Appendix 2
Methodology
Ch.2 Audit Bureau of Circulation disc analysis
Analysis involved:
- Extract relevant data from ACCESS database from Audit Bureau of Circulations for period ended December 2005.
- Extract a list of publications in each category listed above. "Publ_selection.xls'
- Three entries for circulation for each publication were summed for each date listed.
- For some years, there are two entries (mostly prior to 1994) and other years there were 4 entries (1994 to 2005).
- Some circulations were left blank and so an "NA" was used to fill these fields.
- The year was extracted from the date field and changed to a factor variable for calculation purposes.
- Define a mean function to operate on values which allows "NA" entries and then calculated a mean appropriately over each year and publication.
- Create data file where each year from 1938 to 2005 for each publication has mean entry for the specified year, even if "NA". The resulting data is in "AlldailysData.txt', "AllSundaysData.txt', and "AllRegionalData.txt'.
- Selected the relevant publications and listed as a "1" if selected or a "0" if excluded. List is given in "Publ_selection.xls" for each category.
- For all selected publications, the total for each year was calculated. That is, the mean for each selected publication over each year was taken and summed to obtain a total for each year. These values are given in the Word document for each category.
- A time series plot was created for Total circulation for each year. (See Word document for each category)
- A subset of data was then taken to be the total circulation at each five year interval (ie 1940, 1945, 2000, 2005).
- A time series plot was created for the 5 yearly interval data (See Word document for each category).
Other issues:
- The selection of the publications to include in the analysis affects the results. This task was especially difficult for the Metro/ Nationals where it seems that there are several publications listings for the "same" newspaper.
- The data seems to be incomplete for some years. For 1958, for many publications there were many "0" entries, even though the publications had non-zero entries in years either side. Zero values for a particular date affected the means for that year. The effect is quite obvious in the plots.
A peculiar value for 30/ 09/ 84 for regional daily (id=27 Barrier Daily Truth, Broken Hill) appeared in the data. The circulation value was given as 81574, however all other entries close to that date for the same publication are approximately 8100. As it seemed an obvious data entry error, it was changed to 8157 and recalculated.
Ch.4 News content analysis
Structure
The Steering Committee selected the newspapers to be sampled, based on their potential for reflecting as accurately as possible the Australian print media array. The 14 newspapers selected were:
| The Australian | ND | News |
| The Age | MD | Fairfax |
| The Daily Telegraph | MD | News |
| The Sydney Morning Herald | MD | Fairfax |
| Herald Sun | MD | News |
| The Canberra Times | MD | Rural Press |
| Sun Herald | MS | Fairfax |
| Sunday Tasmanian | MS | News |
| Sunday Herald Sun | MS | News |
| Sunday Mail (SA) | MS | News |
| The Sunday Mail (QLD) | MS | News |
| Newcastle Herald | RD | Fairfax |
| Ballarat Courier | RD | Rural Press |
| Northern Territory News | MD | News |
The high number of Sunday newspapers reflects circulation figures; Sunday newspapers are the highest selling in the country.
Twenty-eight days from throughout 2005 were then randomly generated by computer. The dates were:
- Saturday, 1 January
- Sunday, 9 January
- Sunday, 23 January
- Friday, 4 February
- Monday, 7 February
- Saturday, 26 February
- Thursday, 10 March
- Friday, 18 March
- Friday, 4 March
- Sunday, 27 March
- Monday, 28 March
- Monday, 11 April
- Saturday, 14 May
- Sunday, 29 May
- Monday, 27 June
- Sunday, 17 July
- Thursday, 25 August
- Monday, 5 September
- Saturday, 10 September
- Monday, 12 September
- Sunday, 18 September
- Sunday, 2 October
- Friday, 28 October
- Wednesday, 2 November
- Sunday, 13 November
- Tuesday, 15 November
- Sunday, 27 November
- Wednesday, 14 December
Four coders were employed to analyse the first five news pages of each edition of the newspapers on the following days. Coder training was held over two weeks. Coding took place between March and July, 2006 resulting in the analysis of 2448 articles. Coders worked with a Microsoft Excel spreadsheet, developed by academic researchers with assistance from the Spatial Analysis Unit (SPAN) of Charles Sturt University.
Criteria
- Topics in the news
- Changes were made to the US report template to suit the Australian environment. Politics was divided into Federal, State and Local. The categories were;
- Federal Politics
- federal issues/ politicians
- State Politics
- state issues/ politicians
- Local politics
- local council
- Crime
- murder, robbery, assault, fraud etc, where a criminal investigation has been launched or charges laid
- Foreign Affairs
- political issues of overseas countries that are not related to military, terrorism or war
- Military
- issues of military organisations that are not terrorism or war
- Terrorism
- bombings or acts of violence that are not part of traditional warfare
- War
- conflicts that are generally regarded as a war, for example, a clash between US military and Iraqi insurgents is war, the London bombings are terrorism
- Entertainment
- stories involving film, television and other celebrities
- Lifestyle
- 'softer" magazine style stories on living
- Business
- financial stories
- Science
- reports of technological developments and scientific discoveries
- Health
- medical stories that are more consumer than science oriented
- Accident/ Emergency
- incidents where emergency services, police, fire, ambulance, SES are required but do not involve a criminal act
- Sport
- reports involving sports matches and/ or identities
- Weather
- general weather stories that do not involve cataclysmic events
- Other
- stories that do not fit into the other categories
Coders were instructed to identify the dominant topic of the news report so, for example, a report on the devastation of Hurricane Katrina would be an Accident/ Emergency story while an accompanying article about the factors contributing to the US hurricane season would be a weather story. Similarly a technological development for a tsunami warning system would be a science story.
Article length
The number of words appearing on the page for each article were added and divided into the four categories. The length of a story on page one, that spilled onto another page, was calculated as the length appearing on page one, and not the total length. This was done to give an indication of the volume of reports on each page.
Sources
Initially stories were coded for the number of sources quoted. A source was identified as a person or a statement or "an anonymous source'. Furthermore, each source was then categorised under the following.
- Anonymous
- A source identified as "anonymous" or where no identification was given
- Government
- A politician or person employed by a government department
- Non Government Organisation (NGO)
- A non political or corporate organisation
- Company head
- The executive, managing director etc of a corporation
- Emergency
- A member of the emergency services including police, ambulance, fire, SES
- Academic
- A university employee engaged in academic research
- Medical
- A doctor, nurse or other health provider
- PR
- Where the source was identified as a Public Relations spokesperson or statement
- Statement
- Where the source was only identified as coming from "a statement'
- Celebrity
- Where the source was only known for the status as a film, television or other "star'
- Member of public
- Where the source had no other qualification for comment other than their public citizenship
- Other
- Where the source could not be identified by other categories
If a source corresponded to more than one category, coders were instructed to identify their dominant purpose in the article and select that category. The PR source category could only be used where the source was identified as a PR spokesperson or statement. This does not reveal the level of involvement of Public Relations in media content. For example, many of the company heads sourced may have been provided by Public Relations staff to talk to the media.
Sources were further identified by gender.
While the statistics reveal a strong male bias a further breakdown of Member of Public sources, where there should be an even chance of male/ female sources, still demonstrates a male bias at 61.4 per cent compared to 38.6 per cent female. This is not, however, as great as the overall differential. It is suspected the male bias reflects the number of males in senior positions within government and corporations.
Framing
Framing categories were initially based on the US study. It was found, however, many were steeped in US vernacular, such as "reality check." Instead a mix of US frames and other frames was used, with a strong bias towards traditional news values as frames. It is hoped to further develop framing guidelines for future reports. The frames used were the following.
- Conflict
- Stories that emphasised a clash of ideas or forces
- Consequence
- Stories that emphasised the impact of a decision or event
- Winner/ loser
- Where the fortunes of a topic were emphasised
- Problem to be solved
- Where the main issue was presented as a difficulty to be overcome
- Fear
- Stories that emphasised a reason to be fearful
- Human Interest
- Stories that presented no other reason for publication apart from their appeal to the unusual
- Proximity
- Stories that highlighted the closeness of the topic to the publication's population
- Prominence
- Where a story was framed around the reputation of a person
- Gender
- Stories that highlighted gender as the main focus
- Religion
- As above but where religion was the dominant focus
- Multiframe
- Where more than one of the above categories applied
- No frame
- A straight, inverted pyramid style report with no obvious frame.
Coders were instructed to identify a single frame where it was clearly dominant over others. Where several frames were used without a dominant frame stories were deemed to be multiframe.
Data analysis
Once coding was complete the Spatial Analysis Unit of Charles Sturt University, which had assisted in database design, collated data into a final spreadsheet. In all 2448 articles were coded including 1573 from metropolitan newspapers, 442 from regional dailies and 435 from Sunday newspapers. Analysis was then performed by academic researchers.
Ch.5 Credibility
The "Opinion Leaders" sample was drawn from the "top end" of Roy Morgan Single Source—a well constructed Australia-wide sample of people of income $80,000+ or in the top occupational categories, and included booster samples from Henry Thornton Readers, Crikey Readers, Australian Institute of International Affairs Members, Marcus Oldham Associates, Davos Connection Associates and the Australian American Association members.