Qualitative methods are increasingly used in randomised controlled trials (RCTs) of complex health-behaviour interventions at the various stages of complex intervention development and evaluation, including process evaluation [1, 2]. Qualitative methods can inform the understanding of a problem, the development of an intervention, and the understanding of how an intervention is delivered by agencies and received by participants. In process evaluations, qualitative data can contribute insights into how interventions operate and how outcomes are reached, although in practice, qualitative research is not always used to inform the trials they are part of [1]. Interviews and focus groups are commonly used qualitative methods in process evaluations; for example, they are often used to explore the acceptability of an intervention to participants [3]. Ethnography is a methodology which largely, though not exclusively, employs qualitative methods; however, it has a distinctive approach over and above the particular methods it employs, which could be useful in process evaluations to explore the detail of how complex interventions operate. Despite its benefits, the potential contribution of ethnography to process evaluation has not been realised. This article briefly introduces ethnography as a methodology and then discusses three useful features that are relevant to process evaluations: (1) producing valid data, (2) understanding data within social contexts, and (3) building theory productively. The comments made in this article could be applicable to other types of complex interventions besides those targeting health-behaviour change. The focus here on health-behaviour change and public health is because there is an increasing recognition of the social determinants of health in public health research; studies are consequently addressing the social, environmental, and organisational contexts to a greater degree. Ethnography has traditionally examined social contexts and is, therefore, a very relevant methodology for this field.
Ethnography is characterised by long-term participant observation as a central method, where the researcher spends an extended period of time in a social group in order to collect data. The term ‘ethnography’ is often used interchangeably with the term ‘participant observation’, but it is actually a much broader methodology than this, both because of the range of methods it employs and because it encompasses an overall orientation to research, which is detailed below. It comprises a collection of different ways of eliciting and collecting data, including the observation of individuals and groups of individuals, unstructured interviews, documentary analysis, and the use of a researcher’s field notes. It employs these methods within a long-term, holistic, and flexible approach to data collection:
The ethnographic researcher participates, overtly or covertly, in people’s daily lives for an extended period of time, watching what happens, listening to what is said, asking questions; in fact collecting whatever data are available to throw light on the issues with which he or she is concerned [4].
Engagement with a particular social or cultural group is also a distinguishing feature of ethnography, as reflected in Curry et al.’s definition:
Ethnography is a form of field research that seeks to learn the culture of a particular setting or environment. It often relies on participant observation through prolonged field work and may include other qualitative and quantitative methods. The researcher becomes embedded in ongoing relationships with research participants for the purpose of observing and recording talk and behavior. In such cases, the researcher (as opposed to, for instance, surveys or questionnaires) is the primary instrument for data collection and analysis. The researcher seeks to place specific events into a broader, more meaningful context, with a focus on the culture and social interaction of the observed people or groups. Ethnography is particularly valuable in understanding the influence of social and cultural norms on the effectiveness of health interventions [5].
Through living and working with communities through extended periods of time, often months or years, ethnographers aim to see and describe the world through the eyes of members of that community. They pay particular attention to the everyday life, narratives of events, social interactions, and the cultural meanings and practices of a community. Ethnographies, unlike some observation studies, are of a social group and are often geographically bound. One exception, however, would be a digital ethnography which explores an online social group. The social group could be a class of school children, a choir, a general practice, or scientists working within a laboratory: the key requirement of the social group is that they share a common culture within the environment in which they are being studied (children in a class may have a different home cultures, but while they are in the class, they share in the culture of the class). The term ‘ethnography’ also refers to the product of the research, as well as the method: it is a distinct piece of writing which summarises an ethnographic study of a particular community or people, usually by an anthropologist. For examples of classic ethnographies, see ‘The Forest People’ [6] or ‘The Innocent Anthropologist’ [7], which describe and explain the social worlds of tribes in ‘exotic’ locations for Western audiences. As well as ethnographic monographs, outputs of ethnography can also include images, particularly in the field of visual ethnography [8].
Ethnographic methods developed within the field of social anthropology in the nineteenth and twentieth centuries. Among the most notable studies, under what became to be known as the ‘British School’ of social anthropology, are those by Bronislaw Malinowski and E. E. Evans-Pritchard [9, 10]. These studies involved long periods of intensive fieldwork and participant observation with small tribes of indigenous people and advocated an approach to anthropology which sought to understand the culture from their perspective. Whereas anthropology had traditionally visited discrete communities in remote areas of the globe, the ‘Chicago School’ of sociology and criminology in the mid-twentieth century adapted the commitments of the early social anthropologists, turning attention to social problems within urban settings closer (geographically) to home. For example, in their study of ‘Street Corner Society’ (first published in 1943), Whyte and colleagues produced vivid portraits of city life by recording the social worlds of street gangs [11]. Two decades later, sociologists from the Chicago School undertook observations of groups of health professionals and healthcare organisations. The most notable of these, perhaps, includes a participant observation of hospital life conducted by Roth in 1963, in which Roth himself was admitted as a patient for tuberculosis (TB) [12], and a study of the transition of novice medical students to aspiring doctors, which was conducted by Howard Becker and colleagues in 1961 [13]. More recent commentaries on the use of ethnography in the health field are available [14–17], as well as ethnographies of health-service professionals and organisations [18–20] and settings such as intensive care units [21]. Studies in medical anthropology have also explored health and illness in different cultures more broadly in terms of how well-being, physical health and mental illness are defined, their meaning, and how symptoms are experienced from the perspectives of people and communities themselves [22]. However, ethnography has been relatively underused in trials, health psychology, and social science research on health behaviour change [23].
Ethnography is a rich and detailed methodology and, thus, well suited to the challenges of understanding how complex interventions work; process evaluations of health-behaviour interventions could therefore benefit from adopting ethnographic methods. Process evaluations complement traditional randomised controlled trials (RCTs) by attempting to understand how interventions operate. They illuminate the ‘black box’ of the processes of an intervention, which are not addressed by the classic RCT design that examines the relationship between a limited set of variables (at baseline and outcome and in intervention and control groups). The recent Medical Research Council (MRC) guidance [24] has conceptualised process evaluation in terms of examining three principal elements: the delivery of the intervention, the mechanisms of impact that occur in participants (how changes in individuals’ health behaviours occur), and contextual factors which interact with both the delivery and receipt of the intervention. The challenge is not just to include these additional elements in a trial of an intervention but to understand the intervention as a whole in terms of how it produces outcomes and how the different elements are causally related to each other. This involves understanding a complex set of processes and events and how they are linked together in causal pathways. These causal pathways articulate the underpinning theory(ies) of the intervention, as the theory of how an intervention works necessarily explains how the intervention causes change. For example, an intervention that uses a food diary to promote weight loss might be based on a theory of self-monitoring of dietary intake and how this affects eating behaviour, as well as a theory of how diaries might be best designed to encourage participants to use them. A description of the causal pathways would explain how the food diary design and delivery influenced participants to engage with the intervention, how the food diaries were actually used and influenced dietary behaviour in participants, whether and how self-monitoring occurred, contextual factors such as family mealtimes, and how these factors contributed to any weight loss in participants (or not). Process evaluations are challenging because the interventions they analyse usually have several layers of complexity due to the challenging problems they are trying to address, the many elements and processes that may occur during the intervention and which are likely to interact with each other, the different levels in which they operate and interact (e.g. intrapersonal, family, and community), and the number of outcomes they may be trying to bring about [25]. One weakness of quantitative research can be the tendency to fill in gaps in explanation with the researchers’ own assumptions. This occurs because of the distance between the researcher and the data in quantitative research [26] or because statistical results only show association between two variables, and researchers may attempt to fill in the missing explanation of how the intervention worked with speculation [27]. Traditional epidemiological methods can establish relationships among variables, but these are not able to fully explain why outcomes occur. As a result, the understanding of processes in an intervention may be flawed or incomplete. Qualitative methods, such as interviews and focus groups, make an important contribution to process evaluations because they can produce rich, detailed information about processes, contexts, and causal pathways in ways that quantitative data cannot. For example, interviews with participants can capture rich narratives describing how individuals experience and react to an intervention and why they may change their behaviour as a result (or not). Qualitative methods also capture the depth and complexity of topics, and are flexible enough to capture unexpected data which may further contribute to the understanding of how interventions operate. All of these aspects of qualitative data are necessary to develop the comprehensive understanding of how interventions operate, which a good process evaluation will aim to produce; furthermore, these aspects contribute to theory-development in the field through explaining causal pathways occurring in interventions. This article argues that ethnography has further strengths over and above the benefits of standard qualitative methods, just outlined here, that could improve process evaluation methods. First, ethnography can be useful in acquiring valid data on intervention processes through collecting data in naturalistic settings and through observing behaviours and events as they occur in real-life settings, rather than through post-hoc interviews and self-report. Second, ethnography collects and analyses data in a way that is inherently embedded within the subjects’ cultures and social worlds and could thus contribute to incorporating contextual factors of the delivery and receipt of interventions. Third, ethnography is an iterative, theory-building approach which is ideally suited to working with and developing theory, another important function of process evaluations that contribute to wider knowledge building within a field. The way in which ethnography is best employed in process evaluations is partly dependent on the type of trial being conducted. In process evaluations for feasibility trials, ethnographic methods could be employed to assess trial methods or to develop the intervention - especially for unfamiliar contexts or hard-to-reach groups - in order to inform a definitive trial. In the ethnography of Garcia et al. [28], which was an intervention for prophylactic HIV medication for black men who have sex with men (MSM) in the USA, the study was used to investigate what ‘usual care’ was in order to successfully integrate the intervention with ‘usual care’, and also uncovered factors such as mistrust of medical services and medications among participants, which helped inform how the main trial was designed [28]. In definitive trials, ethnography could be used to develop and build intervention theory, explain why different outcomes occur for different subgroups, or explain recruitment or retention issues. The three ways in which ethnography can be applied to improve process evaluation methodology, and their applicability to feasibility and definitive trials, are discussed below.