Actor-network theory and ethnography: Sociomaterial approaches to researching medical education

Medical education is complex. Our work occurs in multiple, ever-changing contexts, and involves diverse people, tools, and resources. Sociomaterial approaches offer new perspectives on the rich messiness of everyday medical education, prompting questions like: how do chosen assessment tools influence OSCE assessment practices [1]? ; how do videoconferencing technologies work to shape distributed medical education [2, 3]? ; how do clinical spaces influence role negotiation amongst interprofessional teams [4‐8]? ; how do social and material elements structure simulation practices, particularly the debrief [9, 10]? ; and, how do health technologies impact nursing practice in clinical contexts [11]?

A strand connecting these works is the idea that medical education is brought about in multifaceted and unpredictable ways by diverse human and non-human, material and immaterial, elements. In other words, the practice of medical education is shaped by both social and material forces, and is therefore sociomaterial [12]. Sociomaterial perspectives necessitate a shift in focus, offering new ways to conceptualize and scrutinize taken for granted medical education practices. For example, material elements—like smartphones, stethoscopes, storm systems, and scorecards—are conceptualized as agentic rather than neutral. This means that materials actively produce teaching and learning, and are therefore subject to rigorous inquiry. Foregrounding the material simply makes sense in medical education research. After all, no assessment strategy exists outside the physical tools we use to measure performance (checklists, testing software and more). No ethical challenge takes place apart from the room in which it happens (clinical spaces, classrooms and more). No competency decision can be detached from the tools used to demonstrate the skill in question (scalpels, gloves, and more).

‘Sociomaterial’ refers to theoretical approaches linked by two related ideas. First, the separation between human and non-human, material and immaterial, is artificial; second, the privileging of the human, or social, over the material limits what we can know. Examples of sociomaterial approaches include Cultural-Historical Activity Theory [13, 14], Actor-Network Theory [15, 16], Practice Theory [17, 18], and Complexity Theory [19, 20]. There is considerable diversity amongst, and even within, sociomaterial approaches. For example, a study conducted from one particular sociomaterial perspective, like cultural-historical activity theory, might look quite different than another informed by practice theories. Regardless of the differences, however, these approaches share a commitment to taking materials seriously in their theorizing [21].

In recognition of the subtle, but significant, differences amongst the various sociomaterial traditions, and in order to avoid vagueness or over-generalization, here we will focus specifically on one sociomaterial approach: actor-network theory. Broadly speaking, this theory acknowledges, and foregrounds, the agency of non-humans within practice; however, we want to make clear that actor-network theory itself is a complex theoretical construction with many adaptations. Thus, our use here is in the interest of exploring some of the methodological implications and possibilities of a general actor-network theory approach, rather than offering a definitive statement on how to ‘do’ actor-network theory.

Operating from an actor-network theory perspective means conceptualizing practices as heterogeneous and spontaneous gatherings of natural, technological, human and non-human actors (i.e., assemblages; see Tab. 1 for a glossary of key terms). Objects, and even individuals, are not static, pre-formed substances [22] but rather surface though a series of negotiations between an ever-evolving assemblage of actors. Rather than considering a medical education issue as a matter of individual human skill or cognition, we focus on untangling a heterogeneous web of human and non-human, material and immaterial, factors bringing about the issue. This also makes uncertainty, via the concept of emergence, an important element in sociomaterial theorizing, given the perspective that medical education consists of continuously changing assemblages [23]. As an action being observed can always unfold in several different directions [24, 25], as Van Dijk and Rietveld noted, ‘What is done is yet to be determined—and can only be determined by the observer after the observed activity has been performed’ [26, p. 6]. For example, an accreditation standard is developed to address comparability among distributed (multi-site) medical campuses to ensure quality, consistency and fairness. This standard prompts medical schools to outfit lecture theatres with sophisticated videoconferencing systems, hire new audiovisual professionals, and adapt existing curriculum formats and processes, to facilitate comparability. No amount of preparation and planning, however, can fully predict what will actually happen in these new spaces moment to moment. For example, we observed that the videoconferencing system sparked several unintended activities, including students not asking questions to teachers (both sites), informally ‘electing’ class spokespeople comfortable with the technology (distant site), turning off the microphone and discussing questions with peers (distant campus) and asking questions privately after class (host campus).

One of the most significant features of actor-network theory is its theorization of agency as distributed and relational [27]; it follows, then, that non-humans have agency. Rather than being conceptualized as the backdrop of human activity or as tools created for and used by humans, an actor-network theory perspective treats things as having the capacity to act (agentic) and having the capacity to make things happen (productive). This means, as Latour declared, ‘things might authorize, allow, afford, encourage, permit, suggest, influence, block, render possible, forbid, and so on’ [15, p. 72]. Building on the example above, although the videoconferencing system was intended to link sites, at times the system belied this intention, prompting (and at times permitting) certain students (and teachers) to ask questions outside the confines of the system.

Non-human agency is not as farfetched as it might sound—of course, a mundane object, like a thermometer, does not find its own way and ‘jump’ into a person’s mouth, and ‘decide’ whether that person has a fever. A fever is determined through multiple relationships between people and things: a person feeling unwell, knowledge of thermometer use and normal temperatures, digital technologies, batteries, and innumerable other factors. So, while actor-network theory allows us to account for non-human agency, ultimately no actor, human or non-human, could exist completely on its own, isolated from the networks of relations in which they come to be [28]. Thus, agency is conceptualized as something other than causal, uncoupled from concepts like intention, subjectivity, and free-will [27].

Considering this perspective on agency, actor-network theory and other sociomaterial studies are symmetrical with respect to theorizing both humans and non-humans. Rather than being concerned with understanding individual people or things, we explore the relationships among people and between people and things, and the productivities of these relationships [15, 29]. This allows for description of various social and material mechanisms at work to hold together a phenomenon. Returning to our earlier example, focusing on the role of the videoconferencing system itself in the distributed lecture was crucial in detecting how these new tools and spaces limit, allow, and transform what actually happens in these settings.

However, the principle of symmetry does not necessarily mean that humans and non-humans are assumed to be identical [27, 30, 31]. Operating from an actor-network theory position means recognizing that non-humans exercise agency; however, this agency does not operate as human agency does. Thus, while we posit ontological, epistemological, and methodological symmetry, we refrain from passing judgment on the extent to which this symmetry is concretely actualized [27, 32]. Also, a deliberate focus on exploring, following, and documenting the active work of things does not preclude paying attention to humans, but rather reorients our inquiry to take into account myriad ways human and non-human elements come together [12, 15, 17].

Theorizing human and non-human agency provides a useful starting point for investigating complex relationships producing everyday practices. If we took the alternative position, that non-humans were a neutral backdrop to human activity, we would be missing important, and as Sayes wrote, ‘sociologically relevant’, aspects of our analyses [27]. Thus actor-network theory, and other sociomaterial perspectives, provide tools to attend to the messiness of the everyday world—and all of its minute negotiations, translations, and processes. In its simplest form, an actor-network theory informed approach asks us to remain open to the possibility that non-humans add something worth studying.

Given these principles, it is quite logical, then, that ethnography is considered the methodology of choice for actor-network theory informed researchers [32]. But, what does it mean to do ethnography? Hammersley [33, p. 4] points out that most ethnographic work shares the following features:

However, he acknowledges that each feature might, in fact, be troubled and differently engaged in ethnographic work, highlighting the huge range of approaches that qualify as ‘ethnography.’

Ethnographic work is ‘messy’ and relies on the skill of the researcher to make sense of a complex field. Ethnographers’ work is highly inductive as they pursue openness to understanding the worlds of their participants, rather than relying on the safety of prescribed analytic approaches. There are parallels between the ethnographic goal of thick description [34] and actor-network theory theorist John Law’s ‘non-coherent realities’ [35, p. 92], both of which anticipate ‘unruliness’ in bringing together heterogeneous actors in order to richly describe everyday practices.

A fundamental connection between actor-network theory and ethnography is their focus on practices: everyday sayings, doings, and relations with objects that make up what people do in their everyday lives [30]. Practices themselves are multilayered and heterogeneous; therefore, understanding practices requires carefully tracing multiple actors that assemble and give meaning to human worlds, activities and lives [32, 36]. There are multiple perspectives on practices; however, they share a common belief that practices are ‘inherently contingent, materially mediated, and […] cannot be understood without reference to a specific place, time and concrete historical context’ [37, p. 1394]. Practices can be described as a ‘mangle’ [38] of people, things, intent, knowledge, processes, and many other factors. Effectively studying the mangle requires avoiding ‘side-step purification processes in the construction of knowledge’ [32, p. 113] and focusing instead on generating thick ontology, a complex task. In addition to making practices visible, one of the most significant strengths of actor-network theory informed work is its ability to move beyond reliance on traditionally human-centred methods like interviews, focus groups and surveys. Ethnographic approaches with an actor-network theory sensibility require researchers to attune to the mundane—specifically lending an eye to everyday objects and practices that we may otherwise not notice, nor bring to the fore [39].

In sum, there are certainly distinctions specific to various actor-network theory- and sociomaterially-informed approaches. However, commonalities between sociomaterial and ethnographic sensibilities include: eschewing neat analytic categories in order to theorize messiness and contingency; acknowledging that practices are both social and material; and, focusing on detailed description of what is actually happening in the field, including the mundane.

There are central epistemological concerns regarding researcher positioning in actor-network theory informed ethnography. Most ethnographic approaches recognize that a researcher’s preconceptions shape their analysis, description, and interpretation [40, 41]. Ethnographers walk the line between insider and outsider, participant and observer. The ethnographer’s work is to document an existing set of social and cultural practices independent of researcher presence. While the researcher will undoubtedly influence the activities taking place [42], the primary ontological principle here is that a scenario exists independently of the researcher.

The work of actor-network theory informed ethnography is to unravel an assemblage perpetually evolving and emerging in everyday life. The ethnographer him/herself is therefore part of the assemblage under study. This means that a situation is only brought about through intermingling of particular social and material elements, of which the researcher is a productive part. In other words, no assemblage exists independently of the researcher. Positioning the researcher within the phenomenon means that researchers actually (re)configure the world under study [43].

These principles are apparent in the types of issues taken up in actor-network theory informed ethnographic work. Rather than focusing primarily on human perspectives, those operating from an actor-network theory informed perspective consider culture symmetrically: that is, as a social and material assemblage performed into existence through constantly evolving negotiations between people and things. Our work, thus, involves becoming immersed in the field, and working to tease apart those connections.

Whatever the entry point, the research process often grows from an observation that a material element accomplishes something important—whether intended or not—and that this accomplishment is worthy of investigation. This accomplishment can be anything, ranging from a policy document that encourages people to develop a set of workarounds [3], the physical layout of the beds, curtains, and other equipment in a critical care unit [44], bodily fluids that alter the ways in which healthcare professionals opt to deliver care [45], a videoconferencing screen that discourages people from participating in classroom conversations [2], a checklist that constrains an assessor’s judgments and ratings [1], or any number of others. Once the element of interest has been identified, our work involves figuring out how to best understand it.

Importantly, actor-network theory informed ethnography is not linear, prescriptive, or predictable. Whilst the steps, or methods, we describe here are characteristic of actor-network theory informed ethnography, they are not, in fact, prescriptive. Actor-network theory informed research involves exploring and ‘unravelling’ agentic elements, and exploring how these elements relate to other humans and non-humans [46]. It involves collecting and analyzing multiple data points—including documents and artefacts, observations, and interviews—as they become relevant or important.

For example, a researcher might begin by identifying a concern: something that feels important or even uncomfortable, and is worthy of our attention. In order to learn more, they may arrange to ‘hang around’: engaging in observation, watching the agents in action, collecting notes and asking questions. They may determine that a particular element—human or non-human—is an important feature, or in other words, the element is ‘determined’ to be an agent. They might then hone in on this actant, collecting formal observations, following it and documenting actions through focused fieldnotes [47‐49]. In order to understand the actant in depth, they may photograph it, video record it in action, collect instruction manuals for it (if relevant), and read policy documents about how the item is intended to be used. They may ask people who are using the item to talk about what they are doing with it, and why, in real time. They may also ask people to reflect on how they have used the item in practice. The specific methods, and in order in which they are invoked, will be determined based on what the researcher is hoping to learn.

Whatever the series of methods engaged, the work of a sociomaterial ethnographer is being open, following actants, and working toward an in depth understanding of the relationships generating the situation under study. Ultimately, the task is to create a coherent account of a complex assemblage without separating the natural from the social, economic, and political— ‘in other words without making the very distinctions which normally constitute the structure of a social scientific explanation’ [32, p. 112].

How might ethnographic methods build a coherent and symmetrical description? We will describe three types of methods generally used in actor-network theory informed ethnography: 1) document and artefact analysis; 2) observation; and, 3) interviews. We will offer a discussion of the principles underpinning each method, provide examples of how these particular methods have been used, and offer concrete suggestions about how these methods might be used in other actor-network theory informed ethnographies.

Data analysis is inherently messy within any ethnographic approach. The real work of actor-network theory informed data analysis lies in attempting to construct an account reflective of our ontological and epistemological foundations while documenting the complexity encountered in the field [60]. Actor-network theory informed approaches orient us to the assemblage as the unit of analysis: the network of connections between social and material elements.

Actor-network theory-informed data analysis strategies are rarely discussed or described [61]. Amongst the small existing literature, there is general consensus that while there is not one right data analysis strategy, it may be useful to connect actor-network theory informed analysis to well-established methods, in order to make visible the rigour involved.

For example, one method we have used evolved from Wolcott’s classic ‘description, analysis, interpretation’ ethnographic frame [62]. While not specifically sociomaterial, Wolcott’s approach provides the scope for actor-network theory informed ethnographers to reconcile multiple human and non-human data points and various contexts and agents producing the field under study. This diversity is particularly useful for data collected by multiple researchers, where field notes are ‘a reinterpretable and contradictory patchwork of perspectives’ [63, p. 90].

Building from Wolcott [62], actor-network theory informed analysis involves telling a story and then, drawing on actor-network theory principles, explicitly stating why we opted to tell the story in a particular way. As researchers work through multiple data sources, they relate description, analysis and interpretation of data collected through multiple strategies back to the guiding principles of symmetry and relational agency. Researchers transform field description into workable data by entering data, developing coding frameworks, and coding via qualitative data analysis software. However, the next phase, interpretation, is especially key in centring the materiality of a given environment. Actor-network theory informed interpretation involves both foregrounding the stuff of the environment, and considering how these things come together with people through networks of relations to produce the phenomenon at hand.

Constant comparative method is another established technique used by actor-network theory informed ethnographers. Booth and colleagues [11] note that an important goal of this theory is tracing actors—social and material—through networks. This tracing process requires an analytic technique that helps tease out focused concepts and themes from large amounts of interconnected data.

In actor-network theory informed ethnographies, data collection and analysis are iterative and concurrent. Rooted in Grounded Theory, constant comparison allows actor-network theory informed researchers to follow the actors and facilitates ongoing reflection and revision, in line with the fluid, uncertain, and emergent nature of practice. This back and forth method is particularly helpful in early data collection, when researchers may not yet have a good sense of the scenario at hand. Booth and colleagues [11] highlight that when observing a clinical environment, an acute care unit for example, the ‘blackboxed’ nature of the work itself may obscure some of the actors, and their co-mingling. In a blackbox (i.e. stabilized) network, actants may coordinate in such a way that at first glance, they may appear as one actant rather than as a collection working in tandem. It is through repeated, reflective observation that a finer-grained picture emerges.

Constant comparative method encourages multiple layers of coding—termed ‘open’ and ‘axial’ by Corbin and Strauss [64]. From an actor-network theory informed perspective, multiple coding levels provide structure and, to some degree, boundaries to the analysis process. Limiting the scope of analysis and focusing on specific actants and activities can be particularly useful in actor-network theory informed inquiry. Because assemblages stretch across space and time, and include infinite possible actants, researchers must identify and adhere to clear parameters aimed to balance analytical richness and practical manageability.

Aligning actor-network theory research with an established analytical approach like those described above—constant comparison, or description/analysis/interpretation—also may help assuage critics questioning the rigour of actor-network theory studies. Medical educators have come to expect systematic and structured research. There is limited literature exploring sociomaterial methods and processes, including actor-network theory informed approaches, and in particular, data analysis [61]. Aligning actor-network theory with more established and recognizable analytical techniques can offer more rigour and methodological transparency.

Having said this, we recognize that there is a risk in being overly ‘technique’ driven [65], encouraging instead that researchers play with data. Hopwood [66] offers the use of ‘synoptic units’, i.e. summaries of how key bits of data relate to something of interest or importance, as an alternative. In this approach, rather than engaging in prescriptive, technique-driven coding, researchers ‘are a strength, not a hazard’ [66, p. 1]. ‘Getting into the data’—being both systematic but also playful—involves looking for patterns, exploring oddities or outliers, looking for silences, etc. This approach has been particularly useful in sociomaterial ethnographies (see Clerke and Hopwood [67]).

Whatever data analysis techniques you choose, the key is to be careful, systematic and clear, while recognizing that analysis is in itself a sociomaterial practice. Focusing on sociomaterial principles of symmetry, agency and emergence in data analysis can help researchers attune to non-human actants in the research setting, making space for new insights to emerge.

There is much to be learned by attuning to what has been largely obscured in medical education research: materiality. We believe that materials are not a neutral backdrop to human activity, but rather of central importance and interwoven with human agency and meaning making. Yet, engaging in actor-network theory informed ethnography presents theoretical and methodological challenges. The idea of foregrounding the material may be conceptually straight forward; however, in action, it can be challenging to avoid focusing on human thoughts and feelings. Doing justice to non-human actants in the research setting requires continual work, even when symmetry between human and non-human elements is built into the research design from the beginning.