Processes for analysing qualitative data
Data analysis, in qualitative research, entails systematically organising and making sense of qualitative data that are collected by the researcher to gain a better understanding of the phenomenon under investigation (Figure 6.1). The process of analysing qualitative data involves coding large amounts of data, which can
Figure 6.1 Processes and levels of abstraction in qualitative data analysis. Source: Original.
include academic literature, interview transcripts, observational notes, document reviews, or other non-textual material (e.g. pictures and videos). Codes are tags or labels attached to chunks of data of varying size, which could be words, phrases, sentences, or whole paragraphs.
Coding will depend on whether the themes being developed are more theory-driven, following deductive approaches, or data-driven, using inductive approaches. In the former, themes would be pre-determined and the researcher would code data that fit with the themes, i.e. a “top-down” approach to theorising. While in the latter, themes would be emergent as coding would be conducted in an inductive manner without the researcher being influenced by prior literature or experience, i.e. a “bottom-up” approach to theorising. In other cases, it might be useful for the researcher to use or integrate both approaches (Ali and Birley, 1999). For example, in a systematic literature review, Sarhan et al. (2019) adopted a “deductive-inductive” reasoning approach to data coding and analysis (Figure 6.2) to explore the synergies and inconsistencies between “Lean and Sustainable Construction” theories and practices. In the study, Sarhan et al. (2019) followed a “lean coding” approach (Creswell, 2007, p. 152) as opposed to purely inductive coding approaches, where researchers might struggle to reduce the numerous lists of generated codes to the five or six main categories or themes that they should arrive at for most publications. In lean coding, the researcher would start the data-coding process by developing a shortlist of five or six main themes with shorthand codes, and then expand and refine the coding structure inductively as they continue to review the literature.
Figure 6.2 A deductive-inductive (integrated) reasoning approach for data coding and analysis.
Coding requires researchers to make sense out of the data being analysed. This includes asking themselves the following questions:
- • What is this data saying?
- • What does this represent?
- • What is this an example of?
- • What kind of events are at issue here?
- • What do I see is going on here?
- • How has this happened?
- • Why has this happened?
- • What is trying to be conveyed?
Details of the process that should he followed when analysing qualitative data, be it for primary or secondary qualitative data, are summarised in Box 6.3.