Coding in grounded theory analysis

Coding is an essential procedure in a qualitative analysis, which highly influences the excellence of qualitative research (Strauss, 2003, p. 27). Strauss (2003) provides a thorough explanation of three different types of coding: open coding, axial coding, and selective coding.

Open coding is the initial, unrestricted coding performed by the close examination of data ‘line by line, or even word by word’ (Strauss, 2003, p. 28) in order to open up the inquiry and produce concepts that seem to fit the data. At this point, the concepts are provisional, and every interpretation is tentative - it may or may not work, or may be modified, and the coding is grounded not only in the data but also on the experiential data and knowledge of literature which the researcher brings into the inquiry. However, open coding forces the researcher to break the data apart analytically. It is important for the researcher to ask a set of questions continually aimed at the generation of a core category that will be at the centre of the theory, such as:

• What study are these data pertinent to? – to remind the researcher that an original idea may not turn out to be that all.
• What category does this incident indicate? – to keep the researcher from getting lost in the rich data by forcing the generation of codes that relate to other codes.
• What is actually happening in the data? – to help the researcher define the main problem.

The data should be analysed thoroughly, and carefully coded to ensure conceptual density, and minimize the overlooking of important categories, and when a code seems relatively saturated and nothing new is happening, the researcher will find himself/herself moving quickly through the data, scanning pages until something new catches the eye. If some portions of the total analysis are not satisfying, another open coding, and even new data collection may become necessary.

At the point of open coding, coding should be frequently interrupted in order to write a theoretical memo, and the accumulated memos will move the analyst further from the data and into a more analytic realm. The analytic relevance of common variables should not be assumed – they should force their way into the grounded theory.

It is important for the analyst not to be too committed to the first codes, and to ensure that individual codes are verified and saturated. Open coding proliferates codes quickly, but the continual verifying that each code really fits slows down the process (Strauss, 2003: 32).

Axial coding – the analysis revolves around the axis of one category at a time, i.e. the intense anlysis of one category at a time is performed resulting in cumulative knowledge about the relationships between that category and other categories and subcategories. According to Strauss (2003), axial coding is an essential aspect of the open coding, which alternates with looser kinds of open coding, especially when the analyst examines new aspects of the phenomena under study.

Selective coding – coding systematically for the core category, i.e. the analyst delimits coding to only those codes that relate to the core codes , and the core code serves as a guide to further theoretical sampling and data collection. The analytic memos become more focused and help to achieve the integration of the theory (Strauss, 2003: 33).

The findings of a grounded theory analysis are supposed to be unique, tightly anchored in the data collected for the particular research purposes, i.e. in the words and experiences of the research participants. It introduces a new phenomenon and emphasises its discovery, whereas the description and verification are not its primary concern. So, the success of the analysis to a great extent depends on the researcher’s sensitivity and analytical skills, and it is important that the researcher constantly be aware of his/her own position in relation to understanding and conveying the messages of the participants.