Purpose of the Research
The purpose of this content analysis is to find out to what extend did the UK Media personalise the NHS crisis to the person of Patricia Hewitt?
It is commonly known that the media influences dramatically on audience attitudes to different problems. Analysing the way media reported on the latest NHS crisis will help to find out what impressions audiences were getting through the media which obviously helped to form their opinion. This analysis might also help to find out who was to blame for this crisis according to the UK Media. The assumption is that the UK Media intend to relate the NHS financial crisis to the person of Patricia Hewitt.
The specific types of research considered appropriate for deriving meanings that respond to the nature of the problem are (a) Pragmatical, (b) Semantical, and (c) Single-Vehicle. These three different types of content analysis apply differently according to the purpose of the investigation (Krippendorff, 1980:33).
In fact, the specific content analysis (NHS in crisis) requires the “semantical” method considering that the purpose of the research is to analyse to what extent the NHS crisis is associated with Patricia Hewitt and to what extent the NHS crisis is personalised in the UK media.
Media Sample
The next step in the content analysis is the media sample. Sampling is appropriate for economical and undoubtedly temporal reasons. Despite the fact that a wide sample gives more reliable results, the implementation is not always possible. However, an effort will be implemented to have a representative media sample.
The media sample used consists of UK newspaper articles only. Newspapers in the media sample will be stratified by their level of importance (stratified sampling). Therefore, articles from UK National newspapers only are used in the analysis.
The timeframe for the content analysis is one month from 10th of April 2006 to 10th of May 2006. Although the decision for the reduction in NHS jobs and NHS expenses was made two months ago (in March 2006), the most intensive media coverage appears only after 10th of April. The “NHS Crisis” story lasted for the whole month (with different intensity) until the latest government elections where it was made clear that the Health Secretary will keep the job (7th May 2006). After that point the crisis can be considered to be over (or to be out of the media interest), so further analysis of media coverage is not necessary.
With the use of Athens Lexis-Nexis the media sample was created. The search was narrowed to the specific dates (10/04/2006 - 10/05/2006) and to UK National newspapers only (as they are a good representative of the UK population). The political alignment of newspapers is not an essential factor for the sample. Newspapers that are loyal to the labour party (Guardian, The Sun, Daily Mirror, etc) and newspapers from the opposition party (Daily Telegraph, Daily Mail, Daily Star etc) and neutral newspapers (e.g. The Independent) will not be divided into different categories as, according to our research question, we are not trying to justify the media coverage by measuring who was more aggressive or neutral and who paid more attention to the crisis (i.e. left wing, right wing).
As there is no such option in Lexis-Nexis as “UK National newspapers only”, an option “Major world newspapers only” was chosen. The search criteria was first stated as “NHS”. As this problem can be considered as of UK national interest only, only UK National newspapers appeared in the search result (The Times, The Guardian, Observer, The Mirror, Daily Mail, The Daily Telegraph, Financial Times, The Independent, etc.). By narrowing the search to “major stories only” about the NHS, 915 articles were found (written by UK national newspapers in the chosen time period). But this was the number of articles about the NHS in general, not about the current crisis specifically. As we do not try to find out how much attention the crisis received according to other NHS news, this media sample is not appropriate for our purposes. Therefore, the search was narrowed to articles containing the words “NHS” and “Crisis”. Although we are trying to justify how much the media related Patricia Hewitt to the crisis, we are not going to put her name into the search (e.g. “NHS”, “Crisis” and “Patricia Hewitt”) as this will result in articles about her only. We would then not be able to find out how many more articles there were about the crisis without mentioning her name. And without that it is not possible to justify to what extent the crisis was personalised. So, the final search criteria used were “NHS and Crisis”.
Although it is possible that some of the articles went missing as journalists did not use the word crisis while writing about current NHS problems, it is not going to influence on the result of present content analysis dramatically as we ended up with 342 articles. This is a big enough sample of the media to be reliable. It also simplifies our coding in the future as the media sample from the very beginning consists of articles about “NHS and Crisis” only.
Coding Devices
Coding is the process whereby raw data are systematically transformed and aggregated into units which permit a precise description of relevant content characteristics. The general problem in any research design is the selection and definition of categories into which content units are to be classified. The most important requirement of categories is that they must adequately reflect the investigator’s research question (1; p.94).
As the question (purpose) of our analysis is to find out to what extent the UK media personalises the NHS Crisis to Patricia Hewitt, the following coding devices are required (further categories):
Category 1:
Personal attribution to Patricia Hewitt
Both positive and negative attributions included as the research question does not require specifying the tone of personal attributions.
Category 2
Personal attribution to somebody else
Both positive and negative attribution is also required. Personal attribution to absolutely any other person BUT not Patricia Hewitt will be classified in this category.
Category 3
Political issues without attribution to personal issue
Attribution to economical problems of the country; lack of physical training in schools with bad health as a result; unhealthy food in Britain, and anything else applied to the current crisis without any personal attribution will go under this category.
Simply counting the number of mentions of Hewitt’s name won’t make any sense as there might be many more articles using another person’s name or just political issues. The number of articles on the NHS crisis without mentioning Patricia Hewitt’s name is a very important indicator. Her name could also be used in an article with 10 other names which will reveals that the crisis is not personalised to her. The occurrence of the three categories defined above in the media sample will be counted and the relative proportion of each category determined. In order to be successful, coding devices of all of the categories need to be unitised.
Unitisation
In addition to defining the categories into which content data are to be classified, the analyst must designate the units to be coded. Our choice is going to be a recording unit, the specific segment of content that is characterised by placing it in a given category (1; p.116).
The units of meaning to be used in the content analysis are themes. In order for the unitisation to be specific, it needs to be implemented using the following thematic divisions for Categories 1, 2 and 3 defined above:
Category 1:
Theme I: Judgment of Patricia Hewitt’s actions (both positive and negative).
Any statement regarding Hewitt or the Health Secretary related to the crisis will go under this theme.
Category 2:
Theme II: Judgment of the government’s actions (both positive and negative).
All statements regarding the government’s (in general) role in the crisis will go under this theme.
Theme III: Judgment of Tony Blair’s actions regarding the current NHS Crisis (both positive and negative).
All statements regarding Blair and Prime Minster will go under this theme.
Theme IV: Judgment of Labor party decisions (both positive and negative).
All statements regarding the Labor party particularly (including Downing Street) go under this theme.
Theme V: Judgment of doctor’s and Health Department’s roles in the crisis (positive and negative).
Category 3:
Theme VI: Analysis of the economic reasons for the crisis.
Theme VII: Budget problems as a reason of the crisis.
Theme VIII: Logical result of years of poor management.
In our research we will count presence of each unit in the newspaper articles content. One unit is a theme of each paragraph related the specific theme (sometimes a sentence is a paragraph in a text and it will be counted as a unit as well). For example, three paragraphs/sentences about Blair and one about Ms Hewitt will be counted as three occurrences of Theme III and one occurrence of Theme I. Later all of the units will be placed into categories and analysed.
Reliability test
Having stated the sample, the recording devices, and the segmentation of units, the reliability of our content analysis needs to be checked. There are three distinct types of reliability. In our test we are going to check the stability of our design. Stability is the degree to which a process is invariant or unchanging over time. Stability becomes manifest under test-retest conditions (same coder codes a set of data twice, at a different points in time) (2; p.120).
The first phase of the reliability test was implemented on 22nd of May, and the second phase on the 24th of May. We are doing the reliability test on a random sample (every 10th article was chosen from our media sample) on 6 articles from UK National Newspapers chosen in the context of the dates that we have defined as time framework.
In the tables below you can see number of themes coded in each article:
22/05/06
24/05/06
In order to calculate our reliability we will calculate the coefficient of reliability (C.R.) which is the ratio of coding agreements to the total number of coding devices:
C.R. = 2M/N1+N2
In this formula, M is the number of coding decisions on which the coder is in agreement in both tests, and N1 and N2 refer to the number of coding decisions made by coder in the 1st and 2nd test respectively. In our example the coder identified and categorised 43 themes in the 1st test and 41 themes in the second test. But only 32 of them were in agreement according to this formula:
N1=43
N2=41
M=32
C.R=2(32)/43+41
The coefficient of reliability of our analysis is therefore C.R. = 0.76.
Conclusion
According to the reliability test, our design is reliable enough to be taken further. This content analysis has a good chance to bring reliable results. In order to improve the reliability, more specific identifications of units are suggested. The researcher must further specify the unit (as the theme has a broad meaning and can be made up of many concepts) and specify how to distinguish between one or two mentions of a theme in an article context in more detail.
The results of our test also disprove the theory that the UK media personalises the crisis with the person of Patricia Hewitt. According to our reliability test, the crisis is related to the person of Tony Blair and Labor party to a much higher extent. Nevertheless, no conclusions can be made out of this test as it is based on only six articles (there is still a chance that in the remaining 336 articles the UK media relates the crisis to the person of Ms. Hewitt) and it was not the purpose of this test.
Bibliography
- Holsti Oler R. “ Content Analysis for the Social Sciences and Humanities” Addison-Wesley, London, 1969
- Krippendorff K. “Content Analysis. An introduction to its Methodology” Sage Publication Inc. London 1980
- Tumber, H. & Palmer J. “Media at War” SAGE Publications Ltd. London 2004
- www.healthmatters.org.uk
Appendix