Chapter 4. Watching the news: researching EBF and PIE reporting on BBC1 and ITV1
4.7 Level 2 Content Analysis: examining PIE issues in detail
The challenges associated with selecting appropriate texts for discourse analysis (Phillips and Di Domenico 2009) were overcome by Level 2 Content Analysis, which was designed to establish key themes, contributors and discourses. Pertinent stories identified by Level 1 were subjected to Level 2 analysis, the focus shifting from news generally to stories containing PIE references. As described, these themes are rarely central within news stories. Accordingly, measuring journalist or social actor contributions by time was impractical, since their interventions were often made in
7 A number of broadcasts were selected and recoded a number of times as the categories were refined.
95 | P a g e passing. For example, a citizen might appear on screen for 30 seconds discussing
increasing education costs, but only 5 seconds of this might be specifically about how a family’s lifestyle had been compromised because of a reduction in discretionary income. Consequently, only spoken and visual data pertaining to PIE issues was captured, and not any extraneous data surrounding it.
Some variables contained well-defined categories. For example, whether contributors were political actors, third sector spokespeople or journalists, the geographical locations associated with stories, and so on. Other less clearly defined variables were populated inductively; the rationale was that by capturing every discourse, subtle variances would not be missed. It became clear that the range of causes, consequences, advocated actions and so on were diverse, and would have not comfortably fitted into a compressed range of pre-set choices. The next chapter (Chapter 5) discusses results in terms of categories that were conveniently merged at the conclusion of the data collection process. Level 2 Content Analysis added a range of information regarding the journalistic treatment of PIE issues:
Journalistic conventions – PIE news stories were deconstructed by news convention. This was a deductive (pre-set) category embracing anchor reports, reporter packages, live interviews and so on (see Cushion and Thomas 2013).
News stories sometimes included multiple items. For example, a story introduced by an anchor and the edited package8 following it might both include references to poverty. This meant a smaller unit of analysis, enabling judgements about whether issues were covered in passing, in detail, or with some sense of immediacy (as in the case of live news, for example).
Name of reporter - This category was populated inductively, adding names as they occurred.
Context - If PIE issues were embedded within other stories, their level of prominence enabled conclusions about how such issues are covered. This was a deductive category; issues were considered “implied/in passing” or “substantive”
to reflect the level of attention paid to the PIE topic within the news item. In the event of rich-poor comparisons, income inequality was the default category (full details in Appendix 2).
8 Short anchor introductions lasting a few seconds were included as part of the convention succeeding them – in other words, they were considered part of the edited package or live interview.
96 | P a g e
Area - The geographical location referred to within the item was added inductively. If there was no specific area, stories were marked “generic”.
Causes - If mentioned, PIE cause(s) were added inductively and as they occurred.
As has been mentioned, causation, or theories of causation are key to critical realist research (Ackroyd and Karlsson 2014).
Blame - If PIE phenomena were blamed on some person, group or thing, these were added inductively. However, some explanation is required to differentiate
“cause” from “blame”. The assigning of “blame” is achieved through language (Wiggins and Riley 2010; Sims-Schouten and Riley 2014) and occurs where a specific person or institution is ascribed agency. This might include for example, suggestions that a population had been “plunged into poverty” because of the specific actions of a political leader. “Cause” however, involves no personal agency or a more general reason - for example this might be increasing interest rates leading to reduced lifestyles.
Consequences - The consequences of any of PIE issues were added inductively.
Actions - If journalists advocated actions to counteract or solve PIE issues, these were also added inductively.
Other discourses - If there were other sections of commentary not qualifying as cause, blame, consequence or action, then these were again added inductively.
For example, suggestions that “the wealthy are greedy” would qualify in this category.
Images - Any identifiable moving or static images clearly used to index PIE issues were added inductively. These may include food queues, expensive cars or homes, and so on.
Framing - News items were categorised as being presented thematically, episodically, or using a mixture of both (see Chapter 2). For example, if poverty was expressed entirely through the lens of one family relying on food banks, then this was considered “episodic”. If the same issue was addressed entirely using a series of charts outlining global trends and policies, then “thematic” was chosen.
Alternatively, if income inequality was described in terms of global data but included a short section where a specific wealthy person was juxtaposed with a poor one, this would be “mainly thematic”.
Metrics - If PIE issues were expressed using numeric measures, these were added deductively. For example, if ordinary people were pushed nearer poverty by
97 | P a g e spiralling interest rates, then “interest rates” would be added as a metric (and in
such a case, it would also be coded as “a cause”).
In addition, if other social actors besides journalists were featured, the following information was collected:
Type of actor - This inductive category described whether contributors were politicians, businesspeople, ordinary citizens and so on, facilitating understandings to what extent influential voices or “primary definers” dominate (Allan 2004, p.71).
Political party - For politicians, affiliated political parties were added inductively.
Causes - As described above.
Blame - As described above.
Consequences - As described above.
Actions - As described above.
Other discourses - As described above.
Images - As described above.
Framing - As described above.
Metrics - As described above.
Notes - Additional notes were made where appropriate; this was to capture extra-ordinary information not otherwise accommodated by the coding framework.
The items subjected to Level 2 Content Analysis were scrutinised in detail several times, and a coding sheet was developed during piloting. The final coding sheet is detailed in Appendix 4. A separate sheet was physically completed by hand for each discrete news item, and the data was entered into the same master SPSS spreadsheet described in the explanation for Level 1 Content Analysis. Where categories were developed inductively, all were entered into the SPSS spreadsheet, category merging taking place at the analysis stage9. As the major instrument of data
9 The third party who carried out the intercoder checking was given a spreadsheet containing the narrower range of refined, merged categories which are used for the data analysis in Chapters 5 and 7.
98 | P a g e storage, when in use, the SPSS spreadsheet was saved regularly in three separate
locations10 which were all password protected.
4.8 Qualitative Phase: wider understandings of discourse, and the choice of