|Anne Adams during a presentation|
While writing up my findings for my pilot study, I am lucky enough to follow a workshop on grounded theory here at the Open University. This post reflects some of the points raised by the presenter Anne Adams, so live blogging from the workshop.Starting from previous qualitative experiences of all the participants, Anne gets an idea of her audience. Anne who is very energetic and clearly so knowledgeable that she is open to questions from the floor at any moment. Anne’s approach is starting from the data.
Premise: it can start from mixed methods, as well as purely qualitative
Important for PhD justifying the methodology used.
Quantitative versus qualitative; either way reflective
With grounded theory the results come from the data of the participants, so the subjectivity of the qualitative is in dialogue with the predetermination of the quantitative.
Quantitative challenge: imposing external system of meaning for internal subjective structures, whereas grounded theory comes from the participants.
Qualitative challenge: generates working hypothesis by producing concepts from data, representing participants reality in its complex context. But here the researcher’s assumption does add to framing the data.
So research always has challenges through the instruments used. In ALL research challenges emerge. So being reflective is the answer to reach validity.
Grounded theory Background
Glaser and Strauss (1967) Glaser comes from quantitative, and Strauss from qualitative.
The theory is the end goal of their Grounded Theory approach. (Henwood and Pidgeon, 1992, p. 101): “both qualitative and quantitative approach…”
An important view when writing up the verification/argumentation: important view: looking at who will examine your research: educationalists, technologists…
Another important idea to remember is that Grounded Theory (GT) is an iteration. So the coding is not done linear, but iterative, where the linear is occurring in stages to find depth and meaning, after which the whole argumentation is thought through again.
Structured/focused approach to theory building
Integrating mixed data sources
It is a skill as a researcher that you can continually manage to combine detail to theory in a valid justification.
Quality rules (Henwood and Pidgeon, 1992)
- Importance of fit with the data
- Integrated at all levels of abstraction
- Reflexivity (always look for the why, justification)
- Keep documentation (field notes, memo’s, notes taking during the exploration)
- Theoretical sampling and negative case analysis. The sampling is very central to the process, because the experiment is not designed, the reason for selecting your participants becomes more important and should be clearly mentioned in your PhD account.
- Theoretical sensitivity (the methodological approach, you should not go in with a prior theory – in theory – this is seen by Glaser as polluting, but there are different flavours of GT. So you need to take a position on what you use, which GT you follow, Anne went in not with a framework, but being guided by some theory (Inge, wondering if this is more Charmaz?).
- Transferability: how far can your findings be transferred beyond your group. The GT purists would say that any theory coming out can only be related to that specific group, but there is an element of transferability to different contexts. So the sample might be generalizable to other contexts. This might also be of importance to your PhD dissertation, but you must be very clear on it not to ignite more discussion than necessary.
A qualitative approach to HCI by Adams: http://oro.open.ac.uk/11911/ (2008) and another one but not typing quick enough to get that one, scholar googled Anne Adams here.
Data in whatever form is: broken down, conceptualised, and put back together in new ways
Analysis stage – 3 levels of coding: open, axial and selective.
Open (concepts are identified, grouped into categories – more abstract concepts and hooks, properties and dimensions of the category identified. Each category might cover a specific property, and will have a dimension – and dimension range - either frequency it occurs, or scope that it has, the intensity with which it is mentioned, and the duration it refers to). A rule of thumb with open coding is to look for frequency and if it is only mentioned infrequently, than it might be fundamental (e;g. after that I never learned anything online again). The idea of saturation is the moment that you know you have gone far enough. Saturation will emerge where you no longer find fundamental new ideas.
Axial: start to move up from the categories, looking for high-level phenomena and conditions: causal conditions, contextual condition, intervening conditions. Phenomena action/interaction strategies and consequences identified by your participants. Where phenomena are central ideas or events. Whereas conditions are events that lead to occurrence or development of a phenomenon. The context is a set of properties (location, e.g.) that pertain to the phenomenon. Intervening conditions provides a light coming from a broader structural contexts (e.g. is it the individual, the organisational, the societal … which depends on the research question you started from and which you are searching an answer to). Action/interaction strategies: devised to manage, handle, carry out, respond to a phenomenon under a specific set of perceived conditions. Consequences – outcomes or results of action/interaction.
(e.g. when I want o have (context) a personal conversation (phenomenon), I encrypt the message (strategy), I think this makes the email private (consequence). )
Selective (latter with process effects):
· Select the core category (central phenomenon around which all the other categories are integrated) and high level story line (a descriptive narrative about the central phenomenon). The high level story line comes from your core category, so just a couple of sentences at the very most, that which goes into your abstract.
· Related subsidiary categories by its properties (the other things are all related – in many cases by the properties – to your core category)
· Relate categories at the dimensional level (it might be related dimensionally: some a lot, some less)
· Iterative validation of relationships with data
· Identify category gaps (telling what might be related, but is NOT what you are addressing with your research – patching holes provides the boundaries of your research)
At the end you – as a researcher – need to find a missing part of the puzzle
Lines between each type of coding are artificial
· Data presented at dimensional level
· Action/interaction and conditions present
The solution to come to these results are tools.
Very important: keep relationship coding notes in open coding/analysis without loss of detail, and code both open and axial together.
My additional question: who area non-purist GT theorists
Gilbert Grounded design flavour? (not sure about this, might have misheard it)
Focus is very variable, because GT has been adapted by many disciplines resulting in different GT flavors : so best is to look at your discipline and look at papers from that area.
Charmaz is really good for a intermediate approach that allows staring from some theoretical assumptions.
Atlas TI fits perfectly with grounded theory (more than NVIVO), within CREET group of licences for Atlas TI (Inge, you must ask whether CREET has licences to offer you as a PhD student)
Using an analysis tool makes it easier to keep the codes up-to-date through the overall process, even if you are changing the code names. Atlas works better with multimedia files (easy coding). BUT the tool should never stand in the way of rigor and personal work/research, if the tool feels restrictive, it is better to use physical options that work for you post-it notes, Word…