There is a growing recognition that current research has failed to adequately address persistent societal challenges, which are complex, uncertain, and evaluative in nature (Ferraro et al., 2015; Loorbach et al., 2017; Saltelli et al., 2016). Along with this recognition come calls for science to help address these increasingly urgent and complex challenges faced at a global and local level, such as biodiversity loss, climate change, or social inequalities (Future Earth, 2014; Parks et al., 2019; WBGU, 2011). This call is echoed from within academia (Bradbury et al., 2019; Fazey et al., 2018; Norström et al., 2020) and has also translated into corresponding research funding (Arnott et al., 2020; Gerber et al., 2020; Vermeer et al., 2020). The fundamental premise is that addressing complex societal challenges requires more than disciplinary knowledge alone and extends beyond the confines of academia (Gibbons et al., 1994; Hirsch Hadorn et al., 2008; Lang et al., 2012). That is, addressing them necessitates interactive knowledge co-production and social learning with societal actors to produce actionable and contextually embedded knowledge for societal transformations (Chambers et al., 2021; Hessels et al., 2009; Schäpke et al., 2018). This trend has prompted a (re)surge of socially engaged approaches to research, including transdisciplinary research, phronetic social sciences, participatory research, action- and impact-oriented research, and transformative research. These approaches involve collaboration between academics and various societal stakeholders, such as policymakers, communities, enterprises, and civil society organisations.

However, often, such socially engaged research approaches are at odds with the institutional traditions designed for monodisciplinary knowledge production. Transformative research, for instance, does not claim an objective observer position; instead, it explicitly embraces a normative orientation. Its goal, as many have argued, is to facilitate transformative societal change towards justice and sustainability by recognising and addressing the deep and persistent socio-ecological challenges inherent in our current society (Mertens, 2007; Wittmayer et al., 2021). This motive to transform existing systems through collaborative research, in our view, obliges researchers to be more critical and vigilant in their decisions (Fazey et al., 2018). As we will present later in this paper, many of these decisions constitute ethical dilemmas, such as who decides what ‘good’ research is, whose knowledge to prioritise, or who should engage and under which circumstances. These ethical dilemmas are only poorly addressed by the ethical review processes in place at most universities, which remain dominated by linear and positivist framings of knowledge production and research design (Wood and Kahts-Kramer, 2023). Consequently, transformative researchers are often left struggling to choose “between doing good (being ethically responsive to the people being researched) and doing good research (maintaining pre-approved protocols)” (Macleod et al., 2018, p. 10). The translation of the values and principles of transformative research into formal and informal ethical guidelines is only starting (Caniglia et al., 2023; Fazey et al., 2018; West and Schill, 2022).

Confronting these ethical dilemmas calls for greater reflexivity and dialogue with ourselves, among researchers, between researchers and their collaborators (including funders and professionals), and between researchers and the institutions within which they operate (Finlay, 2002; Horcea-Milcu et al., 2022; Pearce et al., 2022). Attesting to this call, the authors of this paper engaged in a ‘collaborative autoethnography’ (Lapadat, 2017; Miyahara & Fukao, 2022; Phillips et al., 2022) to explore the following research question: Which ethical dilemmas do researchers face in research collaborations that seek to catalyse transformations? And how do they navigate these in their collaborative practice? Thus, as an interdisciplinary group of researchers affiliated with academic research institutes, we shared, compared, and discussed our experiences concerning ethical dilemmas in our transformative research endeavours. In these discussions, we considered our interactions, engagements, and relationships with collaborators along with how institutional rules and norms influence or constrain our practices and relations.

This paper begins with an overview of transformative research and the challenges that arise when working collaboratively. It also testifies to the formal and informal attempts to support researchers in navigating those challenges (“Ethics in transformative research”). From there, we develop the argument that formal or informal guidelines are most meaningful when they have become a part of the practical wisdom of researchers. When they are, they support researchers in leaning into the uncertainty of what constitutes morally good behaviour and in navigating collaboration ‘in situ’. Inspired by Mertens (2017), we relate our own dilemmas to the three philosophical commitments that comprise a research paradigm: axiology, ontology, and epistemology (“Transformative research practice investigated through collaborative autoethnography”, also for an elaboration of the terms). We share concrete dilemmas while embedding and relating them to a broader body of knowledge around similar dilemmas and questions (“Collaboration in transformative research practice”). We close the paper by pointing to the importance of bottom-up ethics and the need to embed those into revalued and redesigned ethical standards, processes, and assessments that can provide external guidance and accountability (“Concluding thoughts”).

Ethics in transformative research

In this section, we first introduce transformative research (TR) in terms of its underlying values and its ontological and epistemological premises (Mertens, 2007, 2017) (“Introducing transformative research”). We then connect it to its institutional context, where ethical standards and procedures fit the linear production of knowledge, leading to tensions with TR practices (“Institutional context: Formal ethical standards and processes”). Finally, we outline how the research community tries to address this misfit and the felt need for understanding what constitutes morally ‘right’ behaviour by providing peer guidance on the ethical conduct of TR (“Peer context: Informal heuristics for transformative research”).

Introducing transformative research

TR refers to a broad and loosely connected family of research disciplines and approaches, with the explicit normative ambition to fundamentally question the status quo, change the dominant structures, and support just sustainability transitions (Hölscher et al., 2021; Jaeger-Erben et al., 2018; Mertens, 2021; Schneidewind et al., 2016; Wittmayer et al., 2021). Transformative researchers thus start from the basic premise that “all researchers are essentially interveners” (Fazey et al., 2018, p. 63). Consequently, they are explicit about the kind of normative orientation of their interventions to further a social justice and environmental sustainability agenda. There is no denying the fact that such research approaches can also be used with a different normative mindset and value orientation, which will have other ethical consequences.

TR builds on methodological and theoretical pluralism that knits together kindred, or even conflicting, perspectives to complement disciplinary specialism (Hoffmann et al., 2017; Horcea-Milcu et al., 2022; Midgley, 2011). As such, it also comes as a diverse phenomenon, and where such diversity is “not haphazard […] we must be cautious about developing all-embracing standards to differentiate the ‘good’ from the ‘bad’” (Cassell and Johnson, 2006, p. 783). Such an ontological stance involves letting go of the idea of absolute truth and the need to tightly control the research process and outcomes (van Breda and Swilling, 2019). Instead, TR encourages continuous societal learning to generate actionable knowledge and transformative action that manifests in real-world changes in behaviours, values, institutions, etc. (Bartels and Wittmayer, 2018; Hölscher et al., 2021). In doing so, TR is often based upon pragmatist assumptions about the ways knowledge and action inform one another, generating contingent knowledge in a process of action and experimentation (Harney et al., 2016; Popa et al., 2015). The research process serves as a means to assess ideas in practical application, blending a critical realist stance on socially constructed reality with acknowledging subjectivism and the existence of multiple realities (Cassell and Johnson, 2006).

TR also represents an epistemological shift from the notion of the distanced, presumably unbiased, and all-knowing researcher and recognises individuals as sense-makers, agency holders, and change agents (Horcea-Milcu et al., 2022; Hurtado, 2022). Collaboration enables the elicitation of different kinds of knowledge, including scientific knowledge across disciplines as well as phronetic and tacit knowledge from practice. It aims at capturing the plurality of knowing and doing that is relevant to specific contexts and actors (Frantzeskaki and Kabisch, 2016; Nugroho et al., 2018; Pohl, 2008). This sort of mutual social learning supports joint sense-making and experimental processes. These then invite us to rethink existing situations, (re)define desired futures, and (re)position short-term action (Fazey et al., 2018; Lotz-Sisitka et al., 2016; Schneider et al., 2019). The co-creation of knowledge and action can increase ownership, legitimacy, and accountability and can help facilitate trust-building among diverse societal groups (Hessels et al., 2009; Lang et al., 2012). The latter is an essential ingredient for tackling complex societal problems during times of discrediting science and the rise of populist, antidemocratic movements (Saltelli et al., 2016).

Institutional context: formal ethical standards and processes

The institutional environment is challenging for researchers engaging in TR for multiple reasons; one challenge is the formal ethical standards and processes. Current approaches to ethical assessment in social science emerged from several international conventions in the field of medical ethics (BMJ, 1996; General Assembly of the World Medical Association, 2014; National Commission for the Protection of Human Subjects of Biomedical, & Behavioural Research, 1979). Most formal research ethics reviews adopt the four principles of Beauchamp and Childress (2001), which include: (1) non-maleficence by attempting to not harm others; (2) respect for autonomy by attempting to provide information about the research that allows decisions to be taken; (3) beneficence by attempting to achieve useful outcomes outweighing the risks of participation; and (4) justice by attempting fairness in participation and distribution of benefits. These principles have found their way into formal ethical reviews, often practicing value-neutral and utilitarian ethics. This approach is debatable for TR approaches (Detardo-Bora, 2004) and seems more effective at protecting research institutions (foregrounding bureaucratically controllable compliance) than research participants (Christians, 2005). Indeed, many engaged in TR have raised concerns that neither these principles nor their formal translation account for the particularity, situatedness, epistemic responsibilities, and relationality that are key to the conduct and ethics of TR (Cockburn and Cundill, 2018; Lincoln, 2001; Parsell et al., 2014; Wijsman and Feagan, 2019). In the following paragraphs, we highlight several tensions between the understanding of research, as it informs many ethical standards in place, and an understanding of TR.

First, a pre-defined versus an emerging research design. Due to its real-world orientation, TR needs to be able to deal flexibly with changing contexts and windows of opportunity that might arise (Hurtado, 2022). Due to the relationality of TR, it requires ongoing interaction and negotiation between researchers and their collaborators (Bartels and Wittmayer, 2018; Bournot-Trites and Belanger, 2005; Williamson and Prosser, 2002). One-off general consent at the start (e.g., through informed consent forms), as is common for ethical review processes, is thus at odds with the emergent design of TR and is also argued to be insufficient in maintaining participants’ autonomy (Smith, 2008). As an alternative, Locke et al. (2013) posit that informed consent should be seen as a collective, negotiated, continuous process, especially in collaborative action research.

Second, assumed neutrality versus dynamic aspects of researchers’ positionalities. Ethical review protocols are geared towards upholding the objective position of researchers as outsiders in the investigated context, ensuring that they will not influence this research context in any way. However, TR explicates its ambition to influence real-world problems through engagement, acknowledging that research needs to confront existing hegemonic orders and emancipate those involved through a democratic process (Cassell and Johnson, 2006). Furthermore, researchers co-design, facilitate, and participate in the process of knowledge co-production, making them also participants and subjects of their own research (Janes, 2016). To enhance the validity and integrity of the research, Wood, and Kahts-Kramer (2023), among others, suggest that transformative researchers explicitly state their positionality. This involves reflecting on their assumptions, values, and worldviews.

Third, the primacy of knowledge generation versus the importance of action. Ethical review protocols, given their historical roots in medical practice, assume that the act of falsifying, generating, or improving theories alone would benefit participants, collaborators, and the public at large. Yet, researchers engaged in TR take a step further, seeking to develop both scientific and actionable knowledge in a way that addresses persistent societal problems and stimulates social change (Bartels and Wittmayer, 2018; Caniglia et al., 2021; Greenwood and Levin, 2007). As put by Wood and Kahts-Kramer (2023, p. 7), “the ethical imperative of participatory research is to bring about positive change and generate theory from reflection on the purposeful action”. This approach strengthens the responsiveness of research to societal and political needs (Stilgoe et al., 2013).

Transformative researchers thus perceive a lack of utility and guidance from ethical standards and processes in place that have institutionalised a certain understanding of research and related sets of principles. Following Clouser and Gert (1990), one might question whether such institutionalisation of a moral consciousness is possible in the first place. They argue that so-called ‘principlism,’ “the practice of using ‘principles’ to replace both moral theory and particular moral rules and ideals in dealing with the moral problems that arise in medical practice” (Clouser and Gert, 1990, p. 219), has reduced the much-needed debates on morality vis-à-vis research and results in inconsistent and ambiguous directives for morally ‘right’ action in practice. In response to the vacuum left by institutionalised ethics standards and processes and the perceived necessity of defining morally ‘right’ behaviour, the research community is turning inward to develop peer guidance on ethical conduct in TR. The subsequent section highlights several contributions to this endeavour.

Peer context: Informal heuristics for transformative research

Transformative researchers have started offering general principles or frameworks as informal heuristics for what constitutes ‘ethical’ TR. Caniglia et al. (2023), for example, argue that practical wisdom can serve as a moral compass in complex knowledge co-production contexts, and propose four central ‘wills’ for researchers to follow: committing to justice, embracing care, fostering humility, and developing courage. Under the framing of post-normal or Mode-2 science (Funtowicz and Ravetz, 1994; Gibbons et al., 1994; Nowotny et al., 2003), Fazey et al. (2018) present ten ‘essentials’ of action-oriented research on transforming energy systems and climate change researchFootnote 1. One of these essentials highlights that, as researchers, we intervene, and that failing to acknowledge and engage with this reality opens the doors to sustaining unjust power relations or positioning science as apolitical. To address this, they echo Lacey et al.’s (2015, p. 201) assertion that such acknowledgment means “be[ing] transparent and accountable about the choices made about what science is undertaken, and how it is funded and communicated”.

Looking beyond sustainability scholarship, other researchers have also developed practical actions or strategies for enhancing their ethical behaviours in the research collaboration. Taking the unique attributes of community-based participatory research, Kwan and Walsh (2018, p. 382) emphasise a “focus on equity rather than equality” and on practicing a constructive or generative use of power “rather than adopting a power neutral or averse position”. Others provide guiding questions to think about the forms and quality of relationships between researchers and participants (Rowan, 2000) and to support the navigation of the relationship between action research and other participants (Williamson and Prosser, 2002). Such questions should cover not only process-focused questions but also the risks and benefits of the intended outcomes, as well as questions around purpose, motivation, and directionalities (Stilgoe et al., 2013). Others also propose broader guidelines in which they pay attention to non-Western and non-human-centred virtue ethics, such as ‘Ubuntu’ (I am because we are) (Chilisa, 2020). In forwarding climate change as a product of colonisation, Gram-Hanssen et al. (2022) join Donald’s (2012) call for an ethical relationality and reiterate the need to ground all transformation efforts on a continuous process of embodying ‘right relations’ (see also Chilisa, 2020; Wilson, 2020).

Yet, as argued before, ethics in collaboration cannot be approached through developing principles and strategies alone. Not only might they not be at hand or on top of one’s mind when being immersed in a collaborative practice, which often requires a certain reaction on the spot. They also cannot or should not replace the quest for what morality means within that collaboration (cf. Clouser and Gert, 1990). Further questions have been prompted about the necessary skillsets for realising ethical principles in practice (Jaeger-Erben et al., 2018; Pearce et al., 2022; West and Schill, 2022). Caniglia et al. (2023), for example, propose that researchers need skills such as dealing with plural values with agility and traversing principles and situations with discernment. Others focus on competency building among research participants (Menon and Hartz-Karp, 2023). The subsequent section turns to the point of supporting researchers in navigating collaboration ‘in situ’ and in leaning into the uncertainty around what morally good behaviour constitutes—in concrete TR contexts that are plural and uncertain.

Transformative research practice investigated through collaborative autoethnography

Transformative research as a situated practice

The aforementioned institutionalised ethical standards and procedures, as well as the informal peer heuristics, are two vantage points for guidance on what constitutes morally good behaviour for transformative researchers. These existing vantage points are either developed based on theoretical and philosophical framings or based on researchers’ actual experiences of doing TR. They do offer a repertoire to explain and justify positions and decisions in ethical dilemmas during research collaborations. However, it is not until such heuristics or principles have become part of the practical knowledge of researchers that they are useful for actual TR in situ.

Considering research more as a practice situates it as a social activity in a ‘real-world context’. In such a practice, researchers often make decisions on the spot. Moreover, due to the constraints posed by available time and resources, researchers often engage in what Greenwood and Levin (2007, p. 130) term “skilful improvisation” or “pragmatic concessions” (Greenwood and Levin, 2007, p. 85). This “improvisational quality” (Yanow, 2006, p. 70) of the research process does not mean it is not carried out systematically. Such systematicity is based on “action repertoires” (Yanow, 2006, p. 71) that researchers creatively use and remake (Malkki, 2007). This improvisation is thus neither spontaneous nor random; rather, it builds on and is based on the practical knowledge of researchers (formed through their experiences and their situatedness) guiding their behaviours in normatively complex situations. Using ‘organic design’ (Haapala et al., 2016), the researchers blend real-world settings into formal spaces, fostering bricolage and driving sustainable institutional evolution over time. Such practical knowledge includes “both ‘know how’ knowledge (techne), […] and ethical and political-practical knowledge (phronesis)” (Fazey et al., 2018, p. 61). Research can thus be considered a craft (Wittmayer, 2016): the skilful mastery of which develops over time through learning based on experience and reflection (Kolb, 1984).

Such experiential learning should go beyond reflecting on what lies in view to include seeing how attributes of the viewer shape what is being viewed (cf. Stirling, 2006). Engaging in TR includes being one’s own research instrument, which puts a researcher’s positionality, i.e., their social, cultural, and political locations, centre stage. It reminds us that researchers are “located within networks of power and participate in the (re)configuration of power relations” (Wijsman and Feagan, 2019, p. 74). This positionality, the sum of what makes a person and how this informs their actions (Haraway, 1988; Kwan and Walsh, 2018; Marguin et al., 2021), is increasingly being acknowledged in academia. It has a long history in feminist theories, participatory action research, and the critical pedagogy of decolonisation. Positionality refers to the “researcher’s self-understanding and social vision” (Coghlan and Shani, 2005, p. 539) as well as their motivation to ‘better society’ (Boyle et al., 2023; Kump et al., 2023) and how these affect how researchers interpret ethical guidelines, conduct research, interpret data, and present findings. Consequently, one’s positionality can make certain research choices seem unethical. Mertens (2021, p. 2), for example, considers “continuing to do research in a business-as-usual manner” unethical as it makes the researcher “complicit in sustaining oppression”.

Acknowledging one’s positionality and normative role is part of a broader reflexive practice of critically questioning, reflecting on, and being transparent about values, as well as taking responsibility and accountability for research processes and outcomes (Fazey et al., 2018; Pearce et al., 2022; Wijsman and Feagan, 2019). Such a reflexive practice can support individual researchers to act ethically, but more so, to improve our collective ways of being and doing (i.e., an ethically informed research community) by constantly connecting what should be (i.e., the guidelines) and how it has been done (i.e., the practices) through critical reflexive practices. This improvement at the collective level includes a re-valuation and redesign of existing processes and guidelines for morally good research.

A collaborative autoethnography

Responding to this need for critical reflexivity, we engaged with our storied experience in navigating concrete and immediate ethical dilemmas that we have encountered when collaborating with others for TR in practice. We did so through collaborative autoethnography, a multivocal approach in which two or more researchers work together to share personal stories and interpret the pooled autoethnographic data (Chang et al., 2016; Lapadat, 2017; Miyahara and Fukao, 2022). Collaborative autoethnography is appropriate for our inquiry as it broadens the gaze from the dilemmas of the self to locate them within categories of experience shared by many. Interrogating our personal narratives and understanding the shared experiences through multiple lenses not only facilitates a more rigorous, polyvocal analysis but also reveals possibilities for practical action or intervention (Lapadat, 2017). Collaborative auto-ethnography can thus be considered an approach that moves “beyond the clichés and usual explanations to the point where the written memories come as close as they can make them to ‘an embodied sense of what happened’” (Davies and Gannon, 2006, p. 3). It also supports developing researcher reflexivity (Miyahara and Fukao, 2022).

Overall, we engaged in two types of collaborative activities over the course of a period of 18 months: writing and discussing. In hindsight, this period can be divided into three phases: starting up, exploring, and co-working. The first phase was kicked off by an online dialogue session with about 30 participants convened by the Design Impact Transition Platform of the Erasmus University Rotterdam in April 2022. The session was meant to explore and share experiences with a wide range of ethical dilemmas arising from TR collaboration in practice. Following this session, some participants continued deliberating on the questions and dilemmas raised in differing constellations and developed the idea of codifying and sharing our experiences and insights via a publication. In a second phase, we started writing down individual ethical dilemmas, both those we had discussed during the seminar and additional ones. These writings were brought together in an online shared file, where we continued our discussions. This was accompanied by meetings in differing constellations and of differing intensity for the researchers involved.

A third phase of intense co-work was framed by two broader online sessions. During a session in May 2023, we shared and discussed a first attempt at an analysis and sense-making of our individual dilemmas. During this session, we discerned the heuristic by Mertens et al. (2017) and discussed how it could be helpful in structuring our different experiences. Inspired by Mertens et al. (2017), we re-engaged with the three critical dimensions of any research paradigm to scrutinise our philosophical commitments to doing TR. A re-engagement with issues of axiology (the nature of ethics and values), ontology (the nature of reality), and epistemology (the nature of knowledge), as illustrated in Table 1, allowed us to reconcile our ethical dilemmas and opened a space for a more nuanced understanding and bottom-up approach to the ethics of collaboration in TR. In moving forward, the heuristic also helped to guide the elicitation of additional dilemmas. This session kicked off a period of focused co-writing leading up to a second session in December 2023, where we discussed writing progress and specifically made sense of and related the ethical dilemmas to existing literature and insights.

Table 1 The heuristic guiding our collaborative autoethnography.

Especially in this last phase, as we interacted dialogically to analyse and interpret the collection of storied experiences of ethical dilemmas, our thinking about the ethics of collaboration has evolved. It went beyond considering the inadequacy of institutional rules and how we navigated those, towards acknowledging their interplay with individual positionality and a researcher’s situated practice. Closer attention to the contexts within which the ethical dilemmas have arisen has led us to return to our philosophical commitments as transformative researchers and reflect on our assumptions about collaboration and research from a transformative standpoint.

The author team thus comprises a high proportion of those participating in the initial session, as well as others who joined the ensuing collective interpretation and analysis resulting in this paper. An important characteristic of the authors is that we are all affiliated with academic research institutions and that all but one of these institutions are based in high-income countries. It is in this context that we have shared our experiences, which is also limited by it. As such, this paper will mainly speak to other researchers affiliated with academic institutions in comparable settings. Acknowledging these limitations, we are from different (inter)disciplinary backgroundsFootnote 2, nationalities, and work in different national settings and urban and rural locations. This diversity of contexts impacts the constellation of ethical dilemmas that we were faced with. We thus synthesise lessons from disparate yet still limited contexts, whilst remaining cognisant of the ungeneralisable nature of such a study.

Collaboration in transformative research practice

At the heart of our collaborative autoethnographic experience was the sharing and sensemaking of ethical dilemmas. In this section, we share those dilemmas (see Tables 24) clustered along the three philosophical commitments that served to deepen the analysis and interpretation of our storied experience. We embed our dilemmas with the broader body of knowledge around similar issues to discuss ways forward for practical knowledge around ‘what is good’ TR practice and ‘how to’ navigate ethical dilemmas.

Table 2 Axiological concerns in our experiential encounters.
Table 3 Ontological concerns in our experiential encounters.
Table 4 Epistemological concerns in our experiential encounters.

Axiological dimension

Axiology is the study of value, which concerns what is considered ‘good’, what is valued, and most importantly, what ‘ought to be’. The axiological standpoint of TR is to address persistent societal problems and to contribute to transitions towards more just and sustainable societies. The commitment to knowledge development and transformative actions is also shaped by different personal judgements, disciplinary traditions, and institutional contexts. Together, these raise ethical concerns around the shape and form of research collaborations, the research lines being pursued, and where and for whom the benefits of the research accrue. Table 2 provides the details of the ethical dilemmas (described as encounters) that we discuss in the following.

Taking up a transformative stance goes hand in hand with individual researchers holding different roles at the same time (Hoffmann et al., 2022; Horlings et al., 2020; Jhagroe, 2018; Schut et al., 2014). Often resulting from this, they also perceive a wide range of responsibilities towards diverse groups (stakeholders, peers, the academic community, etc.). This is why transformative researchers face questions of who is responsible for what and whom in front of whom, and these questions influence and are influenced by what they consider the ‘right’ thing to do in relation to others in a collaborative setting. As a result, their axiological position is constructed intersubjectively in and through interactions unfolding in the communities of important others. It is thus relational and may differ depending on ‘the other’ in the research collaboration (Arrona & Larrea, 2018; Bartels and Wittmayer, 2018). Encounter 1 illustrates this through a constellation of the research collaboration that holds the potential to become a conflict of interest.

Such conflicts of interest can also occur in the very choice of which ‘community’ is being considered as the main beneficiary of the collaboration. The emphasis on action in TR, especially with regards to the principles of beneficence and justice that we mentioned in “Ethics in transformative research”, can increase this dilemma. Researchers are to continuously evaluate their (perceived) obligations. This includes, for example, obligations towards the scientific community (contributions to the academic discourse via publications) vs. obligations towards stakeholders (being a provider of free practical advice or consultant) vs. scientific requirements (academic rigour and independence) vs. stakeholder requests (answering practical questions). Researchers have to position themselves in this contested field of what ‘good research’ and ‘useful outcomes’ mean and sometimes question or challenge their peers or the academic system at large (see also Kump et al., 2023). This is the very question raised by Encounter 2, where researchers are forced to decide which stakeholders’ values and needs should be prioritised in transforming clinical practice and improving the lives of patients.

Moreover, a similar prioritisation between the interests of different groups needs to be made between whether to create knowledge according to traditional scientific standards of systematicity and rigour or supporting collaborators in developing usable knowledge. This is surely a dilemma that arises from being embedded in an institutional context that judges according to different standards, but it also arises from the double commitment of TR to knowledge development and transformative action (Bartels et al., 2020). Huang et al. (2024) for example show how axiological assumptions serve as the base from which different notions of research excellence (e.g., scientific rigour, ‘impactful’ scholarship) are operationalised and supported institutionally. Encounter 3 reflects a similar dilemma as the lecturer juggles conflicting priorities that are inherent to the axiological concerns of TR. That is, can the goals of knowledge development in the traditional academic sense and transformative action be achieved simultaneously? The answer provided by Encounter 3 seems to suggest a redefinition of what ‘good’ scientific knowledge is, for immediate action to be possible.

Yet, perceived responsibilities—towards human and non-human actors, but also towards the own university, the institutional arrangements in which we partake, and what we understand as ethical behaviours—exist in a close, interdependent relationship with our inner ethical standards. Creed et al. (2022, p. 358) capture this “collection of sedimented evaluations of experiences, attachments, and commitments” as an ‘embodied world of concern’. This can illustrate the complexity of how an individual researcher’s values, emotions, or sentiments tend to intertwine, and can sometimes clash, with the concerns of their communities and the social-political situation where they operate. Given that one’s embodied world of concern is not fixed but characterised by emerging pluralism, as Encounter 4 illustrates, the consequence of an ethical decision tends to fall more heavily on those with less axiological privilege, such as early career researchers or those located in regions where the opportunity for scientific publishing is limited (Kruijf et al., 2022).

As transformative researchers seek systemic change, their values cannot help but influence their research collaboration, including the choice of whom they work with and which methods to use. However, the intention of strengthening the responsiveness of research to societal and political needs through TR collaborations risks being co-opted by the interests of those funding research activities (Bauwens et al., 2023; Strydom et al., 2010). As illustrated in Encounter 5, this might cause dilemmas when being approached by stakeholders (e.g., oil and gas companies) to do research, which may not sit well with the subjective judgements of the researcher or with an overall need for transformative change. Researchers can be caught in an odd position and left to wonder whether a compromise of values is worth the risks and end gain, depending on whether a positive contribution can still be achieved. Negotiating our axiological stances with collaborators thus allows researchers to be seen as social beings embedded in patterns of social interdependence, who are not only “capable and can flourish” but also “vulnerable and susceptible to various kinds of loss or harm [and] can suffer” (Sayer, 2011, p. 1).

Ontological dimension

Ontology is the philosophical study of being, which concerns the nature of reality and what really exists. TR can start from diverse ontological stances, including critical realist, pragmatist, or subjectivist perspectives. This includes a strong acknowledgement that “there are multiple versions of what is believed to be real” (Mertens, 2017, p. 21). Yet, such a pluralist stance remains a theoretical exercise up until the point that researchers ought to define what are ‘the things’ that need to be transformed and into what. In this situation, at least two debates arise: Do ‘the things’ exist based on a specific ontological commitment, such as the divide between measurable constructs and socially constructed understandings of risks and inequities. And is the existence of ‘the things’ universal or merely a construct of a specific time, space, or social group? As the researcher illustrated in Encounter 6 (see Table 3 for the detailed encounters), if maths anxiety and eco-anxiety are recognised as ‘real’ because of growing clinical research, why can’t the research team accept the construct of ‘science anxiety’ that their teacher collaborators have perceived in their classrooms? Collaboration thus remains especially challenging when researchers strive for academic rigour from an empiricist standpoint while having to cross paths or work with individuals from different ontological positions (Midgley, 2011).

Commitments to working collaboratively with members of ‘marginalised’ and ‘vulnerable’ communities add to this dilemma, as researchers are bound to encounter the ethical dilemmas of whose reality is privileged, whose reality can or should be legitimised and considered ‘true’ in a TR process (Kwan and Walsh, 2018). In Encounter 7, for instance, research participants do not recognise themselves as ‘climate displaced persons’ or ‘climate migrants’ because they have a long history of migration for a plethora of reasons. Now, should researchers continue using this term with a view to gain political attention to the issues of climate change, or should they abstain from doing so? How does this relate to their commitment to transformative action, including shaping political agendas? The intention to target system-level change in TR (Burns, 2014; Kemmis, 2008) also means that researchers ought to interrogate the mechanisms that inflict certain perceived realities on the powerless in the name of good causes (Edelman, 2018; Feltham-King et al., 2018), the ways in which these narratives are deployed by powerful stakeholders (Thomas and Warner, 2019) and how these are translated into (research) action.

Moreover, research and action on ‘scientific’ problems can deflect attention from other problems that local communities most care about or lead to unexpected, even negative, implications for some stakeholders. With increasing pressure on the societal impact of research and funding tied to certain policy goals, the issues of labelling and appropriation might only perpetuate a deficit perspective on specific groups (Eriksen et al., 2021; Escobar, 2011; van Steenbergen, 2020). Encounter 8 highlights that, without caution, well-intended efforts risk perpetuating harm and injustice —upholding a certain deficit perspective of the community in question. Communities accustomed to ‘helicopter’ research, where academics ‘fly-in, fly-out’ to further their careers at the expense of the communities, may be reluctant to collaborate. This necessitates transparency, active listening, deliberative involvement, and trust building (Adame, 2021; Haelewaters et al., 2021). It also reminds us of the ‘seagull syndrome’,’ which attests to the frustration felt by community members towards outsider ‘experts’ making generalisations and false diagnoses based on what is usually a superficial or snapshot understanding of local community dynamics (Porter, 2016). In some incidents, transformative researchers may need to redesign collaboration processes in TR that centre on the realities of people in the study (Hickey et al., 2018).

Epistemological dimension

Epistemology is the philosophical study of knowledge, and its primary concern is the relationship between the knower and what can be known. Transformative researchers usually work at the interface of disciplines, each with their own ideas on what constitutes ‘scientifically sound’ but also ‘socially robust’ or ‘actionable’ knowledge (Mach et al., 2020; Nowotny et al., 2003). Many thus hold the epistemological assumption that knowledge is created through multiple ways of knowing, and the processes of knowledge generation need to recognise how power inequities may shape the normative definition of legitimate knowledge. This stance raises ethical concerns about whose knowledge systems and ways of knowing are included, privileged, and/or legitimised in TR practice. Moreover, it raises concerns about ways of ensuring a plurality of knowledge spaces (Savransky, 2017).

Using an epistemological lens to interrogate collaborative practice in TR can illuminate a wide range of ethical dilemmas associated with longstanding critiques of Western norms and ‘scientific superiority’ (Dotson, 2011; Dutta et al., 2022; Wijsman and Feagan, 2019). It also brings to the fore the power dynamics inherent within collaborative processes of TR for sustainability (de Geus et al., 2023; Frantzeskaki and Rok, 2018; Kanemasu and Molnar, 2020; Kok et al., 2021; Strumińska-Kutra and Scholl, 2022). A particular ethical challenge is related to the fact that it is typically researchers from the Global North who design and lead research collaborations, even when these take place in the Global South. This immediately creates “an inequality that is not conducive to effective co-production” and requires “dedicated commitment to identify and confront the embodied power relations [and] hegemonic knowledge systems among the participants in the process” (Vincent, 2022, p. 890). See Table 4 for details on the ethical dilemmas that we discuss in the following.

Concerns about epistemic justice (Ackerly et al., 2020; Harvey et al., 2022; Temper and Del Bene, 2016) and interpretation of voices (Komulainen, 2007) are largely rooted in the deficit narratives about the capacity of certain groups for producing knowledge or for being knowers. Encounter 9 shows how easily certain voices can be muted as not being considered to speak from a position of knowledge. Research processes can usefully be expanded to include disinterested or disengaged citizens (Boyle et al., 2022), or those opposing a project or initiative so as to lay bare the associated tensions of knowledge integration and co-production (Cockburn, 2022). Encounter 10 illustrates that such silencing also relates to the question of who holds legitimate knowledge. This research has three parties that may hold legitimate knowledge: the researcher, the corporation, and the local community. However, the extent to which the researchers’ knowledge is heard remains unclear since the corporation does not consider it in its actions. It also illustrates common insecurities about what one can attain using certain research methods. The reliance of political institutions and citizens on expert advice, particularly when dealing with acute crises (e.g., Covid-19 pandemic), also tends to exacerbate the depoliticisation of decisions (Rovelli, 2021).

Moreover, TR practice nearly inevitably results in privileging certain ways of knowing and knowledges. Researchers make space for shared action or dialogue around a certain issue, inviting certain groups but not others, and choosing certain methods and not others. Encounter 11 illustrates the issue of favouritism in research collaboration. It elaborates on how thoughtful facilitation can intervene to level the playing field and provide a way out of the dilemma going beyond the question of whose benefit it serves. This facilitation enables meaningful collaboration among all parties involved. Particularly in policy sectors dominated by political and economic considerations, which exhibit strong vested interests, there is a need to foster meaningful and safe participation (Nastar et al., 2018). Skilled facilitation is crucial for uniting marginalised groups, preparing them to deal with the intricacies of scientific jargon and technological hegemony (Djenontin and Meadow, 2018; Reed and Abernethy, 2018). The contextual dimensions of collaborators, their associated worldviews, and the social networks in which they are situated are important epistemological foundations. Yet, these are not static and can shift over time throughout collaborative partnerships.

As explicated in “Introducing transformative research”, TR represents an epistemological shift to recognise researchers as sense-makers, agency holders, and change agents. This philosophical commitment can create dilemmas for ‘embedded researchers’ seeking to strengthen the science-policy interface. Encounter 12 illustrates how occupying a dual role — to dive into action and to publish scientifically — can be at odds. This encounter alludes to the fact that transformative researchers often navigate different roles, which come with different, at times conflicting, epistemological priorities and ways of knowing (e.g., roles as a change agent and a reflective scientist, the approach of ‘Two-Eyed Seeing’ by Indigenous scholars) (Bulten et al., 2021; Temper et al., 2019; Wittmayer and Schäpke, 2014). Importantly, such roles change over time in a TR practice and over the course of a researcher’s career (McGowan et al., 2014; Pohl et al., 2017).

Involving diverse stakeholders in knowledge co-production also inevitably leads to ethical questions concerning how to integrate diverse knowledge systems, especially those using multi-method research designs or models to aid decision-making (Hoffmann et al., 2017). Models can be useful in providing scenarios, however, they are constructed by people based on certain assumptions. These assumptions serve as the fundamental lenses through which complex real-world systems are simplified, analysed, and interpreted within the model framework. Despite the well-intention of researchers, the practice of establishing a shared understanding and reaching consensus about key constructs in a model is often unattainable. As Encounter 13 illustrates, participatory model building requires the capacity and willingness of all involved to knit together kindred, or even conflicting, perspectives to complement disciplinary specialism.

We explored the dilemmas of researchers pertaining to knowing ‘how to’ act in a certain situation and considering ‘what is doing good’ in that situation. Transformative researchers (re)build their practical knowledge of what doing research means through cultivating a reflexive practice that puts experiences in context and allows to learn from them. From a meta-perspective, doing TR is a form of experiential learning (Kolb, 1984) and doing TR involves traversing an action research cycle: experiencing and observing one’s action research practice, abstracting from it, building knowledge, and experimenting with it again to cultivate what has been referred to as first person inquiry (Reason and Torbert, 2001).

Concluding thoughts

In this article, we set out to explore which ethical dilemmas researchers face in TR and how they navigate those in practice. We highlighted that researchers engaging in TR face a context of uncertainty and plurality around what counts as ethically acceptable collaboration. With TR emphasising collaboration, it becomes important to discern the notion of ‘right relations’ with others (Gram-Hanssen et al., 2022), to attend to the positionality of the researcher, and to reconfigure power relations. Importantly, with TR emphasising the need for structural and systematic changes, researchers need to be aware of how research itself is characterised by structural injustices.

Using a collaborative autoethnography, we shared ethical dilemmas to uncover the messiness of collaborative TR practice. We established how guidance from institutionalised reference systems (i.e., ethical review boards and procedures) currently falls short in recognising the particularities of TR. We described how the research community generates informal principles, or heuristics to address this gap. However, we also appreciated that in actual collaboration, researchers are often ‘put on the spot’ to react ‘ethically’ in situ, with limited time and space to withdraw and consult guidelines on ‘how to behave’. Such informal heuristics are thus but a start and a helpful direction for developing the practical knowledge of researchers on how to navigate a plural and uncertain context.

This practical knowledge is based on an awareness of the uncertainty around what constitutes morally good behaviour and builds through experience and a critical reflexive practice. Our aim is not to share another set of principles, but rather to highlight the situatedness of TR and the craftsmanship necessary to navigate it and, in doing so, build practical knowledge through experiential learning and insight discovery (Kolb, 1984; Pearce et al., 2022). Such a bottom-up approach to research ethics builds on the experiences of researchers engaging in TR as a situated practice vis-à-vis their personal motivations and normative ambitions and the institutional contexts they are embedded in. This approach nurtures the critical reflexivity of researchers about how they relate to ethical principles and how they translate this into their normative assumptions, practical hypotheses, and methodological strategy.

Next to continuous learning, this critical reflexivity on TR as craftmanship can enhance practical wisdom not only for the individual but also for the broader community of researchers. We envision such wisdom not as a set of closed-ended guidelines or principles, but rather as a growing collection of ethical questions enabling the TR community to continuously deepen the interrogation of their axiological, ontological, and epistemological commitments (see Table 5). Only through this ongoing process of reacting, reflecting, and questioning—or as referred to by Pearce et al. (2022, p. 4) as “an insight discovery process”—can we collectively learn from the past to improve our future actions.

Table 5 Ethical questions emerging from this collective autoethnography.

However, such a bottom-up approach to ethics can only form one part of the answer, set in times of an evolving research ethics landscape. Researchers engaging in transformative academic work cannot and should not be left alone. Additionally, researchers’ ethical judgements cannot be left to their goodwill and virtuous values alone. Therefore, another important part of the answer is the carving out of appropriate institutions that can provide external guidance and accountability. This will require nothing less than structural and cultural changes in established universities and research environments. Rather than having researchers decide between doing good and doing ‘good’ research, such environments should help to align those goals.

From this work, questions arise on how institutional environments can be reformed or transformed to be more conducive to the particularities of TR, and to help nurture critical reflexivity. We highlight the critical role that ethic review boards can play in starting to rethink their roles, structures, and underlying values. Practical ideas include employing mentors for transformative research ethics, having ethical review as a process rather than as a one-off at the start of the project, or continuously investing in moral education. Thus, we underscore the importance of individual reflexivity and learning. However, we would like to set this in the broader context of organisational learning, and even unlearning, among academic institutions to overhaul our academic systems in response to the urgent imperative of tackling socio-ecological challenges globally. In this transformative endeavour, careful consideration of how the ethics of research and collaboration shape academics’ socially engaged work is indispensable.