All entries for July 2018
July 22, 2018
Data analysis has dominated the past couple of weeks, although, whilst engaging with data analysis, I have been continuously engaged with other areas of thought and practice:
· The characteristics of the phenomenon of interest
· The nature, complexity, nuances, and functionality of the specific data source, including comparisons with other sources
· The nature and functions of the social learning context in comparison with other contexts
· Evaluation, critiques and reflections of thematic analysis so far, and comparative observations with other methodologies and methods
Essentially this encompasses four levels of thinking: the phenomenon itself, the data source, the more general learning context, and the research design. All thoughts and processes of evaluations, etc. are situated not just within the research context but also within the context of my philosophical beliefs.
Everything is a work in progress. As I progress through the data analysis phase, my thoughts, interpretations, observations, hypotheses and questions shall be continuously refined in order to more effectively reflect the true reality (remember, I am a realist) of what is occurring in the data. Coding is always a work in progress and all that I am thinking about, observing, hypothesising, questioning etc has developed from earlier coding efforts in the Ph.D.
As I shall be explaining more in the thesis, coding is not just a mechanistic act of labelling meanings and activities in the data, but is an active, engaging, dynamic, nuanced, flexible and adaptable method for analysing qualitative data that (I shall argue) plays a part in understanding the truth of what is happening in the data.
Currently, therefore, I am progressing through the “opening” stage of the analysis phase. This “opening” stage is based on the coding and reanalysis of the data corpus. I am continuously revisiting what I have coded before, and continuously reanalysing and recoding, in order to ensure that the codes are as reflective of the nature and function of the data segments as possible. This shall then help to develop themes that, although constructed on a more theoretical plane, are as close to the data as possible.
I am breaking the context of the data corpus down stage by stage. In the first stage that has been ongoing for a few months on and off, I coded all the way through the data corpus without much thought for nuances and context. It was simply a matter of initially understanding the meanings and functions of the data segments though if nuances and contextual influences were immediately obvious then these would be considered.
What I am doing currently is the next level: I am breaking the data corpus down and really exploring the context and nuances of each data segment, along with developing an understanding of the way in which these segments logically connect with and relate to each other on various levels and various purposes. Additionally, this level involves the rechecking of codes to ensure they reflect the reality of what is being expressed in the segment, and to alter the codes if necessary. This deeper approach to understanding the data is in my view more relative to the research questions.
The study of the nuances and contexts is based on what I have observed during my time of using grounded theory, and which led to moving away from grounded theory as has been documented on this blog and which is being documented in the thesis methodology chapter. It is all ultimately based on what I perceive and interpret within the data, but this is not a subjective, relativist approach. As a part of the theme development I shall be exploring the codes and segments again and test all that I observe. E.g., just because I have coded a segment to represent a particular feature or activity does not mean that I am objectively correct: this correctness, perhaps, comes from repeatable observations of similar data characteristics. This idea is taken from the abductive reasoning method. This shall be discussed further at the time of theme development.
Along with the coding, I have been writing theoretical memos (an aspect of Grounded Theory I have liked, so have included it in my own approach), which serve the purpose of documenting and recording all my thinking, observations, thoughts, hypotheses and questions about each data segment, and also of the meaning, nature, function and representativeness of each code.
This coding level is ongoing and work in progress, but there are already some interesting insights and points of discussion. Nevertheless, my understanding of the relationship between segments, the impact of contextual and situated conditions, and the emergence or development of meaning and activities shall continue to develop and refine as I progress through this analytical phase.
All this shall lead onto the development of themes, which operate and are constructed at the latent level and are constructed through combining, in some way, multiple, different, though similar codes (as discussed in the previous post: I shall be talking more about the development of themes soon). My understanding of themes so far is leading me to think of a theme as a core aspect of a phenomenon of interest that describes and explains the phenomenon’s behaviour and helps to characterise its theoretical existence. Thematic theoretical insights are drawn from the data, and tested against the data.
Speaking of themes, I have made enough observations in the data to tentatively suggest the existence of two themes, and the way in which these themes could relate to each other. At a push I could suggest I have observed four themes, but I am not convinced or at least not as certain about two of the themes as I am with the first two themes I came to observe. These themes, and possibly more, shall be identified, defined, developed, and established following this coding phase. At the moment I have put the thoughts of these themes aside as I do not want to restrict my thinking and open mindedness during the rest of the coding phase. There is a danger that if I did become too fixated in the idea of exploring to prove these themes, I might miss out on something that might be obscure but is equally as important.
That’s over a thousand words and I haven’t scratched the surface!
I intend on writing some more posts during the week related to the four points made at the beginning of the post, but honestly, I’d rather focus on data analysis. But when I get the chance I shall post up more posts!
‘Till next time!
July 07, 2018
I have managed to code through the entire data corpus, involving the development and assignment of codes to relevant data segments; codes that capture the meaning of the assigned data segments, along with embedded theoretical memos within the data. These memos explain the nature, function, context and meaning of the code and the segment’s content and any other relevant thoughts, hypotheses and theories related to the content. However, as I was thinking about the next stage I stated to doubt myself and asked myself the main questions:
What is the real meaning of a theme?
How is a theme really constructed?
What type of theme should I be constructing?
These questions reflected the doubts that I had at the time of my understanding of what a theme really is, and the depth and breadth of which I should involve myself with theme analysis and development for the purposes of my research. These questions are continuously asked but I appear to have some clarity in my rereadings and exploration of the literature. I knew at the time the process of making themes but there was something that bought doubt into my mind: is there really no step between developing codes and developing themes? I wasn’t convinced, and hence the formation of the questions and the subsequent reading of literature. Doubt in this case has been used as a means, a process, of developing questions and of endeavouring to explore topics further.
From what I can understand of the literature, there is varying terminology to refer to the same type of theme but for the sake of brevity I shall focus on a couple of authors who are becoming key writers for my understanding and application of thematic analysis.
Braun and Clarke (2006) define the themes as semantic or latent. Semantic refers to theme development based on just the surface level meaning of the data; essentially, the researcher is not interested in anything beyond what is said literally within the text. There is therefore, from what I can understand, no attempt at understanding context, nuances, variety, diversity and deeper meaning at the semantic level. Semantic level is essentially considered to be a descriptive level of meaning.
At the latent level of theme development, however, there are attempts at going beyond the semantic level and into the realm of interpretation, assumptions, concepts, conceptualisations, meaning making, hypothesis making and theorisation. From what I can understand, Braun and Clarke (2006) describe theme analysis and development as a progress from the descriptive level to the level of interpretation and theorisation. What is identified at the semantic level is taken beyond the obvious and observable to what can be known and understood through theories and interpretations. The latent level, however, is not grounded on hairy fairly assumptions as the latent level assumptions and theorisation processes are grounded in the semantic level. Therefore, what I find or observe at the semantic level I can theorise, hypothesise, assume, and make meaning of their existence, functionality, purpose and context.
This actually makes sense, because how can I possibly stop at just a simple observation? How can I simply consider the existence and meaning of something at only the semantic level and not at the latent level? It doesn’t make any sense to me just to observe and know something at the semantic level: I am immediately drawn to theories, well grounded assumptions, hypotheses, and meaning behind existence and function. Is that because I have an academic mind? Can I perceive beyond the observable? Can I understand meaning and function beyond what is right in front of me and clearly observable? Surely I can if I am drawn to this level of understanding?
Moving forwards, I have this understanding now of semantic and latent themes so surely it is common sense that thematic analysis consists of both themes? That my research would involve the construction of both? According to the approach to thematic analysis by Braun and Clarke (2006), I would be correct.
But wait, there’s more!
Category or Theme? Should we consider both?
After spending a long time pouring over methodological papers about thematic analysis and the idea of theme development, I had more questions than answers. I came across literature that was not only encouraging me to doubt and question what to do in the next stage (I shall discuss this more in a future blog post), but also encouraged me to question my own understanding of what a theme is, and also what a category is. Is it not true that categories are an integral part of grounded theory and therefore I should not worry about them? If only our attempts at understanding the world, of social reality and all the components of social reality were that easy!
Methodological authors differ in their description and discussion of the theme development level and of the definition of categories and themes. After a long while of reading however, I am beginning to lean towards discussions around the likes of Vaismoradi et al (2016), who suggested that the thematic analyst considers both the category and theme, where the category represents the semantic content whilst the theme overarch the category and represents therefore latent data.
What is interesting here therefore is that a theme could consist of multiple categories although some authors name categories as sub-themes. Categories or sub-themes themselves are constructed through the grouping of codes; categories therefore describe and functionalise a group of codes and describe their general meaning. From what I understand of the literature and particularly Braun and Clarke (2006) is that categories (or sub-themes) are constructed first before they are them grouped into themes. But it’s not as clear cut as that, because I’ve just recently read another paper and the author suggests that there is no need for theme development and automatically considered their codes to be themes………..
It is a minefield, but the way my mind works I like the idea of progressing from codes to categories to themes (and, therefore, from semantic or manifest data, different authors label them differently, to latent data; from observation to interpretation and theorisation).
What did I learn from that process? That the whole idea of building themes is to move from semantic or manifest level to the level of interpretation and theorising and this makes a lot more sense to me now and comes to me really as quite obvious. Also reflecting back on the process I have used so far this is something that I have always done, I just wasn’t familiar with the terminology! Also, categories themselves are complex and used in different contexts. Previously I thought categories were terms and features exclusive to grounded theory, but categories are general terms but it appears to me that categories are used differently depending on the research method used. Within grounded theory, they are used to build towards a theory whilst in thematic analysis they are used as part of building understanding and not a theory.
I was right to doubt, because I was able to realise and recognise where I have to build my own understanding. This is an ongoing process, but the more I use thematic analysis and read the relevant literature the more I can understand the way in which it relates to the coding process I am carrying out, and the way in which themes can be used for the next stage of the research.
‘Till next time!
Braun, V., Clarke, V (2006): Using Thematic Analysis in Psychology, Qualitative Research in Psychology, 3 (2), 77 - 101
Vaismoradi, M., Jones, J., Turunen, Shelgrove, S (2016): Theme Development in Qualitative Content Analysis and Thematic Analysis, Journal of Nursing Education and Practice, 6 (5), 100 - 110
July 02, 2018
Like a toddler running back and forth into the arms of those that love that child, ideas and visions that were previously considered irrelevant or perhaps not suitable for this project but might be for another project have been running back to me like that happy little toddler. Everyone say aww……..
(Oh by the way, I’m not at all suggesting that toddlers are irrelevant! Even if they turn into screaming delightful door slamming teenagers…………..)
The day has been a productive coding session. As I have been coding the data and observing patterns and meanings within the data, I have come to realise that certain patterns and meanings that were once considered irrelevant are now becoming more relevant and, also, I have observed new patterns and meanings that I had not previously observed when previous sets of data were coded. Or at least, new patterns and meanings that have not made themselves obvious till now, even though I might have observed them before but had not consciously acknowledged them, for whatever reason. I think this is a psychological thing: the more you become sensitised to a particular pattern or meaning you start to think later in the coding process that you have observed similar before in different contexts and then you start to identify the bigger picture or wider pattern of behaviour. It’s a very interesting and a very involving process. What I have found during the day is making me rethink what I have coded previously, and the way in which I have interpreted and perceived what is occurring in the data, which might lead to recoding the data again as I go through a more deeper coding phase as I go further into building an understanding of the phenomenon of interest. I’ll be talking more about this in another post later this week.
In the meantime however it is clearer to me now more than ever, and what might be good practice for other Ph.D. candidates to adopt, not to throw away any old ideas and visions that were previously considered irrelevant. This is an approach that I have adopted from the beginning of the Ph.D., as I have folders upon folders of books and research papers and thesis related documents and notes, and a fair percentage has been sent back and forth between the archive folders and the working folders as they were continuously examined for relevance at particular times of the project so far.
Now some of the oldest ideas and visions I had right at the earlier stages of the Ph.D. are becoming more relevant for answering my research questions and addressing the research problems. But more than that: what I was writing about earlier in a theoretical memo that documented my thinking of what I was observing was an attempt at building upon those earlier visions. It’s really interesting when you have built your earliest visions upon a section of existing literature and then to observe what you thought was irrelevant within the data brings back home the thinking that nothing is really impossible. There is a slight problem, however.
It is a fair way into the reanalysis and coding phase that these older ideas and visions have occurred, so this leaves me with a couple of questions. Do I carry on with the coding and analysis and simply suggest at what point I observed a new aspect of a phenomenon to be relevant? Or, do I reanalyse the data again and code for these additional observations that I made later in the coding?
Methodological literature that I have come across so far has not been clear on this subject although it is a subject I shall read more about. I have come across a paper that did suggest that you don’t have to reanalyse the data to code any new observations but this from what I remember was associated with grounded theory based Open Coding, where you are basically coding to build a theory and not coding to identify and relate themes. I am leaning towards yes, I would have to recode the data to code for more instances and examples of what I have observed in order to validate and authenticate the existence of what it is I have been observing.
Of course this then leads onto other philosophical questions such as does repeatability really represent truth? If you observe something often enough does it really exist in an external reality or does it exist within our own interpretations? What about if others are not able to perceive or observe what a researcher finds observable? In what way can I tell that something might exist in an external reality? In what way can I possibly know what I know to be true? These, and more, are challenging questions, but the key I think is to keep everything grounded in the data and make sure that arguments and observations are built from the data. You cannot build from existing theory; you can, however, build from a relationship between data observations and existing theory, but I shall cover that point at a later time.
With all that in mind, what I am thinking about is to analyse the data but keep the original copy of the data and embed evidence of a change in perspective or the observation of a potentially key new theme. This would be in the form of a theoretical note embedded within the data that would mark precisely the point that I began to observe the importance and relevance of an event or meaning that could form a part of a theme. This would show and evidence the progression of thinking and the way in which my thinking and thought pattern progressed to the point that I began to observe the importance and relevance of what it was I was observing. I am not really sure what the literature says on this subject, but I am becoming convinced that this might be the best approach.
The key lesson here really is, don’t throw out your old ideas. Whether that idea is represented as a few lines of writing on a scrappy piece of paper or rushed serious of paragraphs on the word processor, keep it! Archive it or put it in some relevant folder or whatever storage system you have so that you can refer back to those ideas in the future if they prove to be relevant. Another lesson is don’t focus your mind exclusively on what you found previously.
In other words, don’t code one set of data and then focus the next set of data on what you have discovered before (I know this is rather a contentious point in academic discussion from what I can understand about coding approaches and debates) (another contentious point is whether or not anything is actually discovered at all, but is actually interpreted), but keep an open mind. Of course what you find whilst you are coding and thinking about the data is exciting, overwhelmingly exciting, but keep a level head, keep an open mind, and don’t be distracted by what you have observed previously. If you become too focussed on what you have observed previously you’ll begin to lose the meaning of innovation and originality, and become potentially enslaved by previous observations. Keep an open mind and keep coding for original insights and meanings, and think and plan carefully to determine if there is a real need to reanalyse the data when you find something new a fair way into your data analysis process. This really depends on your research questions, research problem, and the way in which what you have observed relates to explaining the phenomenon of interest.
‘till next time!