950 were?
It is entirely understandable that 1900 is so far back as to be almost alien to contemporary culture – but the same can’t be said about 1950.
So why, we can legitimately ask – has the volume of negative coverage not reduced more dramatically since 1950? The mental health systems of both the US and the UK have altered almost beyond recognition – the demise of old asylums being just one indicator – so why not press coverage?
Why – despite the best efforts of campaigners – has mental health coverage not fundamentally changed?
I had to be careful when exploring the older sample. The history geek in me was getting dangerously close to being totally distracted by it.
This was from The Guardian in 1950 – it alludes to two topics much discussed today. How people with mental illness can access appropriate support within their local community away from clinical environments - and peer support as a substitute for institutionalisation – something gaining particular traction in the US.
Then there was this Times article in 1900. Allowing for linguistic quirks and extreme nature of the story, what it nonetheless tackles is the ill-treatment of people with mental illness and the bringing to justice of an individual who was guilty of it.
In the US circa 1900 a theme that emerged (and perhaps a portent of today’s coverage of something like Britney Spears problems for example) was people of social status – the celebrities of the day – being institutionalized. Hence headlines like:
Both dealt with people in the public eye being diagnosed and treated.
The stories written in relation to mental illness and suicide in years gone by tell us something more about the coverage than if it was negative or positive. They also offer a window into the editorial priorities of the day and that indeed many were concerned with innovations in mental health care.
One of my favourites is this one from The Guardian in 1950.
Another that caught my eye was in relation to suicide. This was also in The Guardian in 1950 (must have been a stellar year!)
So – IN CONCLUSION - I won’t go over what I’ve said in this presentation – if for no reason other than it would take up more time. But – here is what is clear from this research.
More research on trends in mental health coverage is, I think, warranted – especially in relation to the impact of new media.
Further investigation of the editorial practices and cultural influences that produce the kind of uniformity of coverage our research has found is also worthy of further investigation.