Constant comparison / joint coding and analysis

A key component of Glaser’s method is that processes of generating data (theoretical sampling), coding that data, and analysing occur in parallel (constantly).

The constant comparative method is designed to aid the analyst… in generating a theory that is integrated, consistent, plausible, close to the data – and at the same time is in a form clear enough to be readily, if only partially, operationalised for testing in quantitative research”-p103

Glaser describes the constant comparative method as 4 concurrent stages:

  1. Comparing incidents applicable to each category,
  2. Integrating categories and their properties,
  3. Delimiting the theory, and
  4. Writing the theory

In a little more depth:

Comparing incidents applicable to each category:

This involves allocating each incident the researcher encounters to the categories in which it fits, and to create new categories if they become apparent. In my case I will consider new categories as draft – but won’t demark them differently – I’ll assess later down the line if a category saw very little uptake, whether it needs further saturation, or reflect on whether my creation of the category was useful.

In Glaser’s interviews about care of dying – he noted that nurses would say things like “he was so young” or “what will the children and her husband do without her?” – which he allocated to the category of social loss.

RULE 1. “while coding an incident for a category, compare it with the previous incidents in same and different groups coded in the same category”- p106

2 types of categories will emerge – those I create myself – and those that emerge from the common parlance of the people I’m talking to. According to Glaser, as theory develops, its likely that my category words – are sort of – explanation words, where the lay terms are more likely to refer to behaviours and processes.

As I allocate a few incidents to a category – I will start to note their incongruencies. To capture my thinking and also to ‘allow’ those conflicts to persist – Glaser has a second rule:

RULE 2. “stop coding and record a memo on your ideas” p107

Field notes can also have memo’s added to them – even directly on the page – as it essential to capture all the thoughts and ideas as they appear.

The act of coding, noting ‘parts’ of a story as incidents – breaks that story into components. Glaser sees this as an essential part of how these parts – incidents – allocated to categories and properties can then be regrouped into theory. My natural inkling is to resist reducing a storyline to a series of parts – so I’ll have to see how this goes in practice.

2. Integrating Categories and their properties

The activity of integration begins by comparing initial incidents, as the research progresses, incidents are compared with properties of the categories (that emerged from the initial incident-to-incident comparisons).

This latter activity initiates the integration of properties of a category – i.e. indicating how these are related – from which a sense of a unified ‘whole-picture’ can begin to emerge as the researcher establishes some theoretical sense of the data.

3. Delimiting Theory

The process of constant comparison could easily become an overwhelming task. In picking up a system by a handle one is rapidly confronted by the system in it’s un-ending entirety!

Delimiting is the process which bounds and constrains the system – at both the level of theory and categories.

At the theoretical level, as the whole picture solidifies, fewer major alterations will be made to it. Later edits tending to be centred on further integrating, clarifying, reducing and ‘cleaning’ up any incongruence.

Reducing at the theoretical level is a process of simplification and clarification – crushing down the categories in light of underlying similarities. When reducing terminology, it’s possible in this process to formulate some formal theory from substantive data. In the case of Glaser’s research into dying – he explains that the nurse’s attitudes and story creation about the value of the person and social loss that might be associated with them – “could be generalised to discuss the care of all patients (not just dying ones) by all staff (not just nurses)”. Generalising further, this could be used to underpin a theory of how professionals values underpins their distribution of service to clients. (p110).

[nb reminder – where I’ve typed Glaser / Glaser’s I need to edit this to say Glaser & Strauss]

The continued act of reducing takes the researcher towards 2 essential ‘needs’ of theory –

1. Parsimony of variables and formulation,

2. Scope in the applicability of the theory to a wide range of situations – p111

At the theoretical level it is my intention to additional delimit across 3 strata – based on Adomssent (2013):

1. Analytical Level, 2. Normative Level, 3. Operational Level

[include section on Adomssent 2013]

Reducing at the category level is a part of the process of developing congruence for the researcher. Initially wide categories are proposed to take into account every little difference – because the researcher has yet to have access to data-based-confidence in their fledgling theory. As further data emerges and categories can be reconciled, the researcher is able to see that the theory ‘fits’ the data and can ‘believe’ in it. This seems to be a significant process – from doubt and chaos towards increasing structure and certainty. During this timeline, the researcher can reduce the breadth of constant comparison – and become more targeted in which incidents to analyse and code.

In parallel categories become saturated – for the researcher this means – once several incidents have been allocated (coded as Glaser calls it) to a category, further incidents can be considered thus:

Will it add something new / different? No? – no need to code. Yes – code and compare. The decision not to allocate / code an incident to a category is a reminder that -the point is to seek data to inform the theory, not to add bulk to it – i.e. this is a methodology for developing a theory, not for testing and validating it statistically / quantitatively. If I decide that counting incidents is important – then of course I can allocate and compare incidents always, however given the time consuming nature of this, and that my purpose setting out is to develop an initial theory, I doubt I will be coding everything I encounter.

[reminder to go back through and reconcile where I’ve used the term allocate and where I’ve used the term code]

[reminder that I want this to be legible for lay people – and that Glaser’s language can be impermeable – there is a lot of rewriting to be done here]

There is also a mechanism here for coding incidents that are remembered but were not recorded at the time (p112). If the incident refers to a now saturated category – no need to record it. If the category isn’t saturated, the incident can be contributed as a memo. Later – it may be deemed necessary to follow up on the memo and go back to the field – but if the category isn’t of major significance to the theory-under-development – it may be adequate to just have the memo.

[Reminder to ensure clarity around page number and referencing Glaser – page numbers may need to be turned into fuller references if there is any scope for confusion]

4. Writing Theory

At this stage in the process of qualitative analysis , the analyst possesses coded data, a series of memos, and a theory” p113

Once the researcher feels that they their structure provides a reasonably accurate model of the system, and that it is in a form that others trying to look into the same thing – could use, then writing can begin.

Here the memos about each category can be used as content to explanations of the categories which provide the major themes of the theory. The researcher can use details from the data to back up their theory and to connect the reader to the appropriate place in the dataset.

Barney G. Glaser & Anselm L. Strauss. 1967. The Discovery of Grounded Theory. London: Weidenfeld and Nicolson.