How to code qualitative data? (PART2)

The first part of this series was about the logic of grounded theory coding, where the point was to emerge new theories out of qualitative data. It required a fully open approach and a flexible coding strategy. 

Although ‘grounded theory’ is an excellent, well-structured, and precisely defined qualitative approach to discovering new constructs and enriching existing theories, it is one of the best methods to explore and understand a topic in a context. It is especially fruitful if we enter the research field without significant previous literature. Grounded theory methods enhance not only the researcher’s theoretical sensitivity, but provide an opportunity to develop broad open questions. 

However, sometimes we have quite definite research questions. For example, in the case of a mixed-methods design, you might start with a quantitative strand, and test the initial hypothesis based on quantitative data. Here you have already some idea about the investigated phenomena and the relationships between the variables, and the qualitative phase is about seeking explanations (it is called explanatory research design). Instead of the grounded theory approach, more structured and formal analytical schemes are necessary. Using the phrases of Graham R. Gibbs (2007), instead of data-driven or open coding, you can apply concept-driven coding. 

Benjamin Crabtree and William Miller (1999) offer different techniques in their book, Doing qualitative research, to code and analyze qualitative data:  

Template method

The template organizing style uses a codebook that contains relevant categories and themes according to an already identified theoretical perspective. It is a flexible but structured technique, where you use an ‘a priori’ coding template, usually based on a subset of your data that you revise and reapply (King, 2012). During the process, you narrow, expand or merge your initial code set to get the final template. The template method usually uses hierarchical coding, including broad themes encompassing successively narrower, more specific ones. This coding method is predominantly a deductive, or top-down type of coding in the sense, that you start with a set of codes, however, it makes deductive and inductive analysis both possible.

After having an initial codebook, the first step is reading through the data, and marking the segments that provide answers to the research questions. You will search for relevant perceptions, and experiences also called as ‘themes’ within the texts. Coding is the process of detecting these themes and attaching labels to index them. Where you can identify the segments based on the ‘a priori’ codes, you can code them as such. The codebook can be modified or some codes can be even deleted if they seem to be inappropriate. You are ready when you coded your whole data successfully, and then you can continue with phrasing your findings.

Immersion/crystallization

This research strategy is not different from the template method from the aspect of coding. Here the researcher involves however not only focus group discussions or in-depth interviews but other information too into the analysis, thus e.g. observation, introspection, and media content analyses as supplementary methodologies. This means, that the database can be bigger and more diverse, thus, more challenging from an analytical perspective.

Editing method

Editing is the hermeneutical approach of the ‘60s. Here, the researcher identifies text segments within interview transcripts and arranges texts until a reduced summary reveals. This method stands close to the grounded theory approach. There is no initial focus on any aspects or phenomena, consequently, there is not any set of a priori themes. You, the researcher have the time-consuming task to find the most significant features. The advantage of this method is, that you do not try to fit the data into predefined categories, so there is a lower risk of overlooking significant but new considerations.  

In sum, coding is an important step in qualitative data analysis, because of three reasons: 1) it enables other researchers/colleagues to review your analysis methodologically or systematically; 2) you can be aware of the potential bias; 3) coding can provide higher efficiency in meeting your research goals.

This two-part series gave insight into the wide range of methods and approaches you can apply and also provided some viewpoints on selecting the most suitable for your very own research purposes. Don’t forget, that applying robust research methods is the cornerstone of accurate and unbiased results. Only robust methodology can ensure that the conclusions are as valid and scientifically sound as possible.

Recommended literature:

Crabtree, B. F., & Miller, W. L. (eds.) (1999). Doing qualitative research (2nd ed.). Newbury Park, CA: Sage Publication.

Gibbs, G. R. (2007). Thematic coding and categorizing. Analyzing qualitative data, 703, 38-56.

King, N. (1998). Template analysis, in Symon, G. and Cassell, C.(eds.) Qualitative Methods and Analysis in Organizational Research. London: Sage

King, N. (2012). Doing template analysis, in Symon, G. and Cassell, C. (eds.) Qualitative Organizational Research: Core Methods and Current Challenges. London: Sage