Coding in confirmatory research

A topic in research methodology

Coding is usually the first phase in data analysis, where codes are assigned to features in the data.

The origin of the codes used is usually distinct in confirmatory research, and in discovery research.

Coding in confirmatory research:

In confirmatory research, a codebook may be developed which specifies the codes and precise rules for applying them. Using an analytical framework drawn from the literature provides grounds our work in previous scholarship; and allows us to develop operational definitions of categories in advance to support reliable coding.

However, Using an analytical framework drawn from the literature limits what we can discover in the data to what can be represented in the category suggested by existing research; privileges the researcher’s way of thinking over participants’ ideas and concerns; and so limits the building of new theory.

The reliability of coding may be tested in various ways, to check that data would coded the same way at different times, or by different coders. Studies often quote the ‘inter-rater reliability’ when several coders are involved in coding data, showing how similar the coders code a sample of data. Often iterative approaches are adopted to ensure discrepancies are solved and reliability is improved.

The risk of 'rhetorical' research

Some studies which analyse qualitative data according to analytical frameworks drawn from the literature are not testing theory in a strict sense but simply testing whether the theory offers a (i.e. one, not the) viable way of making sense of the data:

This is a limited test of a theory!