By Andra Ioana Horcea-Milcu
When answering to reviews of qualitative or semi-qualitative interdisciplinary papers, I noticed researchers may be sometimes challenged to explain how they arrived to the qualitative results they present. This made me think several times about how I analyze my own interview data, which often comes down to explaining how I code the data, which is potentially more difficult than the coding itself. I would like to deconstruct the way I perform or rather see coding in this blog entry. More specifically, I would like to emphasize one of the aspects that, to my mind, is often overlooked. As a disclaimer, I would like to mention there is a lot of literature on the different technicalities, types, cycles, etc. of coding, and I believe there is no such thing as a single recipe to these. Here, I aim to provide an intuitive meta-view on coding while I am still learning about it.
Coding: a 3D process
Broadly speaking, qualitative coding is a way of operationalizing the analysis of qualitative data by labeling fragments of it (Bryman 2012). I like to look at coding as a 3D process (Fig. 1) having its origin (0,0) in the research question. Simply put, one dimension are the themes, another one the concepts, and the third dimension is the interpretation.
On the OX axis there is the actual content, the expressed themes, the substance. Coding along this axis goes more in the direction of thematic coding and content analysis. It is the more descriptive part of coding and in my opinion, it is also the most grounded stage of it. Themes may stem from prompted and unprompted content, i.e. emergent themes which may remain as a separate category.
I see the OY axis complementary to the first one. Here we find the underlying content, its form, the way it is expressed and constructed, the more abstract notions. This goes more in the direction of discourse analysis, axial coding, abstraction, conceptualization. It is the more analytical stage of coding. Here, I try to look in my data for more abstract concepts, sometimes guided by the literature, based on the refined codes pertaining to the first dimension.
Finally, I use my own knowledge and creativity to link the above together, and meaningfully “lift” the results above the data, in relation to the research question. Although generally more “invisible”, depending on the extent to which coding is aimed at building a theory, the importance of this axis (OZ) increases. It is thanks to this dimension that we are able to interpret data (i.e. attribute meaning to codes, intermediate points on OX and OY). This is also one of the reasons why qualitative data analysis software cannot fully replace the researcher. The importance of this “interference” of the researcher with the result became clear to me when a more senior social scientist recommended me two editorials: Pratt 2009 and Suddaby 2006. The third axis is, in my opinion, the salt and pepper of coding. It is an area of creativity and where the art of coding becomes possible.
Qualitative results don’t just happen. Although qualitative results presented by papers may seem just as sudden as the black dot on Fig. 1, in reality they represent several dimensions. My understanding is that, depending on the research question, the result, what we report as a finding, is somewhere situated along these three axes (or more, as each of them can be further decomposed). Along the axes, we focus, we aggregate, we synthesize, we reduce, ultimately we extract. Processing and analyzing data along each of them is part of the result. For simplification, going along the axes could correspond to different iterative cycles of coding. This (at least) bimodal coding process is quite commonly described by several authors; e.g.: first-order and second-order codes in Pratt 2009, first and second cycle coding in Seldaña 2009, first-level and second-level codes in Tracy 2013. In reality, I think these dimensions are not so clearly separated and may happen simultaneously.
When coding, the researcher is inherent to the results. In my opinion this has interesting implications for the review process. It needn’t mean that one should cease looking for objectivity or repeatability or that coding isn’t pragmatic, but rather to acknowledge and recognize the characteristics of qualitative research. Although I am still learning about this, I think that the existence in the research design of elements indicating concern for quality criteria, internal coherence, triangulation, feed-backs, and transparency could increase the credibility of results and make the third dimension more embedded (and justifiable). Saving the coding tree at different moments in time allows for tracking back the way codes evolved through the analysis, if necessary. Keeping a record of iterations or metadata about codes, could be also useful ways to transform coding into a “HD” process.
I tried to briefly discuss the way I think about coding, and convey how I broadly code qualitative data. Probably, every researcher has a different way of approaching coding, and has built and adapted through time his or her own understanding of qualitative coding. There may be many other ways to decompose this process and explain how one arrived to specific results. However, openly acknowledging the third dimension of “this movement of data” would make communication among the more quantitative driven and the more qualitative driven scientists more amicable and enjoyable.
Bryman, A. (2012). Social research methods. Oxford university press.
Pratt, M. G. (2009). From the editors: For the lack of a boilerplate: Tips on writing up (and reviewing) qualitative research. Academy of Management Journal, 52(5), 856-862.
Saldaña, J. (2012). The coding manual for qualitative researchers (No. 14). Sage.
Suddaby, R. (2006). From the editors: What grounded theory is not. Academy of management journal, 49(4), 633-642.
Tracy, S. J. (2012). Qualitative research methods: Collecting evidence, crafting analysis, communicating impact. John Wiley & Sons.