Chapter 3 Methodology
3.8 The thematic analysis framework for this study
3.8.2 How did the themes emerge and how were they reported for this study?
According to Alhojailan (2012), any qualitative data collection is also reliant on its interpretation. This means that the data firstly requires to be organised. This is important as a large amount of qualitative evidence was gathered from the thirty nine interviews conducted for this particular study. Gibbs (2008), has also articulated some the steps which were followed alongside the framework provided by Braun and Clarke (2006), to generate the themes for this study. This helped to generate more efficient and robust outcomes.
The first step is to compact extensive and diverse raw data into a succinct structure. This could be achieved by organising data into charts and tables. This provides the researcher the opportunity to identify, compare and determine the data upon which to focus. As shown in
figure 18 the responses of each interview topic for the study were clustered together from
each of the interview transcripts. This enabled the researcher to get a feel of the data and conduct an initial interpretation. This was an important stage in terms of organising the data in a structured manner.
Cohen, Manion and Morrison (2011), suggest that in a thematic analysis there is often little distinction between data collection and its analysis, as both of these can be simultaneous. The second step hence followed for generating themes in this study was data familiarisation. This was done by using colour coding to highlight the key segments of data for each interview topic. This enabled the researcher to get closer to the data and recognise the patterns that were beginning to emerge. At this stage extensive handwritten notes were also made on the transcripts to capture the initial interpretations of the researcher in more detail.
According to Gibbs (2008), the third step for generation of themes was data labelling. This is done by coding the data by systematically marking it. In the case of this particular study as shown in figure 18 each informant was allocated a unique label to anonymise their identify. The data was further organised by using the Microsoft word software to allocate numbers for each line of text. As recommended by Ryan and Bernard (2003), data was then coded to capture repetitions, typologies, metaphors and any analogies used by the informants. The codes also included words that captured any particular significance of something to the
informants. The coding of data was done manually rather than using recommended software such as NVivo. This manual approach to coding was preferred as Welsh (2002), had concluded that software is a useful tool to organise the data and provide surface level analysis. It is however restrictive in the case of a thematic data analysis. This is because of the fluid and creative manner in which themes can emerge. The manual coding allowed the researcher to get closer to data by studying it repeatedly. It also allowed for refining the codes to ensure that they were valid and reliable, and to ensure that they stood up to the scrutiny of two independent reviewers.
According to Braun and Clarke (2006), irrespective of whether the data was coded using software or manually, it is more important that consideration is given to what comprises a theme. The fourth step therefore for this particular study was of ensuring that emerging themes were interpreted and connected to tell a coherent story. This is demonstrated in figure
18 where initial themes were coded to ensure that they reflect something important about
the data in relation to the overall research question for this study. These initial themes are representative of patterns occurring within the data set. In terms of what constituted an initial theme prevalence in terms of space within each interview topic and across the entire data set was an important criteria. The criteria of ‘keyness’ where a theme is not necessarily reliant upon quantifiable measures, but in terms of whether it captures something important in relation to the overall research question was also applied. This ensured that the emerging themes were not just based on the number of occurrences, but also balanced with equal importance given to smaller yet significant patterns.
The fifth step that underpins themes generation for this study according to Gibbs (2008), is the need to now review and organise the themes. This Braun and Clarke (2006), suggested involves moving now from semantic to latent themes. The semantic themes are mainly descriptive with some interpretation, whilst latent themes are rich in interpretation and analysis. The semantic approach in this instance was applied in terms of understanding the links between the more prevalent themes, and the less prevalent sub-themes. As shown in
figure 18 the themes were organised at three levels. The levels one and two comprised of the
smaller sub-themes, whereas level three were qualified as a major theme emerging from a particular interview topic. This review and organisation of themes in such a manner along with
another objective scrutiny from the two independent reviewers, further added to their reliability and validity.
The sixth step utilised for generating themes was finally the reporting of themes from transcribed data, an example of which can be found in Appendix 3. According to Bazeley (2009), this is when the themes attain full significance as they are linked to tell a coherent story. The figure 18 depicts that themes are reported in Chapter 4 of this study with level one and two being as the sub-themes, whilst level 3 as the major theme from the specific interview topics. Bazeley (2009), suggests that a coherent story can be arrived at following a three step approach of describe, compare and relate. The ‘describe’ is to articulate the characteristics and boundaries of the theme. The ‘compare’ is to establish if the theme occurs with varying frequency for or how expressed by different groups. The ‘relate’ is about evaluating if the theme arises under particular conditions, the actions involved, and possible implications.
Process Example – Interview topic: Universities becoming like businesses
Step 1
Transcription – a raw transcribed exact
from an interview
I think universities have become like business for some time, they operate in a particular market place, with different issues, they are not about selling goods, but they do sell a service, that service is not a degree, it is the opportunity to study for a degree, I think they are autonomous and not public sector, although traditionally marginally funded by the public sector, so it is right to regard them as businesses, I don’t have a problem with that, what I would have a problem with is if you lose sight of what your business objective is – which is delivering your programmes to students.
Step 2
Data familiarisation – starting to highlight some key segments
I think universities have become like business for some time, they operate in a particular market place, with different issues, they are not about selling goods, but they do sell a service, that service is not a degree, it is the opportunity to study for a degree, I think they are autonomous and not public sector, although traditionally marginally funded by the public sector, so it is right to regard them as businesses, I don’t have a problem with that, what I would have a problem with is if you lose sight of what your business objective is – which is delivering your programmes to students.
Step 3
Label
Data labelling - into categories of pre-92, post 92, private and other informants and individuals labelled, lines numbered for data location
Name – Person interviewed University
Job title
Interview number Post-92 universities Code - MU/MG/Staffs/VC/1 =
1Answer: I think universities have become like 2business for some time, they operate in a particular market place, with different 3issues, they are not about selling goods, but they do sell a service, that service is 4not a degree, it is the opportunity to study for a degree, I think they are autonomous 5and not public sector, although traditionally marginally funded by the public sector, 6so it is right to regard them as businesses, I don’t have a problem with that, what I 7would have a problem with is if you lose sight of what your business objective is –
8which is delivering your programmes to student Step 4
Generating initial themes
Not losing sight of core mission(linked to values dilemma) important
Shift in government policy wishes to see students being consumers
Formalisation of HR, professional marketing, ‘big things aren’t they universities’, shift last 10 years, more noticeable now – Economic footprint
more business like, policy shift – fee regime ‘Cost is now with the consumer’ ‘big change’ Step 5
Reviewing and organising themes
Literature debate – Universities have become like businesses
Level 1 themes based around what different categories of informants think Level 2 the key overall subthemes
Level 3 the key overall theme
Step 6
Emerging themes
Level 1: Post 92s (inevitable), Pre 92s (identity), Private (clarify), Others (Polarisation)
Level 2: Relational shift, Cultural shift, Values conundrum Level 3: Business like