The myth of the emerging themes: why qualitative research needs theory

In 2016, in response to an apparent decision by the British Medical Journal (BMJ) to deprioritise qualitative research papers for publication, 76 senior academics from countries across the world signed an open letter setting out their objections to that decision.¹ In response, the journal’s editors responded with a number of statements about why qualitative research was no longer a priority for them.² These statements embodied a number of claims, which are set out in Box 1.

Box 1. BMJ editor statements².

Qualitative research is by its nature exploratory, not ‘definitive’ or generalisable

Qualitative research is less relevant to an international audience

Qualitative research is unlikely to change clinical practice and help doctors make better decisions

Qualitative studies are not as widely accessed, downloaded, or cited as other research

All of these propositions can, of course, be challenged. However, it is not my purpose here to point out the many and various ways in which the BMJ editors were wrong. My concern is with our collective obligation to make our qualitative research as good as we possibly can, and to reflect upon how the BJGP can support researchers and authors in this endeavour. With this in mind, I think we need to ask ourselves what it is about the work that we do that fails to convince our colleagues as to its value.

Lisa Morriss, an editor of the journal Qualitative Social Work, set out her concerns about the quality of some qualitative papers submitted to the journal in an editorial, making the point that ‘themes do not ‘emerge from the data’, a sentence we often see in articles submitted to our journal. Instead, the researcher actively constructs each of the themes through their engagement with the material.³ This reflects my own experience as a reviewer, with a distressingly large number of papers suggesting that themes somehow pop up out of the data, without any active role for the researcher. Morriss goes on to emphasise the importance of theory in the qualitative research process, and it is here that I think we need to focus if we want to give ourselves a firm base upon which to counter the criticisms made by the BMJ editors.

Theory is of course a broad and elastic concept, and there are many different ways in which theory can be used. Frustratingly for the novice researcher, this means there is no manual that can be applied which tells you how to use theory in qualitative research. However, it is possible to tease out a number of different types of theoretical ideas and constructs with which researchers will need to make themselves familiar. The first is what I call ‘high theory’ — ontology and epistemology. These concepts often seem obscure and forbidding, but engagement with them is essential if our research is to make sense. Ontology refers to our ideas about the nature of the social world and of reality: positivists believe that there is a truth out there waiting to be discovered, and that the role of the researcher is to find that truth; constructivists believe that there is no absolute truth, and that researchers and research participants actively construct meaning in their interactions; and critical realists believe that there is an underlying reality, but we cannot perceive it other than through the prism of our human consciousness.⁴ These things matter because they inform our epistemology — what we think counts as valid knowledge about the social world. Positivists see the researcher as a source of bias and of error — the goal of positivist research methods is to seek objectivity and reduce bias.⁵ For a constructivist, the researcher is a tool in the research process, actively engaging with the situation being researched and co-creating knowledge about it.⁶ For critical realists, the goal is to explore what we can observe about the world and use this to deduce something about the deeper underlying reality that we cannot observe.⁷ For positivists, the goal is to reduce the impact of the researcher on the research situation, while for both critical realists and constructivists the researcher’s subjectivity is a vital resource, although its use must be conscious and reflexive, with explicit consideration of the impact of the researcher’s position and experiences.⁸

All of this can seem quite remote from the concrete task of doing research, but it does matter. Different epistemological assumptions lead to different approaches to data collection. For example, a positivist approach to focus groups will position the facilitator as a neutral observer, endeavouring to limit their impact on the discussion; a constructivist, meanwhile, might actively engage and offer personal insights, using these to trigger deeper engagement from the other participants. A positivist approach to thematic analysis will emphasise the need for independent coders who come together to agree a ‘correct’ set of codes for each data source, while a constructivist will see different coding decisions as a source of reflection and added richness. So that is the first way in which we must engage with theory — we need to understand and reflect upon the ontological assumptions underpinning our work, and teams need to discuss this before the project starts.

The second way in which theory is important in qualitative research is in developing our theoretical frameworks. This again can seem like a forbidding task if you are new to research. What actually is a ‘theoretical framework’? In a recent paper about the use of theory in qualitative research, Collins and Stockton⁹ provide a useful heuristic (Figure 1). In this framing, a theoretical framework represents a synthesis of our understanding of what is important or interesting about a topic arising out of existing literature with our underlying ontological and epistemological assumptions. Pulling these two things together supports us in designing our study (Who do we need to speak to? What do we need to observe? What questions should we ask?) and in analysing our data. To give an example, colleagues and I undertook a study of the introduction of new types of clinical staff into general practices.¹⁰ In preparing our grant application we thought about the topic and looked at the literature. We knew that when new types of professionals work together there may be problems with professional jurisdiction and professional boundaries.¹¹ We therefore read some of this literature and used it to think about what our research design might look like and what type of questions we might ask. We also thought about the practicalities — if you have a wider range of professionals at work then someone will have to decide who is going to see which professional. This led us to look at the organisational studies literature on organisational routines,¹² and this in turn led us to decide that we needed to incorporate some ethnographic methods in our study, spending time observing in reception and back-office areas where decisions were made about who patients should be booked in to see.¹³ Our framework on this occasion was quite loose — some broad ideas about relevant literature, coupled with an underlying orientation towards a constructivist approach. This loose approach allowed us to adapt as the study went on, bringing in new theoretical ideas in response to what we found, and adapting our existing ideas. This approach is often called a hermeneutic circle — in which the researcher goes backwards and forwards between theory and data, developing theoretical ideas and adapting data collection accordingly¹⁴ (Figure 2).

Figure 1.

The role of theory in qualitative research. From Collins and Stockton.⁹

Figure 2.

The hermeneutic circle of research. From Agyemang-Benneh.¹⁴

Sometimes theoretical frameworks are a little tighter than this. In another study colleagues and I evaluated the implementation of the Vanguard programme in England. These policy pilots involved local groups of organisations receiving considerable additional funding (over £250 million in total) in order to implement more integrated ways of working. A straightforward evaluative question would be ‘Did the Vanguards achieve what they set out to achieve?’ However, we are policy researchers, and so we are also interested in the policymaking process. There is considerable literature on what constitutes ‘success’ in policy terms — sometimes it is actually a political measure of success, in which little of practical value is achieved but a particular policy confers a political advantage on the government.¹⁵ We brought this literature together with the literature on policy pilots. This literature asks the question: ‘Why do governments use piloting approaches to introducing new policy?’¹⁶ Although one might think that the objective is simply to test something prior to wider roll-out, in practice, pilots are used for a variety of different reasons — sometimes simply as a way of getting something implemented, with roll-out taking place regardless of whether or not the pilot ‘worked’. Bringing these two literatures together, we developed a framework for judging whether or not any particular policy pilot has been a ‘success’, exploring different dimensions of success, including political success, alongside a more straightforward assessment of whether or not the pilot achieved its ostensible goals.¹⁷ We tested this using the Vanguard programme as an example, and the resulting paper has been cited 28 times in a wide variety of fields, including climate policy and education. This example illustrates one of the most important reasons why theory is so important in qualitative research: if we take theory seriously, we can generate findings that can be generalised beyond our specific research situation.

Sometimes, rather than developing or modifying our own theories, we might use explanatory frameworks developed by others. Normalisation Process Theory (NPT) is a good example of a broad theory (sometimes called a ‘middle-range theory’) developed out of a programme of empirical work and designed to explain why implementation of new ways of working sometimes succeeds, with the innovation settling in and become ‘normalised’, and sometimes it fails.¹⁸ NPT allows researchers to systematically explore what is happening and to draw lessons about why.¹⁹ Importantly, the use of an explicit framework allows comparison to be made between the implementation of different types of innovation, allowing more general conclusions to be drawn as to how to approach particular implementation challenges. This again illustrates the importance of theory in allowing generalisation — without a structuring theory, accounts of implementation challenges will never build beyond the specifics of a particular context.

Thus, theory is the essential underpinning for qualitative research, determining not just what we do and how we do it, but also providing the firm foundation upon which knowledge can build. Importantly, without theory it is hard for us to counter the criticisms made by the BMJ editors. A number of consequences flow from this. First, it is incumbent on us as researchers to think about it. Researchers need to understand the difference between paradigms and the consequences that arise, and they need to know where they stand. Second, we need to communicate the theoretical ideas and assumptions on which our work is based to our readers. That means that qualitative papers will, by necessity, be quite long — setting out how theory was used in the research takes words. Finally, I would argue that we need to be cautious in the use of checklists to judge the quality of qualitative research. Consolidated criteria for reporting qualitative research (COREQ), for example, contains more than 20 criteria, only one of which references theory. My criteria for judging a qualitative study start with theory — what theoretical tradition and sets of ideas is the study situated within, how did those theories inform the design and conduct of the study, and how do the findings and conclusions relate to these wider theories? The other checklist items — around sampling, approach to analysis, and so on — will all be determined by the answer to those questions. Like many attempts to standardise a complex field, COREQ tends to miss the point.

In summary, while I strongly believe that the BMJ editors’ contentions about the limitations of qualitative research are wrong, as a qualitative research community I think we do need to do better. Thinking about, talking about, and considering how we use theory should become a normal feature of research team meetings, and no qualitative paper should ever claim that themes ‘emerged’ from the data.