Chapter Four: Methodology
4.4 Research method one: Semi-structured interviews
4.4.2 Data collection
The way in which I collected data was from the perspective of constructivism- the concept of constructing grounded theory was adopted. Braun and Clarke (2013:184) argue that the philosophical paradigm of grounded theory has been changing. There are positivists (Glaser and Strauss 1967), contextualists (Pidgen and Henwood 1997) and constructionists (Madill et al. 2000). In the beginning, Glaser and Strauss (1967) adopt grounded theory to discover meanings in data. However, Pidgen and Henwood (1997) situate them in the contextualist framework; furthermore, Charmaz (2006) discusses ‘constructing grounded theory’. Following the philosophical underpinnings of this research, the constructionist grounded theory was adopted to construct the research data via collecting, coding and analysis. This will be further discussed in section 4.4.5 and
section 4.4.6.
In fact, the journalists who I interviewed are experts, and plenty of data were provided (Altheide 2002: 411). Meanwhile, as a researcher who has worked in journalism, I am familiar with the industrial knowledge and I understood the contexts that the
interviewees described. Therefore, throughout the interviews, practical knowledge was communicated clearly. However, the journalists were not familiar with academic knowledge; for example, I asked them some concepts about the elements of journalists (Kovach and Rosenstiel 2001). Sometimes, I did not realise that the phrase, ‘it must serve as an independent monitor of power’ (Kovach and Rosenstiel 2001: 5) was too academic for them to understand, and I explained the concepts of the term for the interviewees. It is suggested that in the future, researchers prepare for the situation that they might need to clarify ‘academic concepts’ to interviewees.
4.4.3 Transcription
After conducting the interviews, I transcribed and interpreted the meanings of the interview data. As Braun and Clarke (2013:162) state, the process of transcription is a ‘representation’, and it is necessary to ‘clearly’ and ‘consistently’ transcribe. I
transcribed in Chinese at first, and then translated selected parts into English in order to analyse in detail. Furthermore, Braun and Clarke (2013) mention that ‘a transcript needs to both signal what is said and who is speaking’ (p.163). I transcribed the language of interviewees and clearly pointed out the specific spoken data were from specific interviewees.
4.4.4 Coding
As the interview data were transcribed, I began to work them into meaningful data. This involved transforming the data into useful codes. As Braun and Clarke (2013) define, a code is ‘a word or brief phrase’ that captures the concepts of researchers with regard to which data are useful (p.207).‘Familiarisation’, which is specifically mentioned by Braun and Clarke (2013), means to understand the meanings of the data not
superficially but ‘actively, analytically and critically’ (p.205). Braun and Clarke (2013) argue that the more researchers engage with the data, the more content can be explored and ‘opened up’ (p.205). By coding, researchers can define what the data are and what they mean (Charmaz 2006:46).
Coding is to relate the data to the research questions, and coding approaches include ‘selective coding’ and ‘complete coding’ (Braun and Clarke 2013:206). Selective coding is mostly for narrative, discursive and conversation analysis. Researchers are interested in specific instances and select them from the data; therefore, ‘pre-existing theories and analytic knowledge’ are required to identify the concepts in the data (Braun and Clarke 2013:206). In contrast, complete coding requires researchers to look for everything in the data that is relevant to the research questions. Only later in the
analytical process might there be a more selective process (Braun and Clarke 2013:206). Codes can be semantic, such as ‘data-derived codes’, or latent, such as, ‘researcher- derived codes’ (Braun and Clarke 2013:207). Semantic codes refer to
‘participant-generated data’ that ‘mirror’ the concepts of the participants; latent codes use theoretical frameworks and the concepts of researchers to identify meanings embedded in the data (Braun and Clarke 2013:207). This research adopted complete
coding to identify anything related to the research questions, and then selective coding was conducted. Furthermore, both data-derived codes and researcher-derived codes were adopted to identify semantic and latent meanings.
Charmaz (2006) suggests that at least two coding phases should be included: first, the ‘initial coding’ codes the interview data ‘word- by- word’ or ‘line- by- line’. Second, ‘focused coding’ integrates ‘the most significant codes’; theoretical integration starts with focused coding (p.46). Although Charmaz (2006) mentions ‘axial coding’ to relate categories to subcategories (Strauss and Corbin 1990, 1998; Strauss 1987 as cited in Charmaz 2006: 60), but the frame of ‘axial coding’ might limit visions; therefore, if researchers prefer simple or flexible guidelines rather than a framed structure, ‘axial coding’ is not necessary (Charmaz 2006: 61). My research codes into essential phases, ‘initial coding’, ‘focused coding’ and then integrates them into theoretical level.
4.4.4.1 The first stage: Initial coding
‘Line- by- line’ initial coding, as Charmaz (2006) suggests, allows researchers who are open to the data to see subtle dissimilarities. Furthermore, it helps researchers to define actions and the implication of meanings, in order to compare the data and identify the gaps (p.50). The concept of the table is from Charmaz (2006).
Table1. Initial coding
The reason for producing governmental paid-for news.
Participant Ming Wu, QTV, 18 years’ experience, male
Lin: You have worked in Journalism for 18 years, from your viewpoint, why did the media produce governmental paid-for news? Ming Wu: They downgraded themselves.
Media downgraded themselves. The attitude of journalists: criticise. The reason for producing
governmental paid-for news Media do not serve the public. The attitude of ownership. Financial Issue.
Ming Wu: The point was…you know why… Journalists were not the main point but ownership.
Lin: Okay. You mentioned the point…But why?’
Ming Wu: Owners only care about
revenues…The news media do not serve the public, but business. Due to a profit-oriented stance, give ownership profits…everything is acceptable.
4.4.4.2 The second stage: Focused coding
After the first stage of coding is established, ‘focused coding’ synthesises the most significant codes. The crucial part of focused coding distinguishes which initial codes are worthy of categorising (Charmaz 2006:58). The concept of the table is from Charmaz (2006).
Table2 Focused coding
The reason for producing governmental paid-for news. The attitude of journalists.
The reason for producing governmental paid-for news.
Who journalists serve. The attitude of ownership. Financial Issue.
Participant Ming Wu, QTV, 18 years’ experience, male
Lin: You have worked in Journalism for 18 years, from your viewpoint, why did the media produce governmental paid-for news? Ming Wu: They downgraded themselves. Ming Wu: The point was…you know why… Journalists were not the main point but ownership.
Lin: Okay. You mentioned the point…But why?’
Ming Wu: Owners only care about
revenues…The news media do not serve the public, but business. Due to a profit-oriented stance, give ownership profits…everything is acceptable.
4.4.4.3 The third stage: Theoretical coding
Theoretical coding integrated from focused coding leads the researcher to tell an ‘analytical story in a theoretical direction’ (Charmaz 2006: 63). Birks and Mills (2011) argue that if a study does not get ‘a high level of conceptual abstraction that is beyond the level of description’, it is not grounded theory (p.119 as cited in Wilson 2011: 35). Glaser (1978) describes 18 theoretical coding families at a sophisticated analytical level, such as the ‘Six Cs: Causes, Contexts, Contingencies, Consequences, Covariances, and Conditions ’ (p.74 as cited in Charmaz 2006:63), ‘degree,’ ‘dimension,’ ‘interactive,’ ‘theoretical,’ and ‘type’ (Charmaz 2006:63). Glaser’s coding families distinctively refer to ‘a specific category’ at a theoretical level (Charmaz 2006:63). The concept of
theoretical coding was adopted; for example, focused coding ‘the reason for accepting governmental paid-for news’ was related to the theory of critical political economy. Focused coding, ‘who journalists serve’, was related to the theory of public service.