The medical-scientific enterprise has benefited from implementation science’s (IS) fundamental insight: that contextual factors contribute to the success or failure of evidence-based interventions. Yet, IS often fails to account for the institutional commitments of research as an essential element of ‘context.’ In doing so, it may fail to acknowledge that its own institutional commitments to scientific research are themselves key power dynamics affecting how social change is thought of or how interventions are designed. This paper leverages two long-term ethnographic projects studying behavioral interventions to examine how structural conditions shape service delivery. We draw from two intervention projects: e-mental health research in Australia, and an overdose prevention clinical trial in the US. We demonstrate how the prioritization of funders’ interests, the political economy of evidence-based interventions, and the epistemic commitments of intervention research dictate what seems feasible or possible, conflicting with the moral and ethical values of intervention researchers themselves. This paper contributes to critical public health scholarship by demonstrating how IS risks legitimizing structural inequities by absorbing institutional logics rather than challenging them. The disconnect between researchers’ private critiques and the field’s focus on intervention adoption merits ongoing research that examines how IS knowledge production itself reinforces systemic injustice.
Although randomized controlled trials (RCTs) hold epistemic authority within health and social policy, the evidence-based interventions they produce often fail to translate into practice settings. In the context of clinical trials, this dilemma has effectively been outsourced to implementation science (IS), a field that has emerged to manage, in a rather technical fashion, the complexity of the ‘real world’ of clinical practice to help facilitate the uptake of evidence-based practices into routine care (Boulton et al. 2020; Rhodes et al. 2016). The medical-scientific enterprise has greatly benefited from one of the fundamental insights of IS: that contextual factors, unique to a given setting, contribute to the success or failure of evidence-based interventions (Bauer & Kirchner 2020, Eccles & Mittman 2006). In this sense, IS rejects the soft chauvinism of top-down, one-size-fits-all research, and posits attentiveness to local realities as a central tenet of professional science. Relatedly, a significant body of critical scholarship has attempted to broaden IS’s conception of ‘context’ and refocus its attention ‘upstream,’ towards the structural inequities which largely determine life-chances and health outcomes (Brownson et al. 2021, Shelton et al. 2021).
Despite this expanding conception of context, the IS field tends to treat context as external to interventions, (i.e., as ‘facilitators’ or ‘barriers’ thereof). Consequently, the field risks obscuring how its own institutional commitments to scientific research operate as power dynamics in themselves, actively shaping how social change is conceptualized and interventions are designed. By treating context as external to intervention research, IS struggles to interrogate the context of the research itself. The field has rarely articulated the ways in which an underlying intervention naturalizes or obscures structural violence because it implicitly grants the premise of the intervention. This article examines how institutional research commitments in implementation science actively shape, and often constrain, the possibilities of how social change is thought of and conducted. Far from a neutral backdrop, research settings and the institutional commitments they invite (e.g., partnership choices, funding structures, and trial designs), may actively ‘co-produce’ (Jasanoff 2004) the structural conditions the field tends to frame as exogenous ‘barriers’ to the success of a given intervention. The field often prioritizes analyzing contextual factors in intervention outcomes over interrogating the underlying tenets of intervention science,1 tenets that may themselves operate as key barriers to health outcomes. The constraints researchers negotiate are not incidental to, but rather constitutive of, the social order IS helps sustain. This naturalization obscures how the scientific enterprise itself narrows intervention possibilities, embedding political and economic commitments (e.g., cost-effectiveness) into what counts as feasible or valid.
As ethnographers of science, we were struck by a marked similarity in our findings when studying researchers implementing evidence-based interventions: the tendency of individual researchers to express an awareness of – and unease about – the way in which the intervention they were studying obscured a more urgent social problem because doing so went against their institutional and professional interests. These crucial disclosures, which potentially call the morality of their work into question, tended to be relegated to ‘off the record’ admissions. This is typical of the current paradigm, in which attempts to reckon with this feeling of ‘absurdity’ are often fleeting, with no mechanism to marshal these concerns in an organized way that might influence implementation.
We elucidate this phenomenon through our guiding concept of ‘what-are-we-doing here?’ moments – a term we use to describe instances of somber realization among researchers that the research apparatus is, in some fashion, preventing their work from doing lasting social good. In doing so, we highlight how intervention research itself constitutes a critical dimension of context – one that must be examined to understand not only intervention development practices but also how the diverse set of actors involved mediate the competing values, priorities, and structural constraints inherent to social change work.
Each of the two authors shares ethnographic vignettes from their fieldwork in two markedly different behavioral health research contexts: one among an overdose prevention trial in the American Midwest, and another among digital mental health researchers in urban Australia. These disparate examples are paired together to highlight the way in which “what-are-we-doing-here” moments provide insights into the ways in which institutional research commitments actively shape and constrain the possibilities for change.
Implementation science’s efforts to cohere as a scientific discipline and promote evidence-based practice are often at odds with a more politically-oriented analysis of science as an instrument of power that shapes how society’s problems are framed and its resources are distributed (Boulton et al. 2020, Brownson et al. 2021). Evaluations of health services may fail to mobilize this more radical conception of the ‘context’ framing, tending to take as a given certain results of dominant ideology, obscuring what evidence is made visible and thus used (Murdoch et al. 2023). Like most disciplines, IS has its own unique interpretation of context—that is, context is itself contextual (Dourish 2004). Scholars both within and outside of IS have noted its tendency to treat context as ‘representational’, whereby issues like culture and climate are abstracted into discrete categories and framed as either facilitative of, or barriers to, intervention effectiveness (Dourish 2004, Dryden-Palmer et al. 2020, Kislov et al. 2019, May et al. 2016, Wensing & Wilson 2023). This differs from the way many interpretive social scientists are trained to conceive of context, which is less a bounded set of local circumstances than an ongoing ‘co-construction’ between disparate actors. From this perspective, intervention research is not only a practice of finding evidence, but of making it (Rhodes & Lancaster 2019).
This deeply relational understanding of context is a mainstay of anthropology and science and technology studies (Jasanoff 2004). Ethnographers of science and medicine have long demonstrated how biomedical knowledge is shaped by the positionality of the people involved in its making, across a diverse array of projects: research on diabetes genetics in France (Rabinow 2002) and Texas (Montoya 2011), sickle-cell anemia in Senegal (Fullwiley 2011), bipolar disorder in an Argentine hospital (Lakoff 2006), nuclear science in Ghana (Osseo-Asare 2019), among many others. These ethnographic studies of science and medicine share a focus on how research is shaped by the lived experiences, cultural contexts, and power dynamics of the people involved, whether scientists, research subjects, or affected communities. They reveal that scientific knowledge is not neutral but emerges from social, economic, and political conditions, often reinforcing or challenging existing inequalities. In so doing, these works demonstrate its contested nature, ethical dilemmas, and the ways it can serve particular interests (e.g., colonial legacies, economic exploitation, or state control). Collectively, they highlight how research itself is a political act. Open-ended, embedded ethnography (‘deep hanging out’ [Geertz 1998]) with these individuals can yield insights about the larger political-economic world that the science is taking place in.
IS is a polyvocal discipline, and many IS scholars have called on the field to use a more reflexive approach that might interrogate its own practices and institutional commitments. In doing so, they have called for critical ethnography to understand ‘interactions between the ensembles of actors, agencies, interventions, and other contextual variables’ and how such interactions might impact the people IS is meant to serve (Brownson et al. 2022, p. 12, Finley et al. 2018, Kislov et al. 2019, Snell-Rood et al. 2021). A handful of IS researchers have identified the research context as itself an important factor influencing the outcomes of implementation processes, as this determines how problems are defined and whose knowledge is prioritized (Stanton et al. 2022, Rhodes & Lancaster 2019). Yet, to date, few sustained ethnographies have spoken directly to IS, documenting how investigators negotiate the structural impediments encountered during intervention research.
Across disciplines, we see how institutionalized practices of knowledge production systematically exclude lived realities that fall outside their methodological purview, particularly in clinical trials. For example, social anthropologist Paul Geissler has noted the way in which hunger – a daily reality for participants in the African HIV trial he observed – was systematically ignored because it was not part of the study. Crucially, this omission was not lost on study participants or researchers, but was instead forced epistemologically underground, into the margins: ‘Only occasionally, “off the record” in moments of personal conversation not planned by study documents, did participants admit to staff that they were hungry and tried to establish a relation between their own needs and staff’s presumed resources and responsibilities’ (Geissler 2013). This closely follows Michael Taussig’s (1999) notion of the ‘public secret’ – that which is widely known but cannot be openly acknowledged. The narrow purview of scientific studies often renders extremely apparent and important social facts invisible or irrelevant.
Unlike the interpretive social sciences from which it has productively borrowed, IS currently has no mechanism to describe such ‘public secrets’ when they call into question the underlying logics of the intervention being studied. We address this gap by applying ethnographic methods to IS research itself (EC) and to intervention researchers grappling with complex contexts (AN). In doing so, we examine how researchers and actors deliberate on the institutional priorities of research, revealing the unexamined power dynamics that perpetuate implementation gaps.
Following insights that science (Franklin 1995), medicine (Taylor 2003), and implementation science itself (Snell-Rood et al. 2021) are sites rich in cultural activity, we draw on two long-term ethnographic projects. Ethnography allows us to widen the scope of implementation analysis to examine the research process, funding structures and epistemological frameworks that are often otherwise elided in accounts of intervention and implementation science (Hamilton 2013). Ethnographic analysis is particularly useful for observing and analyzing how macro-level structural conditions shape the details of scientific practice and service delivery.
Methodologically we follow the same basic toolkit used by many social scientists in their quest to ‘study up’ (Nader 1972), i.e., focus the ethnographic gaze ‘upward’ towards society’s powerful institutions: participant observation with an established organization, semi-structured interviews with a wide range of actors, exchanging occasional acts of volunteer labor for generous access to the working lives of our interlocutors. One of us (AN) is trained in anthropology, a field where participant observation, among other ethnographic techniques, remains de rigueur. The other (EC) has a background in social work with training in science and technology studies. For purposes of confidentiality, both authors have anonymized their field sites by creating pseudonyms for individuals as well as studies, in some cases changing key identifying details.
EC’s vignettes draw from an ethnographic study of a randomized controlled trial in the American Midwest that evaluates the effectiveness of a harm reduction peer recovery-based (HaRP) intervention designed to connect people who use drugs to treatment upon their release from jail. Institutional review board (IRB) approval to study the trial was granted in the summer of 2020. EC attended daily meetings for over two years; these included formal intervention trainings, clinical case consultations, community advisory board meetings, implementation research team meetings, and ‘all team’ meetings where directors, managers, and clinical supervisors convened with the rest of the academic research staff. Key-informant interviews were conducted using a purposive sampling method and, whenever granted permission, all meetings were recorded, transcribed, and coded.
Data were analyzed using an iterative process that began with familiarization, reviewing fieldnotes, memos, and transcripts throughout the data collection period to identify initial patterns. Next, a provisional codebook was developed to systematically code the data,employing a process of continuous comparison to collate codes into potential themes. These themes were subsequently reviewed and refined against the dataset to ensure distinctness and internal consistency. Member checking with the HaRP implementation science team was conducted during the final year of data collection to ensure the defined themes aligned with informants’ reported realities. This was accomplished by creating space for conversations with the implementation science team on the tensions and structural realities they faced. She approached the HaRP project in a way that invited the research team to ‘sit with’ the tensions that we are otherwise quick to try and resolve through more methods, frameworks, and adaptations. Many of the researcher-interlocutors were frustrated by the limitations of implementation science to contend with the stark realities we faced, even if they were actively using the newest models and latest frameworks to tame the complexities of context.
AN’s vignettes draw from embedded ethnographic fieldwork in Australia conducted as part of a doctoral dissertation in anthropology on computer-automated mental health programs. Australia was selected because of its unique policy of providing Government funding for the testing, development, and dissemination of self-guided treatments for ‘mild-to-moderate’ depression and anxiety. Australia’s handful of large cities are home to a number of digital mental health research institutions which develop apps and web programs with these funds. Between June 2018 and November 2020, approximately 18 months of embedded research and 30 interviews were conducted with researchers across several of these organizations. Because these institutions share a number of important characteristics, and in order to maximize de-identification, the study deploys a composite of all of them, an accepted methodology in institutional ethnography (Corman 2021). The ‘Centre for Psyberspace Research’ (CPR) described below is the name of a single fictional institute to which all observations are ascribed.
AN was granted broad access to speak with any CPR employee who wished to speak with him, and to sit in on any meeting to which the attendees consented and in which no personal health information was being shared. Ethics approval for the formal study was obtained via AN's home university and a second IRB approval was obtained from an Australian host university. Interlocutors ranged from the highest-ranking directors of the organization to the support staff working only indirectly in the world of e-mental health. All interviews were recorded and transcribed, and all field notes were either typed into a word processor as they were taken or were taken by hand in a notebook and subsequently typed. Seemingly significant comments were highlighted in the moment, then returned to later for further analysis during the writing stage. Because many interlocutors became close friends who would discuss these very same ethical issues in social settings, and because of the highly reflexive nature of the interviews, organic member-checking was continuously performed.
We turn now to four short ethnographic vignettes– two from each co-author– which highlight real-world instantiations of our titular concept. We explain how ‘what are we doing here’ logic pervades each empirical case. The first-person is used, reflecting the authors’ positionality and physical presence as witnesses.
During my time at CPR, a trial was underway to develop an app for new doctors (known as Junior Medical Officers, or JMOs, in Australia) to better manage their psychological well-being. Medical students and residents are frequently the target of mental health interventions because it is well known that their jobs are highly stressful and prone to produce high incidence of anxiety and depression (Aljuwaiser et al. 2024). The app was called WellDoc, and featured a mixture of CBT exercises, guided meditations, a mood-tracking function, and a list of additional resources. The trial evaluating its efficacy was called HOSPiTAL (Helping Out Struggling Physicians-in-Training with Anxiety and Loneliness). One day, I observed an interaction among the HOSPiTAL team that was exemplary of the friction between its ostensible mission and the on-the-ground realities of intervention science. I describe the scene in detail:
Seven individuals, myself excluded, sit around a rectangular conference room table facing a projector. The meeting is to discuss the overall design of WellDoc, including its aesthetics and the menu options available to users. Such meetings are commonplace at CPR, as researchers and IT staff often revise the capabilities of their programs based on what is most beneficial to the target population. Typical of these meetings, there are members present from the several different CPR departments involved in bringing one of these products to market. Susan, a high-ranking IT administrator sits to my left. Across the table, near the projector, sits Greta, an idealistic and outspoken research psychologist. Up at the front, the project lead is walking the group through the results of a preliminary ‘usability’ test of the app. There is some laughter in the room when she reads that 0% of respondents said they would be willing to use the app every day, meaning that the planned “30-day challenge” feature will have to be reworked. It is then the user experience (UX) manager’s turn to address the room, and she motions to a screenshot of the app on the projector. She points to the traced lines and dots on the screenshot, colored in varying levels of intensity, showing a composite image of where each user testing out the program put their fingers and for how long. A conversation then ensues about how to best redesign the app layout using this data.
Things take an interesting turn, however, when Greta suggests adding a simple feature: allow the JMOs to log and track their working hours, and then compare these against the World Health Organization’s recommended maximum. After all, she reminds the group, the data shows that, after controlling for all other factors, working hours are the biggest risk factor for depression for junior doctors. Susan bristles at this: “that’s a bit of a political risk,” she says. She proceeds to explain that the rostered hours of physicians-in-training is “an old political question,” in other words a longstanding labor dispute that the group should not wade into. Further, she points out, the JMOs’ employer– the State Ministry of Health– is the one putting up the money for the study: it simply wouldn’t do to put such a feature in the app which opposes the fundamental interests of its funders.
If the feature is implemented, Susan says, “participants will basically just say ‘thanks for this [behavioral intervention/meditation], but I want to work fewer hours.” The very premise of the intervention, and in a way CPR’s entire raison d’être, would be called into question. The user would start demanding the wrong thing. Greta pushes back: “but the research supports [this conclusion]. It agrees with the medical students.” The conversation quickly moves on to other topics and remains polite and professional the entire time.
Here Greta gives voice to ‘what-are-we-doing-here’ moral reasoning, calling attention to an as-yet unspoken contradiction at the heart of the team’s work, between the interests of their funders and the interests of their target population. Note that when such a contradiction starts to surface, rather than lead to any kind of interpersonal conflict or concerted follow-up, those ranked higher move quickly to politely but firmly put the genie back in the bottle. This is characteristic of the way in which larger political and social debates lurked just under the surface of countless discussions and cordial disagreements I witnessed at CPR.
Class antagonisms and state interests play out in real time in banal discussions of app, website, and trial design. The features made available (and unavailable) in each program are a response to what the Centre (and by extension, its backers) deem to be ‘the problem’. For example, in the case of HOSPiTAL, those who commissioned the creation of WellDoc see the problem of junior doctor suicide and poor mental health as one of lifestyle management, psychoeducation, and basic emotional regulation. The role of long and taxing working hours for JMOs is not only obscured but rendered invisible as a social issue in the intervention itself. All CPR programs function with some version of this mechanism: being prescriptive about ‘the problem’ by being prescriptive about ‘the solution.’
Despite robust evidence linking excessive work hours to physician depression, the team excluded labor tracking from the app after a senior administrator deemed it ‘too political’, a decision directly tied to the State Ministry of Health’s role as funder. Though a critical piece of context, there were no tools to document how the interests of funders themselves circumscribe intervention research. This research project thus serves a depoliticizing function, and its implementation reifies the narrow scope of ‘acceptable’ solutions in health research.
HaRP was a harm reduction-based intervention to connect people with substance use disorder to treatment as they exit jail. The training and support offered to practitioners through HaRP, as Amy, the clinical supervisor once put it, was ‘luxurious’. In fact, the training, technology, and support required by the HaRP trial did not seem feasible in real-world settings, where resources were often limited and unevenly distributed – particularly in the context of community-based services for people who use drugs, a context in which the clinical supervisor had decades of experience.
Many of these ‘luxurious’ elements of HaRP – the months-long training, the clinical supervision and case consultations, hundred-dollar Uber rides for participants – were plainly ‘infeasible’ outside of the clinical trial context. Concern over HaRP’s future prompted the implementation team, those tasked with evaluating whether HaRP was feasible and sustainable among many other tasks, to ask the PIs about their goals for the intervention beyond the trial. The following exchange took place during a research team meeting a few months before recruitment began:
Amy (Clinical supervisor): How would [HaRP] actually be replicated? We’re wanting to know from the PIs, like what do you envision after this study is completed? [Do you envision] being able to, or wanting to provide training and technical assistance to people who want to replicate this intervention? What’s your vision, post-study?
Robert (HaRP Principal Investigator): Wow, that's a hard one. [...] I guess I didn't envision that we would personally be doing the sort of activities that you just described, I envision more that we would implement this in a very transparent and manualized way, so that others could draw from it.
Amy (Clinical Supervisor): Yeah, I mean, how will people know that they’re replicating what we’ve created if they don't have access to experts to consult with?
This conversation illustrates a certain self-terminating quality to the trial. At this point, there were few plans in place for thinking about the afterlife of the HaRP intervention beyond the trial; the PIs had myriad other issues to consider while fielding a complex intervention across the state, in partnership with the carceral system. Their duty, in other words, was to accomplish the experiment.
Amy suggests that a training manual or list of strategies, and that positions like hers (i.e., the clinical supervisor), and other built-in elements like training, case-consults and supervision would need to be funded beyond the trial. Her work, and the ongoing support and training for practitioners, was critical to its success and not a dispensable ‘extra’ that could be shed after the experiment. This discrepancy prompted sociological discussions about the scarcity of social services, and comments about the critical needs of participants that went beyond the scientific logic of investigation. For example, Carl, the implementation scientist, was disturbed by this unjust discrepancy between the well-resourced trial and the underfunded state it would almost surely return to: ‘[peers and case managers will] deserve support, right? Not just top-down supervision, but support and investment in their skill set and their well-being’.
In witnessing the extensive resources allocated to HaRP, I, like my informants, came to think of the ‘unfeasible’ as instead the ‘ethical.’ Amy and others wanted to change this customary trial practice of short-lived prosperity, and invest in peer workers and case managers. Importantly, even if the implementation team understood that this level of support, training and consultation did not mirror the ‘real world’, or wasn’t feasible, they wanted to show that something like this should exist for peers and case managers in the ‘real world’. Its lack of feasibility took the form of a political critique about under-resourced social service organizations and under-supported lay personnel. The implementation team met a few days after this conversation to discuss the results of the prior team meeting:
Amy (Clinical Supervisor): I get it, but what is the point of developing an intervention for replication if people can't get the information to replicate it?
Carl (Implementation Scientist): Especially given that we're doing all of this implementation work.
Amy (Clinical Supervisor): … it's definitely super labor intensive. I mean, we're going to make an implementation guide and manual. But that's not enough for people.
Carl (Implementation Scientist): I feel like we're making a strong ethical argument to push or to gently support thinking about if this is effective, that there should be technical assistance for this to live in the real world. (...) a lot of my work in the past has been doing intervention studies, publishing it and moving to another thing. We want this stuff to live in the real world, and we need to build an infrastructure around it. [emphasis added].
This conversation reveals a growing disenchantment with intervention science and the pressure to take on more studies and publish more articles without thinking about the ‘real world’ significance or impact. As Amy put it, the focus of PIs seemed to be accomplishing the experiment at all costs, even when it overrode the goals of the implementation science team: to ensure the intervention could exist without the trial infrastructure. For the implementation scientists, ensuring that HaRP lived on, and ‘building an infrastructure around it’ became an ethical cause worth pursuing. The clinical trial put into sharp relief the working conditions and support offered to community-based service providers, particularly those with lived experience – leaving my interlocutors wondering how they might go about making this support a reality.
The HaRP trial shows how the implementation science concept of feasibility naturalizes austerity. While the study temporarily provided ‘luxury’ supports like Uber rides and intensive training, these were framed as unsustainable; and, there were no plans in place to ensure their continued implementation after the study. By reinforcing artificially resourced trials and accepting their subsequent dismantling, the research enterprise converts ethical obligations (housing, living wages for peer workers) into ‘unaffordable’ extras, while recasting behavioral interventions as clinically and politically pragmatic. This positions our interlocutors, once again, into a space of ‘what-are-we-doing-here’ moral reasoning.
One of the largest and most ambitious studies going on at CPR during my time there was GASKET (Giving Australian School Kids Even Tempers). This large-scale longitudinal study sought to follow schoolchildren throughout four states in Australia over a period of five years, assessing the efficacy of a calming mindfulness meditation app at improving emotional regulation among young people. One day, I was interviewing Xīn yí, the project lead on GASKET, about the trial in her office. I caught her in an especially reflective mood:
The conversation turns to the pathologizing of children and childhood behaviors, things like school refusal and oppositional defiant disorder. I ask her what she thinks about the project in relation to changing ideas about “neurodiversity”, and the widening range of ways of being for young people in today’s world. She grows contemplative, and I’m taken slightly aback by what she says next: “CPR’s mission is to eliminate mental illness. But do we never want that illness at all? Do we want to eradicate all illness from our lives?” She goes on to ask rhetorically, “I wonder if in our messaging… is that the right approach?”
It's not the first time I’ve heard this sentiment expressed, that CPR may be attempting to fix a problem whose contours remain poorly understood (this especially because its programs target mild-to-moderate presentations of mental disorders which may be “situational” rather than “pathological”). As one administrator mused to me in my first week at CPR, speaking about the GASKET trial, “sometimes I wonder if what we call depression is really just life — we just kind of go along without knowing what exactly depression is”. This is perhaps the most fundamental critique one can have of any e-mental health research project, let alone a project that one is themself working on: that the target of the intervention, and by extension its purpose, is uncertain.
Xīn yí continues with another familiar line of thinking: “it might not be appropriate to talk about mental health apps [like GASKET] when people in poverty have their mind on more basic questions of survival”. But Xīn yí doesn’t know that I’ve heard these same concerns before from her colleagues. She tells me as much: “I sometimes have these conversations with my peers, but we never talk about why we’re in research with each other”, Xīn yí tells me. “There’s a sense of embarrassment if we talk about anything philosophical”.
This exchange demonstrates Xīn yí’s political consciousness, one which was widely shared by other CPR researchers: broadly liberal/progressive, expressing a desire for robust social services, and with a skepticism towards medicalization despite working within the biomedical enterprise. It also reveals a sense that such discussions are for higher pay grades. As she told me, ‘The programs may not work that well but they make the right people happy – donors, doctors, school officials’, her voice growing quieter as she considered the implications of her words. We see here a clear example of the ‘what-are-we-doing-here’ logic at work: a scientist expresses a sincere concern about the structure of society, coupled with an acknowledgment that their research is not the right tool for the job nor, implicitly, the highest priority for its resources.
Xīn yí's reflexivity and doubt (‘Do we want to eliminate all mental illnesses?’) reveal how scientists become complicit in systems they may also privately question. The trial moved forward not because researchers fully believed in the underlying social problem diagnosis, but because the infrastructure of grants, publications, careers, and institutional priorities kept alternative approaches off the table. This example shows how mental health intervention research helps determine which forms of suffering get recognized as intervenable and, by extension, shows the significant amount of context that remains inaccessible to IS inquiry.
The HaRP project manager began the all-team meeting with reports on how many participants had been recruited into the study during the past two weeks. There were cheers and excitement from meeting attendees at the prospect of increased enrollment since the previous meetings. She relayed bureaucratic updates on hiring, turnover, and progress on establishing partner relationships, then turned to reports from the field. She described how some jails that served as recruitment sites lacked enough people with OUD (opiod use disorder), a primary eligibility criterion. Participants also stayed in jail longer than researchers had expected, meaning people were often still incarcerated even as they were enrolled into the study. This meant that intervention practitioners needed to maintain contact with participants to continue building a relationship. Practitioners also needed to gather information on participants’ release dates to efficiently connect them to services upon their release.
While initial Zoom enrollment in the study was free for participants, texting and emails were managed through a third-party, for-profit company that charged inmates for all messages (called ‘Chirps’) they sent and received. Charging inmates for communication and other daily necessities is one of the more punitive ways jails have dealt with budget shortfalls (Wang 2018), and proved to be prohibitively expensive for many HaRP participants. So although enrollments in the study had increased, practitioners were initially unable to keep in contact with participants in the way originally outlined by the study. That said, the project manager encouraged the PIs to write an op-ed to advocate against the practice of charging incarcerated people money to contact the outside world. However, Robert, one of the PIs, was concerned that any type of advocacy might damage the research team’s relationship with the jails and other carceral stakeholders who he relied upon for trial recruitment. Research collaborations with carceral systems often require framing clinical trials as value-neutral, technical endeavors rather than as challenges to existing institutional structures. By emphasizing measurable outcomes and avoiding overt political critique, RCTs can facilitate partnerships between traditionally opposed actors who might otherwise find little common ground. This strategic depoliticization allows for operational cooperation while leaving fundamental disagreements about systems of punishment unaddressed.
This brief case illustrates how the research apparatus actively participates in sustaining carceral logics while conducting the HaRP clinical trial. Robert recognized that gaining the trust of the criminal-legal system required tailoring the presentation of the HaRP trial’s methodology to align with criminal-legal priorities, even if they were not his priorities as someone outwardly committed to principles of human rights and social justice. To achieve this, Robert tended to highlight the RCT’s rigorous, objective, and transparent design, emphasizing how pre-registration and randomization ensure unbiased results (Claypool, in press). Robert reconciled this impasse through what he termed ‘evidence-based optimism’, articulating a dual commitment to immediate, albeit short-term interventions alongside the long-term struggle for structural change. The PI’s reluctance to advocate against fees demonstrates co-production’s self-reinforcing dynamic, and one that is straightforwardly political: maintaining jail partnerships requires accepting their punitive logics, and thus also perpetuating them. This creates ethical tensions where measurable recruitment metrics take precedence over challenging systems that harm participants. The clinical trial creates provisional space for cooperation between carceral systems and public health actors, while also systematically excluding challenges to punitive logics endemic to the carceral system.
The above vignettes demonstrate how researchers privately critique systemic inequities yet are pressured to treat these conditions as fixed constraints rather than as mutable injustices. Because of the epistemic and institutional commitments of intervention and implementation science, structural issues are primarily understood as non-negotiable, contrary to the personal feelings and beliefs of the researchers and scientists conducting intervention work itself. Instead, these actors see such issues as in urgent need of resolution, before the ‘real’ work of implementation can begin. This tension between researchers’ critical moral and political awareness and their field’s institutionalized constraints underscores the need to examine how IS knowledge production may reproduce the very structures it has recently sought to address (Brownson et al. 2021, Shelton et al. 2021). The primary contribution of the vignettes is to demonstrate this specific kind of positionality, as well as the frustrations, tensions, and contradictions that come from navigating them in professional research settings.
We argue that implementation science’s emphasis on pragmatism may sideline deeper scrutiny of unjust social dynamics in healthcare. IS is, by definition, invested a priori in the institutionalization of evidence-based programs; its lot is tied to the legitimacy of the institutions it studies. Yet IS rarely acknowledges this vested interest and the way in which it limits what IS scholars can study, critique, and ultimately change. This makes the field less powerful and more susceptible to uncritically accepting the forms of violence already entrenched within intervention settings. We follow feminist philosophy in treating scientific knowledge as ‘situated’, produced ‘from somewhere’ by actors with positionalities, ideologies, and interests (Haraway 1988). Without a rigorous investigation of IS’ own positionality as a science the field is made less sensitive to uncovering the power dynamics which have laudably become a recent focus in IS scholarship. Of course, this tension is not specific to IS alone. The broader health services scholarship must mold its accountabilities to healthcare institutions, similarly narrowing its scope of inquiry to questions that, while improving healthcare systems, may fail to challenge the underlying distributions of power and resources that drive health inequities.
While IS intends to bridge the gap between research and practice, it may also inadvertently reproduce social hierarchies, thereby legitimizing the very structural conditions that are known to lead to poor health outcomes. To be sure, some IS scholars have interrogated and developed frameworks for ascertaining these power dynamics, critiquing the dominance of Western epistemologies and highlighting how structural inequities and colonial legacies shape what knowledge is viewed as credible, effectively sustaining the very hierarchies of power that implementation science claims to disrupt (Stanton et al. 2022, Adsul et al. 2024). Others, though less focused on the context of research, have brought a ‘health equity lens’ to the field to interrogate how social determinants of health affect the ways in which innovations are differentially taken up and implemented (Baumann & Cabassa 2020, Woodward et al. 2021). Lastly, a growing cohort of IS scholars have started to focus the field upstream away from behavioral interventions toward ‘policy implementation science’ and a concerted effort to focus the field more on the ‘outer setting’ (Emmons & Chambers 2021, Purtle et al. 2023, Warner et al. 2023). Centering the outer context is critical for health equity because it moves beyond individualist paradigms to address the structural drivers that fundamentally mediate the stratification of health outcomes. Our vignettes show how these power dynamics may play out during the course of research.
Within both studies at the center of these ethnographies, there were no productive avenues to reconcile the tensions captured in vignettes or pathways to ensure such structural ‘barriers’ were meaningfully addressed. Within the everyday constraints of research and implementation work, opportunities for open-ended discussion are often scarce. While routine team meetings are common, they tend to focus on task completion rather than deeper reflection (Finley et al. 2018). The large-scale neglect of these issues generates feelings of absurdity and frustration- i.e. our titular ‘moments’. These dynamics reveal two critical imperatives for implementation science: first, to interrogate how research frameworks co-produce knowledge by legitimizing certain interventions while marginalizing alternatives; and second, to embrace situated reflexivity, and critically examining the internal logics of research itself, including funding structures, institutional alliances, and epistemic norms. This article calls for continued exploration into how IS’s institutional commitments shape implementation. For example, relying on ‘optimism’ and the rewards of incremental change were cited as an antidote to the frustrations of pursuing long-term structural change. Specifically, we must examine the unacknowledged political negotiations that occur in practice yet vanish from the scientific record, as well as how researchers navigate the tension between the social change they value and the systems they perpetuate.
Finally, this article calls for developing further avenues for exploring how the institutional commitments of IS research affect the implementation process, including what types of questions arise but are left unheard, as well as what political negotiations perennially appear during implementation but are left off the scientific record as ‘public secrets’. One simple way researchers can do this is by drawing on existing ethnographic methods. We propose leveraging tools like periodic reflection’ to strengthen team sensemaking through open-ended, iterative discussion, encouraging teams to document and explore ongoing challenges and contextual shifts in real time (Finley et al. 2018). These periods of reflection could feature opportunities for the anonymous sharing of concerns or special guarantees of freedom from reprisal. Such approaches may create spaces for addressing ‘what are we doing here’ moments – helping to transform these ‘public secrets’ into actionable insights. They also create a necessary alternative to short-term optimism by providing a structured pathway for researchers to align their methods and intervention design with their long-term ethical commitments, ensuring that the pursuit of ‘what works’ (or making ‘what works’ work) does not come at the cost of challenging the ways in which structural inequities persist.
Both authors are grateful to Leopoldo Cabassa, Roman Kislov, and Byron Powell for their comments on our work. We extend additional gratitude to the implementation science community for its warmth and willingness to engage with our ideas.
EC was supported by The Fahs Beck Fund for Research and Experimentation to conduct her research, and was supported by the Agency for Healthcare Research and Quality (AHRQ), (T32 HS000084) National Research Service Award, Health Services Research Training Program throughout data collection and study completion. Her contributions to this article were supported by a Ruth L. Kirschstein National Research Service Award (NRSA) Postdoctoral Research Training Grant
AN was supported by fieldwork grants from the National Science Foundation (#BCS-1917569) and the Wenner-Gren Foundation (#5387093910), and a writing fellowship at the Stanford Humanities Center via the Andrew W. Mellon Foundation. His contributions to this article were supported by an NIMH T32 postdoctoral grant (#MH19960) through the Center for Mental Health Services Research.
The authors declare they have no competing interests.
Emily Claypool https://orcid.org/0000-0002-5516-9589
Aaron Neiman https://orcid.org/0000-0001-6928-6690
Adsul, P., Shelton, R. C., Oh, A., Moise, N., Iwelunmor, J., & Griffith, D. M. (2024) Challenges and opportunities for paving the road to global health equity through Implementation Science. Annual Review of Public Health, 45(1), 27–45. https://doi.org/10.1146/annurev-publhealth-060922-034822
Aljuwaiser, S., Brazzelli, M., Arain, I., & Poobalan, A. (2024) Common mental health problems in medical students and junior doctors – an overview of systematic reviews. Journal of Mental Health, 33(6), 779–815. https://doi.org/10.1080/09638237.2023.2278095
Bauer, M. S., & Kirchner, J. (2020) Implementation science: What is it and why should I care? Psychiatry Research, 283, 112376. https://doi.org/10.1016/j.psychres.2019.04.025
Baumann, A. A., & Cabassa, L. J. (2020) Reframing implementation science to address inequities in healthcare delivery. BMC Health Services Research, 20(1), 190. https://doi.org/10.1186/s12913-020-4975-3
Boulton, R., Sandall, J., & Sevdalis, N. (2020) The cultural politics of ‘implementation science’.
The Journal of Medical Humanities, 41(3), 379–394. https://doi.org/10.1007/s10912-020-09607-9
Brownson, R. C., Kumanyika, S. K., Kreuter, M. W., & Haire-Joshu, D. (2021) Implementation science should give higher priority to health equity. Implementation Science, 16(1), 28. https://doi.org/10.1186/s13012-021-01097-0
Brownson, R. C., Shelton, R. C., Geng, E. H., & Glasgow, R. E. (2022) Revisiting concepts of evidence in implementation science. Implementation Science, 17(1), 26. https://doi.org/10.1186/s13012-022-01201-y
Claypool, E. (in press) The double bind of evidence: Harm reduction, structural change and the institutional constraints of randomized controlled trials in criminal-legal settings. Social Service Review
Corman, M. K. (2021) Using composites to craft institutional ethnographic accounts. In P. C. Luken & S. Vaughan (Eds.), The Palgrave handbook of institutional ethnography (pp. 465–482). Springer International Publishing. https://doi.org/10.1007/978-3-030-54222-1_24
Dourish, P. (2004) What we talk about when we talk about context. Personal and Ubiquitous Computing, 8(1), 19–30. https://doi.org/10.1007/s00779-003-0253-8
Dryden-Palmer, K. D., Parshuram, C. S., & Berta, W. B. (2020) Context, complexity and process in the implementation of evidence-based innovation: A realist informed review. BMC Health Services Research, 20(1), 81. https://doi.org/10.1186/s12913-020-4935-y
Eccles, M. P., & Mittman, B. S. (2006) Welcome to Implementation Science. Implementation Science, 1(1), 1. https://doi.org/10.1186/1748-5908-1-1
Emmons, K. M., & Chambers, D. A. (2021) Policy implementation science – an unexplored strategy to address social determinants of health. Ethnicity & Disease, 31(1), 133–138. https://doi.org/10.18865/ed.31.1.133
Finley, E. P., Huynh, A. K., Farmer, M. M., Bean-Mayberry, B., Moin, T., Oishi, S. M.,…& Hamilton, A. B. (2018) Periodic reflections: A method of guided discussions for documenting implementation phenomena. BMC Medical Research Methodology, 18(1), 153. https://doi.org/10.1186/s12874-018-0610-y
Franklin, S. (1995) Science as culture, cultures of science. Annual Review of Anthropology, 24, 163–184. https://doi.org/10.1146/annurev.an.24.100195.001115
Fullwiley, D. (2011) The enculturated gene: Sickle cell health politics and biological difference in West Africa. Princeton University Press.
Geertz, C. (1998, October 22nd) Deep hanging out. The New York Review. http://www.nybooks.com/articles/1998/10/22/deep-hanging-out/ Available at http://hypergeertz.jku.at/GeertzTexts/Deep_Hanging.htm
Geissler, P. W. (2013) Public secrets in public health: Knowing not to know while making scientific knowledge. American Ethnologist, 40(1), 13–34. https://doi.org/10.1111/amet.12002
Hamilton, A. (2013) Bridging anthropology and health services research. Anthropology News, 54(3), e1–e14. https://doi.org/10.1111/j.1556-3502.2013.54301.x
Haraway, D. (1988) Situated knowledges: The science question in feminism and the privilege of partial perspective. Feminist Studies, 14(3), . 575-599. https://doi.org/10.2307/3178066
Jasanoff, S. (Ed.). (2004) States of knowledge: The co-production of science and the social order. Routledge.
Kislov, R., Pope, C., Martin, G. P., & Wilson, P. M. (2019) Harnessing the power of theorising in implementation science. Implementation Science, 14, 103. https://doi.org/10.1186/s13012-019-0957-4
Lakoff, A. (2006) Pharmaceutical reason: Knowledge and value in global psychiatry. Cambridge University Press.
May, C. R., Johnson, M., & Finch, T. (2016) Implementation, context and complexity. Implementation Science, 11, 141. https://doi.org/10.1186/s13012-016-0506-3
Montoya, M. (2011) Making the Mexican diabetic: Race, science, and the genetics of inequality. University of California Press.
Murdoch, J., Paparini, S., Papoutsi, C., James, H., Greenhalgh, T., & Shaw, S. E. (2023) Mobilising context as complex and dynamic in evaluations of complex health interventions. BMC Health Services Research, 23(1), 1430. https://doi.org/10.1186/s12913-023-10354-5
Nader, L. (1972) Up the anthropologist: Perspectives gained from studying up. Available from https://eric.ed.gov/?id=ED065375
Osseo-Asare, A. D. (2019) Atomic junction: Nuclear power in Africa after independence. Cambridge University Press. https://doi.org/10.1017/9781108557955
Purtle, J., Moucheraud, C., Yang, L. H., & Shelley, D. (2023) Four very basic ways to think about policy in implementation science. Implementation Science Communications, 4, 111. https://doi.org/10.1186/s43058-023-00497-1
Rabinow, P. (2002) French DNA: Trouble in purgatory. University of Chicago Press.
Rhodes, T., Closson, E. F., Paparini, S., Guise, A., & Strathdee, S. (2016) Towards ‘evidence-making intervention’ approaches in the social science of implementation science: The making of methadone in East Africa. International Journal of Drug Policy, Drug Use and Drug Policies in Sub-Saharan Africa, 30, 17–26. https://doi.org/10.1016/j.drugpo.2016.01.002
Rhodes, T., & Lancaster, K. (2019) Evidence-making interventions in health: A conceptual framing. Social Science & Medicine, 238, 112488. https://doi.org/10.1016/j.socscimed.2019.112488
Shelton, R. C., Adsul, P., & Oh, A. (2021) Recommendations for addressing structural racism in implementation science: A call to the field. Ethnicity & Disease, 31(Suppl 1),357–364
Snell-Rood, C., Jaramillo, E. T., Hamilton, A. B., Raskin, S. E., Nicosia, F. M., & Willging, C. (2021) Advancing health equity through a theoretically critical implementation science. Translational Behavioral Medicine, 11(8), 1617–1625. https://doi.org/10.1093/tbm/ibab008
Stanton, M. C., Ali, S. B., & The Sustain Center Team (2022) A typology of power in implementation: Building on the exploration, preparation, implementation, sustainment (EPIS) framework to advance mental health and HIV health equity. Implementation Research and Practice, 3. https://doi.org/10.1177/26334895211064250
Taussig, M. (1999) Defacement: Public secrecy and the labor of the negative. Stanford University Press.
Taylor, J. S. (2003) Confronting ‘culture’ in medicine’s ‘culture of no culture’. Academic Medicine, 78(6), 555–559. https://doi.org/10.1177/2633489521106425
Wang, J. (2018) Carceral capitalism. Semiotext(e).
Warner, E. T., Huguet, N., Fredericks, M., Gundersen, D., Nederveld, A., Brown, M. C., … & Emmons, K. M. (2023) Advancing health equity through implementation science: Identifying and examining measures of the outer setting. Social Science & Medicine, 331, 116095. https://doi.org/10.1016/j.socscimed.2023.116095
Wensing, M., & Wilson, P. (2023) Making implementation science more efficient: Capitalizing on opportunities beyond the field. Implementation Science, 18(1), 40. https://doi.org/10.1186/s13012-023-01298-9
Woodward, E. N., Singh, R. S., Ndebele-Ngwenya, P., Melgar Castillo, A., Dickson, K. S., & Kirchner, J. E. (2021) A more practical guide to incorporating health equity domains in implementation determinant frameworks. Implementation Science Communications, 2(1), 61. https://doi.org/10.1186/s43058-021-00146-5
Journal of Critical Public Health, Volume 3 (2026), Issue 3 CC-BY-NC-ND