Conceptualizing Qualitative Data

Why Do I Need Concepts?

“Do I really need to infuse my article with concepts?”; “I consider my data to be novel and useful, so why do I need to add in abstract concepts or an overarching theory?”; “I don’t really consider myself theoretical in orientation, rather, focused on practice, policy and real-world consequences. Isn’t that enough?”; “I have some interesting concepts which elevate the data, is this sufficient, or do I need a full-blown theory?” These are the types of questions that often come up, and these and other questions speak to the importance of recognizing the spectrum of approaches to data conceptualization, and the importance of fostering diversity, and in turn ensuring that we challenge some of the misunderstandings that circulate this terrain.

In essence, conceptual development in some form or another is central to quality qualitative analysis—at least in most instances—and it is thus much more than merely coding or theme development (Charmaz, 2015). The process of conceptualization essentially works toward an explanation of what is going on with your data, rather than merely summarizing it. It is useful to view conceptual development in qualitative research as operating on a continuum. One way of articulating this, although certainly not the only way, is that theory exists on a spectrum—from the use of macro (sometimes articulated as “grand”) theory to more meso or micro levels of conceptualization. Some find it useful to articulate this using Merton’s notion of the middle-range theory—(Merton, 1949). Mid-range theories are probably the most common in the qualitative health research literature, representing a form of conceptual elevation—from the data itself or “theme”—but not a universal explanation as such, or as Merton stated:

theories that lie between the minor but necessary working hypotheses that evolve in abundance during day-to-day research and the all-inclusive systematic efforts to develop a unified theory that will explain all the observed uniformities of social behavior, social organization, and social change.

Such conceptual twists and turns, as it were, often form what might be described as a bricolage approach to conceptualization within an article, which often involves introducing a variety of concepts, which speak to the varying themes identified of these data and provide an additional explanatory layer (and a link to existing scholarship in the field).

A common refrain in discussions about qualitative research and conceptual development is whether we even really need concepts. Isn’t this merely intellectual gymnastics? Isn’t this just making the results more abstract, or less accessible? This is probably most acute when qualitative data are part of a mixed-methods study. Implicit, often, in this context, is that the qualitative data collected will further describe, albeit in people’s own words or experiences, the quantitative data. Often run by a dedicated qualitative researcher, theoretical development can become quite complex in these mixed-methods contexts, because the deployment of critical interpretation—the development of concepts—can challenge the “empirical realities” that the quantitative data/project reports. This can lend itself to qualitative researchers producing descriptive analyses, rather than fully conceptualized papers. It generally means that the interpretation has been overrun by a form of qualitative empiricism, which lacks identification of the broader meaning of the data and a critical eye. This can be rationalized in many different ways, including via the logics underpinning the questions posed earlier, but ultimately interpretative inquiry is conceptual work, and cannot be depoliticized.

The institutionalization and expansion of NVIVO and similar computer packages has been incredibly positive in many respects, and it would be naïve not to recognize their importance and usefulness for researchers across disciplines and fields. Such devices have, however, come at a cost. As qualitative “analysis” gradually became systematized, automated, and centered on the peculiarities of the treatment of data, rather than the interpretive sophistication brought to the process, the problem of inadequate conceptualization has been accelerated. The automation turn in qualitative research has meant that it has become acceptable across many fields to embark on qualitative analysis as a purely empirical exercise, rather than an interpretative process. Practicing such things as inter-rater reliability and the like has become proxies for rigor. And it is not that there is not a place for multidimensional rigor. Rather, such formulaic proxies misunderstand qualitative, interpretative practice, and most problematically, miss what conceptualizing data actually mean or achieve and its importance for quality in qualitative research. This problem is not aided by a proliferation of journals which valorize certain forms of “qualitative rigor” that are legible for an empirically orientated audience. In essence, what is required is merely evidence of the practical “sorting” rather than interpreting data.

In an interpretive world, adequate conceptualization of data elevates the article to speak to what connects the data, and in a much broader sense than empirical similarities across qualitative accounts. Applying and developing concepts is the act of figuring out and naming the important processes and dynamics—often political, cultural, social, economic, interpersonal—that bind experiences of your participants together (if indeed, we are talking qualitative data of this kind). One needs to embark on conceptual development to understand the broader significance of what is being expressed or documented. To develop concepts, one needs to read around relevant and previous conceptualizations, and those deployed in and beyond the discipline you are working in. In that sense, qualitative analysis, more than merely coding and sorting data, is an iterative practice in conversation with existing conceptual and empirical scholarship, which requires both abstraction and (re)embedding in data. It also speaks to study “generalizability” in the sense that conceptualization helps us illustrate what our data mean at a broader level, thus allowing it to speak across realms, sites, and contexts. In Charmaz’s (1990) articulation, and as she notes in her article I draw on frequently, we are tasked with asking of what larger process is this action, belief, or experience we are witnessing, a part? If we can’t or won’t answer this, we are not practicing interpretive research. Relatedly, I find the common assumption of qualitative work as lacking generalizability—acknowledging the baggage of this word—is generally reflective of a culture of under-conceptualized qualitative data. If well conceptualized, qualitative studies have much to say about many other contexts.

Maintaining the Spectrum

Research practice should be diverse and funneling everyone into one model or the another is reductive and harmful. Similarly, the practice of conceptualization in qualitative work will and should exist on a spectrum. Some articles may have incredibly vivid and novel empirical data, which may require a conceptual “soft touch” or indeed, may be squarely aimed at policy or practice change, with considerable emphasis on practical implications. Other articles will be focused on a unique conceptual argument from a data set which seems, at least at first glance, as relatively routine and or empirically similar to work done previously in the field. Similarly, and at the other end of the spectrum, scholars may wish to write an entire piece largely dedicated to a particular theoretical position, with some relatively minimal qualitative data or selective case studies presented to tease out the idea(s). The concepts or theory developed, in this latter genre, come to the fore, and the case studies or data presented become exemplars of the detailed, extensive theoretical case. Every now and then we get a very theoretical article which is then utilized by other scholars across a wide variety of empirical contexts for many years to come, often teased out in more empirical contributions. And this is the way in which scholarship is in conversation, and how different mixes of conceptualization and empirical data work to ensure progression, innovation, and diversity. Problems emerge when there is a complete absence of any conceptualization, and/or a lack of understanding of the importance of concepts. Unfortunately, from an editorial perspective, this is relatively common.

Common Problems

There are a range of common problems that are worth thinking about before submission to a journal. As common situation is where there is no interpretative leap. Very often data presented are extremely interesting, but authors make little or no attempt to think through which concepts might be relevant or pertinent. This is always surprising and somewhat disappointing as there are generally many that are really useful, and yet the author has chosen not to attempt any elevation, at perhaps, misunderstands what this process actually entails. Sometimes this relates to reluctance or lack of confidence in conceptual elevation, and I would argue strongly that some is better than none, and reviewers will often jump on board to help out with the process of elevation if given a chance. However, it can also be a product of institutionalized practices. One of the underlying causes of this lack of interpretative leap is the construction of basic layers of data organization (practices of “sorting” qualitative data) which act as a substitute for elevation (i.e., higher order thinking), without actually elevating the discussion in any meaningful way. This is when coding processes interfere with quality interpretative inquiry. For instance, maintaining lay/participant phrases as thematic labels can—although not always—keep “analysis” at the level of data, often un-linked to concepts. Such forms of analysis remain segmented and local, losing broader application. By missing this analytic step, the research remains analytically undeveloped, incomplete, and unfinished. Thematic analysis – and acknowledging this is now a very “broad church” it its own right – simply enables the first step in the analytic process.

Another common problem is the presence of an interpretive gap but there is a gap between the data presented and the concepts introduced. This may be linked to the use of qualitative analysis software, in some instances, and the act of post hoc and somewhat belatedly delving into the literature to retrofit a concept or two to the empirical coding. This comes across as a half-hearted attempt at conceptualization, often embedded in the previous points about whether indeed this is really necessary in qualitative research. This is quite solvable with a more in-depth exploration of the concepts themselves, and some work linking the data more effectively to them.

In other instances, authors claim things as concepts, when they are really just typologies or themes from their empirical observation. The naming of themes or categories doesn’t actually explore the meaning(s) of these data, it merely summarizes. This is probabaly the most problematic zone, because it very often results in a “desk reject”, when there were plenty of opportunities for conceptual development. This gives the reviewers no opportunity to “help out” with the work of conceptualisation, as it hasn’t even really begun yet.

Finally, sometimes authors seek to “test” theory with their data. While this is perfectly appropriate for particular data sets and areas of research, this often jars with interpretive designs. Why? Because the nature of inductive research is it develops concepts rather than tests them. Sometimes “testing” can work, but it has to be set up in the right way, and also suit the empirical problem being examined. More often than not, qualitative data suit explication of concepts in dialogue—and in response to—fieldwork.

“Handling” Concepts

There are many different views on how to work with concepts, and how to present ones conceptualization, and I certainly don’t cover all views here. Some disciplines, fields, and traditions can be quite dogmatic in their view of how theory and/or concepts should be used and treated within a qualitative study. I take the approach that there are many modes of conceptualization, and these include, but are not limited to, (a) concepts as a toolkit within the qualitative analysis; (b) concepts as iteratively developed and emergent (i.e., grounded in) data; (c) concepts as a means to theoretical development; and (d) concepts as under scrutiny (i.e., “tested”) in the data analysis process. The latter approach is often used in more positivist traditions, but ultimately, still does appear within the field of qualitative research. Because of these and other genres of working with concepts in qualitative research, the order in which we present concepts can be challenging. One can get into hot water, for instance, outlining a theoretical idea up front, when in fact one is taking a “grounded” approach to theory development. In this sense, there needs to be a match between one’s approach to conceptualization, and one’s presentation of analysis/article.

Qualitative research is a broad interdisciplinary church, and this means there are multiple ways and means of conceptualizing qualitative data. I prefer a less dogmatic approach, accepting that much of the spectrum outlined above is a matter of preference, taste, and tradition. In that sense, being overly restrictive denies the fact that qualitative research exists within and emerges from the habitus (Bourdieu, 1977) of particular disciplinary areas, and thus at least in part, articulates these. In this sense, practice is imbued with power and politics (Foucault, 1991), just as any other sphere of life, and governing research practice is murky territory. And recognizing this, and not being overly prescriptive, is important to retaining vitality and diversity in qualitative research.